diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,87027 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.1376971841368126,
+  "eval_steps": 10000,
+  "global_step": 14500,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9999999958813268e-05,
+      "loss": 1.875,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.999999983525307e-05,
+      "loss": 2.1562,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.999999962931941e-05,
+      "loss": 1.6953,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9999999341012284e-05,
+      "loss": 1.3047,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9999998970331697e-05,
+      "loss": 4.0938,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9999998517277653e-05,
+      "loss": 1.3281,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.999999798185015e-05,
+      "loss": 1.1641,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9999997364049193e-05,
+      "loss": 0.1885,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9999996663874784e-05,
+      "loss": 1.0234,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.999999588132693e-05,
+      "loss": 0.1816,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9999995016405633e-05,
+      "loss": 1.0938,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.99999940691109e-05,
+      "loss": 0.8359,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.999999303944273e-05,
+      "loss": 0.2773,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9999991927401137e-05,
+      "loss": 0.8555,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9999990732986124e-05,
+      "loss": 1.0234,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9999989456197696e-05,
+      "loss": 0.957,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.999998809703586e-05,
+      "loss": 0.8945,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9999986655500623e-05,
+      "loss": 0.6484,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9999985131591996e-05,
+      "loss": 0.8516,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.999998352530998e-05,
+      "loss": 2.5469,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.99999818366546e-05,
+      "loss": 2.7188,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9999980065625845e-05,
+      "loss": 1.1094,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9999978212223736e-05,
+      "loss": 1.6875,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9999976276448283e-05,
+      "loss": 2.3438,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.99999742582995e-05,
+      "loss": 1.0156,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9999972157777387e-05,
+      "loss": 1.5859,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9999969974881965e-05,
+      "loss": 0.5273,
+      "step": 27
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9999967709613243e-05,
+      "loss": 1.4922,
+      "step": 28
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9999965361971227e-05,
+      "loss": 1.4141,
+      "step": 29
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9999962931955942e-05,
+      "loss": 0.4473,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9999960419567397e-05,
+      "loss": 0.5703,
+      "step": 31
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.99999578248056e-05,
+      "loss": 0.9219,
+      "step": 32
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.999995514767057e-05,
+      "loss": 1.5234,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.999995238816232e-05,
+      "loss": 1.7734,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.999994954628087e-05,
+      "loss": 0.4766,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9999946622026234e-05,
+      "loss": 1.7891,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.999994361539842e-05,
+      "loss": 0.3711,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.999994052639745e-05,
+      "loss": 0.1455,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9999937355023348e-05,
+      "loss": 0.2471,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.999993410127612e-05,
+      "loss": 0.0728,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.999993076515579e-05,
+      "loss": 0.3594,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9999927346662373e-05,
+      "loss": 0.0815,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9999923845795893e-05,
+      "loss": 0.0732,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9999920262556367e-05,
+      "loss": 0.0217,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9999916596943814e-05,
+      "loss": 1.1016,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.999991284895825e-05,
+      "loss": 0.0282,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.99999090185997e-05,
+      "loss": 1.5156,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9999905105868186e-05,
+      "loss": 1.3516,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9999901110763727e-05,
+      "loss": 1.1875,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9999897033286344e-05,
+      "loss": 1.3672,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9999892873436067e-05,
+      "loss": 1.7109,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9999888631212907e-05,
+      "loss": 0.5742,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.99998843066169e-05,
+      "loss": 0.1934,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.999987989964806e-05,
+      "loss": 1.8906,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.999987541030641e-05,
+      "loss": 1.7422,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9999870838591985e-05,
+      "loss": 0.6914,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9999866184504804e-05,
+      "loss": 1.0469,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.999986144804489e-05,
+      "loss": 0.9102,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9999856629212272e-05,
+      "loss": 0.5117,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9999851728006976e-05,
+      "loss": 0.3633,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9999846744429034e-05,
+      "loss": 0.6602,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9999841678478462e-05,
+      "loss": 0.1689,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9999836530155298e-05,
+      "loss": 2.6719,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9999831299459562e-05,
+      "loss": 0.8906,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9999825986391292e-05,
+      "loss": 0.3516,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.999982059095051e-05,
+      "loss": 0.5391,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.999981511313725e-05,
+      "loss": 1.5234,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9999809552951543e-05,
+      "loss": 0.1157,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9999803910393412e-05,
+      "loss": 0.1973,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9999798185462893e-05,
+      "loss": 1.125,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.999979237816002e-05,
+      "loss": 0.1533,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.999978648848482e-05,
+      "loss": 1.3125,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9999780516437325e-05,
+      "loss": 1.0703,
+      "step": 73
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9999774462017573e-05,
+      "loss": 0.2266,
+      "step": 74
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9999768325225598e-05,
+      "loss": 0.3398,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9999762106061427e-05,
+      "loss": 0.0251,
+      "step": 76
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9999755804525094e-05,
+      "loss": 0.7852,
+      "step": 77
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9999749420616637e-05,
+      "loss": 0.5,
+      "step": 78
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9999742954336092e-05,
+      "loss": 0.2676,
+      "step": 79
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.99997364056835e-05,
+      "loss": 0.5742,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9999729774658884e-05,
+      "loss": 0.0466,
+      "step": 81
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9999723061262284e-05,
+      "loss": 0.0054,
+      "step": 82
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9999716265493747e-05,
+      "loss": 0.5234,
+      "step": 83
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.99997093873533e-05,
+      "loss": 0.0076,
+      "step": 84
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9999702426840982e-05,
+      "loss": 0.7266,
+      "step": 85
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9999695383956836e-05,
+      "loss": 0.6523,
+      "step": 86
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9999688258700895e-05,
+      "loss": 0.0281,
+      "step": 87
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9999681051073205e-05,
+      "loss": 0.3223,
+      "step": 88
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9999673761073797e-05,
+      "loss": 0.0535,
+      "step": 89
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9999666388702715e-05,
+      "loss": 0.0312,
+      "step": 90
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9999658933960004e-05,
+      "loss": 0.1147,
+      "step": 91
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.99996513968457e-05,
+      "loss": 0.0014,
+      "step": 92
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9999643777359843e-05,
+      "loss": 0.0361,
+      "step": 93
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9999636075502476e-05,
+      "loss": 0.7266,
+      "step": 94
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9999628291273648e-05,
+      "loss": 1.1094,
+      "step": 95
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.999962042467339e-05,
+      "loss": 1.1641,
+      "step": 96
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9999612475701756e-05,
+      "loss": 0.8203,
+      "step": 97
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9999604444358784e-05,
+      "loss": 0.8516,
+      "step": 98
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9999596330644514e-05,
+      "loss": 1.0312,
+      "step": 99
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9999588134559e-05,
+      "loss": 1.25,
+      "step": 100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9999579856102284e-05,
+      "loss": 1.4609,
+      "step": 101
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.999957149527441e-05,
+      "loss": 0.668,
+      "step": 102
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.999956305207542e-05,
+      "loss": 1.7422,
+      "step": 103
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9999554526505367e-05,
+      "loss": 0.1416,
+      "step": 104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9999545918564296e-05,
+      "loss": 0.4883,
+      "step": 105
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9999537228252253e-05,
+      "loss": 0.3574,
+      "step": 106
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9999528455569284e-05,
+      "loss": 0.6719,
+      "step": 107
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.999951960051544e-05,
+      "loss": 0.5742,
+      "step": 108
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9999510663090775e-05,
+      "loss": 1.8984,
+      "step": 109
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9999501643295325e-05,
+      "loss": 1.4609,
+      "step": 110
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.999949254112915e-05,
+      "loss": 2.1875,
+      "step": 111
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9999483356592295e-05,
+      "loss": 0.8711,
+      "step": 112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.999947408968481e-05,
+      "loss": 1.6172,
+      "step": 113
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9999464740406754e-05,
+      "loss": 0.9922,
+      "step": 114
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.999945530875817e-05,
+      "loss": 0.6992,
+      "step": 115
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.999944579473911e-05,
+      "loss": 0.1826,
+      "step": 116
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.999943619834963e-05,
+      "loss": 0.0041,
+      "step": 117
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.999942651958978e-05,
+      "loss": 0.0718,
+      "step": 118
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9999416758459613e-05,
+      "loss": 0.2061,
+      "step": 119
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9999406914959185e-05,
+      "loss": 0.1025,
+      "step": 120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.999939698908855e-05,
+      "loss": 0.8633,
+      "step": 121
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.999938698084776e-05,
+      "loss": 0.1602,
+      "step": 122
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9999376890236873e-05,
+      "loss": 1.7578,
+      "step": 123
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.999936671725594e-05,
+      "loss": 0.0071,
+      "step": 124
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9999356461905022e-05,
+      "loss": 0.2871,
+      "step": 125
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9999346124184173e-05,
+      "loss": 0.3809,
+      "step": 126
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9999335704093446e-05,
+      "loss": 0.252,
+      "step": 127
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9999325201632904e-05,
+      "loss": 0.418,
+      "step": 128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9999314616802603e-05,
+      "loss": 0.0227,
+      "step": 129
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.99993039496026e-05,
+      "loss": 0.0066,
+      "step": 130
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9999293200032956e-05,
+      "loss": 0.6719,
+      "step": 131
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.999928236809373e-05,
+      "loss": 1.6406,
+      "step": 132
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.999927145378498e-05,
+      "loss": 1.4375,
+      "step": 133
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9999260457106758e-05,
+      "loss": 0.0031,
+      "step": 134
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9999249378059136e-05,
+      "loss": 0.0084,
+      "step": 135
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9999238216642174e-05,
+      "loss": 0.0025,
+      "step": 136
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9999226972855928e-05,
+      "loss": 0.4258,
+      "step": 137
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9999215646700463e-05,
+      "loss": 0.0303,
+      "step": 138
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.999920423817584e-05,
+      "loss": 0.5703,
+      "step": 139
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.999919274728212e-05,
+      "loss": 0.054,
+      "step": 140
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.999918117401937e-05,
+      "loss": 0.0028,
+      "step": 141
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9999169518387648e-05,
+      "loss": 0.0125,
+      "step": 142
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9999157780387028e-05,
+      "loss": 0.0013,
+      "step": 143
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.999914596001756e-05,
+      "loss": 1.8047,
+      "step": 144
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9999134057279324e-05,
+      "loss": 0.9375,
+      "step": 145
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9999122072172376e-05,
+      "loss": 0.7695,
+      "step": 146
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9999110004696784e-05,
+      "loss": 0.9766,
+      "step": 147
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.999909785485261e-05,
+      "loss": 0.2676,
+      "step": 148
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.999908562263993e-05,
+      "loss": 1.0781,
+      "step": 149
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9999073308058803e-05,
+      "loss": 1.1406,
+      "step": 150
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9999060911109305e-05,
+      "loss": 0.3652,
+      "step": 151
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9999048431791498e-05,
+      "loss": 1.375,
+      "step": 152
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9999035870105448e-05,
+      "loss": 0.2119,
+      "step": 153
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9999023226051228e-05,
+      "loss": 0.6367,
+      "step": 154
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.999901049962891e-05,
+      "loss": 2.25,
+      "step": 155
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9998997690838554e-05,
+      "loss": 0.4922,
+      "step": 156
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9998984799680246e-05,
+      "loss": 1.8047,
+      "step": 157
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9998971826154042e-05,
+      "loss": 0.4434,
+      "step": 158
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.999895877026002e-05,
+      "loss": 0.4941,
+      "step": 159
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9998945631998253e-05,
+      "loss": 0.7227,
+      "step": 160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9998932411368807e-05,
+      "loss": 0.5469,
+      "step": 161
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.999891910837176e-05,
+      "loss": 1.4375,
+      "step": 162
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9998905723007187e-05,
+      "loss": 0.1611,
+      "step": 163
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9998892255275154e-05,
+      "loss": 1.0234,
+      "step": 164
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9998878705175743e-05,
+      "loss": 0.1187,
+      "step": 165
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.999886507270902e-05,
+      "loss": 0.0114,
+      "step": 166
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.999885135787507e-05,
+      "loss": 1.0078,
+      "step": 167
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9998837560673957e-05,
+      "loss": 0.5469,
+      "step": 168
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.999882368110577e-05,
+      "loss": 0.0144,
+      "step": 169
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.999880971917057e-05,
+      "loss": 2.8125,
+      "step": 170
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9998795674868447e-05,
+      "loss": 0.054,
+      "step": 171
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9998781548199465e-05,
+      "loss": 0.0825,
+      "step": 172
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9998767339163716e-05,
+      "loss": 0.2988,
+      "step": 173
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.999875304776127e-05,
+      "loss": 0.457,
+      "step": 174
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9998738673992206e-05,
+      "loss": 0.0008,
+      "step": 175
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9998724217856603e-05,
+      "loss": 0.8516,
+      "step": 176
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.999870967935454e-05,
+      "loss": 0.5508,
+      "step": 177
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.99986950584861e-05,
+      "loss": 0.0031,
+      "step": 178
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9998680355251357e-05,
+      "loss": 0.2598,
+      "step": 179
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.99986655696504e-05,
+      "loss": 0.0008,
+      "step": 180
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9998650701683304e-05,
+      "loss": 0.1738,
+      "step": 181
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.999863575135015e-05,
+      "loss": 0.6875,
+      "step": 182
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.999862071865103e-05,
+      "loss": 0.5391,
+      "step": 183
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9998605603586013e-05,
+      "loss": 0.5547,
+      "step": 184
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9998590406155192e-05,
+      "loss": 0.9219,
+      "step": 185
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9998575126358646e-05,
+      "loss": 0.0079,
+      "step": 186
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9998559764196458e-05,
+      "loss": 0.0066,
+      "step": 187
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9998544319668714e-05,
+      "loss": 0.543,
+      "step": 188
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.99985287927755e-05,
+      "loss": 1.0234,
+      "step": 189
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.99985131835169e-05,
+      "loss": 0.0016,
+      "step": 190
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9998497491893e-05,
+      "loss": 0.0011,
+      "step": 191
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9998481717903888e-05,
+      "loss": 0.5664,
+      "step": 192
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9998465861549644e-05,
+      "loss": 0.0019,
+      "step": 193
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9998449922830363e-05,
+      "loss": 0.6211,
+      "step": 194
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9998433901746127e-05,
+      "loss": 0.2715,
+      "step": 195
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9998417798297027e-05,
+      "loss": 0.0003,
+      "step": 196
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.999840161248315e-05,
+      "loss": 0.9922,
+      "step": 197
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.999838534430459e-05,
+      "loss": 0.6328,
+      "step": 198
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9998368993761425e-05,
+      "loss": 0.4688,
+      "step": 199
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9998352560853758e-05,
+      "loss": 0.0072,
+      "step": 200
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9998336045581667e-05,
+      "loss": 1.0625,
+      "step": 201
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9998319447945255e-05,
+      "loss": 0.5547,
+      "step": 202
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9998302767944603e-05,
+      "loss": 1.4141,
+      "step": 203
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9998286005579808e-05,
+      "loss": 1.0469,
+      "step": 204
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9998269160850956e-05,
+      "loss": 0.6328,
+      "step": 205
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9998252233758148e-05,
+      "loss": 0.5898,
+      "step": 206
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9998235224301473e-05,
+      "loss": 1.4453,
+      "step": 207
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9998218132481018e-05,
+      "loss": 1.0547,
+      "step": 208
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.999820095829689e-05,
+      "loss": 0.8398,
+      "step": 209
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9998183701749175e-05,
+      "loss": 1.1172,
+      "step": 210
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.999816636283797e-05,
+      "loss": 0.8359,
+      "step": 211
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.999814894156337e-05,
+      "loss": 0.6875,
+      "step": 212
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9998131437925464e-05,
+      "loss": 1.6562,
+      "step": 213
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.999811385192436e-05,
+      "loss": 1.7031,
+      "step": 214
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9998096183560146e-05,
+      "loss": 0.8633,
+      "step": 215
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.999807843283292e-05,
+      "loss": 0.7266,
+      "step": 216
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9998060599742787e-05,
+      "loss": 0.0688,
+      "step": 217
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.999804268428984e-05,
+      "loss": 0.1592,
+      "step": 218
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.999802468647417e-05,
+      "loss": 0.1572,
+      "step": 219
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9998006606295883e-05,
+      "loss": 0.5156,
+      "step": 220
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9997988443755083e-05,
+      "loss": 0.0771,
+      "step": 221
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9997970198851865e-05,
+      "loss": 0.6289,
+      "step": 222
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9997951871586324e-05,
+      "loss": 0.3477,
+      "step": 223
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9997933461958566e-05,
+      "loss": 0.1133,
+      "step": 224
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.999791496996869e-05,
+      "loss": 0.5469,
+      "step": 225
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9997896395616803e-05,
+      "loss": 0.0359,
+      "step": 226
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9997877738903e-05,
+      "loss": 0.9023,
+      "step": 227
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.999785899982739e-05,
+      "loss": 0.0918,
+      "step": 228
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9997840178390067e-05,
+      "loss": 1.25,
+      "step": 229
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.999782127459114e-05,
+      "loss": 0.2773,
+      "step": 230
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.999780228843072e-05,
+      "loss": 1.5469,
+      "step": 231
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9997783219908896e-05,
+      "loss": 0.2715,
+      "step": 232
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9997764069025783e-05,
+      "loss": 0.5664,
+      "step": 233
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9997744835781486e-05,
+      "loss": 0.1396,
+      "step": 234
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9997725520176103e-05,
+      "loss": 0.1875,
+      "step": 235
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.999770612220975e-05,
+      "loss": 2.1719,
+      "step": 236
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9997686641882527e-05,
+      "loss": 0.082,
+      "step": 237
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9997667079194544e-05,
+      "loss": 0.0082,
+      "step": 238
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9997647434145904e-05,
+      "loss": 0.3203,
+      "step": 239
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9997627706736725e-05,
+      "loss": 0.0198,
+      "step": 240
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.99976078969671e-05,
+      "loss": 0.0038,
+      "step": 241
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9997588004837147e-05,
+      "loss": 0.0369,
+      "step": 242
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.999756803034698e-05,
+      "loss": 0.0092,
+      "step": 243
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9997547973496698e-05,
+      "loss": 0.125,
+      "step": 244
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.999752783428642e-05,
+      "loss": 0.0125,
+      "step": 245
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9997507612716256e-05,
+      "loss": 0.0613,
+      "step": 246
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9997487308786308e-05,
+      "loss": 0.7617,
+      "step": 247
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9997466922496694e-05,
+      "loss": 0.0004,
+      "step": 248
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.999744645384753e-05,
+      "loss": 0.5039,
+      "step": 249
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9997425902838917e-05,
+      "loss": 0.2734,
+      "step": 250
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.999740526947098e-05,
+      "loss": 1.9453,
+      "step": 251
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9997384553743827e-05,
+      "loss": 0.5898,
+      "step": 252
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.999736375565757e-05,
+      "loss": 0.7969,
+      "step": 253
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9997342875212324e-05,
+      "loss": 0.6953,
+      "step": 254
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9997321912408203e-05,
+      "loss": 1.9453,
+      "step": 255
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9997300867245334e-05,
+      "loss": 0.1914,
+      "step": 256
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9997279739723814e-05,
+      "loss": 1.5469,
+      "step": 257
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9997258529843764e-05,
+      "loss": 1.9844,
+      "step": 258
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9997237237605308e-05,
+      "loss": 0.668,
+      "step": 259
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.999721586300856e-05,
+      "loss": 0.7227,
+      "step": 260
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9997194406053637e-05,
+      "loss": 0.7617,
+      "step": 261
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.999717286674065e-05,
+      "loss": 0.6758,
+      "step": 262
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.999715124506973e-05,
+      "loss": 0.0022,
+      "step": 263
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9997129541040984e-05,
+      "loss": 0.7969,
+      "step": 264
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.999710775465454e-05,
+      "loss": 1.7969,
+      "step": 265
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9997085885910513e-05,
+      "loss": 0.4336,
+      "step": 266
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.999706393480902e-05,
+      "loss": 0.4414,
+      "step": 267
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9997041901350192e-05,
+      "loss": 0.0049,
+      "step": 268
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.999701978553414e-05,
+      "loss": 0.4473,
+      "step": 269
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9996997587360987e-05,
+      "loss": 0.0126,
+      "step": 270
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9996975306830857e-05,
+      "loss": 1.7578,
+      "step": 271
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9996952943943877e-05,
+      "loss": 0.9102,
+      "step": 272
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9996930498700158e-05,
+      "loss": 0.5156,
+      "step": 273
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9996907971099838e-05,
+      "loss": 0.4414,
+      "step": 274
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9996885361143027e-05,
+      "loss": 0.3945,
+      "step": 275
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.999686266882986e-05,
+      "loss": 0.3457,
+      "step": 276
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9996839894160455e-05,
+      "loss": 0.4688,
+      "step": 277
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9996817037134934e-05,
+      "loss": 0.1167,
+      "step": 278
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9996794097753434e-05,
+      "loss": 0.6055,
+      "step": 279
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9996771076016072e-05,
+      "loss": 0.3828,
+      "step": 280
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9996747971922974e-05,
+      "loss": 0.3418,
+      "step": 281
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.999672478547427e-05,
+      "loss": 0.6328,
+      "step": 282
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9996701516670087e-05,
+      "loss": 0.1709,
+      "step": 283
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9996678165510556e-05,
+      "loss": 0.0664,
+      "step": 284
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.99966547319958e-05,
+      "loss": 0.0041,
+      "step": 285
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9996631216125954e-05,
+      "loss": 0.0103,
+      "step": 286
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9996607617901136e-05,
+      "loss": 0.1084,
+      "step": 287
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9996583937321486e-05,
+      "loss": 0.0048,
+      "step": 288
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.999656017438713e-05,
+      "loss": 0.1787,
+      "step": 289
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9996536329098197e-05,
+      "loss": 0.1216,
+      "step": 290
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9996512401454822e-05,
+      "loss": 0.0038,
+      "step": 291
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.999648839145713e-05,
+      "loss": 0.0028,
+      "step": 292
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9996464299105265e-05,
+      "loss": 0.0359,
+      "step": 293
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9996440124399347e-05,
+      "loss": 0.4883,
+      "step": 294
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9996415867339513e-05,
+      "loss": 0.0007,
+      "step": 295
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9996391527925893e-05,
+      "loss": 1.0,
+      "step": 296
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.999636710615863e-05,
+      "loss": 0.0012,
+      "step": 297
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9996342602037847e-05,
+      "loss": 0.5312,
+      "step": 298
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9996318015563687e-05,
+      "loss": 0.0003,
+      "step": 299
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9996293346736285e-05,
+      "loss": 0.9336,
+      "step": 300
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9996268595555765e-05,
+      "loss": 0.0287,
+      "step": 301
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9996243762022275e-05,
+      "loss": 0.582,
+      "step": 302
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9996218846135948e-05,
+      "loss": 0.707,
+      "step": 303
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.999619384789692e-05,
+      "loss": 0.6562,
+      "step": 304
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9996168767305332e-05,
+      "loss": 0.5547,
+      "step": 305
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9996143604361315e-05,
+      "loss": 1.5156,
+      "step": 306
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.999611835906501e-05,
+      "loss": 1.2734,
+      "step": 307
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9996093031416554e-05,
+      "loss": 0.5195,
+      "step": 308
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9996067621416095e-05,
+      "loss": 0.4844,
+      "step": 309
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.999604212906376e-05,
+      "loss": 0.7266,
+      "step": 310
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.99960165543597e-05,
+      "loss": 1.2734,
+      "step": 311
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9995990897304046e-05,
+      "loss": 1.0391,
+      "step": 312
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.999596515789694e-05,
+      "loss": 0.7539,
+      "step": 313
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9995939336138532e-05,
+      "loss": 0.6055,
+      "step": 314
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9995913432028956e-05,
+      "loss": 0.6875,
+      "step": 315
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9995887445568356e-05,
+      "loss": 1.4219,
+      "step": 316
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9995861376756875e-05,
+      "loss": 0.2178,
+      "step": 317
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.999583522559466e-05,
+      "loss": 0.0547,
+      "step": 318
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9995808992081848e-05,
+      "loss": 0.0195,
+      "step": 319
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9995782676218588e-05,
+      "loss": 0.8633,
+      "step": 320
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.999575627800502e-05,
+      "loss": 1.125,
+      "step": 321
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9995729797441293e-05,
+      "loss": 0.0042,
+      "step": 322
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9995703234527553e-05,
+      "loss": 0.8203,
+      "step": 323
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9995676589263944e-05,
+      "loss": 0.1611,
+      "step": 324
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9995649861650607e-05,
+      "loss": 0.707,
+      "step": 325
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9995623051687696e-05,
+      "loss": 0.0011,
+      "step": 326
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9995596159375358e-05,
+      "loss": 0.0009,
+      "step": 327
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9995569184713735e-05,
+      "loss": 0.0053,
+      "step": 328
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9995542127702983e-05,
+      "loss": 0.4121,
+      "step": 329
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9995514988343243e-05,
+      "loss": 0.0004,
+      "step": 330
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9995487766634673e-05,
+      "loss": 0.0001,
+      "step": 331
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9995460462577417e-05,
+      "loss": 0.2236,
+      "step": 332
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9995433076171617e-05,
+      "loss": 0.4492,
+      "step": 333
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.999540560741744e-05,
+      "loss": 0.4375,
+      "step": 334
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9995378056315023e-05,
+      "loss": 0.0002,
+      "step": 335
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9995350422864527e-05,
+      "loss": 0.5703,
+      "step": 336
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9995322707066097e-05,
+      "loss": 0.7188,
+      "step": 337
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9995294908919886e-05,
+      "loss": 0.2695,
+      "step": 338
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.999526702842605e-05,
+      "loss": 0.0004,
+      "step": 339
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.999523906558474e-05,
+      "loss": 0.4961,
+      "step": 340
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9995211020396107e-05,
+      "loss": 0.332,
+      "step": 341
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.999518289286031e-05,
+      "loss": 0.0036,
+      "step": 342
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9995154682977503e-05,
+      "loss": 0.0079,
+      "step": 343
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.999512639074784e-05,
+      "loss": 0.0014,
+      "step": 344
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9995098016171473e-05,
+      "loss": 0.0008,
+      "step": 345
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9995069559248562e-05,
+      "loss": 0.8594,
+      "step": 346
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.999504101997926e-05,
+      "loss": 0.5977,
+      "step": 347
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9995012398363725e-05,
+      "loss": 0.457,
+      "step": 348
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9994983694402117e-05,
+      "loss": 0.543,
+      "step": 349
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.999495490809459e-05,
+      "loss": 0.5977,
+      "step": 350
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9994926039441304e-05,
+      "loss": 0.4082,
+      "step": 351
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9994897088442415e-05,
+      "loss": 0.3965,
+      "step": 352
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9994868055098083e-05,
+      "loss": 0.6641,
+      "step": 353
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.999483893940847e-05,
+      "loss": 1.1328,
+      "step": 354
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9994809741373736e-05,
+      "loss": 1.0938,
+      "step": 355
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.999478046099404e-05,
+      "loss": 0.6016,
+      "step": 356
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9994751098269543e-05,
+      "loss": 1.8438,
+      "step": 357
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9994721653200398e-05,
+      "loss": 0.7266,
+      "step": 358
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9994692125786782e-05,
+      "loss": 0.8672,
+      "step": 359
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9994662516028845e-05,
+      "loss": 1.4531,
+      "step": 360
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9994632823926756e-05,
+      "loss": 1.2188,
+      "step": 361
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9994603049480674e-05,
+      "loss": 0.5312,
+      "step": 362
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9994573192690765e-05,
+      "loss": 0.7617,
+      "step": 363
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9994543253557195e-05,
+      "loss": 1.0938,
+      "step": 364
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9994513232080127e-05,
+      "loss": 0.0918,
+      "step": 365
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.999448312825972e-05,
+      "loss": 0.1406,
+      "step": 366
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9994452942096147e-05,
+      "loss": 1.8359,
+      "step": 367
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.999442267358957e-05,
+      "loss": 1.1719,
+      "step": 368
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9994392322740156e-05,
+      "loss": 1.7109,
+      "step": 369
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9994361889548077e-05,
+      "loss": 0.6836,
+      "step": 370
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9994331374013486e-05,
+      "loss": 0.0698,
+      "step": 371
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9994300776136566e-05,
+      "loss": 0.0281,
+      "step": 372
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.9994270095917475e-05,
+      "loss": 1.1953,
+      "step": 373
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9994239333356388e-05,
+      "loss": 0.8203,
+      "step": 374
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.999420848845347e-05,
+      "loss": 0.1455,
+      "step": 375
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.999417756120889e-05,
+      "loss": 0.7344,
+      "step": 376
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9994146551622818e-05,
+      "loss": 0.2363,
+      "step": 377
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.999411545969543e-05,
+      "loss": 0.1475,
+      "step": 378
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9994084285426885e-05,
+      "loss": 0.0008,
+      "step": 379
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9994053028817368e-05,
+      "loss": 0.0815,
+      "step": 380
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9994021689867043e-05,
+      "loss": 0.373,
+      "step": 381
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9993990268576083e-05,
+      "loss": 0.0161,
+      "step": 382
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9993958764944656e-05,
+      "loss": 0.0047,
+      "step": 383
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9993927178972947e-05,
+      "loss": 0.5703,
+      "step": 384
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.999389551066112e-05,
+      "loss": 0.0021,
+      "step": 385
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9993863760009344e-05,
+      "loss": 0.1455,
+      "step": 386
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.999383192701781e-05,
+      "loss": 0.2246,
+      "step": 387
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9993800011686677e-05,
+      "loss": 0.0077,
+      "step": 388
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9993768014016132e-05,
+      "loss": 0.3965,
+      "step": 389
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9993735934006344e-05,
+      "loss": 0.0022,
+      "step": 390
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9993703771657486e-05,
+      "loss": 0.0176,
+      "step": 391
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9993671526969738e-05,
+      "loss": 0.0077,
+      "step": 392
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9993639199943283e-05,
+      "loss": 0.0005,
+      "step": 393
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9993606790578293e-05,
+      "loss": 0.4375,
+      "step": 394
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9993574298874948e-05,
+      "loss": 0.0005,
+      "step": 395
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.999354172483342e-05,
+      "loss": 0.5039,
+      "step": 396
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.99935090684539e-05,
+      "loss": 0.543,
+      "step": 397
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9993476329736555e-05,
+      "loss": 0.0002,
+      "step": 398
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9993443508681568e-05,
+      "loss": 0.668,
+      "step": 399
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9993410605289125e-05,
+      "loss": 0.6328,
+      "step": 400
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9993377619559404e-05,
+      "loss": 0.9453,
+      "step": 401
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.999334455149258e-05,
+      "loss": 0.668,
+      "step": 402
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9993311401088847e-05,
+      "loss": 0.5352,
+      "step": 403
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9993278168348377e-05,
+      "loss": 0.7656,
+      "step": 404
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9993244853271355e-05,
+      "loss": 0.2695,
+      "step": 405
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.999321145585796e-05,
+      "loss": 1.8438,
+      "step": 406
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.999317797610838e-05,
+      "loss": 0.4863,
+      "step": 407
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9993144414022803e-05,
+      "loss": 0.6094,
+      "step": 408
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9993110769601407e-05,
+      "loss": 0.373,
+      "step": 409
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.999307704284438e-05,
+      "loss": 0.8203,
+      "step": 410
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9993043233751905e-05,
+      "loss": 0.7617,
+      "step": 411
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9993009342324165e-05,
+      "loss": 1.6328,
+      "step": 412
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9992975368561353e-05,
+      "loss": 0.8711,
+      "step": 413
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9992941312463652e-05,
+      "loss": 1.9766,
+      "step": 414
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.999290717403125e-05,
+      "loss": 0.7578,
+      "step": 415
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.999287295326433e-05,
+      "loss": 0.0082,
+      "step": 416
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9992838650163086e-05,
+      "loss": 0.8438,
+      "step": 417
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9992804264727706e-05,
+      "loss": 0.0344,
+      "step": 418
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9992769796958372e-05,
+      "loss": 0.4336,
+      "step": 419
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.999273524685528e-05,
+      "loss": 0.0006,
+      "step": 420
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9992700614418616e-05,
+      "loss": 0.1216,
+      "step": 421
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9992665899648577e-05,
+      "loss": 0.2432,
+      "step": 422
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9992631102545345e-05,
+      "loss": 0.2334,
+      "step": 423
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9992596223109112e-05,
+      "loss": 0.5312,
+      "step": 424
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9992561261340078e-05,
+      "loss": 0.9023,
+      "step": 425
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9992526217238424e-05,
+      "loss": 0.2676,
+      "step": 426
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.999249109080435e-05,
+      "loss": 1.375,
+      "step": 427
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.999245588203805e-05,
+      "loss": 0.0369,
+      "step": 428
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9992420590939707e-05,
+      "loss": 0.1875,
+      "step": 429
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9992385217509526e-05,
+      "loss": 0.0006,
+      "step": 430
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9992349761747695e-05,
+      "loss": 0.5625,
+      "step": 431
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.999231422365441e-05,
+      "loss": 1.3438,
+      "step": 432
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9992278603229872e-05,
+      "loss": 0.1543,
+      "step": 433
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.999224290047427e-05,
+      "loss": 1.2812,
+      "step": 434
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9992207115387795e-05,
+      "loss": 0.0051,
+      "step": 435
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9992171247970655e-05,
+      "loss": 0.0015,
+      "step": 436
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.999213529822304e-05,
+      "loss": 1.6484,
+      "step": 437
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9992099266145153e-05,
+      "loss": 0.3164,
+      "step": 438
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9992063151737187e-05,
+      "loss": 0.0317,
+      "step": 439
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.9992026954999342e-05,
+      "loss": 0.2041,
+      "step": 440
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.999199067593181e-05,
+      "loss": 0.5938,
+      "step": 441
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9991954314534803e-05,
+      "loss": 0.009,
+      "step": 442
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9991917870808515e-05,
+      "loss": 0.0042,
+      "step": 443
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.999188134475314e-05,
+      "loss": 0.0027,
+      "step": 444
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.999184473636889e-05,
+      "loss": 0.2832,
+      "step": 445
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9991808045655955e-05,
+      "loss": 0.0093,
+      "step": 446
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9991771272614542e-05,
+      "loss": 0.0003,
+      "step": 447
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9991734417244858e-05,
+      "loss": 0.4531,
+      "step": 448
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9991697479547098e-05,
+      "loss": 0.3047,
+      "step": 449
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9991660459521462e-05,
+      "loss": 0.3867,
+      "step": 450
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.999162335716816e-05,
+      "loss": 0.5703,
+      "step": 451
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.99915861724874e-05,
+      "loss": 2.0781,
+      "step": 452
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.999154890547937e-05,
+      "loss": 0.625,
+      "step": 453
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.999151155614429e-05,
+      "loss": 0.5469,
+      "step": 454
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9991474124482355e-05,
+      "loss": 1.0859,
+      "step": 455
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.999143661049378e-05,
+      "loss": 2.0938,
+      "step": 456
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9991399014178765e-05,
+      "loss": 0.6406,
+      "step": 457
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.999136133553752e-05,
+      "loss": 0.5234,
+      "step": 458
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9991323574570246e-05,
+      "loss": 2.6094,
+      "step": 459
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9991285731277158e-05,
+      "loss": 0.6641,
+      "step": 460
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9991247805658458e-05,
+      "loss": 2.0625,
+      "step": 461
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9991209797714355e-05,
+      "loss": 1.3906,
+      "step": 462
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9991171707445057e-05,
+      "loss": 1.3516,
+      "step": 463
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9991133534850777e-05,
+      "loss": 0.6094,
+      "step": 464
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9991095279931725e-05,
+      "loss": 0.7695,
+      "step": 465
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9991056942688108e-05,
+      "loss": 0.7266,
+      "step": 466
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9991018523120136e-05,
+      "loss": 1.5312,
+      "step": 467
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.999098002122802e-05,
+      "loss": 0.6875,
+      "step": 468
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9990941437011975e-05,
+      "loss": 0.3516,
+      "step": 469
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.999090277047221e-05,
+      "loss": 0.8633,
+      "step": 470
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9990864021608938e-05,
+      "loss": 0.0189,
+      "step": 471
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9990825190422367e-05,
+      "loss": 0.5508,
+      "step": 472
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9990786276912722e-05,
+      "loss": 0.418,
+      "step": 473
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.999074728108021e-05,
+      "loss": 3.0781,
+      "step": 474
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9990708202925038e-05,
+      "loss": 0.1514,
+      "step": 475
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9990669042447432e-05,
+      "loss": 0.0017,
+      "step": 476
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9990629799647603e-05,
+      "loss": 0.0005,
+      "step": 477
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9990590474525763e-05,
+      "loss": 0.1162,
+      "step": 478
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9990551067082133e-05,
+      "loss": 0.0065,
+      "step": 479
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9990511577316925e-05,
+      "loss": 0.0003,
+      "step": 480
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.999047200523036e-05,
+      "loss": 0.707,
+      "step": 481
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9990432350822654e-05,
+      "loss": 0.0005,
+      "step": 482
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.999039261409402e-05,
+      "loss": 0.0162,
+      "step": 483
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9990352795044686e-05,
+      "loss": 0.9375,
+      "step": 484
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9990312893674863e-05,
+      "loss": 0.5625,
+      "step": 485
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.999027290998477e-05,
+      "loss": 0.0046,
+      "step": 486
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.999023284397463e-05,
+      "loss": 0.8047,
+      "step": 487
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9990192695644664e-05,
+      "loss": 0.0001,
+      "step": 488
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9990152464995093e-05,
+      "loss": 0.3965,
+      "step": 489
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.999011215202613e-05,
+      "loss": 0.0004,
+      "step": 490
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9990071756738e-05,
+      "loss": 0.0005,
+      "step": 491
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.999003127913093e-05,
+      "loss": 0.0006,
+      "step": 492
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.998999071920514e-05,
+      "loss": 0.6758,
+      "step": 493
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9989950076960847e-05,
+      "loss": 0.003,
+      "step": 494
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9989909352398283e-05,
+      "loss": 0.2949,
+      "step": 495
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9989868545517663e-05,
+      "loss": 0.8984,
+      "step": 496
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9989827656319223e-05,
+      "loss": 0.9023,
+      "step": 497
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9989786684803174e-05,
+      "loss": 0.4258,
+      "step": 498
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9989745630969745e-05,
+      "loss": 0.4492,
+      "step": 499
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9989704494819168e-05,
+      "loss": 0.3301,
+      "step": 500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9989663276351662e-05,
+      "loss": 1.2422,
+      "step": 501
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9989621975567455e-05,
+      "loss": 0.8477,
+      "step": 502
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.998958059246678e-05,
+      "loss": 1.2969,
+      "step": 503
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.998953912704985e-05,
+      "loss": 0.5859,
+      "step": 504
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9989497579316905e-05,
+      "loss": 1.5781,
+      "step": 505
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.998945594926817e-05,
+      "loss": 1.5703,
+      "step": 506
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9989414236903873e-05,
+      "loss": 0.8125,
+      "step": 507
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.9989372442224248e-05,
+      "loss": 0.6992,
+      "step": 508
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9989330565229514e-05,
+      "loss": 0.957,
+      "step": 509
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.998928860591991e-05,
+      "loss": 0.3613,
+      "step": 510
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9989246564295664e-05,
+      "loss": 0.4629,
+      "step": 511
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9989204440356998e-05,
+      "loss": 0.8242,
+      "step": 512
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.998916223410416e-05,
+      "loss": 0.027,
+      "step": 513
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9989119945537373e-05,
+      "loss": 1.0469,
+      "step": 514
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9989077574656863e-05,
+      "loss": 0.6289,
+      "step": 515
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9989035121462874e-05,
+      "loss": 0.1338,
+      "step": 516
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.998899258595563e-05,
+      "loss": 0.0903,
+      "step": 517
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9988949968135373e-05,
+      "loss": 0.051,
+      "step": 518
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9988907268002332e-05,
+      "loss": 0.0549,
+      "step": 519
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9988864485556743e-05,
+      "loss": 0.0659,
+      "step": 520
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.998882162079884e-05,
+      "loss": 0.1113,
+      "step": 521
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9988778673728857e-05,
+      "loss": 0.1924,
+      "step": 522
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9988735644347032e-05,
+      "loss": 0.0776,
+      "step": 523
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.99886925326536e-05,
+      "loss": 0.2695,
+      "step": 524
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.99886493386488e-05,
+      "loss": 0.3711,
+      "step": 525
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.998860606233287e-05,
+      "loss": 0.0062,
+      "step": 526
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.998856270370604e-05,
+      "loss": 0.543,
+      "step": 527
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9988519262768558e-05,
+      "loss": 0.7461,
+      "step": 528
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9988475739520652e-05,
+      "loss": 0.3867,
+      "step": 529
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9988432133962577e-05,
+      "loss": 0.0053,
+      "step": 530
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9988388446094553e-05,
+      "loss": 0.625,
+      "step": 531
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9988344675916838e-05,
+      "loss": 0.0004,
+      "step": 532
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9988300823429655e-05,
+      "loss": 0.0002,
+      "step": 533
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.998825688863326e-05,
+      "loss": 0.1572,
+      "step": 534
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9988212871527888e-05,
+      "loss": 0.0007,
+      "step": 535
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9988168772113775e-05,
+      "loss": 0.2969,
+      "step": 536
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9988124590391174e-05,
+      "loss": 0.0052,
+      "step": 537
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9988080326360318e-05,
+      "loss": 0.0015,
+      "step": 538
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9988035980021457e-05,
+      "loss": 0.2168,
+      "step": 539
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9987991551374837e-05,
+      "loss": 0.0496,
+      "step": 540
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9987947040420693e-05,
+      "loss": 0.0012,
+      "step": 541
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.998790244715927e-05,
+      "loss": 0.0044,
+      "step": 542
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.998785777159082e-05,
+      "loss": 0.0281,
+      "step": 543
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9987813013715584e-05,
+      "loss": 0.0046,
+      "step": 544
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.998776817353381e-05,
+      "loss": 0.022,
+      "step": 545
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9987723251045743e-05,
+      "loss": 0.0062,
+      "step": 546
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9987678246251625e-05,
+      "loss": 0.9297,
+      "step": 547
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.998763315915171e-05,
+      "loss": 1.0,
+      "step": 548
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.998758798974624e-05,
+      "loss": 0.2461,
+      "step": 549
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9987542738035472e-05,
+      "loss": 0.6133,
+      "step": 550
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9987497404019648e-05,
+      "loss": 1.6406,
+      "step": 551
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9987451987699017e-05,
+      "loss": 0.127,
+      "step": 552
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9987406489073825e-05,
+      "loss": 0.1445,
+      "step": 553
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9987360908144332e-05,
+      "loss": 0.6211,
+      "step": 554
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.998731524491078e-05,
+      "loss": 0.0933,
+      "step": 555
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.998726949937342e-05,
+      "loss": 1.0469,
+      "step": 556
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9987223671532504e-05,
+      "loss": 1.1953,
+      "step": 557
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.998717776138829e-05,
+      "loss": 0.0064,
+      "step": 558
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.998713176894102e-05,
+      "loss": 1.9141,
+      "step": 559
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.998708569419096e-05,
+      "loss": 0.6602,
+      "step": 560
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9987039537138347e-05,
+      "loss": 0.6289,
+      "step": 561
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9986993297783442e-05,
+      "loss": 0.7812,
+      "step": 562
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9986946976126503e-05,
+      "loss": 0.0356,
+      "step": 563
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9986900572167773e-05,
+      "loss": 0.1641,
+      "step": 564
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9986854085907525e-05,
+      "loss": 0.6133,
+      "step": 565
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9986807517345998e-05,
+      "loss": 1.5391,
+      "step": 566
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9986760866483452e-05,
+      "loss": 0.0576,
+      "step": 567
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9986714133320147e-05,
+      "loss": 1.6562,
+      "step": 568
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9986667317856335e-05,
+      "loss": 0.4219,
+      "step": 569
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9986620420092275e-05,
+      "loss": 0.1494,
+      "step": 570
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9986573440028224e-05,
+      "loss": 0.0026,
+      "step": 571
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9986526377664445e-05,
+      "loss": 0.0747,
+      "step": 572
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9986479233001187e-05,
+      "loss": 0.0001,
+      "step": 573
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.9986432006038717e-05,
+      "loss": 0.1465,
+      "step": 574
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.998638469677729e-05,
+      "loss": 0.5742,
+      "step": 575
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.998633730521717e-05,
+      "loss": 1.8047,
+      "step": 576
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.998628983135861e-05,
+      "loss": 0.0156,
+      "step": 577
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9986242275201878e-05,
+      "loss": 0.9375,
+      "step": 578
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9986194636747234e-05,
+      "loss": 1.7812,
+      "step": 579
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9986146915994934e-05,
+      "loss": 0.1147,
+      "step": 580
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9986099112945248e-05,
+      "loss": 0.2354,
+      "step": 581
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.998605122759843e-05,
+      "loss": 1.8203,
+      "step": 582
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9986003259954752e-05,
+      "loss": 0.003,
+      "step": 583
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9985955210014468e-05,
+      "loss": 0.0006,
+      "step": 584
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.998590707777785e-05,
+      "loss": 0.0178,
+      "step": 585
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.998585886324516e-05,
+      "loss": 0.4863,
+      "step": 586
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.998581056641666e-05,
+      "loss": 0.0004,
+      "step": 587
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9985762187292615e-05,
+      "loss": 0.0282,
+      "step": 588
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.99857137258733e-05,
+      "loss": 0.0004,
+      "step": 589
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9985665182158966e-05,
+      "loss": 0.3887,
+      "step": 590
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.998561655614989e-05,
+      "loss": 0.1748,
+      "step": 591
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9985567847846335e-05,
+      "loss": 0.0107,
+      "step": 592
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9985519057248573e-05,
+      "loss": 0.3594,
+      "step": 593
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9985470184356867e-05,
+      "loss": 0.0004,
+      "step": 594
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.998542122917149e-05,
+      "loss": 0.0103,
+      "step": 595
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9985372191692707e-05,
+      "loss": 0.6836,
+      "step": 596
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9985323071920783e-05,
+      "loss": 0.0001,
+      "step": 597
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9985273869856e-05,
+      "loss": 0.0016,
+      "step": 598
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.998522458549862e-05,
+      "loss": 0.2129,
+      "step": 599
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9985175218848913e-05,
+      "loss": 0.4531,
+      "step": 600
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9985125769907155e-05,
+      "loss": 0.5078,
+      "step": 601
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.998507623867361e-05,
+      "loss": 1.0391,
+      "step": 602
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9985026625148563e-05,
+      "loss": 0.4746,
+      "step": 603
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.998497692933227e-05,
+      "loss": 0.5508,
+      "step": 604
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9984927151225016e-05,
+      "loss": 0.707,
+      "step": 605
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9984877290827067e-05,
+      "loss": 0.793,
+      "step": 606
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9984827348138702e-05,
+      "loss": 0.7188,
+      "step": 607
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9984777323160195e-05,
+      "loss": 0.7891,
+      "step": 608
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9984727215891817e-05,
+      "loss": 1.0781,
+      "step": 609
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9984677026333846e-05,
+      "loss": 0.625,
+      "step": 610
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.998462675448656e-05,
+      "loss": 0.2402,
+      "step": 611
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9984576400350226e-05,
+      "loss": 0.8008,
+      "step": 612
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.998452596392513e-05,
+      "loss": 1.5078,
+      "step": 613
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9984475445211548e-05,
+      "loss": 0.9453,
+      "step": 614
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9984424844209752e-05,
+      "loss": 0.7109,
+      "step": 615
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9984374160920023e-05,
+      "loss": 0.5586,
+      "step": 616
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9984323395342636e-05,
+      "loss": 0.0381,
+      "step": 617
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.998427254747788e-05,
+      "loss": 0.3008,
+      "step": 618
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.998422161732602e-05,
+      "loss": 0.1504,
+      "step": 619
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9984170604887347e-05,
+      "loss": 1.25,
+      "step": 620
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9984119510162135e-05,
+      "loss": 0.6562,
+      "step": 621
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.998406833315067e-05,
+      "loss": 0.0057,
+      "step": 622
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9984017073853226e-05,
+      "loss": 0.0028,
+      "step": 623
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9983965732270094e-05,
+      "loss": 0.5781,
+      "step": 624
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9983914308401546e-05,
+      "loss": 0.2676,
+      "step": 625
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9983862802247868e-05,
+      "loss": 0.2949,
+      "step": 626
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9983811213809343e-05,
+      "loss": 1.0078,
+      "step": 627
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9983759543086257e-05,
+      "loss": 0.0005,
+      "step": 628
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.998370779007889e-05,
+      "loss": 0.0005,
+      "step": 629
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9983655954787524e-05,
+      "loss": 0.0014,
+      "step": 630
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.998360403721245e-05,
+      "loss": 0.0003,
+      "step": 631
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.998355203735395e-05,
+      "loss": 0.004,
+      "step": 632
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.998349995521231e-05,
+      "loss": 0.0016,
+      "step": 633
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.998344779078782e-05,
+      "loss": 1.25,
+      "step": 634
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.998339554408076e-05,
+      "loss": 0.054,
+      "step": 635
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9983343215091415e-05,
+      "loss": 0.0713,
+      "step": 636
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.998329080382008e-05,
+      "loss": 0.0041,
+      "step": 637
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9983238310267042e-05,
+      "loss": 0.0019,
+      "step": 638
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9983185734432582e-05,
+      "loss": 0.1709,
+      "step": 639
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9983133076317e-05,
+      "loss": 0.0002,
+      "step": 640
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.998308033592057e-05,
+      "loss": 1.6094,
+      "step": 641
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.99830275132436e-05,
+      "loss": 0.0005,
+      "step": 642
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.9982974608286363e-05,
+      "loss": 0.0002,
+      "step": 643
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.998292162104916e-05,
+      "loss": 0.0011,
+      "step": 644
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.998286855153228e-05,
+      "loss": 0.9102,
+      "step": 645
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9982815399736008e-05,
+      "loss": 0.6055,
+      "step": 646
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9982762165660647e-05,
+      "loss": 0.8984,
+      "step": 647
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.998270884930648e-05,
+      "loss": 0.6055,
+      "step": 648
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9982655450673807e-05,
+      "loss": 0.3008,
+      "step": 649
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.998260196976291e-05,
+      "loss": 0.0003,
+      "step": 650
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9982548406574094e-05,
+      "loss": 1.2266,
+      "step": 651
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.998249476110765e-05,
+      "loss": 0.8945,
+      "step": 652
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9982441033363877e-05,
+      "loss": 1.3203,
+      "step": 653
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.998238722334306e-05,
+      "loss": 0.8828,
+      "step": 654
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.99823333310455e-05,
+      "loss": 0.582,
+      "step": 655
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9982279356471493e-05,
+      "loss": 0.5156,
+      "step": 656
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9982225299621334e-05,
+      "loss": 0.3555,
+      "step": 657
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.998217116049532e-05,
+      "loss": 0.7227,
+      "step": 658
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9982116939093756e-05,
+      "loss": 0.0056,
+      "step": 659
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9982062635416925e-05,
+      "loss": 0.0559,
+      "step": 660
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9982008249465135e-05,
+      "loss": 0.1494,
+      "step": 661
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9981953781238684e-05,
+      "loss": 0.1631,
+      "step": 662
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.998189923073787e-05,
+      "loss": 1.3281,
+      "step": 663
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9981844597962992e-05,
+      "loss": 0.0226,
+      "step": 664
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9981789882914352e-05,
+      "loss": 1.4297,
+      "step": 665
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9981735085592246e-05,
+      "loss": 0.1523,
+      "step": 666
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9981680205996976e-05,
+      "loss": 0.0011,
+      "step": 667
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9981625244128847e-05,
+      "loss": 0.0004,
+      "step": 668
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.998157019998816e-05,
+      "loss": 1.8281,
+      "step": 669
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9981515073575222e-05,
+      "loss": 1.7031,
+      "step": 670
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9981459864890325e-05,
+      "loss": 0.0977,
+      "step": 671
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9981404573933772e-05,
+      "loss": 0.0583,
+      "step": 672
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9981349200705878e-05,
+      "loss": 0.0012,
+      "step": 673
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9981293745206937e-05,
+      "loss": 0.0045,
+      "step": 674
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.998123820743726e-05,
+      "loss": 0.001,
+      "step": 675
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.998118258739715e-05,
+      "loss": 0.6484,
+      "step": 676
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9981126885086913e-05,
+      "loss": 0.0002,
+      "step": 677
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9981071100506853e-05,
+      "loss": 0.0002,
+      "step": 678
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.998101523365727e-05,
+      "loss": 0.1289,
+      "step": 679
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9980959284538487e-05,
+      "loss": 0.8672,
+      "step": 680
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.99809032531508e-05,
+      "loss": 0.1001,
+      "step": 681
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.998084713949452e-05,
+      "loss": 0.0022,
+      "step": 682
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9980790943569948e-05,
+      "loss": 0.4082,
+      "step": 683
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9980734665377406e-05,
+      "loss": 0.5625,
+      "step": 684
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.998067830491719e-05,
+      "loss": 0.0013,
+      "step": 685
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.998062186218962e-05,
+      "loss": 0.0771,
+      "step": 686
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9980565337195e-05,
+      "loss": 0.1523,
+      "step": 687
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9980508729933638e-05,
+      "loss": 0.377,
+      "step": 688
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.998045204040585e-05,
+      "loss": 0.0008,
+      "step": 689
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.998039526861195e-05,
+      "loss": 1.875,
+      "step": 690
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.998033841455224e-05,
+      "loss": 0.0393,
+      "step": 691
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9980281478227043e-05,
+      "loss": 0.0217,
+      "step": 692
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.998022445963666e-05,
+      "loss": 0.0139,
+      "step": 693
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9980167358781416e-05,
+      "loss": 0.0273,
+      "step": 694
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9980110175661618e-05,
+      "loss": 2.7344,
+      "step": 695
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.998005291027758e-05,
+      "loss": 0.0123,
+      "step": 696
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9979995562629615e-05,
+      "loss": 0.0008,
+      "step": 697
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9979938132718046e-05,
+      "loss": 0.0003,
+      "step": 698
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9979880620543177e-05,
+      "loss": 0.7109,
+      "step": 699
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9979823026105335e-05,
+      "loss": 2.6406,
+      "step": 700
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.997976534940483e-05,
+      "loss": 1.3047,
+      "step": 701
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9979707590441978e-05,
+      "loss": 0.6914,
+      "step": 702
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9979649749217095e-05,
+      "loss": 0.9102,
+      "step": 703
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9979591825730504e-05,
+      "loss": 0.5938,
+      "step": 704
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9979533819982527e-05,
+      "loss": 0.6367,
+      "step": 705
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.997947573197347e-05,
+      "loss": 0.207,
+      "step": 706
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9979417561703658e-05,
+      "loss": 0.6797,
+      "step": 707
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.997935930917341e-05,
+      "loss": 0.6406,
+      "step": 708
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9979300974383053e-05,
+      "loss": 0.7539,
+      "step": 709
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9979242557332894e-05,
+      "loss": 0.8516,
+      "step": 710
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.9979184058023263e-05,
+      "loss": 1.9062,
+      "step": 711
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.997912547645448e-05,
+      "loss": 0.6328,
+      "step": 712
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9979066812626865e-05,
+      "loss": 0.4863,
+      "step": 713
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9979008066540737e-05,
+      "loss": 0.0776,
+      "step": 714
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9978949238196428e-05,
+      "loss": 0.3242,
+      "step": 715
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9978890327594252e-05,
+      "loss": 0.1533,
+      "step": 716
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.997883133473454e-05,
+      "loss": 0.7305,
+      "step": 717
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.997877225961761e-05,
+      "loss": 0.918,
+      "step": 718
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9978713102243785e-05,
+      "loss": 0.0771,
+      "step": 719
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9978653862613396e-05,
+      "loss": 0.1387,
+      "step": 720
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9978594540726767e-05,
+      "loss": 0.0208,
+      "step": 721
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9978535136584222e-05,
+      "loss": 0.668,
+      "step": 722
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9978475650186084e-05,
+      "loss": 0.5352,
+      "step": 723
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9978416081532686e-05,
+      "loss": 0.0483,
+      "step": 724
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.997835643062435e-05,
+      "loss": 1.5156,
+      "step": 725
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.997829669746141e-05,
+      "loss": 0.8633,
+      "step": 726
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.997823688204419e-05,
+      "loss": 0.0708,
+      "step": 727
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9978176984373012e-05,
+      "loss": 2.7188,
+      "step": 728
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.997811700444822e-05,
+      "loss": 0.4492,
+      "step": 729
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.997805694227013e-05,
+      "loss": 0.0192,
+      "step": 730
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9977996797839075e-05,
+      "loss": 0.2275,
+      "step": 731
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.997793657115539e-05,
+      "loss": 0.5273,
+      "step": 732
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9977876262219402e-05,
+      "loss": 0.5547,
+      "step": 733
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.997781587103144e-05,
+      "loss": 1.0547,
+      "step": 734
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.997775539759184e-05,
+      "loss": 0.0023,
+      "step": 735
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9977694841900933e-05,
+      "loss": 1.0547,
+      "step": 736
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.997763420395905e-05,
+      "loss": 0.0001,
+      "step": 737
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9977573483766532e-05,
+      "loss": 0.0008,
+      "step": 738
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.99775126813237e-05,
+      "loss": 0.0092,
+      "step": 739
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.997745179663089e-05,
+      "loss": 0.2578,
+      "step": 740
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9977390829688445e-05,
+      "loss": 0.75,
+      "step": 741
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9977329780496694e-05,
+      "loss": 0.625,
+      "step": 742
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.997726864905597e-05,
+      "loss": 0.2471,
+      "step": 743
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9977207435366614e-05,
+      "loss": 0.0003,
+      "step": 744
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.997714613942896e-05,
+      "loss": 0.0025,
+      "step": 745
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.997708476124334e-05,
+      "loss": 0.0359,
+      "step": 746
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9977023300810098e-05,
+      "loss": 0.0013,
+      "step": 747
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9976961758129573e-05,
+      "loss": 0.3242,
+      "step": 748
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.997690013320209e-05,
+      "loss": 0.3926,
+      "step": 749
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.997683842602801e-05,
+      "loss": 0.5625,
+      "step": 750
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9976776636607647e-05,
+      "loss": 0.0544,
+      "step": 751
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9976714764941352e-05,
+      "loss": 1.8047,
+      "step": 752
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9976652811029465e-05,
+      "loss": 1.7344,
+      "step": 753
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.997659077487233e-05,
+      "loss": 0.7578,
+      "step": 754
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.997652865647028e-05,
+      "loss": 0.7227,
+      "step": 755
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9976466455823656e-05,
+      "loss": 0.4922,
+      "step": 756
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.997640417293281e-05,
+      "loss": 0.625,
+      "step": 757
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.997634180779807e-05,
+      "loss": 0.6484,
+      "step": 758
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.997627936041979e-05,
+      "loss": 0.6484,
+      "step": 759
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9976216830798307e-05,
+      "loss": 0.668,
+      "step": 760
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9976154218933968e-05,
+      "loss": 0.5352,
+      "step": 761
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9976091524827108e-05,
+      "loss": 0.8086,
+      "step": 762
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.997602874847808e-05,
+      "loss": 0.6641,
+      "step": 763
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.997596588988723e-05,
+      "loss": 0.0654,
+      "step": 764
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.99759029490549e-05,
+      "loss": 0.0664,
+      "step": 765
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9975839925981434e-05,
+      "loss": 2.0781,
+      "step": 766
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9975776820667176e-05,
+      "loss": 2.25,
+      "step": 767
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.997571363311248e-05,
+      "loss": 0.4941,
+      "step": 768
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.997565036331769e-05,
+      "loss": 0.0457,
+      "step": 769
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.997558701128315e-05,
+      "loss": 0.1934,
+      "step": 770
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.997552357700921e-05,
+      "loss": 0.0056,
+      "step": 771
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.997546006049622e-05,
+      "loss": 0.1123,
+      "step": 772
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9975396461744524e-05,
+      "loss": 0.1348,
+      "step": 773
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.997533278075448e-05,
+      "loss": 0.5625,
+      "step": 774
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.997526901752643e-05,
+      "loss": 0.0625,
+      "step": 775
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9975205172060724e-05,
+      "loss": 0.0713,
+      "step": 776
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.997514124435772e-05,
+      "loss": 0.0052,
+      "step": 777
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9975077234417765e-05,
+      "loss": 0.0006,
+      "step": 778
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9975013142241205e-05,
+      "loss": 0.0013,
+      "step": 779
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.9974948967828402e-05,
+      "loss": 0.1367,
+      "step": 780
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.99748847111797e-05,
+      "loss": 0.7305,
+      "step": 781
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9974820372295454e-05,
+      "loss": 0.0008,
+      "step": 782
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.997475595117602e-05,
+      "loss": 0.0009,
+      "step": 783
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.997469144782175e-05,
+      "loss": 0.8242,
+      "step": 784
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9974626862232998e-05,
+      "loss": 0.0036,
+      "step": 785
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.997456219441012e-05,
+      "loss": 0.0003,
+      "step": 786
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9974497444353468e-05,
+      "loss": 0.0854,
+      "step": 787
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9974432612063402e-05,
+      "loss": 0.0006,
+      "step": 788
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9974367697540277e-05,
+      "loss": 1.7578,
+      "step": 789
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.997430270078445e-05,
+      "loss": 0.0004,
+      "step": 790
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.997423762179627e-05,
+      "loss": 0.5703,
+      "step": 791
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9974172460576105e-05,
+      "loss": 0.2539,
+      "step": 792
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9974107217124303e-05,
+      "loss": 0.3496,
+      "step": 793
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.997404189144123e-05,
+      "loss": 0.2598,
+      "step": 794
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.997397648352725e-05,
+      "loss": 0.3926,
+      "step": 795
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.997391099338271e-05,
+      "loss": 0.0008,
+      "step": 796
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.997384542100797e-05,
+      "loss": 0.3691,
+      "step": 797
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9973779766403397e-05,
+      "loss": 0.332,
+      "step": 798
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.997371402956935e-05,
+      "loss": 0.3691,
+      "step": 799
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.997364821050619e-05,
+      "loss": 0.2832,
+      "step": 800
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9973582309214278e-05,
+      "loss": 0.3379,
+      "step": 801
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.997351632569397e-05,
+      "loss": 0.6484,
+      "step": 802
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.997345025994564e-05,
+      "loss": 1.2031,
+      "step": 803
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.997338411196964e-05,
+      "loss": 0.0928,
+      "step": 804
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.997331788176634e-05,
+      "loss": 0.7266,
+      "step": 805
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9973251569336098e-05,
+      "loss": 1.9141,
+      "step": 806
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9973185174679287e-05,
+      "loss": 0.3164,
+      "step": 807
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.997311869779626e-05,
+      "loss": 0.957,
+      "step": 808
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9973052138687396e-05,
+      "loss": 0.6602,
+      "step": 809
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.997298549735305e-05,
+      "loss": 1.0234,
+      "step": 810
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9972918773793588e-05,
+      "loss": 0.3516,
+      "step": 811
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.997285196800938e-05,
+      "loss": 0.1924,
+      "step": 812
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9972785080000792e-05,
+      "loss": 0.7031,
+      "step": 813
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9972718109768193e-05,
+      "loss": 0.248,
+      "step": 814
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9972651057311944e-05,
+      "loss": 0.9258,
+      "step": 815
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9972583922632425e-05,
+      "loss": 1.75,
+      "step": 816
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9972516705729998e-05,
+      "loss": 0.8984,
+      "step": 817
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9972449406605026e-05,
+      "loss": 0.0039,
+      "step": 818
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.997238202525789e-05,
+      "loss": 0.0022,
+      "step": 819
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9972314561688953e-05,
+      "loss": 0.2812,
+      "step": 820
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9972247015898588e-05,
+      "loss": 0.0028,
+      "step": 821
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.997217938788716e-05,
+      "loss": 0.8047,
+      "step": 822
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9972111677655053e-05,
+      "loss": 0.0038,
+      "step": 823
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9972043885202628e-05,
+      "loss": 0.0002,
+      "step": 824
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9971976010530264e-05,
+      "loss": 0.5547,
+      "step": 825
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9971908053638326e-05,
+      "loss": 0.8125,
+      "step": 826
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9971840014527194e-05,
+      "loss": 0.0214,
+      "step": 827
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9971771893197236e-05,
+      "loss": 0.2559,
+      "step": 828
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9971703689648836e-05,
+      "loss": 0.2373,
+      "step": 829
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9971635403882353e-05,
+      "loss": 0.3652,
+      "step": 830
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.997156703589818e-05,
+      "loss": 0.0001,
+      "step": 831
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9971498585696676e-05,
+      "loss": 0.8281,
+      "step": 832
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9971430053278224e-05,
+      "loss": 0.2129,
+      "step": 833
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9971361438643207e-05,
+      "loss": 1.2656,
+      "step": 834
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.997129274179199e-05,
+      "loss": 0.0011,
+      "step": 835
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.997122396272496e-05,
+      "loss": 0.3789,
+      "step": 836
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9971155101442487e-05,
+      "loss": 0.0006,
+      "step": 837
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9971086157944956e-05,
+      "loss": 0.0028,
+      "step": 838
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9971017132232737e-05,
+      "loss": 0.0004,
+      "step": 839
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.997094802430622e-05,
+      "loss": 0.0854,
+      "step": 840
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9970878834165773e-05,
+      "loss": 0.0087,
+      "step": 841
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9970809561811786e-05,
+      "loss": 0.0017,
+      "step": 842
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.997074020724463e-05,
+      "loss": 0.0028,
+      "step": 843
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9970670770464698e-05,
+      "loss": 0.0006,
+      "step": 844
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9970601251472356e-05,
+      "loss": 0.0005,
+      "step": 845
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9970531650268002e-05,
+      "loss": 0.0012,
+      "step": 846
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.9970461966852002e-05,
+      "loss": 0.5195,
+      "step": 847
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9970392201224754e-05,
+      "loss": 0.7617,
+      "step": 848
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9970322353386628e-05,
+      "loss": 0.6484,
+      "step": 849
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.997025242333802e-05,
+      "loss": 0.1152,
+      "step": 850
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9970182411079306e-05,
+      "loss": 0.9336,
+      "step": 851
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.997011231661087e-05,
+      "loss": 0.3691,
+      "step": 852
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9970042139933096e-05,
+      "loss": 0.6055,
+      "step": 853
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.996997188104638e-05,
+      "loss": 0.6328,
+      "step": 854
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9969901539951095e-05,
+      "loss": 2.2969,
+      "step": 855
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9969831116647635e-05,
+      "loss": 0.7031,
+      "step": 856
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.996976061113638e-05,
+      "loss": 0.5273,
+      "step": 857
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9969690023417725e-05,
+      "loss": 1.0859,
+      "step": 858
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.996961935349205e-05,
+      "loss": 1.2109,
+      "step": 859
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.996954860135975e-05,
+      "loss": 1.2031,
+      "step": 860
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9969477767021218e-05,
+      "loss": 0.4707,
+      "step": 861
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9969406850476825e-05,
+      "loss": 0.6484,
+      "step": 862
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9969335851726974e-05,
+      "loss": 0.7266,
+      "step": 863
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9969264770772053e-05,
+      "loss": 1.7812,
+      "step": 864
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9969193607612453e-05,
+      "loss": 0.6523,
+      "step": 865
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9969122362248562e-05,
+      "loss": 0.4395,
+      "step": 866
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.996905103468077e-05,
+      "loss": 0.2695,
+      "step": 867
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9968979624909477e-05,
+      "loss": 0.0403,
+      "step": 868
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.996890813293506e-05,
+      "loss": 1.0938,
+      "step": 869
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.996883655875793e-05,
+      "loss": 0.0002,
+      "step": 870
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9968764902378468e-05,
+      "loss": 2.0625,
+      "step": 871
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9968693163797064e-05,
+      "loss": 0.0117,
+      "step": 872
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9968621343014127e-05,
+      "loss": 0.5234,
+      "step": 873
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.996854944003004e-05,
+      "loss": 1.2969,
+      "step": 874
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9968477454845197e-05,
+      "loss": 0.4082,
+      "step": 875
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9968405387460005e-05,
+      "loss": 0.0105,
+      "step": 876
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9968333237874847e-05,
+      "loss": 0.0031,
+      "step": 877
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9968261006090122e-05,
+      "loss": 0.0003,
+      "step": 878
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.996818869210623e-05,
+      "loss": 0.168,
+      "step": 879
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.996811629592357e-05,
+      "loss": 0.0957,
+      "step": 880
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9968043817542532e-05,
+      "loss": 0.001,
+      "step": 881
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.996797125696352e-05,
+      "loss": 0.5664,
+      "step": 882
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.996789861418693e-05,
+      "loss": 0.1387,
+      "step": 883
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.996782588921316e-05,
+      "loss": 1.4453,
+      "step": 884
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9967753082042618e-05,
+      "loss": 0.0226,
+      "step": 885
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.996768019267569e-05,
+      "loss": 0.006,
+      "step": 886
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9967607221112787e-05,
+      "loss": 0.0134,
+      "step": 887
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9967534167354303e-05,
+      "loss": 0.0006,
+      "step": 888
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9967461031400644e-05,
+      "loss": 0.0007,
+      "step": 889
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.996738781325221e-05,
+      "loss": 0.0008,
+      "step": 890
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9967314512909402e-05,
+      "loss": 0.0006,
+      "step": 891
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.996724113037262e-05,
+      "loss": 0.1807,
+      "step": 892
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9967167665642274e-05,
+      "loss": 0.2334,
+      "step": 893
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9967094118718762e-05,
+      "loss": 0.3223,
+      "step": 894
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9967020489602493e-05,
+      "loss": 0.0002,
+      "step": 895
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9966946778293867e-05,
+      "loss": 0.2178,
+      "step": 896
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9966872984793287e-05,
+      "loss": 0.2891,
+      "step": 897
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.996679910910116e-05,
+      "loss": 0.0005,
+      "step": 898
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9966725151217896e-05,
+      "loss": 0.0002,
+      "step": 899
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9966651111143897e-05,
+      "loss": 0.2227,
+      "step": 900
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.996657698887957e-05,
+      "loss": 1.3984,
+      "step": 901
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9966502784425328e-05,
+      "loss": 0.5156,
+      "step": 902
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.996642849778157e-05,
+      "loss": 1.2578,
+      "step": 903
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9966354128948704e-05,
+      "loss": 0.6836,
+      "step": 904
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9966279677927143e-05,
+      "loss": 1.4453,
+      "step": 905
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.996620514471729e-05,
+      "loss": 0.0227,
+      "step": 906
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9966130529319565e-05,
+      "loss": 0.8828,
+      "step": 907
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9966055831734368e-05,
+      "loss": 0.7148,
+      "step": 908
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9965981051962112e-05,
+      "loss": 0.6797,
+      "step": 909
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.996590619000321e-05,
+      "loss": 1.8672,
+      "step": 910
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.996583124585807e-05,
+      "loss": 0.8438,
+      "step": 911
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9965756219527104e-05,
+      "loss": 0.9102,
+      "step": 912
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9965681111010727e-05,
+      "loss": 1.0156,
+      "step": 913
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.9965605920309348e-05,
+      "loss": 0.5859,
+      "step": 914
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.996553064742338e-05,
+      "loss": 0.3711,
+      "step": 915
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.996545529235324e-05,
+      "loss": 0.9805,
+      "step": 916
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.996537985509934e-05,
+      "loss": 0.3418,
+      "step": 917
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9965304335662088e-05,
+      "loss": 0.3887,
+      "step": 918
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.996522873404191e-05,
+      "loss": 0.0698,
+      "step": 919
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9965153050239212e-05,
+      "loss": 0.126,
+      "step": 920
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9965077284254414e-05,
+      "loss": 0.0332,
+      "step": 921
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9965001436087935e-05,
+      "loss": 0.0007,
+      "step": 922
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9964925505740185e-05,
+      "loss": 0.9102,
+      "step": 923
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.996484949321158e-05,
+      "loss": 0.0086,
+      "step": 924
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9964773398502545e-05,
+      "loss": 0.0569,
+      "step": 925
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9964697221613495e-05,
+      "loss": 0.5039,
+      "step": 926
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9964620962544845e-05,
+      "loss": 0.0005,
+      "step": 927
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9964544621297018e-05,
+      "loss": 0.4453,
+      "step": 928
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.996446819787043e-05,
+      "loss": 0.3359,
+      "step": 929
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.99643916922655e-05,
+      "loss": 1.3047,
+      "step": 930
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9964315104482654e-05,
+      "loss": 0.6641,
+      "step": 931
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9964238434522312e-05,
+      "loss": 0.2949,
+      "step": 932
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9964161682384884e-05,
+      "loss": 0.2061,
+      "step": 933
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.99640848480708e-05,
+      "loss": 0.0029,
+      "step": 934
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9964007931580485e-05,
+      "loss": 0.1826,
+      "step": 935
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9963930932914358e-05,
+      "loss": 0.0021,
+      "step": 936
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.996385385207284e-05,
+      "loss": 0.2441,
+      "step": 937
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9963776689056356e-05,
+      "loss": 0.4688,
+      "step": 938
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.996369944386533e-05,
+      "loss": 0.0081,
+      "step": 939
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9963622116500182e-05,
+      "loss": 0.0114,
+      "step": 940
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9963544706961344e-05,
+      "loss": 0.2598,
+      "step": 941
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9963467215249237e-05,
+      "loss": 0.0103,
+      "step": 942
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9963389641364286e-05,
+      "loss": 0.3027,
+      "step": 943
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.996331198530692e-05,
+      "loss": 0.0903,
+      "step": 944
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.996323424707756e-05,
+      "loss": 0.0036,
+      "step": 945
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.996315642667664e-05,
+      "loss": 0.0029,
+      "step": 946
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9963078524104585e-05,
+      "loss": 0.375,
+      "step": 947
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9963000539361817e-05,
+      "loss": 0.1523,
+      "step": 948
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9962922472448768e-05,
+      "loss": 0.0008,
+      "step": 949
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.996284432336587e-05,
+      "loss": 0.4316,
+      "step": 950
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9962766092113553e-05,
+      "loss": 0.3789,
+      "step": 951
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9962687778692237e-05,
+      "loss": 0.5352,
+      "step": 952
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9962609383102365e-05,
+      "loss": 1.1562,
+      "step": 953
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9962530905344354e-05,
+      "loss": 1.3281,
+      "step": 954
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.996245234541865e-05,
+      "loss": 0.3164,
+      "step": 955
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.996237370332567e-05,
+      "loss": 0.6523,
+      "step": 956
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.996229497906585e-05,
+      "loss": 0.6836,
+      "step": 957
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9962216172639626e-05,
+      "loss": 0.6133,
+      "step": 958
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9962137284047434e-05,
+      "loss": 1.2734,
+      "step": 959
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.99620583132897e-05,
+      "loss": 0.6094,
+      "step": 960
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.996197926036686e-05,
+      "loss": 0.3457,
+      "step": 961
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9961900125279346e-05,
+      "loss": 1.0078,
+      "step": 962
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9961820908027596e-05,
+      "loss": 0.0742,
+      "step": 963
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9961741608612045e-05,
+      "loss": 1.75,
+      "step": 964
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.996166222703313e-05,
+      "loss": 1.2578,
+      "step": 965
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9961582763291274e-05,
+      "loss": 0.5273,
+      "step": 966
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9961503217386933e-05,
+      "loss": 0.168,
+      "step": 967
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.996142358932053e-05,
+      "loss": 0.7578,
+      "step": 968
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9961343879092512e-05,
+      "loss": 0.1914,
+      "step": 969
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9961264086703308e-05,
+      "loss": 0.8086,
+      "step": 970
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.996118421215336e-05,
+      "loss": 0.377,
+      "step": 971
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9961104255443105e-05,
+      "loss": 0.0171,
+      "step": 972
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9961024216572984e-05,
+      "loss": 0.0052,
+      "step": 973
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9960944095543433e-05,
+      "loss": 0.0435,
+      "step": 974
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.99608638923549e-05,
+      "loss": 1.2344,
+      "step": 975
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.996078360700782e-05,
+      "loss": 0.6602,
+      "step": 976
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9960703239502632e-05,
+      "loss": 0.106,
+      "step": 977
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.996062278983978e-05,
+      "loss": 0.5742,
+      "step": 978
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9960542258019704e-05,
+      "loss": 0.0075,
+      "step": 979
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9960461644042852e-05,
+      "loss": 0.1328,
+      "step": 980
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.996038094790966e-05,
+      "loss": 0.0016,
+      "step": 981
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.9960300169620572e-05,
+      "loss": 0.0079,
+      "step": 982
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.996021930917603e-05,
+      "loss": 0.0053,
+      "step": 983
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9960138366576488e-05,
+      "loss": 0.2305,
+      "step": 984
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9960057341822382e-05,
+      "loss": 0.0208,
+      "step": 985
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9959976234914155e-05,
+      "loss": 0.001,
+      "step": 986
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.995989504585226e-05,
+      "loss": 0.0085,
+      "step": 987
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9959813774637136e-05,
+      "loss": 0.1118,
+      "step": 988
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9959732421269232e-05,
+      "loss": 0.0026,
+      "step": 989
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9959650985749e-05,
+      "loss": 0.0016,
+      "step": 990
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9959569468076873e-05,
+      "loss": 0.3359,
+      "step": 991
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9959487868253316e-05,
+      "loss": 0.0081,
+      "step": 992
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9959406186278764e-05,
+      "loss": 0.0059,
+      "step": 993
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9959324422153674e-05,
+      "loss": 0.2715,
+      "step": 994
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9959242575878488e-05,
+      "loss": 0.0001,
+      "step": 995
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.995916064745366e-05,
+      "loss": 0.5977,
+      "step": 996
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.995907863687964e-05,
+      "loss": 0.3438,
+      "step": 997
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9958996544156874e-05,
+      "loss": 0.21,
+      "step": 998
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.995891436928582e-05,
+      "loss": 0.0005,
+      "step": 999
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9958832112266922e-05,
+      "loss": 0.416,
+      "step": 1000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9958749773100638e-05,
+      "loss": 1.6484,
+      "step": 1001
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9958667351787418e-05,
+      "loss": 1.0781,
+      "step": 1002
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9958584848327706e-05,
+      "loss": 0.5742,
+      "step": 1003
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9958502262721973e-05,
+      "loss": 1.1016,
+      "step": 1004
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9958419594970653e-05,
+      "loss": 0.0271,
+      "step": 1005
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9958336845074217e-05,
+      "loss": 0.9648,
+      "step": 1006
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9958254013033108e-05,
+      "loss": 0.5195,
+      "step": 1007
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9958171098847784e-05,
+      "loss": 0.6016,
+      "step": 1008
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9958088102518706e-05,
+      "loss": 0.6523,
+      "step": 1009
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9958005024046317e-05,
+      "loss": 0.1426,
+      "step": 1010
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9957921863431086e-05,
+      "loss": 0.6172,
+      "step": 1011
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9957838620673462e-05,
+      "loss": 0.5117,
+      "step": 1012
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.995775529577391e-05,
+      "loss": 0.7227,
+      "step": 1013
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9957671888732877e-05,
+      "loss": 0.7578,
+      "step": 1014
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9957588399550828e-05,
+      "loss": 0.832,
+      "step": 1015
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.995750482822822e-05,
+      "loss": 0.0014,
+      "step": 1016
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9957421174765516e-05,
+      "loss": 0.0212,
+      "step": 1017
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9957337439163164e-05,
+      "loss": 0.252,
+      "step": 1018
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9957253621421636e-05,
+      "loss": 0.7617,
+      "step": 1019
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9957169721541388e-05,
+      "loss": 0.0267,
+      "step": 1020
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9957085739522874e-05,
+      "loss": 0.0967,
+      "step": 1021
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9957001675366563e-05,
+      "loss": 0.0022,
+      "step": 1022
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9956917529072917e-05,
+      "loss": 0.0149,
+      "step": 1023
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9956833300642403e-05,
+      "loss": 0.0031,
+      "step": 1024
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9956748990075465e-05,
+      "loss": 0.0,
+      "step": 1025
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9956664597372584e-05,
+      "loss": 0.1426,
+      "step": 1026
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9956580122534215e-05,
+      "loss": 0.8477,
+      "step": 1027
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9956495565560826e-05,
+      "loss": 0.0005,
+      "step": 1028
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9956410926452878e-05,
+      "loss": 0.0004,
+      "step": 1029
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9956326205210836e-05,
+      "loss": 0.0001,
+      "step": 1030
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9956241401835166e-05,
+      "loss": 0.0013,
+      "step": 1031
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.995615651632633e-05,
+      "loss": 0.0004,
+      "step": 1032
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.995607154868481e-05,
+      "loss": 0.6406,
+      "step": 1033
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9955986498911053e-05,
+      "loss": 0.0005,
+      "step": 1034
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9955901367005534e-05,
+      "loss": 0.3555,
+      "step": 1035
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.995581615296872e-05,
+      "loss": 0.7852,
+      "step": 1036
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9955730856801077e-05,
+      "loss": 0.0001,
+      "step": 1037
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9955645478503082e-05,
+      "loss": 0.8516,
+      "step": 1038
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9955560018075194e-05,
+      "loss": 0.0001,
+      "step": 1039
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9955474475517887e-05,
+      "loss": 0.6719,
+      "step": 1040
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.995538885083163e-05,
+      "loss": 0.0386,
+      "step": 1041
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9955303144016895e-05,
+      "loss": 0.2021,
+      "step": 1042
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9955217355074145e-05,
+      "loss": 0.1309,
+      "step": 1043
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9955131484003857e-05,
+      "loss": 0.0001,
+      "step": 1044
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.995504553080651e-05,
+      "loss": 0.2295,
+      "step": 1045
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.995495949548256e-05,
+      "loss": 0.3203,
+      "step": 1046
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9954873378032488e-05,
+      "loss": 0.4277,
+      "step": 1047
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.995478717845677e-05,
+      "loss": 1.1406,
+      "step": 1048
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9954700896755878e-05,
+      "loss": 0.1357,
+      "step": 1049
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.995461453293028e-05,
+      "loss": 0.0065,
+      "step": 1050
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.9954528086980454e-05,
+      "loss": 1.0859,
+      "step": 1051
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.995444155890688e-05,
+      "loss": 0.6172,
+      "step": 1052
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.995435494871002e-05,
+      "loss": 1.0703,
+      "step": 1053
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.995426825639036e-05,
+      "loss": 0.0366,
+      "step": 1054
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9954181481948374e-05,
+      "loss": 1.2969,
+      "step": 1055
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.995409462538454e-05,
+      "loss": 0.3965,
+      "step": 1056
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9954007686699335e-05,
+      "loss": 0.5234,
+      "step": 1057
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.995392066589323e-05,
+      "loss": 1.8125,
+      "step": 1058
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.995383356296671e-05,
+      "loss": 0.7109,
+      "step": 1059
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9953746377920253e-05,
+      "loss": 0.377,
+      "step": 1060
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.995365911075433e-05,
+      "loss": 0.6094,
+      "step": 1061
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.995357176146943e-05,
+      "loss": 1.1719,
+      "step": 1062
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9953484330066026e-05,
+      "loss": 1.6875,
+      "step": 1063
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9953396816544602e-05,
+      "loss": 0.4004,
+      "step": 1064
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.995330922090564e-05,
+      "loss": 1.0938,
+      "step": 1065
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9953221543149615e-05,
+      "loss": 1.9688,
+      "step": 1066
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.995313378327701e-05,
+      "loss": 0.5703,
+      "step": 1067
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9953045941288313e-05,
+      "loss": 0.0786,
+      "step": 1068
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9952958017184002e-05,
+      "loss": 0.3262,
+      "step": 1069
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9952870010964557e-05,
+      "loss": 0.0006,
+      "step": 1070
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9952781922630462e-05,
+      "loss": 1.1328,
+      "step": 1071
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9952693752182208e-05,
+      "loss": 0.0005,
+      "step": 1072
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9952605499620275e-05,
+      "loss": 1.5078,
+      "step": 1073
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.995251716494514e-05,
+      "loss": 1.8828,
+      "step": 1074
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9952428748157302e-05,
+      "loss": 0.0007,
+      "step": 1075
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9952340249257236e-05,
+      "loss": 0.0004,
+      "step": 1076
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.995225166824543e-05,
+      "loss": 0.5703,
+      "step": 1077
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.995216300512238e-05,
+      "loss": 0.4141,
+      "step": 1078
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.995207425988856e-05,
+      "loss": 0.0003,
+      "step": 1079
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.995198543254446e-05,
+      "loss": 0.2158,
+      "step": 1080
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.995189652309057e-05,
+      "loss": 0.001,
+      "step": 1081
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.995180753152738e-05,
+      "loss": 0.0006,
+      "step": 1082
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9951718457855375e-05,
+      "loss": 0.0771,
+      "step": 1083
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9951629302075045e-05,
+      "loss": 0.0002,
+      "step": 1084
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9951540064186885e-05,
+      "loss": 0.0008,
+      "step": 1085
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9951450744191376e-05,
+      "loss": 0.0089,
+      "step": 1086
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9951361342089017e-05,
+      "loss": 0.0004,
+      "step": 1087
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9951271857880295e-05,
+      "loss": 0.0003,
+      "step": 1088
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.99511822915657e-05,
+      "loss": 0.0002,
+      "step": 1089
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9951092643145727e-05,
+      "loss": 0.0002,
+      "step": 1090
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9951002912620864e-05,
+      "loss": 0.0005,
+      "step": 1091
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.995091309999161e-05,
+      "loss": 0.0006,
+      "step": 1092
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9950823205258452e-05,
+      "loss": 0.0002,
+      "step": 1093
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.995073322842189e-05,
+      "loss": 0.0002,
+      "step": 1094
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.995064316948241e-05,
+      "loss": 0.2324,
+      "step": 1095
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9950553028440516e-05,
+      "loss": 0.2695,
+      "step": 1096
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9950462805296693e-05,
+      "loss": 0.373,
+      "step": 1097
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9950372500051444e-05,
+      "loss": 0.252,
+      "step": 1098
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9950282112705263e-05,
+      "loss": 0.332,
+      "step": 1099
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9950191643258645e-05,
+      "loss": 0.2266,
+      "step": 1100
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.995010109171209e-05,
+      "loss": 1.6172,
+      "step": 1101
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.995001045806609e-05,
+      "loss": 0.5625,
+      "step": 1102
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9949919742321143e-05,
+      "loss": 1.7656,
+      "step": 1103
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9949828944477758e-05,
+      "loss": 0.9414,
+      "step": 1104
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.994973806453642e-05,
+      "loss": 0.127,
+      "step": 1105
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.994964710249763e-05,
+      "loss": 0.8164,
+      "step": 1106
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9949556058361897e-05,
+      "loss": 1.1328,
+      "step": 1107
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9949464932129716e-05,
+      "loss": 0.6641,
+      "step": 1108
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.994937372380158e-05,
+      "loss": 1.4219,
+      "step": 1109
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9949282433378006e-05,
+      "loss": 0.5078,
+      "step": 1110
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9949191060859476e-05,
+      "loss": 0.0403,
+      "step": 1111
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9949099606246503e-05,
+      "loss": 0.4258,
+      "step": 1112
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9949008069539588e-05,
+      "loss": 1.5391,
+      "step": 1113
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.994891645073924e-05,
+      "loss": 0.6914,
+      "step": 1114
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.994882474984595e-05,
+      "loss": 0.0815,
+      "step": 1115
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9948732966860226e-05,
+      "loss": 2.6094,
+      "step": 1116
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9948641101782574e-05,
+      "loss": 0.0008,
+      "step": 1117
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.99485491546135e-05,
+      "loss": 0.7578,
+      "step": 1118
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.9948457125353502e-05,
+      "loss": 0.2617,
+      "step": 1119
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9948365014003095e-05,
+      "loss": 1.4219,
+      "step": 1120
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.994827282056278e-05,
+      "loss": 0.3223,
+      "step": 1121
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9948180545033058e-05,
+      "loss": 0.2471,
+      "step": 1122
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9948088187414445e-05,
+      "loss": 0.002,
+      "step": 1123
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9947995747707443e-05,
+      "loss": 0.0728,
+      "step": 1124
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9947903225912557e-05,
+      "loss": 0.0231,
+      "step": 1125
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9947810622030302e-05,
+      "loss": 0.0008,
+      "step": 1126
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9947717936061183e-05,
+      "loss": 0.0092,
+      "step": 1127
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.994762516800571e-05,
+      "loss": 0.3359,
+      "step": 1128
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.994753231786439e-05,
+      "loss": 0.4961,
+      "step": 1129
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9947439385637733e-05,
+      "loss": 0.5625,
+      "step": 1130
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9947346371326254e-05,
+      "loss": 1.0938,
+      "step": 1131
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.994725327493046e-05,
+      "loss": 0.0767,
+      "step": 1132
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9947160096450863e-05,
+      "loss": 1.6562,
+      "step": 1133
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9947066835887973e-05,
+      "loss": 0.5977,
+      "step": 1134
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9946973493242304e-05,
+      "loss": 0.0009,
+      "step": 1135
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9946880068514372e-05,
+      "loss": 0.2793,
+      "step": 1136
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9946786561704682e-05,
+      "loss": 0.0005,
+      "step": 1137
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9946692972813752e-05,
+      "loss": 0.0106,
+      "step": 1138
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9946599301842098e-05,
+      "loss": 0.2695,
+      "step": 1139
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.994650554879023e-05,
+      "loss": 0.9414,
+      "step": 1140
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9946411713658667e-05,
+      "loss": 0.001,
+      "step": 1141
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9946317796447924e-05,
+      "loss": 0.0012,
+      "step": 1142
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9946223797158513e-05,
+      "loss": 0.2402,
+      "step": 1143
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9946129715790952e-05,
+      "loss": 0.2832,
+      "step": 1144
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9946035552345757e-05,
+      "loss": 0.2598,
+      "step": 1145
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9945941306823448e-05,
+      "loss": 0.25,
+      "step": 1146
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9945846979224544e-05,
+      "loss": 0.0002,
+      "step": 1147
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.994575256954955e-05,
+      "loss": 0.1553,
+      "step": 1148
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9945658077799e-05,
+      "loss": 0.249,
+      "step": 1149
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.994556350397341e-05,
+      "loss": 0.0092,
+      "step": 1150
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9945468848073292e-05,
+      "loss": 0.5977,
+      "step": 1151
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9945374110099174e-05,
+      "loss": 1.1406,
+      "step": 1152
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9945279290051568e-05,
+      "loss": 0.2793,
+      "step": 1153
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9945184387931006e-05,
+      "loss": 0.5703,
+      "step": 1154
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9945089403737992e-05,
+      "loss": 0.6484,
+      "step": 1155
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9944994337473067e-05,
+      "loss": 0.0276,
+      "step": 1156
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.994489918913674e-05,
+      "loss": 1.0312,
+      "step": 1157
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9944803958729535e-05,
+      "loss": 0.707,
+      "step": 1158
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9944708646251983e-05,
+      "loss": 0.1406,
+      "step": 1159
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.99446132517046e-05,
+      "loss": 0.8906,
+      "step": 1160
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.994451777508791e-05,
+      "loss": 0.0737,
+      "step": 1161
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.994442221640244e-05,
+      "loss": 1.3672,
+      "step": 1162
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9944326575648716e-05,
+      "loss": 2.4062,
+      "step": 1163
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9944230852827258e-05,
+      "loss": 0.832,
+      "step": 1164
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9944135047938595e-05,
+      "loss": 0.0002,
+      "step": 1165
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9944039160983257e-05,
+      "loss": 0.0002,
+      "step": 1166
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.994394319196176e-05,
+      "loss": 0.625,
+      "step": 1167
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9943847140874642e-05,
+      "loss": 0.0952,
+      "step": 1168
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9943751007722425e-05,
+      "loss": 0.0023,
+      "step": 1169
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.994365479250564e-05,
+      "loss": 1.7188,
+      "step": 1170
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9943558495224807e-05,
+      "loss": 0.0157,
+      "step": 1171
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9943462115880465e-05,
+      "loss": 0.0089,
+      "step": 1172
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9943365654473142e-05,
+      "loss": 0.0004,
+      "step": 1173
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.994326911100336e-05,
+      "loss": 0.1748,
+      "step": 1174
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9943172485471653e-05,
+      "loss": 0.2461,
+      "step": 1175
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9943075777878558e-05,
+      "loss": 0.0923,
+      "step": 1176
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.99429789882246e-05,
+      "loss": 0.6016,
+      "step": 1177
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9942882116510312e-05,
+      "loss": 0.0007,
+      "step": 1178
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.994278516273622e-05,
+      "loss": 0.003,
+      "step": 1179
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9942688126902867e-05,
+      "loss": 0.8242,
+      "step": 1180
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.994259100901078e-05,
+      "loss": 0.1797,
+      "step": 1181
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9942493809060488e-05,
+      "loss": 0.2061,
+      "step": 1182
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.994239652705254e-05,
+      "loss": 0.457,
+      "step": 1183
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.994229916298745e-05,
+      "loss": 0.0002,
+      "step": 1184
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9942201716865768e-05,
+      "loss": 0.4609,
+      "step": 1185
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9942104188688024e-05,
+      "loss": 0.0006,
+      "step": 1186
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.9942006578454753e-05,
+      "loss": 0.0011,
+      "step": 1187
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.994190888616649e-05,
+      "loss": 0.1289,
+      "step": 1188
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9941811111823774e-05,
+      "loss": 0.1699,
+      "step": 1189
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.994171325542714e-05,
+      "loss": 0.2041,
+      "step": 1190
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9941615316977126e-05,
+      "loss": 0.1006,
+      "step": 1191
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.994151729647427e-05,
+      "loss": 0.0091,
+      "step": 1192
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.994141919391911e-05,
+      "loss": 0.0092,
+      "step": 1193
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.994132100931219e-05,
+      "loss": 0.0625,
+      "step": 1194
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9941222742654044e-05,
+      "loss": 0.5391,
+      "step": 1195
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9941124393945204e-05,
+      "loss": 0.543,
+      "step": 1196
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.994102596318622e-05,
+      "loss": 0.0006,
+      "step": 1197
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9940927450377637e-05,
+      "loss": 0.1396,
+      "step": 1198
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9940828855519987e-05,
+      "loss": 0.2432,
+      "step": 1199
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9940730178613812e-05,
+      "loss": 0.2773,
+      "step": 1200
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.994063141965966e-05,
+      "loss": 2.0625,
+      "step": 1201
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9940532578658063e-05,
+      "loss": 0.9609,
+      "step": 1202
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9940433655609572e-05,
+      "loss": 0.1475,
+      "step": 1203
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.994033465051473e-05,
+      "loss": 0.5508,
+      "step": 1204
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9940235563374082e-05,
+      "loss": 0.0688,
+      "step": 1205
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9940136394188165e-05,
+      "loss": 0.957,
+      "step": 1206
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9940037142957524e-05,
+      "loss": 1.5156,
+      "step": 1207
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9939937809682713e-05,
+      "loss": 0.6797,
+      "step": 1208
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9939838394364274e-05,
+      "loss": 0.4238,
+      "step": 1209
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9939738897002746e-05,
+      "loss": 0.4414,
+      "step": 1210
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9939639317598684e-05,
+      "loss": 0.7539,
+      "step": 1211
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9939539656152633e-05,
+      "loss": 0.832,
+      "step": 1212
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9939439912665136e-05,
+      "loss": 0.1807,
+      "step": 1213
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.993934008713674e-05,
+      "loss": 0.5078,
+      "step": 1214
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9939240179568002e-05,
+      "loss": 0.1465,
+      "step": 1215
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9939140189959467e-05,
+      "loss": 0.8477,
+      "step": 1216
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9939040118311677e-05,
+      "loss": 0.3164,
+      "step": 1217
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.993893996462519e-05,
+      "loss": 2.75,
+      "step": 1218
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9938839728900547e-05,
+      "loss": 0.2988,
+      "step": 1219
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9938739411138312e-05,
+      "loss": 0.6445,
+      "step": 1220
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.993863901133903e-05,
+      "loss": 0.4414,
+      "step": 1221
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9938538529503245e-05,
+      "loss": 0.2275,
+      "step": 1222
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9938437965631515e-05,
+      "loss": 0.0043,
+      "step": 1223
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9938337319724392e-05,
+      "loss": 0.019,
+      "step": 1224
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9938236591782434e-05,
+      "loss": 0.0933,
+      "step": 1225
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.993813578180618e-05,
+      "loss": 0.0991,
+      "step": 1226
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.99380348897962e-05,
+      "loss": 0.002,
+      "step": 1227
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9937933915753032e-05,
+      "loss": 0.1543,
+      "step": 1228
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9937832859677246e-05,
+      "loss": 0.0005,
+      "step": 1229
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9937731721569388e-05,
+      "loss": 0.0015,
+      "step": 1230
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9937630501430015e-05,
+      "loss": 0.166,
+      "step": 1231
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.993752919925968e-05,
+      "loss": 0.1465,
+      "step": 1232
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9937427815058944e-05,
+      "loss": 0.0014,
+      "step": 1233
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.993732634882836e-05,
+      "loss": 0.0098,
+      "step": 1234
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9937224800568492e-05,
+      "loss": 0.2676,
+      "step": 1235
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.993712317027989e-05,
+      "loss": 0.0007,
+      "step": 1236
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9937021457963114e-05,
+      "loss": 0.0015,
+      "step": 1237
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9936919663618727e-05,
+      "loss": 0.0747,
+      "step": 1238
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9936817787247282e-05,
+      "loss": 0.0093,
+      "step": 1239
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9936715828849344e-05,
+      "loss": 0.0093,
+      "step": 1240
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9936613788425467e-05,
+      "loss": 0.0115,
+      "step": 1241
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9936511665976216e-05,
+      "loss": 0.1104,
+      "step": 1242
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.993640946150215e-05,
+      "loss": 0.8086,
+      "step": 1243
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9936307175003826e-05,
+      "loss": 0.1196,
+      "step": 1244
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9936204806481815e-05,
+      "loss": 0.0048,
+      "step": 1245
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9936102355936673e-05,
+      "loss": 0.1235,
+      "step": 1246
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9935999823368964e-05,
+      "loss": 0.3242,
+      "step": 1247
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9935897208779253e-05,
+      "loss": 0.1602,
+      "step": 1248
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9935794512168096e-05,
+      "loss": 0.2539,
+      "step": 1249
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.993569173353607e-05,
+      "loss": 0.2295,
+      "step": 1250
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9935588872883728e-05,
+      "loss": 1.3984,
+      "step": 1251
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9935485930211637e-05,
+      "loss": 0.3008,
+      "step": 1252
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9935382905520364e-05,
+      "loss": 0.1572,
+      "step": 1253
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.9935279798810474e-05,
+      "loss": 0.6094,
+      "step": 1254
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.993517661008254e-05,
+      "loss": 0.9688,
+      "step": 1255
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9935073339337116e-05,
+      "loss": 0.5625,
+      "step": 1256
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9934969986574784e-05,
+      "loss": 1.7188,
+      "step": 1257
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9934866551796094e-05,
+      "loss": 0.4668,
+      "step": 1258
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.993476303500163e-05,
+      "loss": 0.5859,
+      "step": 1259
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9934659436191953e-05,
+      "loss": 0.707,
+      "step": 1260
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9934555755367628e-05,
+      "loss": 0.0193,
+      "step": 1261
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9934451992529232e-05,
+      "loss": 1.2422,
+      "step": 1262
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9934348147677335e-05,
+      "loss": 1.3281,
+      "step": 1263
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.99342442208125e-05,
+      "loss": 0.041,
+      "step": 1264
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9934140211935302e-05,
+      "loss": 0.668,
+      "step": 1265
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9934036121046314e-05,
+      "loss": 0.543,
+      "step": 1266
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.993393194814611e-05,
+      "loss": 0.0026,
+      "step": 1267
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9933827693235252e-05,
+      "loss": 0.6875,
+      "step": 1268
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.993372335631432e-05,
+      "loss": 0.4453,
+      "step": 1269
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.993361893738388e-05,
+      "loss": 0.8047,
+      "step": 1270
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.993351443644452e-05,
+      "loss": 0.0491,
+      "step": 1271
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.99334098534968e-05,
+      "loss": 0.5703,
+      "step": 1272
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.99333051885413e-05,
+      "loss": 0.9023,
+      "step": 1273
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.993320044157859e-05,
+      "loss": 0.8398,
+      "step": 1274
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9933095612609253e-05,
+      "loss": 0.0005,
+      "step": 1275
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9932990701633862e-05,
+      "loss": 0.9375,
+      "step": 1276
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.993288570865299e-05,
+      "loss": 0.0058,
+      "step": 1277
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9932780633667212e-05,
+      "loss": 0.21,
+      "step": 1278
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9932675476677112e-05,
+      "loss": 0.0035,
+      "step": 1279
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9932570237683262e-05,
+      "loss": 0.0009,
+      "step": 1280
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9932464916686245e-05,
+      "loss": 0.0052,
+      "step": 1281
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9932359513686634e-05,
+      "loss": 0.2344,
+      "step": 1282
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.993225402868501e-05,
+      "loss": 0.1924,
+      "step": 1283
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.993214846168195e-05,
+      "loss": 0.0008,
+      "step": 1284
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.993204281267804e-05,
+      "loss": 0.3809,
+      "step": 1285
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9931937081673853e-05,
+      "loss": 0.1689,
+      "step": 1286
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9931831268669972e-05,
+      "loss": 0.0056,
+      "step": 1287
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.993172537366698e-05,
+      "loss": 0.6172,
+      "step": 1288
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.993161939666546e-05,
+      "loss": 0.0854,
+      "step": 1289
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.993151333766599e-05,
+      "loss": 2.2188,
+      "step": 1290
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9931407196669153e-05,
+      "loss": 0.0018,
+      "step": 1291
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.993130097367553e-05,
+      "loss": 0.104,
+      "step": 1292
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9931194668685714e-05,
+      "loss": 0.3359,
+      "step": 1293
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9931088281700274e-05,
+      "loss": 0.1094,
+      "step": 1294
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9930981812719805e-05,
+      "loss": 0.0049,
+      "step": 1295
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.993087526174489e-05,
+      "loss": 0.004,
+      "step": 1296
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.993076862877611e-05,
+      "loss": 0.3203,
+      "step": 1297
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.993066191381406e-05,
+      "loss": 0.416,
+      "step": 1298
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9930555116859316e-05,
+      "loss": 0.3008,
+      "step": 1299
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.993044823791246e-05,
+      "loss": 0.3125,
+      "step": 1300
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9930341276974094e-05,
+      "loss": 1.4609,
+      "step": 1301
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9930234234044802e-05,
+      "loss": 1.6016,
+      "step": 1302
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9930127109125163e-05,
+      "loss": 0.1367,
+      "step": 1303
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.993001990221577e-05,
+      "loss": 0.7266,
+      "step": 1304
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9929912613317217e-05,
+      "loss": 0.4824,
+      "step": 1305
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9929805242430086e-05,
+      "loss": 0.4531,
+      "step": 1306
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9929697789554966e-05,
+      "loss": 1.3594,
+      "step": 1307
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.992959025469245e-05,
+      "loss": 1.5859,
+      "step": 1308
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.992948263784313e-05,
+      "loss": 0.6836,
+      "step": 1309
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9929374939007597e-05,
+      "loss": 0.5195,
+      "step": 1310
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.992926715818644e-05,
+      "loss": 0.793,
+      "step": 1311
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.992915929538025e-05,
+      "loss": 1.6484,
+      "step": 1312
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.992905135058962e-05,
+      "loss": 0.2891,
+      "step": 1313
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.992894332381515e-05,
+      "loss": 0.4746,
+      "step": 1314
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9928835215057423e-05,
+      "loss": 0.3242,
+      "step": 1315
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9928727024317036e-05,
+      "loss": 0.0157,
+      "step": 1316
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9928618751594587e-05,
+      "loss": 0.6641,
+      "step": 1317
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9928510396890668e-05,
+      "loss": 0.0728,
+      "step": 1318
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.992840196020587e-05,
+      "loss": 1.0,
+      "step": 1319
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9928293441540796e-05,
+      "loss": 0.4551,
+      "step": 1320
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.992818484089603e-05,
+      "loss": 0.0422,
+      "step": 1321
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.9928076158272185e-05,
+      "loss": 0.0018,
+      "step": 1322
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9927967393669847e-05,
+      "loss": 0.3066,
+      "step": 1323
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9927858547089616e-05,
+      "loss": 0.2363,
+      "step": 1324
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.992774961853209e-05,
+      "loss": 0.0042,
+      "step": 1325
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9927640607997865e-05,
+      "loss": 0.0033,
+      "step": 1326
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9927531515487537e-05,
+      "loss": 0.1123,
+      "step": 1327
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9927422341001713e-05,
+      "loss": 0.0034,
+      "step": 1328
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9927313084540993e-05,
+      "loss": 0.9883,
+      "step": 1329
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9927203746105968e-05,
+      "loss": 0.1367,
+      "step": 1330
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9927094325697243e-05,
+      "loss": 0.0009,
+      "step": 1331
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.992698482331542e-05,
+      "loss": 0.0002,
+      "step": 1332
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.99268752389611e-05,
+      "loss": 0.0029,
+      "step": 1333
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9926765572634883e-05,
+      "loss": 0.0005,
+      "step": 1334
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9926655824337373e-05,
+      "loss": 0.0007,
+      "step": 1335
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.992654599406917e-05,
+      "loss": 0.019,
+      "step": 1336
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9926436081830884e-05,
+      "loss": 0.0039,
+      "step": 1337
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9926326087623112e-05,
+      "loss": 0.025,
+      "step": 1338
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.992621601144646e-05,
+      "loss": 0.6367,
+      "step": 1339
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.992610585330153e-05,
+      "loss": 0.0008,
+      "step": 1340
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9925995613188934e-05,
+      "loss": 0.957,
+      "step": 1341
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.992588529110927e-05,
+      "loss": 0.0019,
+      "step": 1342
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9925774887063145e-05,
+      "loss": 0.1338,
+      "step": 1343
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.992566440105117e-05,
+      "loss": 0.0004,
+      "step": 1344
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9925553833073944e-05,
+      "loss": 0.0002,
+      "step": 1345
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.992544318313208e-05,
+      "loss": 0.3398,
+      "step": 1346
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9925332451226188e-05,
+      "loss": 0.1963,
+      "step": 1347
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.992522163735687e-05,
+      "loss": 0.4141,
+      "step": 1348
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.992511074152474e-05,
+      "loss": 0.0879,
+      "step": 1349
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9924999763730397e-05,
+      "loss": 0.1279,
+      "step": 1350
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9924888703974465e-05,
+      "loss": 0.9375,
+      "step": 1351
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9924777562257545e-05,
+      "loss": 0.6172,
+      "step": 1352
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9924666338580247e-05,
+      "loss": 1.6094,
+      "step": 1353
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9924555032943183e-05,
+      "loss": 0.2812,
+      "step": 1354
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9924443645346964e-05,
+      "loss": 0.3887,
+      "step": 1355
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9924332175792206e-05,
+      "loss": 0.7109,
+      "step": 1356
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9924220624279513e-05,
+      "loss": 0.8711,
+      "step": 1357
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9924108990809508e-05,
+      "loss": 0.6133,
+      "step": 1358
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9923997275382793e-05,
+      "loss": 0.3262,
+      "step": 1359
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.992388547799999e-05,
+      "loss": 0.1855,
+      "step": 1360
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9923773598661707e-05,
+      "loss": 1.9062,
+      "step": 1361
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9923661637368562e-05,
+      "loss": 0.7734,
+      "step": 1362
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.992354959412117e-05,
+      "loss": 0.6172,
+      "step": 1363
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9923437468920147e-05,
+      "loss": 0.0251,
+      "step": 1364
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9923325261766104e-05,
+      "loss": 0.0503,
+      "step": 1365
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9923212972659658e-05,
+      "loss": 0.3672,
+      "step": 1366
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9923100601601432e-05,
+      "loss": 0.0102,
+      "step": 1367
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9922988148592037e-05,
+      "loss": 0.0168,
+      "step": 1368
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.992287561363209e-05,
+      "loss": 2.1094,
+      "step": 1369
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9922762996722216e-05,
+      "loss": 0.0002,
+      "step": 1370
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.992265029786303e-05,
+      "loss": 0.2715,
+      "step": 1371
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9922537517055143e-05,
+      "loss": 0.0815,
+      "step": 1372
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9922424654299188e-05,
+      "loss": 0.9531,
+      "step": 1373
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.992231170959577e-05,
+      "loss": 0.0038,
+      "step": 1374
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9922198682945526e-05,
+      "loss": 0.0018,
+      "step": 1375
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9922085574349062e-05,
+      "loss": 0.0248,
+      "step": 1376
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9921972383807006e-05,
+      "loss": 0.0001,
+      "step": 1377
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.992185911131998e-05,
+      "loss": 0.5391,
+      "step": 1378
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9921745756888606e-05,
+      "loss": 0.6562,
+      "step": 1379
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.99216323205135e-05,
+      "loss": 0.0232,
+      "step": 1380
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9921518802195293e-05,
+      "loss": 0.0002,
+      "step": 1381
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9921405201934604e-05,
+      "loss": 0.0002,
+      "step": 1382
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.992129151973206e-05,
+      "loss": 0.0,
+      "step": 1383
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9921177755588285e-05,
+      "loss": 0.4531,
+      "step": 1384
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.99210639095039e-05,
+      "loss": 0.0007,
+      "step": 1385
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.992094998147953e-05,
+      "loss": 0.0005,
+      "step": 1386
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.992083597151581e-05,
+      "loss": 0.4023,
+      "step": 1387
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9920721879613352e-05,
+      "loss": 0.0,
+      "step": 1388
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9920607705772794e-05,
+      "loss": 1.5234,
+      "step": 1389
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.9920493449994758e-05,
+      "loss": 0.0002,
+      "step": 1390
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9920379112279874e-05,
+      "loss": 0.0001,
+      "step": 1391
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9920264692628765e-05,
+      "loss": 0.0771,
+      "step": 1392
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.992015019104206e-05,
+      "loss": 0.1992,
+      "step": 1393
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9920035607520396e-05,
+      "loss": 0.1611,
+      "step": 1394
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9919920942064392e-05,
+      "loss": 0.0001,
+      "step": 1395
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9919806194674688e-05,
+      "loss": 0.0,
+      "step": 1396
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9919691365351903e-05,
+      "loss": 0.126,
+      "step": 1397
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9919576454096672e-05,
+      "loss": 0.1846,
+      "step": 1398
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9919461460909635e-05,
+      "loss": 0.1387,
+      "step": 1399
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9919346385791407e-05,
+      "loss": 0.457,
+      "step": 1400
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.991923122874263e-05,
+      "loss": 0.5352,
+      "step": 1401
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9919115989763935e-05,
+      "loss": 0.2393,
+      "step": 1402
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9919000668855958e-05,
+      "loss": 1.9219,
+      "step": 1403
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9918885266019328e-05,
+      "loss": 0.6797,
+      "step": 1404
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9918769781254677e-05,
+      "loss": 2.75,
+      "step": 1405
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9918654214562644e-05,
+      "loss": 0.7891,
+      "step": 1406
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.991853856594386e-05,
+      "loss": 0.6406,
+      "step": 1407
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9918422835398965e-05,
+      "loss": 0.2617,
+      "step": 1408
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9918307022928583e-05,
+      "loss": 0.3906,
+      "step": 1409
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9918191128533363e-05,
+      "loss": 0.5859,
+      "step": 1410
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.991807515221394e-05,
+      "loss": 0.7148,
+      "step": 1411
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9917959093970945e-05,
+      "loss": 0.4258,
+      "step": 1412
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9917842953805016e-05,
+      "loss": 0.3281,
+      "step": 1413
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9917726731716795e-05,
+      "loss": 0.6445,
+      "step": 1414
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9917610427706915e-05,
+      "loss": 0.6055,
+      "step": 1415
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.991749404177602e-05,
+      "loss": 0.1152,
+      "step": 1416
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9917377573924747e-05,
+      "loss": 1.2969,
+      "step": 1417
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9917261024153736e-05,
+      "loss": 1.6172,
+      "step": 1418
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.991714439246362e-05,
+      "loss": 1.9688,
+      "step": 1419
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9917027678855053e-05,
+      "loss": 0.5391,
+      "step": 1420
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9916910883328664e-05,
+      "loss": 0.0452,
+      "step": 1421
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9916794005885102e-05,
+      "loss": 0.6406,
+      "step": 1422
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9916677046525004e-05,
+      "loss": 1.0625,
+      "step": 1423
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9916560005249017e-05,
+      "loss": 0.2715,
+      "step": 1424
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.991644288205778e-05,
+      "loss": 0.0566,
+      "step": 1425
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9916325676951932e-05,
+      "loss": 0.1001,
+      "step": 1426
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9916208389932127e-05,
+      "loss": 0.0063,
+      "step": 1427
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9916091020999002e-05,
+      "loss": 0.0006,
+      "step": 1428
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9915973570153207e-05,
+      "loss": 0.625,
+      "step": 1429
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.991585603739538e-05,
+      "loss": 0.8281,
+      "step": 1430
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9915738422726168e-05,
+      "loss": 0.0503,
+      "step": 1431
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9915620726146225e-05,
+      "loss": 1.7812,
+      "step": 1432
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9915502947656186e-05,
+      "loss": 0.4355,
+      "step": 1433
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9915385087256707e-05,
+      "loss": 0.0004,
+      "step": 1434
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9915267144948426e-05,
+      "loss": 0.1836,
+      "step": 1435
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9915149120732e-05,
+      "loss": 0.0005,
+      "step": 1436
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9915031014608068e-05,
+      "loss": 0.5352,
+      "step": 1437
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9914912826577288e-05,
+      "loss": 0.0009,
+      "step": 1438
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9914794556640302e-05,
+      "loss": 0.0001,
+      "step": 1439
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.991467620479776e-05,
+      "loss": 0.0615,
+      "step": 1440
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.991455777105032e-05,
+      "loss": 0.0214,
+      "step": 1441
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9914439255398624e-05,
+      "loss": 0.0043,
+      "step": 1442
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9914320657843325e-05,
+      "loss": 0.0018,
+      "step": 1443
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9914201978385076e-05,
+      "loss": 0.0017,
+      "step": 1444
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9914083217024524e-05,
+      "loss": 0.1367,
+      "step": 1445
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9913964373762328e-05,
+      "loss": 0.2695,
+      "step": 1446
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9913845448599133e-05,
+      "loss": 0.1973,
+      "step": 1447
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9913726441535598e-05,
+      "loss": 0.1709,
+      "step": 1448
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9913607352572376e-05,
+      "loss": 0.3301,
+      "step": 1449
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.991348818171012e-05,
+      "loss": 0.2988,
+      "step": 1450
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.991336892894948e-05,
+      "loss": 0.5508,
+      "step": 1451
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.991324959429112e-05,
+      "loss": 0.6367,
+      "step": 1452
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9913130177735687e-05,
+      "loss": 1.0391,
+      "step": 1453
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9913010679283842e-05,
+      "loss": 1.2969,
+      "step": 1454
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9912891098936238e-05,
+      "loss": 3.4531,
+      "step": 1455
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9912771436693528e-05,
+      "loss": 0.5078,
+      "step": 1456
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9912651692556383e-05,
+      "loss": 0.1748,
+      "step": 1457
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.9912531866525446e-05,
+      "loss": 0.1621,
+      "step": 1458
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9912411958601385e-05,
+      "loss": 0.6016,
+      "step": 1459
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9912291968784854e-05,
+      "loss": 0.3301,
+      "step": 1460
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.991217189707651e-05,
+      "loss": 0.7617,
+      "step": 1461
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9912051743477014e-05,
+      "loss": 0.2373,
+      "step": 1462
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9911931507987026e-05,
+      "loss": 1.3281,
+      "step": 1463
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.991181119060721e-05,
+      "loss": 0.0757,
+      "step": 1464
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9911690791338223e-05,
+      "loss": 0.8555,
+      "step": 1465
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9911570310180722e-05,
+      "loss": 0.0525,
+      "step": 1466
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9911449747135376e-05,
+      "loss": 0.0986,
+      "step": 1467
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9911329102202845e-05,
+      "loss": 0.1436,
+      "step": 1468
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9911208375383787e-05,
+      "loss": 1.625,
+      "step": 1469
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9911087566678872e-05,
+      "loss": 0.1826,
+      "step": 1470
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.991096667608876e-05,
+      "loss": 0.064,
+      "step": 1471
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9910845703614116e-05,
+      "loss": 0.2695,
+      "step": 1472
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9910724649255606e-05,
+      "loss": 0.0588,
+      "step": 1473
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9910603513013886e-05,
+      "loss": 0.3867,
+      "step": 1474
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.991048229488963e-05,
+      "loss": 0.1738,
+      "step": 1475
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.99103609948835e-05,
+      "loss": 0.002,
+      "step": 1476
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.991023961299617e-05,
+      "loss": 0.2178,
+      "step": 1477
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9910118149228292e-05,
+      "loss": 0.2754,
+      "step": 1478
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9909996603580546e-05,
+      "loss": 0.6367,
+      "step": 1479
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9909874976053587e-05,
+      "loss": 0.0004,
+      "step": 1480
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9909753266648097e-05,
+      "loss": 0.2129,
+      "step": 1481
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9909631475364736e-05,
+      "loss": 0.05,
+      "step": 1482
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9909509602204173e-05,
+      "loss": 0.001,
+      "step": 1483
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.990938764716708e-05,
+      "loss": 0.1328,
+      "step": 1484
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9909265610254126e-05,
+      "loss": 0.0011,
+      "step": 1485
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.990914349146598e-05,
+      "loss": 0.1982,
+      "step": 1486
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9909021290803313e-05,
+      "loss": 0.0006,
+      "step": 1487
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9908899008266796e-05,
+      "loss": 0.0008,
+      "step": 1488
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.99087766438571e-05,
+      "loss": 0.0014,
+      "step": 1489
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.99086541975749e-05,
+      "loss": 0.4551,
+      "step": 1490
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9908531669420867e-05,
+      "loss": 0.2227,
+      "step": 1491
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9908409059395666e-05,
+      "loss": 0.0004,
+      "step": 1492
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9908286367499984e-05,
+      "loss": 0.1992,
+      "step": 1493
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9908163593734488e-05,
+      "loss": 0.1484,
+      "step": 1494
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.990804073809985e-05,
+      "loss": 0.124,
+      "step": 1495
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.990791780059675e-05,
+      "loss": 0.2451,
+      "step": 1496
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9907794781225855e-05,
+      "loss": 0.1826,
+      "step": 1497
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9907671679987854e-05,
+      "loss": 0.1855,
+      "step": 1498
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.990754849688341e-05,
+      "loss": 0.1592,
+      "step": 1499
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9907425231913202e-05,
+      "loss": 0.1465,
+      "step": 1500
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9907301885077913e-05,
+      "loss": 1.2812,
+      "step": 1501
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.990717845637821e-05,
+      "loss": 0.2285,
+      "step": 1502
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9907054945814788e-05,
+      "loss": 0.793,
+      "step": 1503
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.990693135338831e-05,
+      "loss": 0.1924,
+      "step": 1504
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9906807679099455e-05,
+      "loss": 0.7773,
+      "step": 1505
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9906683922948912e-05,
+      "loss": 1.6094,
+      "step": 1506
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9906560084937353e-05,
+      "loss": 0.6914,
+      "step": 1507
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9906436165065458e-05,
+      "loss": 0.4199,
+      "step": 1508
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9906312163333914e-05,
+      "loss": 0.8477,
+      "step": 1509
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9906188079743394e-05,
+      "loss": 0.4062,
+      "step": 1510
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9906063914294584e-05,
+      "loss": 0.04,
+      "step": 1511
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9905939666988164e-05,
+      "loss": 0.1133,
+      "step": 1512
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.990581533782482e-05,
+      "loss": 0.6406,
+      "step": 1513
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9905690926805233e-05,
+      "loss": 0.3438,
+      "step": 1514
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.990556643393008e-05,
+      "loss": 0.0086,
+      "step": 1515
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9905441859200053e-05,
+      "loss": 0.3926,
+      "step": 1516
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.990531720261583e-05,
+      "loss": 0.8281,
+      "step": 1517
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9905192464178105e-05,
+      "loss": 0.1582,
+      "step": 1518
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9905067643887554e-05,
+      "loss": 0.6289,
+      "step": 1519
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9904942741744862e-05,
+      "loss": 0.4004,
+      "step": 1520
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.990481775775072e-05,
+      "loss": 0.0018,
+      "step": 1521
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9904692691905812e-05,
+      "loss": 0.2412,
+      "step": 1522
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9904567544210826e-05,
+      "loss": 0.0212,
+      "step": 1523
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9904442314666448e-05,
+      "loss": 0.1611,
+      "step": 1524
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9904317003273366e-05,
+      "loss": 0.2061,
+      "step": 1525
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.9904191610032264e-05,
+      "loss": 0.001,
+      "step": 1526
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.990406613494384e-05,
+      "loss": 0.0007,
+      "step": 1527
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9903940578008778e-05,
+      "loss": 0.2373,
+      "step": 1528
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9903814939227765e-05,
+      "loss": 0.4141,
+      "step": 1529
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9903689218601497e-05,
+      "loss": 0.2412,
+      "step": 1530
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9903563416130656e-05,
+      "loss": 0.0918,
+      "step": 1531
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9903437531815938e-05,
+      "loss": 0.0036,
+      "step": 1532
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9903311565658037e-05,
+      "loss": 0.0003,
+      "step": 1533
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.990318551765764e-05,
+      "loss": 0.0967,
+      "step": 1534
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.990305938781544e-05,
+      "loss": 0.1328,
+      "step": 1535
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.990293317613213e-05,
+      "loss": 0.0085,
+      "step": 1536
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9902806882608406e-05,
+      "loss": 0.1113,
+      "step": 1537
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9902680507244958e-05,
+      "loss": 0.1553,
+      "step": 1538
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9902554050042484e-05,
+      "loss": 0.0703,
+      "step": 1539
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9902427511001673e-05,
+      "loss": 0.0009,
+      "step": 1540
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9902300890123217e-05,
+      "loss": 0.0203,
+      "step": 1541
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9902174187407824e-05,
+      "loss": 0.0195,
+      "step": 1542
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.990204740285618e-05,
+      "loss": 0.8008,
+      "step": 1543
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9901920536468985e-05,
+      "loss": 0.2266,
+      "step": 1544
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9901793588246936e-05,
+      "loss": 0.0036,
+      "step": 1545
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9901666558190723e-05,
+      "loss": 0.2812,
+      "step": 1546
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9901539446301055e-05,
+      "loss": 0.2754,
+      "step": 1547
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9901412252578622e-05,
+      "loss": 0.1904,
+      "step": 1548
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9901284977024123e-05,
+      "loss": 0.2754,
+      "step": 1549
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9901157619638262e-05,
+      "loss": 0.418,
+      "step": 1550
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9901030180421734e-05,
+      "loss": 0.332,
+      "step": 1551
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9900902659375243e-05,
+      "loss": 0.4668,
+      "step": 1552
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.990077505649948e-05,
+      "loss": 1.4688,
+      "step": 1553
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9900647371795156e-05,
+      "loss": 0.707,
+      "step": 1554
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.990051960526297e-05,
+      "loss": 0.2832,
+      "step": 1555
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.990039175690362e-05,
+      "loss": 0.5898,
+      "step": 1556
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9900263826717812e-05,
+      "loss": 0.625,
+      "step": 1557
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.990013581470624e-05,
+      "loss": 0.707,
+      "step": 1558
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.990000772086962e-05,
+      "loss": 0.0128,
+      "step": 1559
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9899879545208644e-05,
+      "loss": 0.6836,
+      "step": 1560
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9899751287724025e-05,
+      "loss": 0.6484,
+      "step": 1561
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.989962294841646e-05,
+      "loss": 0.165,
+      "step": 1562
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.989949452728666e-05,
+      "loss": 1.3828,
+      "step": 1563
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.989936602433532e-05,
+      "loss": 0.5234,
+      "step": 1564
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.989923743956316e-05,
+      "loss": 0.4316,
+      "step": 1565
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9899108772970878e-05,
+      "loss": 0.0015,
+      "step": 1566
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.989898002455918e-05,
+      "loss": 2.4844,
+      "step": 1567
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9898851194328777e-05,
+      "loss": 0.166,
+      "step": 1568
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.989872228228037e-05,
+      "loss": 0.0569,
+      "step": 1569
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9898593288414674e-05,
+      "loss": 0.0405,
+      "step": 1570
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9898464212732392e-05,
+      "loss": 0.0129,
+      "step": 1571
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.989833505523424e-05,
+      "loss": 0.0292,
+      "step": 1572
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9898205815920917e-05,
+      "loss": 0.5195,
+      "step": 1573
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9898076494793143e-05,
+      "loss": 0.6719,
+      "step": 1574
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9897947091851618e-05,
+      "loss": 0.0195,
+      "step": 1575
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9897817607097065e-05,
+      "loss": 0.418,
+      "step": 1576
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.989768804053018e-05,
+      "loss": 0.0002,
+      "step": 1577
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.989755839215169e-05,
+      "loss": 0.457,
+      "step": 1578
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9897428661962292e-05,
+      "loss": 0.3164,
+      "step": 1579
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9897298849962713e-05,
+      "loss": 0.0007,
+      "step": 1580
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9897168956153652e-05,
+      "loss": 0.1631,
+      "step": 1581
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9897038980535835e-05,
+      "loss": 0.0004,
+      "step": 1582
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.989690892310997e-05,
+      "loss": 0.1191,
+      "step": 1583
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9896778783876766e-05,
+      "loss": 0.0018,
+      "step": 1584
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9896648562836946e-05,
+      "loss": 0.1523,
+      "step": 1585
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.989651825999122e-05,
+      "loss": 0.2363,
+      "step": 1586
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9896387875340312e-05,
+      "loss": 0.0023,
+      "step": 1587
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9896257408884928e-05,
+      "loss": 0.1108,
+      "step": 1588
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9896126860625784e-05,
+      "loss": 0.0034,
+      "step": 1589
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9895996230563604e-05,
+      "loss": 0.0049,
+      "step": 1590
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9895865518699105e-05,
+      "loss": 0.0042,
+      "step": 1591
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9895734725032995e-05,
+      "loss": 1.2266,
+      "step": 1592
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9895603849566003e-05,
+      "loss": 0.0005,
+      "step": 1593
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.9895472892298844e-05,
+      "loss": 0.4277,
+      "step": 1594
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.989534185323224e-05,
+      "loss": 0.0004,
+      "step": 1595
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.98952107323669e-05,
+      "loss": 0.0005,
+      "step": 1596
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9895079529703563e-05,
+      "loss": 0.5039,
+      "step": 1597
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9894948245242932e-05,
+      "loss": 0.1196,
+      "step": 1598
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9894816878985735e-05,
+      "loss": 0.3594,
+      "step": 1599
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.989468543093269e-05,
+      "loss": 0.7344,
+      "step": 1600
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9894553901084523e-05,
+      "loss": 1.1328,
+      "step": 1601
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.989442228944196e-05,
+      "loss": 1.5781,
+      "step": 1602
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9894290596005713e-05,
+      "loss": 2.2656,
+      "step": 1603
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9894158820776516e-05,
+      "loss": 0.6211,
+      "step": 1604
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.989402696375508e-05,
+      "loss": 0.5547,
+      "step": 1605
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9893895024942147e-05,
+      "loss": 1.0547,
+      "step": 1606
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9893763004338422e-05,
+      "loss": 0.5117,
+      "step": 1607
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9893630901944643e-05,
+      "loss": 0.8281,
+      "step": 1608
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.989349871776153e-05,
+      "loss": 0.6289,
+      "step": 1609
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.989336645178981e-05,
+      "loss": 0.5859,
+      "step": 1610
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9893234104030215e-05,
+      "loss": 1.7109,
+      "step": 1611
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9893101674483462e-05,
+      "loss": 0.6992,
+      "step": 1612
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9892969163150284e-05,
+      "loss": 0.6406,
+      "step": 1613
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9892836570031406e-05,
+      "loss": 0.0118,
+      "step": 1614
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.989270389512756e-05,
+      "loss": 0.7266,
+      "step": 1615
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.989257113843947e-05,
+      "loss": 0.6875,
+      "step": 1616
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.989243829996787e-05,
+      "loss": 0.0258,
+      "step": 1617
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9892305379713485e-05,
+      "loss": 0.2002,
+      "step": 1618
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.989217237767705e-05,
+      "loss": 0.543,
+      "step": 1619
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9892039293859288e-05,
+      "loss": 0.7852,
+      "step": 1620
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9891906128260938e-05,
+      "loss": 1.875,
+      "step": 1621
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9891772880882726e-05,
+      "loss": 1.7578,
+      "step": 1622
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9891639551725387e-05,
+      "loss": 0.0234,
+      "step": 1623
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9891506140789645e-05,
+      "loss": 0.0344,
+      "step": 1624
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9891372648076245e-05,
+      "loss": 1.4922,
+      "step": 1625
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.989123907358591e-05,
+      "loss": 0.4531,
+      "step": 1626
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9891105417319377e-05,
+      "loss": 0.0006,
+      "step": 1627
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9890971679277384e-05,
+      "loss": 0.0009,
+      "step": 1628
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.989083785946066e-05,
+      "loss": 0.0004,
+      "step": 1629
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9890703957869942e-05,
+      "loss": 0.1206,
+      "step": 1630
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9890569974505964e-05,
+      "loss": 1.0938,
+      "step": 1631
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9890435909369463e-05,
+      "loss": 0.1943,
+      "step": 1632
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.989030176246118e-05,
+      "loss": 0.291,
+      "step": 1633
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.989016753378184e-05,
+      "loss": 0.001,
+      "step": 1634
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.989003322333219e-05,
+      "loss": 0.2451,
+      "step": 1635
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9889898831112964e-05,
+      "loss": 0.4395,
+      "step": 1636
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.98897643571249e-05,
+      "loss": 0.0393,
+      "step": 1637
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.988962980136874e-05,
+      "loss": 0.2236,
+      "step": 1638
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9889495163845217e-05,
+      "loss": 0.4902,
+      "step": 1639
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9889360444555077e-05,
+      "loss": 0.0001,
+      "step": 1640
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.988922564349905e-05,
+      "loss": 0.0003,
+      "step": 1641
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9889090760677885e-05,
+      "loss": 0.0023,
+      "step": 1642
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9888955796092323e-05,
+      "loss": 0.0145,
+      "step": 1643
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9888820749743098e-05,
+      "loss": 0.5391,
+      "step": 1644
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9888685621630957e-05,
+      "loss": 0.1748,
+      "step": 1645
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9888550411756645e-05,
+      "loss": 0.1963,
+      "step": 1646
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9888415120120898e-05,
+      "loss": 0.166,
+      "step": 1647
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.988827974672446e-05,
+      "loss": 0.1494,
+      "step": 1648
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9888144291568073e-05,
+      "loss": 0.1826,
+      "step": 1649
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.988800875465249e-05,
+      "loss": 0.207,
+      "step": 1650
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9887873135978446e-05,
+      "loss": 0.1934,
+      "step": 1651
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9887737435546686e-05,
+      "loss": 0.8398,
+      "step": 1652
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.988760165335796e-05,
+      "loss": 0.4375,
+      "step": 1653
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.988746578941301e-05,
+      "loss": 0.9141,
+      "step": 1654
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.988732984371259e-05,
+      "loss": 1.4453,
+      "step": 1655
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9887193816257437e-05,
+      "loss": 1.8047,
+      "step": 1656
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9887057707048304e-05,
+      "loss": 1.2812,
+      "step": 1657
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9886921516085934e-05,
+      "loss": 0.0835,
+      "step": 1658
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9886785243371074e-05,
+      "loss": 0.7109,
+      "step": 1659
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9886648888904477e-05,
+      "loss": 1.5781,
+      "step": 1660
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.9886512452686884e-05,
+      "loss": 0.832,
+      "step": 1661
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.988637593471906e-05,
+      "loss": 1.3125,
+      "step": 1662
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9886239335001736e-05,
+      "loss": 0.4961,
+      "step": 1663
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9886102653535677e-05,
+      "loss": 0.0933,
+      "step": 1664
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9885965890321626e-05,
+      "loss": 0.5781,
+      "step": 1665
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9885829045360336e-05,
+      "loss": 0.2676,
+      "step": 1666
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9885692118652554e-05,
+      "loss": 0.0008,
+      "step": 1667
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.988555511019904e-05,
+      "loss": 0.012,
+      "step": 1668
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.988541802000054e-05,
+      "loss": 0.0457,
+      "step": 1669
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9885280848057808e-05,
+      "loss": 0.0425,
+      "step": 1670
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.98851435943716e-05,
+      "loss": 0.053,
+      "step": 1671
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.988500625894267e-05,
+      "loss": 0.0008,
+      "step": 1672
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.988486884177177e-05,
+      "loss": 0.0752,
+      "step": 1673
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9884731342859655e-05,
+      "loss": 0.0006,
+      "step": 1674
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9884593762207083e-05,
+      "loss": 0.4902,
+      "step": 1675
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.98844560998148e-05,
+      "loss": 0.1001,
+      "step": 1676
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9884318355683572e-05,
+      "loss": 0.0029,
+      "step": 1677
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.988418052981415e-05,
+      "loss": 0.0981,
+      "step": 1678
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9884042622207298e-05,
+      "loss": 0.0059,
+      "step": 1679
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9883904632863765e-05,
+      "loss": 0.2344,
+      "step": 1680
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.988376656178431e-05,
+      "loss": 0.1396,
+      "step": 1681
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9883628408969695e-05,
+      "loss": 0.2676,
+      "step": 1682
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.988349017442068e-05,
+      "loss": 0.0004,
+      "step": 1683
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9883351858138015e-05,
+      "loss": 0.0022,
+      "step": 1684
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.988321346012247e-05,
+      "loss": 0.4746,
+      "step": 1685
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.98830749803748e-05,
+      "loss": 0.543,
+      "step": 1686
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9882936418895767e-05,
+      "loss": 1.1094,
+      "step": 1687
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.988279777568613e-05,
+      "loss": 0.0031,
+      "step": 1688
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.988265905074665e-05,
+      "loss": 0.0359,
+      "step": 1689
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9882520244078092e-05,
+      "loss": 0.0004,
+      "step": 1690
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9882381355681215e-05,
+      "loss": 0.1021,
+      "step": 1691
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9882242385556785e-05,
+      "loss": 0.002,
+      "step": 1692
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.988210333370556e-05,
+      "loss": 0.1934,
+      "step": 1693
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9881964200128313e-05,
+      "loss": 0.0684,
+      "step": 1694
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9881824984825802e-05,
+      "loss": 0.5391,
+      "step": 1695
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9881685687798784e-05,
+      "loss": 0.1475,
+      "step": 1696
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9881546309048036e-05,
+      "loss": 1.3828,
+      "step": 1697
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9881406848574318e-05,
+      "loss": 0.0752,
+      "step": 1698
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9881267306378398e-05,
+      "loss": 0.1235,
+      "step": 1699
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.988112768246104e-05,
+      "loss": 0.3496,
+      "step": 1700
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9880987976823015e-05,
+      "loss": 0.6992,
+      "step": 1701
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9880848189465083e-05,
+      "loss": 0.2178,
+      "step": 1702
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.988070832038802e-05,
+      "loss": 0.0459,
+      "step": 1703
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9880568369592583e-05,
+      "loss": 0.5039,
+      "step": 1704
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9880428337079553e-05,
+      "loss": 1.1016,
+      "step": 1705
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9880288222849696e-05,
+      "loss": 0.5547,
+      "step": 1706
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9880148026903772e-05,
+      "loss": 0.6562,
+      "step": 1707
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.988000774924256e-05,
+      "loss": 1.0547,
+      "step": 1708
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9879867389866827e-05,
+      "loss": 0.0654,
+      "step": 1709
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9879726948777348e-05,
+      "loss": 0.4121,
+      "step": 1710
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9879586425974888e-05,
+      "loss": 0.0378,
+      "step": 1711
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9879445821460223e-05,
+      "loss": 2.4375,
+      "step": 1712
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9879305135234123e-05,
+      "loss": 0.7734,
+      "step": 1713
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9879164367297364e-05,
+      "loss": 0.7461,
+      "step": 1714
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9879023517650714e-05,
+      "loss": 0.5664,
+      "step": 1715
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.987888258629495e-05,
+      "loss": 0.0427,
+      "step": 1716
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9878741573230846e-05,
+      "loss": 0.4395,
+      "step": 1717
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9878600478459173e-05,
+      "loss": 1.0391,
+      "step": 1718
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.987845930198071e-05,
+      "loss": 0.7148,
+      "step": 1719
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.987831804379623e-05,
+      "loss": 0.3594,
+      "step": 1720
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.987817670390651e-05,
+      "loss": 0.0002,
+      "step": 1721
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9878035282312327e-05,
+      "loss": 0.3223,
+      "step": 1722
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.987789377901445e-05,
+      "loss": 0.1426,
+      "step": 1723
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9877752194013665e-05,
+      "loss": 0.4395,
+      "step": 1724
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9877610527310747e-05,
+      "loss": 0.1973,
+      "step": 1725
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9877468778906472e-05,
+      "loss": 0.0126,
+      "step": 1726
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.9877326948801622e-05,
+      "loss": 0.0018,
+      "step": 1727
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.987718503699697e-05,
+      "loss": 0.0135,
+      "step": 1728
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.987704304349331e-05,
+      "loss": 0.3105,
+      "step": 1729
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.98769009682914e-05,
+      "loss": 0.0002,
+      "step": 1730
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9876758811392034e-05,
+      "loss": 0.0001,
+      "step": 1731
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.987661657279599e-05,
+      "loss": 0.0001,
+      "step": 1732
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.987647425250405e-05,
+      "loss": 0.25,
+      "step": 1733
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9876331850516992e-05,
+      "loss": 0.918,
+      "step": 1734
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9876189366835604e-05,
+      "loss": 0.0006,
+      "step": 1735
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9876046801460663e-05,
+      "loss": 0.0018,
+      "step": 1736
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9875904154392955e-05,
+      "loss": 0.0449,
+      "step": 1737
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.987576142563326e-05,
+      "loss": 0.1914,
+      "step": 1738
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9875618615182366e-05,
+      "loss": 0.0007,
+      "step": 1739
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9875475723041056e-05,
+      "loss": 0.0007,
+      "step": 1740
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9875332749210108e-05,
+      "loss": 0.0011,
+      "step": 1741
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9875189693690315e-05,
+      "loss": 0.0003,
+      "step": 1742
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9875046556482463e-05,
+      "loss": 0.0002,
+      "step": 1743
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9874903337587342e-05,
+      "loss": 0.0026,
+      "step": 1744
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9874760037005722e-05,
+      "loss": 0.001,
+      "step": 1745
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9874616654738407e-05,
+      "loss": 0.0013,
+      "step": 1746
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.987447319078617e-05,
+      "loss": 0.1162,
+      "step": 1747
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9874329645149812e-05,
+      "loss": 1.0703,
+      "step": 1748
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9874186017830116e-05,
+      "loss": 0.2432,
+      "step": 1749
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9874042308827867e-05,
+      "loss": 0.2402,
+      "step": 1750
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9873898518143864e-05,
+      "loss": 0.9844,
+      "step": 1751
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9873754645778884e-05,
+      "loss": 0.8594,
+      "step": 1752
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9873610691733728e-05,
+      "loss": 1.6094,
+      "step": 1753
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9873466656009177e-05,
+      "loss": 0.7539,
+      "step": 1754
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9873322538606033e-05,
+      "loss": 0.8828,
+      "step": 1755
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9873178339525077e-05,
+      "loss": 1.25,
+      "step": 1756
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9873034058767105e-05,
+      "loss": 0.0002,
+      "step": 1757
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9872889696332907e-05,
+      "loss": 2.2812,
+      "step": 1758
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9872745252223284e-05,
+      "loss": 1.6328,
+      "step": 1759
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9872600726439023e-05,
+      "loss": 0.8945,
+      "step": 1760
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9872456118980914e-05,
+      "loss": 0.0072,
+      "step": 1761
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9872311429849758e-05,
+      "loss": 1.4531,
+      "step": 1762
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9872166659046347e-05,
+      "loss": 0.6797,
+      "step": 1763
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9872021806571476e-05,
+      "loss": 0.4355,
+      "step": 1764
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.987187687242594e-05,
+      "loss": 2.1094,
+      "step": 1765
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9871731856610537e-05,
+      "loss": 0.1826,
+      "step": 1766
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9871586759126054e-05,
+      "loss": 2.2031,
+      "step": 1767
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9871441579973304e-05,
+      "loss": 0.8047,
+      "step": 1768
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9871296319153072e-05,
+      "loss": 0.7305,
+      "step": 1769
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9871150976666162e-05,
+      "loss": 0.0476,
+      "step": 1770
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9871005552513367e-05,
+      "loss": 0.2119,
+      "step": 1771
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.987086004669549e-05,
+      "loss": 0.0005,
+      "step": 1772
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9870714459213325e-05,
+      "loss": 0.8359,
+      "step": 1773
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.987056879006768e-05,
+      "loss": 0.6641,
+      "step": 1774
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9870423039259344e-05,
+      "loss": 0.4043,
+      "step": 1775
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9870277206789126e-05,
+      "loss": 0.042,
+      "step": 1776
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9870131292657824e-05,
+      "loss": 1.375,
+      "step": 1777
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9869985296866236e-05,
+      "loss": 1.1562,
+      "step": 1778
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9869839219415174e-05,
+      "loss": 0.1582,
+      "step": 1779
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9869693060305426e-05,
+      "loss": 0.4043,
+      "step": 1780
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9869546819537806e-05,
+      "loss": 0.4609,
+      "step": 1781
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9869400497113112e-05,
+      "loss": 0.3555,
+      "step": 1782
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.986925409303215e-05,
+      "loss": 0.3242,
+      "step": 1783
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.986910760729572e-05,
+      "loss": 0.0806,
+      "step": 1784
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9868961039904628e-05,
+      "loss": 0.0031,
+      "step": 1785
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9868814390859684e-05,
+      "loss": 1.1562,
+      "step": 1786
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9868667660161688e-05,
+      "loss": 0.5234,
+      "step": 1787
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9868520847811443e-05,
+      "loss": 0.252,
+      "step": 1788
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.986837395380976e-05,
+      "loss": 0.6016,
+      "step": 1789
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9868226978157443e-05,
+      "loss": 1.25,
+      "step": 1790
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.986807992085531e-05,
+      "loss": 0.0386,
+      "step": 1791
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9867932781904154e-05,
+      "loss": 0.0869,
+      "step": 1792
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9867785561304785e-05,
+      "loss": 0.001,
+      "step": 1793
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.986763825905802e-05,
+      "loss": 0.1934,
+      "step": 1794
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9867490875164657e-05,
+      "loss": 0.0004,
+      "step": 1795
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9867343409625518e-05,
+      "loss": 0.165,
+      "step": 1796
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9867195862441402e-05,
+      "loss": 0.1562,
+      "step": 1797
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9867048233613125e-05,
+      "loss": 0.0009,
+      "step": 1798
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9866900523141494e-05,
+      "loss": 0.2188,
+      "step": 1799
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9866752731027325e-05,
+      "loss": 0.1475,
+      "step": 1800
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9866604857271425e-05,
+      "loss": 1.3984,
+      "step": 1801
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.986645690187461e-05,
+      "loss": 1.9922,
+      "step": 1802
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.986630886483769e-05,
+      "loss": 0.4668,
+      "step": 1803
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9866160746161477e-05,
+      "loss": 0.6562,
+      "step": 1804
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9866012545846782e-05,
+      "loss": 1.4297,
+      "step": 1805
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9865864263894426e-05,
+      "loss": 0.7305,
+      "step": 1806
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9865715900305223e-05,
+      "loss": 0.0654,
+      "step": 1807
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.986556745507998e-05,
+      "loss": 0.7305,
+      "step": 1808
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9865418928219523e-05,
+      "loss": 0.1709,
+      "step": 1809
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9865270319724653e-05,
+      "loss": 0.6562,
+      "step": 1810
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9865121629596202e-05,
+      "loss": 0.543,
+      "step": 1811
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9864972857834975e-05,
+      "loss": 0.2354,
+      "step": 1812
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9864824004441795e-05,
+      "loss": 0.373,
+      "step": 1813
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9864675069417474e-05,
+      "loss": 0.1245,
+      "step": 1814
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.986452605276284e-05,
+      "loss": 0.0104,
+      "step": 1815
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.98643769544787e-05,
+      "loss": 0.0854,
+      "step": 1816
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.986422777456588e-05,
+      "loss": 0.0435,
+      "step": 1817
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9864078513025194e-05,
+      "loss": 0.0076,
+      "step": 1818
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9863929169857465e-05,
+      "loss": 0.0197,
+      "step": 1819
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9863779745063512e-05,
+      "loss": 0.2891,
+      "step": 1820
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9863630238644162e-05,
+      "loss": 0.0063,
+      "step": 1821
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9863480650600225e-05,
+      "loss": 0.0017,
+      "step": 1822
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9863330980932527e-05,
+      "loss": 0.4766,
+      "step": 1823
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.986318122964189e-05,
+      "loss": 0.7539,
+      "step": 1824
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9863031396729138e-05,
+      "loss": 0.1416,
+      "step": 1825
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9862881482195097e-05,
+      "loss": 0.2637,
+      "step": 1826
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9862731486040583e-05,
+      "loss": 0.0623,
+      "step": 1827
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.986258140826642e-05,
+      "loss": 0.5234,
+      "step": 1828
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.986243124887344e-05,
+      "loss": 0.9727,
+      "step": 1829
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9862281007862463e-05,
+      "loss": 0.1113,
+      "step": 1830
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9862130685234308e-05,
+      "loss": 0.003,
+      "step": 1831
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.986198028098981e-05,
+      "loss": 0.0193,
+      "step": 1832
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.986182979512979e-05,
+      "loss": 0.1064,
+      "step": 1833
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.986167922765507e-05,
+      "loss": 0.0645,
+      "step": 1834
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.986152857856649e-05,
+      "loss": 0.0035,
+      "step": 1835
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9861377847864865e-05,
+      "loss": 0.0005,
+      "step": 1836
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9861227035551032e-05,
+      "loss": 0.0014,
+      "step": 1837
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.986107614162581e-05,
+      "loss": 0.1216,
+      "step": 1838
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9860925166090034e-05,
+      "loss": 0.0004,
+      "step": 1839
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9860774108944533e-05,
+      "loss": 0.0019,
+      "step": 1840
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.986062297019013e-05,
+      "loss": 0.2266,
+      "step": 1841
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9860471749827664e-05,
+      "loss": 0.1523,
+      "step": 1842
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9860320447857962e-05,
+      "loss": 0.083,
+      "step": 1843
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.986016906428185e-05,
+      "loss": 1.3203,
+      "step": 1844
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9860017599100168e-05,
+      "loss": 0.207,
+      "step": 1845
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.985986605231374e-05,
+      "loss": 0.1836,
+      "step": 1846
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9859714423923405e-05,
+      "loss": 0.2158,
+      "step": 1847
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9859562713929987e-05,
+      "loss": 0.1787,
+      "step": 1848
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.985941092233433e-05,
+      "loss": 0.0791,
+      "step": 1849
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9859259049137257e-05,
+      "loss": 0.21,
+      "step": 1850
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9859107094339605e-05,
+      "loss": 1.2891,
+      "step": 1851
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9858955057942214e-05,
+      "loss": 0.9531,
+      "step": 1852
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9858802939945917e-05,
+      "loss": 1.9609,
+      "step": 1853
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.985865074035155e-05,
+      "loss": 1.1719,
+      "step": 1854
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.985849845915994e-05,
+      "loss": 0.5547,
+      "step": 1855
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.985834609637193e-05,
+      "loss": 0.9141,
+      "step": 1856
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9858193651988357e-05,
+      "loss": 2.0781,
+      "step": 1857
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.985804112601006e-05,
+      "loss": 0.5078,
+      "step": 1858
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9857888518437874e-05,
+      "loss": 1.4375,
+      "step": 1859
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9857735829272638e-05,
+      "loss": 0.5234,
+      "step": 1860
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9857583058515184e-05,
+      "loss": 1.9688,
+      "step": 1861
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.985743020616636e-05,
+      "loss": 1.4453,
+      "step": 1862
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9857277272227004e-05,
+      "loss": 0.2178,
+      "step": 1863
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9857124256697954e-05,
+      "loss": 0.7148,
+      "step": 1864
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9856971159580048e-05,
+      "loss": 0.0479,
+      "step": 1865
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9856817980874133e-05,
+      "loss": 0.1553,
+      "step": 1866
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.985666472058104e-05,
+      "loss": 0.8828,
+      "step": 1867
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9856511378701622e-05,
+      "loss": 0.2236,
+      "step": 1868
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.985635795523671e-05,
+      "loss": 0.0082,
+      "step": 1869
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9856204450187157e-05,
+      "loss": 0.0108,
+      "step": 1870
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.98560508635538e-05,
+      "loss": 0.1099,
+      "step": 1871
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9855897195337485e-05,
+      "loss": 0.0723,
+      "step": 1872
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9855743445539053e-05,
+      "loss": 0.0623,
+      "step": 1873
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9855589614159354e-05,
+      "loss": 1.5,
+      "step": 1874
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.985543570119922e-05,
+      "loss": 0.0396,
+      "step": 1875
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9855281706659513e-05,
+      "loss": 0.1963,
+      "step": 1876
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.985512763054107e-05,
+      "loss": 0.4102,
+      "step": 1877
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9854973472844736e-05,
+      "loss": 0.0001,
+      "step": 1878
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9854819233571357e-05,
+      "loss": 0.1367,
+      "step": 1879
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9854664912721784e-05,
+      "loss": 0.0003,
+      "step": 1880
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9854510510296867e-05,
+      "loss": 0.0742,
+      "step": 1881
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9854356026297445e-05,
+      "loss": 0.0002,
+      "step": 1882
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9854201460724373e-05,
+      "loss": 0.0006,
+      "step": 1883
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.98540468135785e-05,
+      "loss": 0.0425,
+      "step": 1884
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.985389208486067e-05,
+      "loss": 1.1328,
+      "step": 1885
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9853737274571737e-05,
+      "loss": 0.1914,
+      "step": 1886
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9853582382712544e-05,
+      "loss": 0.0513,
+      "step": 1887
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9853427409283957e-05,
+      "loss": 1.6172,
+      "step": 1888
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9853272354286815e-05,
+      "loss": 0.0004,
+      "step": 1889
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9853117217721973e-05,
+      "loss": 0.0201,
+      "step": 1890
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.985296199959028e-05,
+      "loss": 1.0547,
+      "step": 1891
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.985280669989259e-05,
+      "loss": 0.2949,
+      "step": 1892
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9852651318629758e-05,
+      "loss": 0.0009,
+      "step": 1893
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9852495855802634e-05,
+      "loss": 0.4043,
+      "step": 1894
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9852340311412075e-05,
+      "loss": 0.1553,
+      "step": 1895
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.985218468545893e-05,
+      "loss": 0.1631,
+      "step": 1896
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9852028977944062e-05,
+      "loss": 0.2031,
+      "step": 1897
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9851873188868317e-05,
+      "loss": 0.1245,
+      "step": 1898
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9851717318232556e-05,
+      "loss": 0.2637,
+      "step": 1899
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.985156136603764e-05,
+      "loss": 0.1973,
+      "step": 1900
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9851405332284412e-05,
+      "loss": 0.7227,
+      "step": 1901
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9851249216973734e-05,
+      "loss": 1.0234,
+      "step": 1902
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9851093020106472e-05,
+      "loss": 0.9102,
+      "step": 1903
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9850936741683473e-05,
+      "loss": 0.668,
+      "step": 1904
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.98507803817056e-05,
+      "loss": 0.8828,
+      "step": 1905
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9850623940173715e-05,
+      "loss": 0.2832,
+      "step": 1906
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.985046741708867e-05,
+      "loss": 1.3203,
+      "step": 1907
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9850310812451328e-05,
+      "loss": 0.8047,
+      "step": 1908
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9850154126262548e-05,
+      "loss": 0.9766,
+      "step": 1909
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.984999735852319e-05,
+      "loss": 0.0918,
+      "step": 1910
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9849840509234117e-05,
+      "loss": 0.1738,
+      "step": 1911
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.984968357839619e-05,
+      "loss": 1.8984,
+      "step": 1912
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9849526566010272e-05,
+      "loss": 0.0203,
+      "step": 1913
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.984936947207722e-05,
+      "loss": 0.208,
+      "step": 1914
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9849212296597902e-05,
+      "loss": 0.0238,
+      "step": 1915
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9849055039573178e-05,
+      "loss": 0.1533,
+      "step": 1916
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9848897701003918e-05,
+      "loss": 0.3477,
+      "step": 1917
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9848740280890973e-05,
+      "loss": 0.0698,
+      "step": 1918
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9848582779235222e-05,
+      "loss": 0.6289,
+      "step": 1919
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9848425196037518e-05,
+      "loss": 0.1719,
+      "step": 1920
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9848267531298733e-05,
+      "loss": 0.457,
+      "step": 1921
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.984810978501973e-05,
+      "loss": 0.2412,
+      "step": 1922
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9847951957201382e-05,
+      "loss": 0.1016,
+      "step": 1923
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9847794047844547e-05,
+      "loss": 0.2266,
+      "step": 1924
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9847636056950095e-05,
+      "loss": 0.8242,
+      "step": 1925
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9847477984518896e-05,
+      "loss": 0.2412,
+      "step": 1926
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9847319830551814e-05,
+      "loss": 0.0009,
+      "step": 1927
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9847161595049727e-05,
+      "loss": 0.0015,
+      "step": 1928
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9847003278013488e-05,
+      "loss": 0.0103,
+      "step": 1929
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9846844879443976e-05,
+      "loss": 0.0618,
+      "step": 1930
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9846686399342067e-05,
+      "loss": 0.0008,
+      "step": 1931
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.984652783770862e-05,
+      "loss": 0.0034,
+      "step": 1932
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.9846369194544506e-05,
+      "loss": 0.0378,
+      "step": 1933
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.9846210469850605e-05,
+      "loss": 0.2227,
+      "step": 1934
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.9846051663627788e-05,
+      "loss": 1.25,
+      "step": 1935
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.9845892775876918e-05,
+      "loss": 0.0017,
+      "step": 1936
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.984573380659887e-05,
+      "loss": 0.0004,
+      "step": 1937
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.9845574755794522e-05,
+      "loss": 0.5742,
+      "step": 1938
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.984541562346475e-05,
+      "loss": 0.1748,
+      "step": 1939
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.9845256409610415e-05,
+      "loss": 0.0002,
+      "step": 1940
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.9845097114232407e-05,
+      "loss": 0.0089,
+      "step": 1941
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.9844937737331587e-05,
+      "loss": 0.0001,
+      "step": 1942
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.9844778278908836e-05,
+      "loss": 0.8633,
+      "step": 1943
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.9844618738965036e-05,
+      "loss": 0.2207,
+      "step": 1944
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.984445911750105e-05,
+      "loss": 0.002,
+      "step": 1945
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.9844299414517767e-05,
+      "loss": 0.0991,
+      "step": 1946
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.9844139630016057e-05,
+      "loss": 0.1885,
+      "step": 1947
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.98439797639968e-05,
+      "loss": 0.0933,
+      "step": 1948
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.9843819816460868e-05,
+      "loss": 0.1436,
+      "step": 1949
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.984365978740915e-05,
+      "loss": 0.252,
+      "step": 1950
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.9843499676842517e-05,
+      "loss": 0.9375,
+      "step": 1951
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.9843339484761854e-05,
+      "loss": 0.9727,
+      "step": 1952
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.984317921116803e-05,
+      "loss": 0.7344,
+      "step": 1953
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.9843018856061937e-05,
+      "loss": 0.5312,
+      "step": 1954
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.984285841944445e-05,
+      "loss": 1.6172,
+      "step": 1955
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.9842697901316453e-05,
+      "loss": 2.5156,
+      "step": 1956
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.9842537301678823e-05,
+      "loss": 2.1719,
+      "step": 1957
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.9842376620532446e-05,
+      "loss": 1.7891,
+      "step": 1958
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.9842215857878203e-05,
+      "loss": 1.5078,
+      "step": 1959
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.9842055013716974e-05,
+      "loss": 1.5938,
+      "step": 1960
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.9841894088049647e-05,
+      "loss": 0.2227,
+      "step": 1961
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.98417330808771e-05,
+      "loss": 2.7656,
+      "step": 1962
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.9841571992200225e-05,
+      "loss": 0.0121,
+      "step": 1963
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.98414108220199e-05,
+      "loss": 2.0781,
+      "step": 1964
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.9841249570337016e-05,
+      "loss": 1.0469,
+      "step": 1965
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.9841088237152456e-05,
+      "loss": 0.4629,
+      "step": 1966
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.98409268224671e-05,
+      "loss": 0.7383,
+      "step": 1967
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.984076532628184e-05,
+      "loss": 0.8945,
+      "step": 1968
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.9840603748597566e-05,
+      "loss": 0.8359,
+      "step": 1969
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.984044208941516e-05,
+      "loss": 0.7148,
+      "step": 1970
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.984028034873551e-05,
+      "loss": 2.6719,
+      "step": 1971
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.9840118526559507e-05,
+      "loss": 0.3496,
+      "step": 1972
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.9839956622888035e-05,
+      "loss": 0.0466,
+      "step": 1973
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.983979463772199e-05,
+      "loss": 0.0464,
+      "step": 1974
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.983963257106226e-05,
+      "loss": 0.0317,
+      "step": 1975
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.9839470422909733e-05,
+      "loss": 0.0005,
+      "step": 1976
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.9839308193265294e-05,
+      "loss": 0.208,
+      "step": 1977
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.9839145882129842e-05,
+      "loss": 0.2695,
+      "step": 1978
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.9838983489504264e-05,
+      "loss": 0.0001,
+      "step": 1979
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.9838821015389454e-05,
+      "loss": 0.5078,
+      "step": 1980
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.9838658459786307e-05,
+      "loss": 0.0003,
+      "step": 1981
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.9838495822695707e-05,
+      "loss": 0.0004,
+      "step": 1982
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.9838333104118552e-05,
+      "loss": 0.0012,
+      "step": 1983
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.9838170304055744e-05,
+      "loss": 0.3164,
+      "step": 1984
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.9838007422508163e-05,
+      "loss": 0.332,
+      "step": 1985
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.983784445947671e-05,
+      "loss": 0.0005,
+      "step": 1986
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.983768141496228e-05,
+      "loss": 0.0008,
+      "step": 1987
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.983751828896577e-05,
+      "loss": 0.0002,
+      "step": 1988
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.983735508148807e-05,
+      "loss": 0.0015,
+      "step": 1989
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.9837191792530082e-05,
+      "loss": 0.1543,
+      "step": 1990
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.9837028422092702e-05,
+      "loss": 0.0012,
+      "step": 1991
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.9836864970176824e-05,
+      "loss": 0.0006,
+      "step": 1992
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.983670143678335e-05,
+      "loss": 0.0996,
+      "step": 1993
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.983653782191317e-05,
+      "loss": 0.0854,
+      "step": 1994
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.9836374125567193e-05,
+      "loss": 0.1118,
+      "step": 1995
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.9836210347746308e-05,
+      "loss": 0.1504,
+      "step": 1996
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.983604648845142e-05,
+      "loss": 0.2334,
+      "step": 1997
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.9835882547683434e-05,
+      "loss": 0.1631,
+      "step": 1998
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.9835718525443237e-05,
+      "loss": 0.1055,
+      "step": 1999
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.9835554421731745e-05,
+      "loss": 0.21,
+      "step": 2000
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.9835390236549846e-05,
+      "loss": 0.291,
+      "step": 2001
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.9835225969898444e-05,
+      "loss": 0.0825,
+      "step": 2002
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.983506162177845e-05,
+      "loss": 0.625,
+      "step": 2003
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.983489719219076e-05,
+      "loss": 0.5469,
+      "step": 2004
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.9834732681136274e-05,
+      "loss": 1.4297,
+      "step": 2005
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.9834568088615902e-05,
+      "loss": 0.5703,
+      "step": 2006
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.9834403414630547e-05,
+      "loss": 0.6289,
+      "step": 2007
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.9834238659181104e-05,
+      "loss": 0.4414,
+      "step": 2008
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.983407382226849e-05,
+      "loss": 0.5312,
+      "step": 2009
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.9833908903893606e-05,
+      "loss": 0.6992,
+      "step": 2010
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.9833743904057353e-05,
+      "loss": 0.0099,
+      "step": 2011
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.9833578822760644e-05,
+      "loss": 0.8359,
+      "step": 2012
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.9833413660004383e-05,
+      "loss": 1.0625,
+      "step": 2013
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.9833248415789475e-05,
+      "loss": 0.8086,
+      "step": 2014
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.9833083090116825e-05,
+      "loss": 0.0052,
+      "step": 2015
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.983291768298735e-05,
+      "loss": 2.2969,
+      "step": 2016
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.9832752194401954e-05,
+      "loss": 0.1865,
+      "step": 2017
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.9832586624361543e-05,
+      "loss": 0.5664,
+      "step": 2018
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.9832420972867027e-05,
+      "loss": 0.0723,
+      "step": 2019
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.983225523991932e-05,
+      "loss": 0.0048,
+      "step": 2020
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.9832089425519326e-05,
+      "loss": 0.7031,
+      "step": 2021
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.9831923529667958e-05,
+      "loss": 0.002,
+      "step": 2022
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.983175755236613e-05,
+      "loss": 0.0002,
+      "step": 2023
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.983159149361475e-05,
+      "loss": 0.0004,
+      "step": 2024
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.983142535341473e-05,
+      "loss": 2.9375,
+      "step": 2025
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.9831259131766988e-05,
+      "loss": 0.1738,
+      "step": 2026
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.983109282867243e-05,
+      "loss": 0.0145,
+      "step": 2027
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.983092644413197e-05,
+      "loss": 0.1377,
+      "step": 2028
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.983075997814652e-05,
+      "loss": 0.0012,
+      "step": 2029
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.9830593430717002e-05,
+      "loss": 0.0005,
+      "step": 2030
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.9830426801844325e-05,
+      "loss": 0.0105,
+      "step": 2031
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.9830260091529406e-05,
+      "loss": 0.4316,
+      "step": 2032
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.983009329977316e-05,
+      "loss": 0.0194,
+      "step": 2033
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.98299264265765e-05,
+      "loss": 0.1143,
+      "step": 2034
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.9829759471940342e-05,
+      "loss": 0.3086,
+      "step": 2035
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.9829592435865612e-05,
+      "loss": 0.0111,
+      "step": 2036
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.9829425318353218e-05,
+      "loss": 0.002,
+      "step": 2037
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.982925811940408e-05,
+      "loss": 0.0019,
+      "step": 2038
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.982909083901912e-05,
+      "loss": 0.0225,
+      "step": 2039
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.982892347719925e-05,
+      "loss": 0.0012,
+      "step": 2040
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.9828756033945395e-05,
+      "loss": 0.0277,
+      "step": 2041
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.9828588509258473e-05,
+      "loss": 0.0004,
+      "step": 2042
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.98284209031394e-05,
+      "loss": 0.0013,
+      "step": 2043
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.98282532155891e-05,
+      "loss": 0.1758,
+      "step": 2044
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.9828085446608495e-05,
+      "loss": 0.0007,
+      "step": 2045
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.9827917596198505e-05,
+      "loss": 0.1953,
+      "step": 2046
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.982774966436005e-05,
+      "loss": 0.1875,
+      "step": 2047
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.9827581651094054e-05,
+      "loss": 0.2041,
+      "step": 2048
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.9827413556401436e-05,
+      "loss": 0.2227,
+      "step": 2049
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.9827245380283127e-05,
+      "loss": 0.1816,
+      "step": 2050
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.9827077122740046e-05,
+      "loss": 1.3438,
+      "step": 2051
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.9826908783773114e-05,
+      "loss": 0.875,
+      "step": 2052
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.982674036338326e-05,
+      "loss": 0.0669,
+      "step": 2053
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.982657186157141e-05,
+      "loss": 1.2266,
+      "step": 2054
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.9826403278338484e-05,
+      "loss": 0.332,
+      "step": 2055
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.9826234613685408e-05,
+      "loss": 0.9023,
+      "step": 2056
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.9826065867613117e-05,
+      "loss": 0.6055,
+      "step": 2057
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.9825897040122525e-05,
+      "loss": 1.0859,
+      "step": 2058
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.982572813121457e-05,
+      "loss": 0.0713,
+      "step": 2059
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.982555914089017e-05,
+      "loss": 0.6445,
+      "step": 2060
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.9825390069150262e-05,
+      "loss": 0.543,
+      "step": 2061
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.982522091599577e-05,
+      "loss": 0.8516,
+      "step": 2062
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.9825051681427624e-05,
+      "loss": 0.3457,
+      "step": 2063
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.982488236544675e-05,
+      "loss": 0.7188,
+      "step": 2064
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.9824712968054086e-05,
+      "loss": 0.6133,
+      "step": 2065
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.982454348925055e-05,
+      "loss": 1.2734,
+      "step": 2066
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.9824373929037088e-05,
+      "loss": 0.0566,
+      "step": 2067
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.9824204287414616e-05,
+      "loss": 1.1875,
+      "step": 2068
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.9824034564384074e-05,
+      "loss": 0.6367,
+      "step": 2069
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.9823864759946392e-05,
+      "loss": 0.0071,
+      "step": 2070
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.9823694874102505e-05,
+      "loss": 0.0038,
+      "step": 2071
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.9823524906853344e-05,
+      "loss": 0.0308,
+      "step": 2072
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.9823354858199837e-05,
+      "loss": 0.3203,
+      "step": 2073
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.982318472814293e-05,
+      "loss": 0.3867,
+      "step": 2074
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.9823014516683542e-05,
+      "loss": 0.1235,
+      "step": 2075
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.9822844223822624e-05,
+      "loss": 0.707,
+      "step": 2076
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.9822673849561098e-05,
+      "loss": 0.0117,
+      "step": 2077
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.982250339389991e-05,
+      "loss": 0.0193,
+      "step": 2078
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.9822332856839984e-05,
+      "loss": 0.0479,
+      "step": 2079
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.9822162238382268e-05,
+      "loss": 0.4473,
+      "step": 2080
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.9821991538527697e-05,
+      "loss": 0.2363,
+      "step": 2081
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.9821820757277204e-05,
+      "loss": 0.0001,
+      "step": 2082
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.9821649894631723e-05,
+      "loss": 0.6367,
+      "step": 2083
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.9821478950592204e-05,
+      "loss": 0.1504,
+      "step": 2084
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.9821307925159578e-05,
+      "loss": 0.0129,
+      "step": 2085
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.9821136818334783e-05,
+      "loss": 0.0591,
+      "step": 2086
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.9820965630118766e-05,
+      "loss": 0.0008,
+      "step": 2087
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.9820794360512463e-05,
+      "loss": 0.418,
+      "step": 2088
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.9820623009516808e-05,
+      "loss": 0.0006,
+      "step": 2089
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.9820451577132756e-05,
+      "loss": 0.0002,
+      "step": 2090
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.9820280063361237e-05,
+      "loss": 0.0009,
+      "step": 2091
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.9820108468203192e-05,
+      "loss": 0.0004,
+      "step": 2092
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.9819936791659572e-05,
+      "loss": 0.0001,
+      "step": 2093
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.9819765033731318e-05,
+      "loss": 0.0011,
+      "step": 2094
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.9819593194419366e-05,
+      "loss": 0.0001,
+      "step": 2095
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.981942127372467e-05,
+      "loss": 0.0001,
+      "step": 2096
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.9819249271648164e-05,
+      "loss": 0.1338,
+      "step": 2097
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.98190771881908e-05,
+      "loss": 0.085,
+      "step": 2098
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.9818905023353515e-05,
+      "loss": 0.167,
+      "step": 2099
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.9818732777137263e-05,
+      "loss": 0.1973,
+      "step": 2100
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.9818560449542987e-05,
+      "loss": 0.5469,
+      "step": 2101
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.9818388040571634e-05,
+      "loss": 1.7656,
+      "step": 2102
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.9818215550224152e-05,
+      "loss": 0.9219,
+      "step": 2103
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.981804297850148e-05,
+      "loss": 1.9297,
+      "step": 2104
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.9817870325404572e-05,
+      "loss": 0.207,
+      "step": 2105
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.9817697590934378e-05,
+      "loss": 0.125,
+      "step": 2106
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.9817524775091843e-05,
+      "loss": 0.1147,
+      "step": 2107
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.9817351877877916e-05,
+      "loss": 0.7578,
+      "step": 2108
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.9817178899293555e-05,
+      "loss": 0.9766,
+      "step": 2109
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.9817005839339692e-05,
+      "loss": 1.2266,
+      "step": 2110
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.9816832698017297e-05,
+      "loss": 0.7383,
+      "step": 2111
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.9816659475327307e-05,
+      "loss": 0.332,
+      "step": 2112
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.9816486171270682e-05,
+      "loss": 0.9883,
+      "step": 2113
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.9816312785848367e-05,
+      "loss": 0.4844,
+      "step": 2114
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.9816139319061317e-05,
+      "loss": 0.6484,
+      "step": 2115
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.9815965770910484e-05,
+      "loss": 1.6797,
+      "step": 2116
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.9815792141396824e-05,
+      "loss": 0.0447,
+      "step": 2117
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.9815618430521286e-05,
+      "loss": 0.5117,
+      "step": 2118
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.9815444638284825e-05,
+      "loss": 0.3379,
+      "step": 2119
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.9815270764688402e-05,
+      "loss": 0.1416,
+      "step": 2120
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.981509680973296e-05,
+      "loss": 0.0001,
+      "step": 2121
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.9814922773419464e-05,
+      "loss": 0.1426,
+      "step": 2122
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.9814748655748868e-05,
+      "loss": 0.0435,
+      "step": 2123
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.9814574456722127e-05,
+      "loss": 0.9609,
+      "step": 2124
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.981440017634019e-05,
+      "loss": 1.7969,
+      "step": 2125
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.9814225814604028e-05,
+      "loss": 2.0781,
+      "step": 2126
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.981405137151459e-05,
+      "loss": 0.0005,
+      "step": 2127
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.9813876847072836e-05,
+      "loss": 0.4004,
+      "step": 2128
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.981370224127972e-05,
+      "loss": 0.0032,
+      "step": 2129
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.981352755413621e-05,
+      "loss": 0.0029,
+      "step": 2130
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.981335278564326e-05,
+      "loss": 0.0002,
+      "step": 2131
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.981317793580183e-05,
+      "loss": 0.0061,
+      "step": 2132
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.981300300461288e-05,
+      "loss": 0.1582,
+      "step": 2133
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.981282799207737e-05,
+      "loss": 0.2393,
+      "step": 2134
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.9812652898196264e-05,
+      "loss": 0.0016,
+      "step": 2135
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.981247772297052e-05,
+      "loss": 0.0014,
+      "step": 2136
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.9812302466401096e-05,
+      "loss": 0.2363,
+      "step": 2137
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.9812127128488965e-05,
+      "loss": 0.5742,
+      "step": 2138
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.9811951709235087e-05,
+      "loss": 1.6719,
+      "step": 2139
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.981177620864042e-05,
+      "loss": 0.3418,
+      "step": 2140
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.9811600626705933e-05,
+      "loss": 0.0038,
+      "step": 2141
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.9811424963432584e-05,
+      "loss": 0.1152,
+      "step": 2142
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.9811249218821347e-05,
+      "loss": 0.1108,
+      "step": 2143
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.981107339287318e-05,
+      "loss": 0.2578,
+      "step": 2144
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.981089748558905e-05,
+      "loss": 0.1494,
+      "step": 2145
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.9810721496969924e-05,
+      "loss": 0.1299,
+      "step": 2146
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.9810545427016768e-05,
+      "loss": 0.1147,
+      "step": 2147
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.981036927573055e-05,
+      "loss": 0.1973,
+      "step": 2148
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.9810193043112234e-05,
+      "loss": 0.3848,
+      "step": 2149
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.981001672916279e-05,
+      "loss": 0.209,
+      "step": 2150
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.9809840333883186e-05,
+      "loss": 0.4746,
+      "step": 2151
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.9809663857274395e-05,
+      "loss": 0.4785,
+      "step": 2152
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.9809487299337375e-05,
+      "loss": 1.0625,
+      "step": 2153
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.980931066007311e-05,
+      "loss": 0.3418,
+      "step": 2154
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.980913393948256e-05,
+      "loss": 1.7031,
+      "step": 2155
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.98089571375667e-05,
+      "loss": 1.4062,
+      "step": 2156
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.9808780254326494e-05,
+      "loss": 0.2812,
+      "step": 2157
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.980860328976292e-05,
+      "loss": 0.5781,
+      "step": 2158
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.980842624387695e-05,
+      "loss": 1.0234,
+      "step": 2159
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.9808249116669553e-05,
+      "loss": 0.377,
+      "step": 2160
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.9808071908141703e-05,
+      "loss": 0.2754,
+      "step": 2161
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.9807894618294376e-05,
+      "loss": 0.7383,
+      "step": 2162
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.9807717247128543e-05,
+      "loss": 1.0547,
+      "step": 2163
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.9807539794645175e-05,
+      "loss": 0.2832,
+      "step": 2164
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.980736226084525e-05,
+      "loss": 1.6719,
+      "step": 2165
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.980718464572974e-05,
+      "loss": 0.0011,
+      "step": 2166
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.980700694929963e-05,
+      "loss": 0.0664,
+      "step": 2167
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.9806829171555885e-05,
+      "loss": 0.4727,
+      "step": 2168
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.9806651312499483e-05,
+      "loss": 0.0237,
+      "step": 2169
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.9806473372131402e-05,
+      "loss": 1.6797,
+      "step": 2170
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.9806295350452624e-05,
+      "loss": 0.1631,
+      "step": 2171
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.980611724746412e-05,
+      "loss": 0.0442,
+      "step": 2172
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.980593906316687e-05,
+      "loss": 0.0014,
+      "step": 2173
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.980576079756185e-05,
+      "loss": 0.0015,
+      "step": 2174
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.9805582450650048e-05,
+      "loss": 0.6172,
+      "step": 2175
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.980540402243244e-05,
+      "loss": 0.1406,
+      "step": 2176
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.9805225512909995e-05,
+      "loss": 0.0017,
+      "step": 2177
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.9805046922083704e-05,
+      "loss": 1.0078,
+      "step": 2178
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.9804868249954546e-05,
+      "loss": 0.0017,
+      "step": 2179
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.9804689496523503e-05,
+      "loss": 0.0015,
+      "step": 2180
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.980451066179155e-05,
+      "loss": 0.957,
+      "step": 2181
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.980433174575968e-05,
+      "loss": 0.0003,
+      "step": 2182
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.9804152748428866e-05,
+      "loss": 0.4043,
+      "step": 2183
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.9803973669800096e-05,
+      "loss": 0.0471,
+      "step": 2184
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.9803794509874353e-05,
+      "loss": 0.0654,
+      "step": 2185
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.9803615268652618e-05,
+      "loss": 0.0003,
+      "step": 2186
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.980343594613588e-05,
+      "loss": 0.0087,
+      "step": 2187
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.980325654232512e-05,
+      "loss": 0.0223,
+      "step": 2188
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.980307705722132e-05,
+      "loss": 0.0032,
+      "step": 2189
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.9802897490825473e-05,
+      "loss": 0.0004,
+      "step": 2190
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.9802717843138562e-05,
+      "loss": 0.0003,
+      "step": 2191
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.9802538114161575e-05,
+      "loss": 0.0003,
+      "step": 2192
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.9802358303895495e-05,
+      "loss": 0.0009,
+      "step": 2193
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.9802178412341318e-05,
+      "loss": 0.1611,
+      "step": 2194
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.980199843950002e-05,
+      "loss": 0.0002,
+      "step": 2195
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.9801818385372597e-05,
+      "loss": 0.125,
+      "step": 2196
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.9801638249960033e-05,
+      "loss": 0.1885,
+      "step": 2197
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.9801458033263324e-05,
+      "loss": 0.2236,
+      "step": 2198
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.9801277735283453e-05,
+      "loss": 0.123,
+      "step": 2199
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.9801097356021412e-05,
+      "loss": 0.3594,
+      "step": 2200
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.98009168954782e-05,
+      "loss": 0.8281,
+      "step": 2201
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.9800736353654795e-05,
+      "loss": 0.8555,
+      "step": 2202
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.9800555730552193e-05,
+      "loss": 0.3047,
+      "step": 2203
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.980037502617139e-05,
+      "loss": 0.4062,
+      "step": 2204
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.980019424051337e-05,
+      "loss": 0.7031,
+      "step": 2205
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.9800013373579136e-05,
+      "loss": 0.8516,
+      "step": 2206
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.9799832425369676e-05,
+      "loss": 1.9062,
+      "step": 2207
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.979965139588598e-05,
+      "loss": 0.5039,
+      "step": 2208
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.979947028512905e-05,
+      "loss": 2.1562,
+      "step": 2209
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.9799289093099876e-05,
+      "loss": 1.4531,
+      "step": 2210
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.9799107819799455e-05,
+      "loss": 0.0025,
+      "step": 2211
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.979892646522878e-05,
+      "loss": 0.8867,
+      "step": 2212
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.9798745029388844e-05,
+      "loss": 0.0549,
+      "step": 2213
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.9798563512280653e-05,
+      "loss": 0.4766,
+      "step": 2214
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.9798381913905196e-05,
+      "loss": 0.0104,
+      "step": 2215
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.9798200234263477e-05,
+      "loss": 0.4551,
+      "step": 2216
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.9798018473356483e-05,
+      "loss": 0.0177,
+      "step": 2217
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.979783663118522e-05,
+      "loss": 0.2715,
+      "step": 2218
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.979765470775069e-05,
+      "loss": 0.4414,
+      "step": 2219
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.9797472703053884e-05,
+      "loss": 0.0003,
+      "step": 2220
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.9797290617095797e-05,
+      "loss": 0.1016,
+      "step": 2221
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.979710844987745e-05,
+      "loss": 0.0947,
+      "step": 2222
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.979692620139982e-05,
+      "loss": 0.126,
+      "step": 2223
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.979674387166392e-05,
+      "loss": 2.5781,
+      "step": 2224
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.979656146067075e-05,
+      "loss": 0.5898,
+      "step": 2225
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.9796378968421308e-05,
+      "loss": 0.0101,
+      "step": 2226
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.9796196394916603e-05,
+      "loss": 0.0004,
+      "step": 2227
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.9796013740157633e-05,
+      "loss": 0.0162,
+      "step": 2228
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.97958310041454e-05,
+      "loss": 0.2578,
+      "step": 2229
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.9795648186880907e-05,
+      "loss": 0.8516,
+      "step": 2230
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.9795465288365164e-05,
+      "loss": 0.0017,
+      "step": 2231
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.9795282308599168e-05,
+      "loss": 0.0003,
+      "step": 2232
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.979509924758393e-05,
+      "loss": 0.083,
+      "step": 2233
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.9794916105320455e-05,
+      "loss": 0.0483,
+      "step": 2234
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.979473288180974e-05,
+      "loss": 0.0016,
+      "step": 2235
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.97945495770528e-05,
+      "loss": 0.1523,
+      "step": 2236
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.979436619105064e-05,
+      "loss": 0.625,
+      "step": 2237
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.9794182723804264e-05,
+      "loss": 0.3105,
+      "step": 2238
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.9793999175314682e-05,
+      "loss": 0.5586,
+      "step": 2239
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.9793815545582904e-05,
+      "loss": 0.0005,
+      "step": 2240
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.9793631834609935e-05,
+      "loss": 0.5391,
+      "step": 2241
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.9793448042396784e-05,
+      "loss": 0.0022,
+      "step": 2242
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.979326416894446e-05,
+      "loss": 0.0014,
+      "step": 2243
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.9793080214253975e-05,
+      "loss": 0.1738,
+      "step": 2244
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.979289617832634e-05,
+      "loss": 1.2031,
+      "step": 2245
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.9792712061162565e-05,
+      "loss": 0.1719,
+      "step": 2246
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.9792527862763657e-05,
+      "loss": 0.1475,
+      "step": 2247
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.9792343583130634e-05,
+      "loss": 0.21,
+      "step": 2248
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.97921592222645e-05,
+      "loss": 0.1562,
+      "step": 2249
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.9791974780166277e-05,
+      "loss": 0.1494,
+      "step": 2250
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.9791790256836965e-05,
+      "loss": 0.8906,
+      "step": 2251
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.9791605652277594e-05,
+      "loss": 0.5312,
+      "step": 2252
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.9791420966489162e-05,
+      "loss": 0.5195,
+      "step": 2253
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.979123619947269e-05,
+      "loss": 0.5977,
+      "step": 2254
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.97910513512292e-05,
+      "loss": 0.5859,
+      "step": 2255
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.9790866421759693e-05,
+      "loss": 0.293,
+      "step": 2256
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.979068141106519e-05,
+      "loss": 1.6562,
+      "step": 2257
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.9790496319146715e-05,
+      "loss": 0.1992,
+      "step": 2258
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.9790311146005273e-05,
+      "loss": 0.6406,
+      "step": 2259
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.979012589164189e-05,
+      "loss": 1.1641,
+      "step": 2260
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.9789940556057574e-05,
+      "loss": 0.5742,
+      "step": 2261
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.9789755139253352e-05,
+      "loss": 0.1143,
+      "step": 2262
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.9789569641230235e-05,
+      "loss": 0.3066,
+      "step": 2263
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.9789384061989245e-05,
+      "loss": 0.5352,
+      "step": 2264
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.9789198401531402e-05,
+      "loss": 0.4785,
+      "step": 2265
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.9789012659857726e-05,
+      "loss": 1.5781,
+      "step": 2266
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.9788826836969233e-05,
+      "loss": 0.2471,
+      "step": 2267
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.9788640932866944e-05,
+      "loss": 0.1074,
+      "step": 2268
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.9788454947551887e-05,
+      "loss": 0.7109,
+      "step": 2269
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.9788268881025077e-05,
+      "loss": 0.0688,
+      "step": 2270
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.9788082733287533e-05,
+      "loss": 0.1235,
+      "step": 2271
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.9787896504340282e-05,
+      "loss": 0.4375,
+      "step": 2272
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.978771019418435e-05,
+      "loss": 0.1133,
+      "step": 2273
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.9787523802820753e-05,
+      "loss": 0.0532,
+      "step": 2274
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.9787337330250516e-05,
+      "loss": 2.7656,
+      "step": 2275
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.978715077647467e-05,
+      "loss": 1.0703,
+      "step": 2276
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.9786964141494227e-05,
+      "loss": 0.2402,
+      "step": 2277
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.9786777425310226e-05,
+      "loss": 0.3164,
+      "step": 2278
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.978659062792368e-05,
+      "loss": 0.0364,
+      "step": 2279
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.9786403749335623e-05,
+      "loss": 0.0493,
+      "step": 2280
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.9786216789547078e-05,
+      "loss": 0.0026,
+      "step": 2281
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.978602974855907e-05,
+      "loss": 0.3965,
+      "step": 2282
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.978584262637263e-05,
+      "loss": 0.1807,
+      "step": 2283
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.9785655422988788e-05,
+      "loss": 0.0079,
+      "step": 2284
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.9785468138408562e-05,
+      "loss": 0.0037,
+      "step": 2285
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.9785280772632988e-05,
+      "loss": 0.1226,
+      "step": 2286
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.9785093325663092e-05,
+      "loss": 0.4395,
+      "step": 2287
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.9784905797499904e-05,
+      "loss": 0.0096,
+      "step": 2288
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.9784718188144458e-05,
+      "loss": 0.014,
+      "step": 2289
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.9784530497597783e-05,
+      "loss": 0.3066,
+      "step": 2290
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.9784342725860903e-05,
+      "loss": 0.0145,
+      "step": 2291
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.9784154872934855e-05,
+      "loss": 0.0219,
+      "step": 2292
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.9783966938820668e-05,
+      "loss": 0.0659,
+      "step": 2293
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.9783778923519376e-05,
+      "loss": 0.0698,
+      "step": 2294
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.978359082703201e-05,
+      "loss": 0.001,
+      "step": 2295
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.9783402649359607e-05,
+      "loss": 0.0981,
+      "step": 2296
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.9783214390503196e-05,
+      "loss": 0.0991,
+      "step": 2297
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.978302605046381e-05,
+      "loss": 0.1377,
+      "step": 2298
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.9782837629242484e-05,
+      "loss": 0.1099,
+      "step": 2299
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.978264912684026e-05,
+      "loss": 0.1865,
+      "step": 2300
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.9782460543258163e-05,
+      "loss": 0.7188,
+      "step": 2301
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.9782271878497235e-05,
+      "loss": 0.4023,
+      "step": 2302
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.9782083132558502e-05,
+      "loss": 0.2852,
+      "step": 2303
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.978189430544302e-05,
+      "loss": 2.2344,
+      "step": 2304
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.9781705397151804e-05,
+      "loss": 0.6641,
+      "step": 2305
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.978151640768591e-05,
+      "loss": 1.1719,
+      "step": 2306
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.978132733704636e-05,
+      "loss": 0.6367,
+      "step": 2307
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.9781138185234206e-05,
+      "loss": 0.3281,
+      "step": 2308
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.9780948952250478e-05,
+      "loss": 1.2031,
+      "step": 2309
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.9780759638096217e-05,
+      "loss": 0.3789,
+      "step": 2310
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.9780570242772465e-05,
+      "loss": 0.5586,
+      "step": 2311
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.9780380766280264e-05,
+      "loss": 0.3516,
+      "step": 2312
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.978019120862065e-05,
+      "loss": 1.1172,
+      "step": 2313
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.9780001569794658e-05,
+      "loss": 0.0219,
+      "step": 2314
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.9779811849803344e-05,
+      "loss": 0.2793,
+      "step": 2315
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.9779622048647737e-05,
+      "loss": 0.3418,
+      "step": 2316
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.9779432166328888e-05,
+      "loss": 0.375,
+      "step": 2317
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.9779242202847832e-05,
+      "loss": 0.001,
+      "step": 2318
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.977905215820562e-05,
+      "loss": 0.1084,
+      "step": 2319
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.9778862032403293e-05,
+      "loss": 0.0264,
+      "step": 2320
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.977867182544189e-05,
+      "loss": 0.4141,
+      "step": 2321
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.9778481537322466e-05,
+      "loss": 0.0009,
+      "step": 2322
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.9778291168046056e-05,
+      "loss": 2.5,
+      "step": 2323
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.9778100717613708e-05,
+      "loss": 0.4473,
+      "step": 2324
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.9777910186026473e-05,
+      "loss": 0.0129,
+      "step": 2325
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.977771957328539e-05,
+      "loss": 0.006,
+      "step": 2326
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.977752887939151e-05,
+      "loss": 0.1289,
+      "step": 2327
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.9777338104345878e-05,
+      "loss": 0.0002,
+      "step": 2328
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.9777147248149546e-05,
+      "loss": 0.0148,
+      "step": 2329
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.9776956310803557e-05,
+      "loss": 0.2217,
+      "step": 2330
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.9776765292308963e-05,
+      "loss": 0.0007,
+      "step": 2331
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.9776574192666813e-05,
+      "loss": 0.0008,
+      "step": 2332
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.977638301187815e-05,
+      "loss": 0.1621,
+      "step": 2333
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.9776191749944035e-05,
+      "loss": 1.9609,
+      "step": 2334
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.9776000406865508e-05,
+      "loss": 0.5352,
+      "step": 2335
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.9775808982643626e-05,
+      "loss": 2.2969,
+      "step": 2336
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.9775617477279437e-05,
+      "loss": 0.0002,
+      "step": 2337
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.9775425890773994e-05,
+      "loss": 0.0004,
+      "step": 2338
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.9775234223128352e-05,
+      "loss": 0.0019,
+      "step": 2339
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.9775042474343553e-05,
+      "loss": 0.0012,
+      "step": 2340
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9774850644420666e-05,
+      "loss": 0.0004,
+      "step": 2341
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9774658733360733e-05,
+      "loss": 0.0005,
+      "step": 2342
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.977446674116481e-05,
+      "loss": 0.001,
+      "step": 2343
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.977427466783395e-05,
+      "loss": 1.9766,
+      "step": 2344
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9774082513369218e-05,
+      "loss": 0.1465,
+      "step": 2345
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9773890277771657e-05,
+      "loss": 0.1641,
+      "step": 2346
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9773697961042323e-05,
+      "loss": 0.1533,
+      "step": 2347
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.977350556318228e-05,
+      "loss": 0.1187,
+      "step": 2348
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.977331308419258e-05,
+      "loss": 0.2031,
+      "step": 2349
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.977312052407428e-05,
+      "loss": 0.1387,
+      "step": 2350
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.977292788282844e-05,
+      "loss": 0.7539,
+      "step": 2351
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9772735160456114e-05,
+      "loss": 1.1094,
+      "step": 2352
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9772542356958364e-05,
+      "loss": 1.125,
+      "step": 2353
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9772349472336244e-05,
+      "loss": 0.8398,
+      "step": 2354
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.977215650659082e-05,
+      "loss": 0.4648,
+      "step": 2355
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.977196345972315e-05,
+      "loss": 1.1016,
+      "step": 2356
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9771770331734287e-05,
+      "loss": 0.793,
+      "step": 2357
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9771577122625294e-05,
+      "loss": 0.053,
+      "step": 2358
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9771383832397246e-05,
+      "loss": 1.1953,
+      "step": 2359
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9771190461051183e-05,
+      "loss": 1.1797,
+      "step": 2360
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.977099700858818e-05,
+      "loss": 0.0154,
+      "step": 2361
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.97708034750093e-05,
+      "loss": 0.6133,
+      "step": 2362
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.97706098603156e-05,
+      "loss": 0.6406,
+      "step": 2363
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9770416164508146e-05,
+      "loss": 0.3438,
+      "step": 2364
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9770222387587998e-05,
+      "loss": 0.5625,
+      "step": 2365
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.977002852955623e-05,
+      "loss": 1.4844,
+      "step": 2366
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9769834590413894e-05,
+      "loss": 1.375,
+      "step": 2367
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9769640570162064e-05,
+      "loss": 1.7734,
+      "step": 2368
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9769446468801802e-05,
+      "loss": 0.3848,
+      "step": 2369
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9769252286334174e-05,
+      "loss": 0.7227,
+      "step": 2370
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9769058022760247e-05,
+      "loss": 0.5,
+      "step": 2371
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9768863678081087e-05,
+      "loss": 0.0005,
+      "step": 2372
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.976866925229776e-05,
+      "loss": 0.004,
+      "step": 2373
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.976847474541134e-05,
+      "loss": 0.3301,
+      "step": 2374
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9768280157422885e-05,
+      "loss": 0.4355,
+      "step": 2375
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9768085488333477e-05,
+      "loss": 0.0044,
+      "step": 2376
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9767890738144174e-05,
+      "loss": 0.0002,
+      "step": 2377
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9767695906856048e-05,
+      "loss": 0.0001,
+      "step": 2378
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9767500994470168e-05,
+      "loss": 1.3516,
+      "step": 2379
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.976730600098761e-05,
+      "loss": 0.0016,
+      "step": 2380
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9767110926409437e-05,
+      "loss": 0.0035,
+      "step": 2381
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9766915770736724e-05,
+      "loss": 0.0001,
+      "step": 2382
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9766720533970548e-05,
+      "loss": 0.3262,
+      "step": 2383
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9766525216111975e-05,
+      "loss": 0.0004,
+      "step": 2384
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9766329817162076e-05,
+      "loss": 0.0002,
+      "step": 2385
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9766134337121928e-05,
+      "loss": 0.0008,
+      "step": 2386
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9765938775992602e-05,
+      "loss": 0.0282,
+      "step": 2387
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9765743133775172e-05,
+      "loss": 0.0171,
+      "step": 2388
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9765547410470716e-05,
+      "loss": 0.0003,
+      "step": 2389
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9765351606080304e-05,
+      "loss": 0.0003,
+      "step": 2390
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9765155720605017e-05,
+      "loss": 0.0003,
+      "step": 2391
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9764959754045923e-05,
+      "loss": 0.3164,
+      "step": 2392
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9764763706404108e-05,
+      "loss": 0.1523,
+      "step": 2393
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.976456757768064e-05,
+      "loss": 0.1006,
+      "step": 2394
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9764371367876597e-05,
+      "loss": 0.168,
+      "step": 2395
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9764175076993062e-05,
+      "loss": 0.1523,
+      "step": 2396
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.976397870503111e-05,
+      "loss": 0.1177,
+      "step": 2397
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9763782251991816e-05,
+      "loss": 0.1387,
+      "step": 2398
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.976358571787627e-05,
+      "loss": 0.2246,
+      "step": 2399
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9763389102685533e-05,
+      "loss": 0.1196,
+      "step": 2400
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9763192406420697e-05,
+      "loss": 2.0156,
+      "step": 2401
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.976299562908284e-05,
+      "loss": 0.7852,
+      "step": 2402
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9762798770673046e-05,
+      "loss": 0.5195,
+      "step": 2403
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.976260183119239e-05,
+      "loss": 0.6211,
+      "step": 2404
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9762404810641958e-05,
+      "loss": 0.5312,
+      "step": 2405
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9762207709022827e-05,
+      "loss": 1.5781,
+      "step": 2406
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 2.9762010526336087e-05,
+      "loss": 0.6328,
+      "step": 2407
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9761813262582815e-05,
+      "loss": 1.25,
+      "step": 2408
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9761615917764092e-05,
+      "loss": 0.4297,
+      "step": 2409
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.976141849188101e-05,
+      "loss": 0.1025,
+      "step": 2410
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9761220984934645e-05,
+      "loss": 0.3945,
+      "step": 2411
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.976102339692609e-05,
+      "loss": 0.0123,
+      "step": 2412
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.976082572785642e-05,
+      "loss": 0.4629,
+      "step": 2413
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9760627977726727e-05,
+      "loss": 0.6094,
+      "step": 2414
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.97604301465381e-05,
+      "loss": 0.0815,
+      "step": 2415
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9760232234291618e-05,
+      "loss": 0.0012,
+      "step": 2416
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9760034240988367e-05,
+      "loss": 0.0003,
+      "step": 2417
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9759836166629444e-05,
+      "loss": 0.0908,
+      "step": 2418
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.975963801121593e-05,
+      "loss": 1.125,
+      "step": 2419
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9759439774748912e-05,
+      "loss": 0.4043,
+      "step": 2420
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9759241457229486e-05,
+      "loss": 0.0031,
+      "step": 2421
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.975904305865873e-05,
+      "loss": 0.0144,
+      "step": 2422
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.975884457903774e-05,
+      "loss": 0.0154,
+      "step": 2423
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9758646018367607e-05,
+      "loss": 0.0001,
+      "step": 2424
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.975844737664942e-05,
+      "loss": 0.0006,
+      "step": 2425
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.975824865388427e-05,
+      "loss": 0.0166,
+      "step": 2426
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9758049850073246e-05,
+      "loss": 0.418,
+      "step": 2427
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9757850965217444e-05,
+      "loss": 0.7969,
+      "step": 2428
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.975765199931795e-05,
+      "loss": 0.293,
+      "step": 2429
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9757452952375866e-05,
+      "loss": 0.3203,
+      "step": 2430
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9757253824392277e-05,
+      "loss": 0.7812,
+      "step": 2431
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.975705461536828e-05,
+      "loss": 0.0001,
+      "step": 2432
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9756855325304965e-05,
+      "loss": 0.0007,
+      "step": 2433
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.975665595420343e-05,
+      "loss": 0.0003,
+      "step": 2434
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.975645650206477e-05,
+      "loss": 0.0003,
+      "step": 2435
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.975625696889008e-05,
+      "loss": 0.0,
+      "step": 2436
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9756057354680458e-05,
+      "loss": 0.0002,
+      "step": 2437
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9755857659437e-05,
+      "loss": 0.0107,
+      "step": 2438
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9755657883160792e-05,
+      "loss": 1.3984,
+      "step": 2439
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9755458025852945e-05,
+      "loss": 0.0002,
+      "step": 2440
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9755258087514548e-05,
+      "loss": 0.0004,
+      "step": 2441
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9755058068146705e-05,
+      "loss": 1.0,
+      "step": 2442
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9754857967750505e-05,
+      "loss": 0.7734,
+      "step": 2443
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.975465778632706e-05,
+      "loss": 0.123,
+      "step": 2444
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9754457523877457e-05,
+      "loss": 0.1216,
+      "step": 2445
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9754257180402805e-05,
+      "loss": 0.5391,
+      "step": 2446
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.97540567559042e-05,
+      "loss": 0.0001,
+      "step": 2447
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9753856250382746e-05,
+      "loss": 0.3711,
+      "step": 2448
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9753655663839538e-05,
+      "loss": 0.1982,
+      "step": 2449
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9753454996275677e-05,
+      "loss": 0.1396,
+      "step": 2450
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.975325424769227e-05,
+      "loss": 0.3965,
+      "step": 2451
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9753053418090417e-05,
+      "loss": 0.8086,
+      "step": 2452
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9752852507471228e-05,
+      "loss": 0.0967,
+      "step": 2453
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9752651515835797e-05,
+      "loss": 0.5898,
+      "step": 2454
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9752450443185233e-05,
+      "loss": 0.2891,
+      "step": 2455
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9752249289520635e-05,
+      "loss": 1.2031,
+      "step": 2456
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.975204805484311e-05,
+      "loss": 1.2031,
+      "step": 2457
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.975184673915377e-05,
+      "loss": 0.3359,
+      "step": 2458
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9751645342453708e-05,
+      "loss": 0.5703,
+      "step": 2459
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.975144386474404e-05,
+      "loss": 0.2715,
+      "step": 2460
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9751242306025864e-05,
+      "loss": 0.125,
+      "step": 2461
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9751040666300297e-05,
+      "loss": 0.5078,
+      "step": 2462
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9750838945568436e-05,
+      "loss": 2.0469,
+      "step": 2463
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.97506371438314e-05,
+      "loss": 0.8398,
+      "step": 2464
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9750435261090287e-05,
+      "loss": 0.3594,
+      "step": 2465
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.975023329734621e-05,
+      "loss": 0.0488,
+      "step": 2466
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.975003125260028e-05,
+      "loss": 0.5352,
+      "step": 2467
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9749829126853603e-05,
+      "loss": 0.0967,
+      "step": 2468
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9749626920107288e-05,
+      "loss": 0.1836,
+      "step": 2469
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9749424632362454e-05,
+      "loss": 0.209,
+      "step": 2470
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9749222263620204e-05,
+      "loss": 2.1094,
+      "step": 2471
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9749019813881644e-05,
+      "loss": 0.2188,
+      "step": 2472
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.97488172831479e-05,
+      "loss": 0.0854,
+      "step": 2473
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9748614671420076e-05,
+      "loss": 0.2871,
+      "step": 2474
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 2.9748411978699284e-05,
+      "loss": 0.0006,
+      "step": 2475
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9748209204986638e-05,
+      "loss": 2.8906,
+      "step": 2476
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9748006350283252e-05,
+      "loss": 0.4844,
+      "step": 2477
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9747803414590242e-05,
+      "loss": 0.0002,
+      "step": 2478
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.974760039790872e-05,
+      "loss": 0.1973,
+      "step": 2479
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9747397300239798e-05,
+      "loss": 0.8555,
+      "step": 2480
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9747194121584602e-05,
+      "loss": 0.0067,
+      "step": 2481
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9746990861944238e-05,
+      "loss": 0.0017,
+      "step": 2482
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.974678752131982e-05,
+      "loss": 0.5,
+      "step": 2483
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.974658409971247e-05,
+      "loss": 0.5391,
+      "step": 2484
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9746380597123312e-05,
+      "loss": 0.3477,
+      "step": 2485
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.974617701355345e-05,
+      "loss": 0.0003,
+      "step": 2486
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.974597334900401e-05,
+      "loss": 0.0006,
+      "step": 2487
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.974576960347611e-05,
+      "loss": 0.3418,
+      "step": 2488
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9745565776970864e-05,
+      "loss": 0.0002,
+      "step": 2489
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9745361869489395e-05,
+      "loss": 0.0005,
+      "step": 2490
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.974515788103283e-05,
+      "loss": 0.0004,
+      "step": 2491
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9744953811602273e-05,
+      "loss": 0.0055,
+      "step": 2492
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9744749661198856e-05,
+      "loss": 0.1572,
+      "step": 2493
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9744545429823697e-05,
+      "loss": 0.0762,
+      "step": 2494
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.974434111747792e-05,
+      "loss": 0.0037,
+      "step": 2495
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9744136724162647e-05,
+      "loss": 0.6797,
+      "step": 2496
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9743932249878994e-05,
+      "loss": 0.1279,
+      "step": 2497
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.974372769462809e-05,
+      "loss": 0.166,
+      "step": 2498
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9743523058411057e-05,
+      "loss": 0.1689,
+      "step": 2499
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.974331834122902e-05,
+      "loss": 0.1826,
+      "step": 2500
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9743113543083098e-05,
+      "loss": 1.3828,
+      "step": 2501
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.974290866397442e-05,
+      "loss": 0.4902,
+      "step": 2502
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9742703703904112e-05,
+      "loss": 1.2578,
+      "step": 2503
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9742498662873297e-05,
+      "loss": 0.0403,
+      "step": 2504
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9742293540883103e-05,
+      "loss": 0.793,
+      "step": 2505
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9742088337934655e-05,
+      "loss": 1.5078,
+      "step": 2506
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.974188305402908e-05,
+      "loss": 0.5312,
+      "step": 2507
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9741677689167508e-05,
+      "loss": 0.582,
+      "step": 2508
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9741472243351068e-05,
+      "loss": 2.0312,
+      "step": 2509
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9741266716580878e-05,
+      "loss": 1.5,
+      "step": 2510
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9741061108858075e-05,
+      "loss": 2.1875,
+      "step": 2511
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9740855420183784e-05,
+      "loss": 0.167,
+      "step": 2512
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.974064965055914e-05,
+      "loss": 0.2617,
+      "step": 2513
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.974044379998527e-05,
+      "loss": 1.9219,
+      "step": 2514
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9740237868463302e-05,
+      "loss": 0.0008,
+      "step": 2515
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.974003185599437e-05,
+      "loss": 1.1016,
+      "step": 2516
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9739825762579606e-05,
+      "loss": 0.0037,
+      "step": 2517
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9739619588220144e-05,
+      "loss": 0.0002,
+      "step": 2518
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.973941333291711e-05,
+      "loss": 0.6133,
+      "step": 2519
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9739206996671635e-05,
+      "loss": 0.8359,
+      "step": 2520
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9739000579484863e-05,
+      "loss": 0.4707,
+      "step": 2521
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9738794081357915e-05,
+      "loss": 0.0297,
+      "step": 2522
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.973858750229193e-05,
+      "loss": 0.0011,
+      "step": 2523
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9738380842288053e-05,
+      "loss": 0.001,
+      "step": 2524
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9738174101347402e-05,
+      "loss": 0.1221,
+      "step": 2525
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.973796727947112e-05,
+      "loss": 0.21,
+      "step": 2526
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9737760376660345e-05,
+      "loss": 0.1709,
+      "step": 2527
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.973755339291621e-05,
+      "loss": 0.1328,
+      "step": 2528
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.973734632823985e-05,
+      "loss": 0.021,
+      "step": 2529
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9737139182632406e-05,
+      "loss": 0.1709,
+      "step": 2530
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9736931956095014e-05,
+      "loss": 0.3281,
+      "step": 2531
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9736724648628813e-05,
+      "loss": 0.0003,
+      "step": 2532
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9736517260234937e-05,
+      "loss": 0.0001,
+      "step": 2533
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.973630979091453e-05,
+      "loss": 0.0287,
+      "step": 2534
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9736102240668732e-05,
+      "loss": 0.1387,
+      "step": 2535
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9735894609498676e-05,
+      "loss": 1.9297,
+      "step": 2536
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.973568689740551e-05,
+      "loss": 0.0001,
+      "step": 2537
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.973547910439037e-05,
+      "loss": 0.2852,
+      "step": 2538
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.97352712304544e-05,
+      "loss": 0.0591,
+      "step": 2539
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.973506327559874e-05,
+      "loss": 0.0232,
+      "step": 2540
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.973485523982453e-05,
+      "loss": 0.053,
+      "step": 2541
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9734647123132915e-05,
+      "loss": 0.0003,
+      "step": 2542
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 2.9734438925525036e-05,
+      "loss": 0.0001,
+      "step": 2543
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.973423064700204e-05,
+      "loss": 0.0002,
+      "step": 2544
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.9734022287565067e-05,
+      "loss": 0.0002,
+      "step": 2545
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.9733813847215265e-05,
+      "loss": 0.0003,
+      "step": 2546
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.9733605325953772e-05,
+      "loss": 0.1465,
+      "step": 2547
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.973339672378174e-05,
+      "loss": 0.1348,
+      "step": 2548
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.973318804070031e-05,
+      "loss": 0.0869,
+      "step": 2549
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.973297927671063e-05,
+      "loss": 0.1035,
+      "step": 2550
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.9732770431813848e-05,
+      "loss": 0.0327,
+      "step": 2551
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.973256150601111e-05,
+      "loss": 1.1562,
+      "step": 2552
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.9732352499303563e-05,
+      "loss": 1.1641,
+      "step": 2553
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.973214341169235e-05,
+      "loss": 1.6641,
+      "step": 2554
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.9731934243178628e-05,
+      "loss": 0.6719,
+      "step": 2555
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.973172499376354e-05,
+      "loss": 0.5078,
+      "step": 2556
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.9731515663448235e-05,
+      "loss": 0.6016,
+      "step": 2557
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.9731306252233863e-05,
+      "loss": 1.2109,
+      "step": 2558
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.973109676012158e-05,
+      "loss": 1.1406,
+      "step": 2559
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.9730887187112527e-05,
+      "loss": 0.6562,
+      "step": 2560
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.973067753320786e-05,
+      "loss": 1.6797,
+      "step": 2561
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.973046779840873e-05,
+      "loss": 1.2891,
+      "step": 2562
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.9730257982716287e-05,
+      "loss": 1.6719,
+      "step": 2563
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.9730048086131687e-05,
+      "loss": 0.0608,
+      "step": 2564
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.972983810865608e-05,
+      "loss": 0.7812,
+      "step": 2565
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.9729628050290617e-05,
+      "loss": 0.7383,
+      "step": 2566
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.9729417911036454e-05,
+      "loss": 0.0014,
+      "step": 2567
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.9729207690894745e-05,
+      "loss": 0.3789,
+      "step": 2568
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.9728997389866648e-05,
+      "loss": 0.334,
+      "step": 2569
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.972878700795331e-05,
+      "loss": 0.0008,
+      "step": 2570
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.972857654515589e-05,
+      "loss": 0.1157,
+      "step": 2571
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.9728366001475543e-05,
+      "loss": 1.1484,
+      "step": 2572
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.9728155376913434e-05,
+      "loss": 0.4453,
+      "step": 2573
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.9727944671470706e-05,
+      "loss": 0.0001,
+      "step": 2574
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.9727733885148522e-05,
+      "loss": 0.6094,
+      "step": 2575
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.972752301794804e-05,
+      "loss": 0.0408,
+      "step": 2576
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.972731206987042e-05,
+      "loss": 0.0105,
+      "step": 2577
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.972710104091682e-05,
+      "loss": 0.5898,
+      "step": 2578
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.9726889931088392e-05,
+      "loss": 0.707,
+      "step": 2579
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.9726678740386306e-05,
+      "loss": 1.7031,
+      "step": 2580
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.972646746881171e-05,
+      "loss": 1.6562,
+      "step": 2581
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.9726256116365772e-05,
+      "loss": 0.0001,
+      "step": 2582
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.9726044683049656e-05,
+      "loss": 0.7266,
+      "step": 2583
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.972583316886451e-05,
+      "loss": 0.3652,
+      "step": 2584
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.9725621573811512e-05,
+      "loss": 0.0398,
+      "step": 2585
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.9725409897891814e-05,
+      "loss": 0.0097,
+      "step": 2586
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.9725198141106574e-05,
+      "loss": 0.0583,
+      "step": 2587
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.9724986303456966e-05,
+      "loss": 0.0008,
+      "step": 2588
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.9724774384944148e-05,
+      "loss": 0.0002,
+      "step": 2589
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.9724562385569282e-05,
+      "loss": 0.3301,
+      "step": 2590
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.9724350305333534e-05,
+      "loss": 0.1157,
+      "step": 2591
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.9724138144238072e-05,
+      "loss": 0.0001,
+      "step": 2592
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.9723925902284056e-05,
+      "loss": 0.7031,
+      "step": 2593
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.9723713579472654e-05,
+      "loss": 0.0002,
+      "step": 2594
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.9723501175805033e-05,
+      "loss": 0.1328,
+      "step": 2595
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.9723288691282353e-05,
+      "loss": 1.5547,
+      "step": 2596
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.972307612590579e-05,
+      "loss": 0.2148,
+      "step": 2597
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.9722863479676505e-05,
+      "loss": 0.1128,
+      "step": 2598
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.9722650752595668e-05,
+      "loss": 0.0938,
+      "step": 2599
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.972243794466445e-05,
+      "loss": 0.6758,
+      "step": 2600
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.9722225055884015e-05,
+      "loss": 1.4219,
+      "step": 2601
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.9722012086255534e-05,
+      "loss": 1.4609,
+      "step": 2602
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.9721799035780177e-05,
+      "loss": 1.3672,
+      "step": 2603
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.9721585904459114e-05,
+      "loss": 1.1719,
+      "step": 2604
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.9721372692293512e-05,
+      "loss": 0.3281,
+      "step": 2605
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.9721159399284545e-05,
+      "loss": 0.4844,
+      "step": 2606
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.9720946025433385e-05,
+      "loss": 0.7383,
+      "step": 2607
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.9720732570741204e-05,
+      "loss": 0.8633,
+      "step": 2608
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.972051903520917e-05,
+      "loss": 0.5938,
+      "step": 2609
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.972030541883846e-05,
+      "loss": 0.6992,
+      "step": 2610
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 2.972009172163025e-05,
+      "loss": 0.3965,
+      "step": 2611
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.97198779435857e-05,
+      "loss": 1.3984,
+      "step": 2612
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.9719664084706e-05,
+      "loss": 0.416,
+      "step": 2613
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.9719450144992317e-05,
+      "loss": 0.4336,
+      "step": 2614
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.9719236124445827e-05,
+      "loss": 0.0869,
+      "step": 2615
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.97190220230677e-05,
+      "loss": 0.2773,
+      "step": 2616
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.971880784085912e-05,
+      "loss": 0.918,
+      "step": 2617
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.971859357782126e-05,
+      "loss": 0.8828,
+      "step": 2618
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.9718379233955297e-05,
+      "loss": 0.4531,
+      "step": 2619
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.9718164809262407e-05,
+      "loss": 0.5469,
+      "step": 2620
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.9717950303743764e-05,
+      "loss": 0.252,
+      "step": 2621
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.9717735717400553e-05,
+      "loss": 0.7734,
+      "step": 2622
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.9717521050233952e-05,
+      "loss": 0.0125,
+      "step": 2623
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.9717306302245134e-05,
+      "loss": 0.2812,
+      "step": 2624
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.9717091473435282e-05,
+      "loss": 0.0547,
+      "step": 2625
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.971687656380557e-05,
+      "loss": 0.0007,
+      "step": 2626
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.9716661573357194e-05,
+      "loss": 0.0006,
+      "step": 2627
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.971644650209132e-05,
+      "loss": 0.0004,
+      "step": 2628
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.9716231350009128e-05,
+      "loss": 0.0068,
+      "step": 2629
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.971601611711181e-05,
+      "loss": 0.1235,
+      "step": 2630
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.9715800803400543e-05,
+      "loss": 0.0011,
+      "step": 2631
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.9715585408876508e-05,
+      "loss": 0.0007,
+      "step": 2632
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.9715369933540885e-05,
+      "loss": 0.0008,
+      "step": 2633
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.9715154377394866e-05,
+      "loss": 0.8594,
+      "step": 2634
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.9714938740439627e-05,
+      "loss": 0.0962,
+      "step": 2635
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.9714723022676354e-05,
+      "loss": 0.2139,
+      "step": 2636
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.971450722410624e-05,
+      "loss": 0.2451,
+      "step": 2637
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.9714291344730456e-05,
+      "loss": 0.0003,
+      "step": 2638
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.97140753845502e-05,
+      "loss": 0.0013,
+      "step": 2639
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.9713859343566645e-05,
+      "loss": 0.0262,
+      "step": 2640
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.9713643221780992e-05,
+      "loss": 0.0009,
+      "step": 2641
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.9713427019194413e-05,
+      "loss": 0.0022,
+      "step": 2642
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.971321073580811e-05,
+      "loss": 0.1885,
+      "step": 2643
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.9712994371623256e-05,
+      "loss": 0.0011,
+      "step": 2644
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.9712777926641052e-05,
+      "loss": 0.1738,
+      "step": 2645
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.9712561400862683e-05,
+      "loss": 0.167,
+      "step": 2646
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.971234479428933e-05,
+      "loss": 0.2109,
+      "step": 2647
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.9712128106922195e-05,
+      "loss": 0.2236,
+      "step": 2648
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.971191133876246e-05,
+      "loss": 0.1089,
+      "step": 2649
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.971169448981132e-05,
+      "loss": 0.1729,
+      "step": 2650
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.9711477560069954e-05,
+      "loss": 0.8672,
+      "step": 2651
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.971126054953957e-05,
+      "loss": 0.1309,
+      "step": 2652
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.9711043458221353e-05,
+      "loss": 0.7422,
+      "step": 2653
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.9710826286116492e-05,
+      "loss": 1.5078,
+      "step": 2654
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.971060903322618e-05,
+      "loss": 0.0251,
+      "step": 2655
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.9710391699551614e-05,
+      "loss": 0.3633,
+      "step": 2656
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.9710174285093985e-05,
+      "loss": 2.2188,
+      "step": 2657
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.970995678985449e-05,
+      "loss": 0.8125,
+      "step": 2658
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.9709739213834316e-05,
+      "loss": 0.332,
+      "step": 2659
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.9709521557034668e-05,
+      "loss": 0.5781,
+      "step": 2660
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.970930381945673e-05,
+      "loss": 0.8398,
+      "step": 2661
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.970908600110171e-05,
+      "loss": 0.2373,
+      "step": 2662
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.9708868101970795e-05,
+      "loss": 0.4102,
+      "step": 2663
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.970865012206518e-05,
+      "loss": 0.0195,
+      "step": 2664
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.9708432061386076e-05,
+      "loss": 1.7969,
+      "step": 2665
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.970821391993466e-05,
+      "loss": 0.7031,
+      "step": 2666
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.970799569771215e-05,
+      "loss": 2.375,
+      "step": 2667
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.970777739471973e-05,
+      "loss": 0.1289,
+      "step": 2668
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.970755901095861e-05,
+      "loss": 0.3203,
+      "step": 2669
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.9707340546429976e-05,
+      "loss": 0.009,
+      "step": 2670
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.970712200113504e-05,
+      "loss": 1.0,
+      "step": 2671
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.9706903375074995e-05,
+      "loss": 0.0082,
+      "step": 2672
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.9706684668251042e-05,
+      "loss": 0.75,
+      "step": 2673
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.9706465880664386e-05,
+      "loss": 0.2109,
+      "step": 2674
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.9706247012316227e-05,
+      "loss": 0.0005,
+      "step": 2675
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.9706028063207767e-05,
+      "loss": 0.0532,
+      "step": 2676
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.9705809033340207e-05,
+      "loss": 0.0006,
+      "step": 2677
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.970558992271475e-05,
+      "loss": 0.6875,
+      "step": 2678
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 2.97053707313326e-05,
+      "loss": 0.7344,
+      "step": 2679
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.970515145919496e-05,
+      "loss": 1.8047,
+      "step": 2680
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9704932106303036e-05,
+      "loss": 0.6914,
+      "step": 2681
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9704712672658032e-05,
+      "loss": 0.0058,
+      "step": 2682
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9704493158261148e-05,
+      "loss": 0.1245,
+      "step": 2683
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9704273563113594e-05,
+      "loss": 0.0011,
+      "step": 2684
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9704053887216584e-05,
+      "loss": 0.2637,
+      "step": 2685
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.970383413057131e-05,
+      "loss": 0.0012,
+      "step": 2686
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.970361429317898e-05,
+      "loss": 0.6445,
+      "step": 2687
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.970339437504081e-05,
+      "loss": 0.0055,
+      "step": 2688
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9703174376158007e-05,
+      "loss": 0.0206,
+      "step": 2689
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9702954296531772e-05,
+      "loss": 0.0006,
+      "step": 2690
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.970273413616332e-05,
+      "loss": 0.0004,
+      "step": 2691
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9702513895053855e-05,
+      "loss": 0.0011,
+      "step": 2692
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.970229357320459e-05,
+      "loss": 0.3438,
+      "step": 2693
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9702073170616733e-05,
+      "loss": 0.0806,
+      "step": 2694
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9701852687291495e-05,
+      "loss": 0.0004,
+      "step": 2695
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.970163212323009e-05,
+      "loss": 0.0869,
+      "step": 2696
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9701411478433726e-05,
+      "loss": 0.1846,
+      "step": 2697
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9701190752903615e-05,
+      "loss": 0.0737,
+      "step": 2698
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9700969946640967e-05,
+      "loss": 0.1338,
+      "step": 2699
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9700749059646995e-05,
+      "loss": 0.1895,
+      "step": 2700
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9700528091922917e-05,
+      "loss": 2.4375,
+      "step": 2701
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9700307043469944e-05,
+      "loss": 0.8047,
+      "step": 2702
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9700085914289284e-05,
+      "loss": 1.4688,
+      "step": 2703
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9699864704382158e-05,
+      "loss": 0.0135,
+      "step": 2704
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9699643413749784e-05,
+      "loss": 1.8359,
+      "step": 2705
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9699422042393368e-05,
+      "loss": 1.1953,
+      "step": 2706
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9699200590314132e-05,
+      "loss": 0.8281,
+      "step": 2707
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.969897905751329e-05,
+      "loss": 0.8867,
+      "step": 2708
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9698757443992055e-05,
+      "loss": 0.8398,
+      "step": 2709
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.969853574975165e-05,
+      "loss": 0.0869,
+      "step": 2710
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.969831397479329e-05,
+      "loss": 0.707,
+      "step": 2711
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9698092119118193e-05,
+      "loss": 0.1787,
+      "step": 2712
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9697870182727582e-05,
+      "loss": 1.3047,
+      "step": 2713
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.969764816562267e-05,
+      "loss": 0.3711,
+      "step": 2714
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.969742606780467e-05,
+      "loss": 2.6875,
+      "step": 2715
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9697203889274818e-05,
+      "loss": 1.6406,
+      "step": 2716
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.969698163003432e-05,
+      "loss": 0.106,
+      "step": 2717
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.96967592900844e-05,
+      "loss": 0.0007,
+      "step": 2718
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9696536869426287e-05,
+      "loss": 2.2188,
+      "step": 2719
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.969631436806119e-05,
+      "loss": 0.3027,
+      "step": 2720
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9696091785990343e-05,
+      "loss": 0.6367,
+      "step": 2721
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.969586912321496e-05,
+      "loss": 0.0781,
+      "step": 2722
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9695646379736264e-05,
+      "loss": 0.1611,
+      "step": 2723
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9695423555555483e-05,
+      "loss": 0.0317,
+      "step": 2724
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9695200650673837e-05,
+      "loss": 0.4355,
+      "step": 2725
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.969497766509255e-05,
+      "loss": 0.8359,
+      "step": 2726
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.969475459881285e-05,
+      "loss": 0.083,
+      "step": 2727
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.969453145183596e-05,
+      "loss": 0.0535,
+      "step": 2728
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9694308224163102e-05,
+      "loss": 0.3574,
+      "step": 2729
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9694084915795506e-05,
+      "loss": 0.3945,
+      "step": 2730
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9693861526734398e-05,
+      "loss": 0.0004,
+      "step": 2731
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9693638056981005e-05,
+      "loss": 0.0012,
+      "step": 2732
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.969341450653655e-05,
+      "loss": 0.0008,
+      "step": 2733
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9693190875402264e-05,
+      "loss": 0.0364,
+      "step": 2734
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9692967163579377e-05,
+      "loss": 0.0005,
+      "step": 2735
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9692743371069114e-05,
+      "loss": 0.5547,
+      "step": 2736
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9692519497872708e-05,
+      "loss": 0.2676,
+      "step": 2737
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9692295543991386e-05,
+      "loss": 0.0391,
+      "step": 2738
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9692071509426374e-05,
+      "loss": 0.1826,
+      "step": 2739
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.969184739417891e-05,
+      "loss": 0.0028,
+      "step": 2740
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9691623198250218e-05,
+      "loss": 0.4785,
+      "step": 2741
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9691398921641533e-05,
+      "loss": 0.0001,
+      "step": 2742
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9691174564354082e-05,
+      "loss": 0.0001,
+      "step": 2743
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.9690950126389107e-05,
+      "loss": 0.0009,
+      "step": 2744
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.969072560774783e-05,
+      "loss": 0.0004,
+      "step": 2745
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.969050100843149e-05,
+      "loss": 0.0037,
+      "step": 2746
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 2.969027632844132e-05,
+      "loss": 0.3125,
+      "step": 2747
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9690051567778546e-05,
+      "loss": 0.1816,
+      "step": 2748
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.968982672644441e-05,
+      "loss": 0.2061,
+      "step": 2749
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.968960180444015e-05,
+      "loss": 0.1729,
+      "step": 2750
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9689376801766997e-05,
+      "loss": 1.0859,
+      "step": 2751
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.968915171842618e-05,
+      "loss": 0.8242,
+      "step": 2752
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9688926554418945e-05,
+      "loss": 0.1621,
+      "step": 2753
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9688701309746524e-05,
+      "loss": 0.2031,
+      "step": 2754
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9688475984410155e-05,
+      "loss": 0.0393,
+      "step": 2755
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9688250578411076e-05,
+      "loss": 0.3926,
+      "step": 2756
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9688025091750527e-05,
+      "loss": 0.5117,
+      "step": 2757
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9687799524429737e-05,
+      "loss": 0.5352,
+      "step": 2758
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.968757387644995e-05,
+      "loss": 0.8867,
+      "step": 2759
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.968734814781241e-05,
+      "loss": 0.9141,
+      "step": 2760
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.968712233851835e-05,
+      "loss": 0.0079,
+      "step": 2761
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.968689644856902e-05,
+      "loss": 0.0393,
+      "step": 2762
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.968667047796564e-05,
+      "loss": 0.4609,
+      "step": 2763
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9686444426709477e-05,
+      "loss": 0.1396,
+      "step": 2764
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9686218294801753e-05,
+      "loss": 0.0156,
+      "step": 2765
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.968599208224372e-05,
+      "loss": 1.0391,
+      "step": 2766
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9685765789036612e-05,
+      "loss": 0.2891,
+      "step": 2767
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.968553941518168e-05,
+      "loss": 0.3789,
+      "step": 2768
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.968531296068016e-05,
+      "loss": 0.083,
+      "step": 2769
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.96850864255333e-05,
+      "loss": 0.0004,
+      "step": 2770
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9684859809742348e-05,
+      "loss": 0.0583,
+      "step": 2771
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9684633113308533e-05,
+      "loss": 1.4141,
+      "step": 2772
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.968440633623312e-05,
+      "loss": 0.0006,
+      "step": 2773
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.968417947851734e-05,
+      "loss": 0.064,
+      "step": 2774
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9683952540162446e-05,
+      "loss": 0.5,
+      "step": 2775
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9683725521169683e-05,
+      "loss": 0.6172,
+      "step": 2776
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.968349842154029e-05,
+      "loss": 0.3418,
+      "step": 2777
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.968327124127553e-05,
+      "loss": 1.4844,
+      "step": 2778
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9683043980376632e-05,
+      "loss": 0.0003,
+      "step": 2779
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9682816638844856e-05,
+      "loss": 0.0031,
+      "step": 2780
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9682589216681452e-05,
+      "loss": 0.1221,
+      "step": 2781
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.968236171388766e-05,
+      "loss": 0.0025,
+      "step": 2782
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.968213413046474e-05,
+      "loss": 0.1455,
+      "step": 2783
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.968190646641393e-05,
+      "loss": 0.0003,
+      "step": 2784
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9681678721736487e-05,
+      "loss": 0.0002,
+      "step": 2785
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.968145089643366e-05,
+      "loss": 0.0011,
+      "step": 2786
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.96812229905067e-05,
+      "loss": 0.4883,
+      "step": 2787
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9680995003956862e-05,
+      "loss": 0.0002,
+      "step": 2788
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9680766936785394e-05,
+      "loss": 0.0005,
+      "step": 2789
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9680538788993554e-05,
+      "loss": 0.3535,
+      "step": 2790
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9680310560582588e-05,
+      "loss": 0.1631,
+      "step": 2791
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.968008225155375e-05,
+      "loss": 0.5508,
+      "step": 2792
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9679853861908294e-05,
+      "loss": 0.0002,
+      "step": 2793
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9679625391647477e-05,
+      "loss": 0.2969,
+      "step": 2794
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.967939684077256e-05,
+      "loss": 0.0002,
+      "step": 2795
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9679168209284783e-05,
+      "loss": 0.1572,
+      "step": 2796
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.967893949718541e-05,
+      "loss": 0.1533,
+      "step": 2797
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.96787107044757e-05,
+      "loss": 0.1602,
+      "step": 2798
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9678481831156902e-05,
+      "loss": 0.1514,
+      "step": 2799
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9678252877230277e-05,
+      "loss": 0.2168,
+      "step": 2800
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9678023842697086e-05,
+      "loss": 1.4844,
+      "step": 2801
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9677794727558577e-05,
+      "loss": 1.3359,
+      "step": 2802
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.967756553181602e-05,
+      "loss": 0.249,
+      "step": 2803
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9677336255470663e-05,
+      "loss": 0.6953,
+      "step": 2804
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.967710689852377e-05,
+      "loss": 0.6641,
+      "step": 2805
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.96768774609766e-05,
+      "loss": 1.8594,
+      "step": 2806
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9676647942830414e-05,
+      "loss": 0.0349,
+      "step": 2807
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.967641834408647e-05,
+      "loss": 0.1162,
+      "step": 2808
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9676188664746032e-05,
+      "loss": 0.6797,
+      "step": 2809
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.967595890481036e-05,
+      "loss": 0.4688,
+      "step": 2810
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9675729064280718e-05,
+      "loss": 0.416,
+      "step": 2811
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9675499143158358e-05,
+      "loss": 0.3457,
+      "step": 2812
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9675269141444553e-05,
+      "loss": 0.8047,
+      "step": 2813
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 2.9675039059140565e-05,
+      "loss": 1.5312,
+      "step": 2814
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9674808896247656e-05,
+      "loss": 0.6836,
+      "step": 2815
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.967457865276709e-05,
+      "loss": 2.1875,
+      "step": 2816
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.967434832870013e-05,
+      "loss": 0.7422,
+      "step": 2817
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9674117924048043e-05,
+      "loss": 1.1094,
+      "step": 2818
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.967388743881209e-05,
+      "loss": 0.0007,
+      "step": 2819
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9673656872993545e-05,
+      "loss": 0.007,
+      "step": 2820
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.967342622659367e-05,
+      "loss": 1.4609,
+      "step": 2821
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9673195499613726e-05,
+      "loss": 0.0014,
+      "step": 2822
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9672964692054985e-05,
+      "loss": 0.0325,
+      "step": 2823
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9672733803918714e-05,
+      "loss": 0.3965,
+      "step": 2824
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.967250283520618e-05,
+      "loss": 0.0053,
+      "step": 2825
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.967227178591866e-05,
+      "loss": 0.6133,
+      "step": 2826
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9672040656057412e-05,
+      "loss": 0.6445,
+      "step": 2827
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9671809445623706e-05,
+      "loss": 0.0262,
+      "step": 2828
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9671578154618816e-05,
+      "loss": 0.0006,
+      "step": 2829
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.967134678304401e-05,
+      "loss": 0.0007,
+      "step": 2830
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.967111533090056e-05,
+      "loss": 0.5781,
+      "step": 2831
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9670883798189735e-05,
+      "loss": 0.2314,
+      "step": 2832
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.967065218491281e-05,
+      "loss": 0.6016,
+      "step": 2833
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9670420491071052e-05,
+      "loss": 0.0001,
+      "step": 2834
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9670188716665737e-05,
+      "loss": 0.4883,
+      "step": 2835
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.966995686169814e-05,
+      "loss": 0.1455,
+      "step": 2836
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9669724926169532e-05,
+      "loss": 0.0031,
+      "step": 2837
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.966949291008118e-05,
+      "loss": 0.0002,
+      "step": 2838
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9669260813434363e-05,
+      "loss": 0.0012,
+      "step": 2839
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9669028636230357e-05,
+      "loss": 0.0013,
+      "step": 2840
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.966879637847044e-05,
+      "loss": 0.0012,
+      "step": 2841
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.966856404015588e-05,
+      "loss": 0.0059,
+      "step": 2842
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.966833162128796e-05,
+      "loss": 0.0003,
+      "step": 2843
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9668099121867955e-05,
+      "loss": 0.9727,
+      "step": 2844
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9667866541897136e-05,
+      "loss": 0.0002,
+      "step": 2845
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9667633881376786e-05,
+      "loss": 0.0006,
+      "step": 2846
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9667401140308178e-05,
+      "loss": 0.0001,
+      "step": 2847
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9667168318692594e-05,
+      "loss": 0.1748,
+      "step": 2848
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9666935416531316e-05,
+      "loss": 0.1738,
+      "step": 2849
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9666702433825614e-05,
+      "loss": 0.2051,
+      "step": 2850
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.966646937057677e-05,
+      "loss": 0.5195,
+      "step": 2851
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.966623622678607e-05,
+      "loss": 0.3008,
+      "step": 2852
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9666003002454788e-05,
+      "loss": 1.5156,
+      "step": 2853
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9665769697584207e-05,
+      "loss": 0.3613,
+      "step": 2854
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9665536312175606e-05,
+      "loss": 1.2031,
+      "step": 2855
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9665302846230276e-05,
+      "loss": 0.418,
+      "step": 2856
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9665069299749483e-05,
+      "loss": 1.4453,
+      "step": 2857
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9664835672734522e-05,
+      "loss": 0.6172,
+      "step": 2858
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9664601965186674e-05,
+      "loss": 0.793,
+      "step": 2859
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.966436817710722e-05,
+      "loss": 0.1631,
+      "step": 2860
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9664134308497445e-05,
+      "loss": 0.5508,
+      "step": 2861
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9663900359358633e-05,
+      "loss": 0.1611,
+      "step": 2862
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9663666329692066e-05,
+      "loss": 2.1406,
+      "step": 2863
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9663432219499036e-05,
+      "loss": 0.052,
+      "step": 2864
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9663198028780823e-05,
+      "loss": 0.2754,
+      "step": 2865
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.966296375753871e-05,
+      "loss": 2.0938,
+      "step": 2866
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9662729405773992e-05,
+      "loss": 0.5586,
+      "step": 2867
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9662494973487956e-05,
+      "loss": 0.2734,
+      "step": 2868
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.966226046068188e-05,
+      "loss": 0.1621,
+      "step": 2869
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9662025867357056e-05,
+      "loss": 0.8984,
+      "step": 2870
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9661791193514777e-05,
+      "loss": 0.0082,
+      "step": 2871
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9661556439156328e-05,
+      "loss": 0.0101,
+      "step": 2872
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9661321604282996e-05,
+      "loss": 0.0004,
+      "step": 2873
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9661086688896075e-05,
+      "loss": 0.6172,
+      "step": 2874
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9660851692996853e-05,
+      "loss": 0.3926,
+      "step": 2875
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.966061661658662e-05,
+      "loss": 0.1484,
+      "step": 2876
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9660381459666674e-05,
+      "loss": 0.0693,
+      "step": 2877
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9660146222238287e-05,
+      "loss": 0.0001,
+      "step": 2878
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.965991090430277e-05,
+      "loss": 0.0835,
+      "step": 2879
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9659675505861413e-05,
+      "loss": 0.3418,
+      "step": 2880
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9659440026915504e-05,
+      "loss": 0.0002,
+      "step": 2881
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 2.9659204467466333e-05,
+      "loss": 0.0918,
+      "step": 2882
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.9658968827515198e-05,
+      "loss": 0.0011,
+      "step": 2883
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.9658733107063394e-05,
+      "loss": 0.0007,
+      "step": 2884
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.9658497306112214e-05,
+      "loss": 0.0017,
+      "step": 2885
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.9658261424662953e-05,
+      "loss": 0.0133,
+      "step": 2886
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.9658025462716906e-05,
+      "loss": 0.0005,
+      "step": 2887
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.965778942027537e-05,
+      "loss": 0.0004,
+      "step": 2888
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.9657553297339637e-05,
+      "loss": 0.0003,
+      "step": 2889
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.965731709391101e-05,
+      "loss": 0.0184,
+      "step": 2890
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.965708080999078e-05,
+      "loss": 0.0003,
+      "step": 2891
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.965684444558025e-05,
+      "loss": 0.0001,
+      "step": 2892
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.9656608000680715e-05,
+      "loss": 0.127,
+      "step": 2893
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.9656371475293477e-05,
+      "loss": 0.0217,
+      "step": 2894
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.9656134869419827e-05,
+      "loss": 0.0003,
+      "step": 2895
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.9655898183061072e-05,
+      "loss": 0.0005,
+      "step": 2896
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.965566141621851e-05,
+      "loss": 0.0003,
+      "step": 2897
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.9655424568893443e-05,
+      "loss": 0.1069,
+      "step": 2898
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.965518764108716e-05,
+      "loss": 0.1191,
+      "step": 2899
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.9654950632800978e-05,
+      "loss": 0.1514,
+      "step": 2900
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.965471354403619e-05,
+      "loss": 1.3828,
+      "step": 2901
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.96544763747941e-05,
+      "loss": 0.3418,
+      "step": 2902
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.9654239125076014e-05,
+      "loss": 0.7812,
+      "step": 2903
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.9654001794883223e-05,
+      "loss": 1.6719,
+      "step": 2904
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.965376438421704e-05,
+      "loss": 0.8281,
+      "step": 2905
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.9653526893078774e-05,
+      "loss": 0.3926,
+      "step": 2906
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.9653289321469715e-05,
+      "loss": 0.6328,
+      "step": 2907
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.9653051669391177e-05,
+      "loss": 0.6797,
+      "step": 2908
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.965281393684446e-05,
+      "loss": 0.1172,
+      "step": 2909
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.965257612383088e-05,
+      "loss": 1.1641,
+      "step": 2910
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.9652338230351733e-05,
+      "loss": 0.0172,
+      "step": 2911
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.9652100256408325e-05,
+      "loss": 0.0688,
+      "step": 2912
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.9651862202001968e-05,
+      "loss": 0.2402,
+      "step": 2913
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.9651624067133963e-05,
+      "loss": 0.0184,
+      "step": 2914
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.9651385851805627e-05,
+      "loss": 0.6602,
+      "step": 2915
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.9651147556018258e-05,
+      "loss": 0.7383,
+      "step": 2916
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.9650909179773174e-05,
+      "loss": 0.2031,
+      "step": 2917
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.9650670723071676e-05,
+      "loss": 0.7773,
+      "step": 2918
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.965043218591508e-05,
+      "loss": 0.0043,
+      "step": 2919
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.965019356830469e-05,
+      "loss": 1.0312,
+      "step": 2920
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.9649954870241823e-05,
+      "loss": 0.625,
+      "step": 2921
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.9649716091727782e-05,
+      "loss": 1.6328,
+      "step": 2922
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.9649477232763887e-05,
+      "loss": 0.4883,
+      "step": 2923
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.964923829335144e-05,
+      "loss": 0.0035,
+      "step": 2924
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.9648999273491764e-05,
+      "loss": 0.001,
+      "step": 2925
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.9648760173186163e-05,
+      "loss": 0.6523,
+      "step": 2926
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.964852099243595e-05,
+      "loss": 0.1396,
+      "step": 2927
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.964828173124245e-05,
+      "loss": 0.7344,
+      "step": 2928
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.9648042389606956e-05,
+      "loss": 0.001,
+      "step": 2929
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.9647802967530807e-05,
+      "loss": 0.1016,
+      "step": 2930
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.96475634650153e-05,
+      "loss": 0.0004,
+      "step": 2931
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.9647323882061757e-05,
+      "loss": 0.9961,
+      "step": 2932
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.964708421867149e-05,
+      "loss": 0.0006,
+      "step": 2933
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.964684447484582e-05,
+      "loss": 0.1475,
+      "step": 2934
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.964660465058606e-05,
+      "loss": 0.0003,
+      "step": 2935
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.964636474589353e-05,
+      "loss": 0.3496,
+      "step": 2936
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.9646124760769543e-05,
+      "loss": 0.6641,
+      "step": 2937
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.964588469521542e-05,
+      "loss": 0.0004,
+      "step": 2938
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.964564454923248e-05,
+      "loss": 0.0003,
+      "step": 2939
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.9645404322822037e-05,
+      "loss": 0.0015,
+      "step": 2940
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.964516401598542e-05,
+      "loss": 0.0562,
+      "step": 2941
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.9644923628723938e-05,
+      "loss": 0.0369,
+      "step": 2942
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.9644683161038918e-05,
+      "loss": 0.0171,
+      "step": 2943
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.9644442612931677e-05,
+      "loss": 0.1504,
+      "step": 2944
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.9644201984403537e-05,
+      "loss": 0.0002,
+      "step": 2945
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.964396127545582e-05,
+      "loss": 0.0269,
+      "step": 2946
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.9643720486089847e-05,
+      "loss": 0.0002,
+      "step": 2947
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.964347961630694e-05,
+      "loss": 0.0005,
+      "step": 2948
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.9643238666108422e-05,
+      "loss": 0.0005,
+      "step": 2949
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 2.9642997635495622e-05,
+      "loss": 0.0001,
+      "step": 2950
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.964275652446985e-05,
+      "loss": 0.5742,
+      "step": 2951
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9642515333032447e-05,
+      "loss": 1.3438,
+      "step": 2952
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9642274061184725e-05,
+      "loss": 1.9375,
+      "step": 2953
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9642032708928012e-05,
+      "loss": 0.207,
+      "step": 2954
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9641791276263635e-05,
+      "loss": 0.8086,
+      "step": 2955
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.964154976319292e-05,
+      "loss": 0.4609,
+      "step": 2956
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9641308169717192e-05,
+      "loss": 0.4902,
+      "step": 2957
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9641066495837777e-05,
+      "loss": 0.7422,
+      "step": 2958
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9640824741556e-05,
+      "loss": 0.6953,
+      "step": 2959
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.96405829068732e-05,
+      "loss": 1.5,
+      "step": 2960
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9640340991790692e-05,
+      "loss": 0.5508,
+      "step": 2961
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.964009899630981e-05,
+      "loss": 1.6406,
+      "step": 2962
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9639856920431883e-05,
+      "loss": 0.041,
+      "step": 2963
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.963961476415824e-05,
+      "loss": 1.875,
+      "step": 2964
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9639372527490208e-05,
+      "loss": 0.7461,
+      "step": 2965
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9639130210429128e-05,
+      "loss": 0.4941,
+      "step": 2966
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9638887812976313e-05,
+      "loss": 0.2539,
+      "step": 2967
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9638645335133106e-05,
+      "loss": 1.375,
+      "step": 2968
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9638402776900838e-05,
+      "loss": 0.1621,
+      "step": 2969
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.963816013828084e-05,
+      "loss": 0.9727,
+      "step": 2970
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9637917419274444e-05,
+      "loss": 0.5938,
+      "step": 2971
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9637674619882978e-05,
+      "loss": 0.8086,
+      "step": 2972
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9637431740107784e-05,
+      "loss": 0.3184,
+      "step": 2973
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9637188779950193e-05,
+      "loss": 0.4785,
+      "step": 2974
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9636945739411533e-05,
+      "loss": 0.3027,
+      "step": 2975
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9636702618493147e-05,
+      "loss": 0.1226,
+      "step": 2976
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9636459417196366e-05,
+      "loss": 0.0305,
+      "step": 2977
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9636216135522527e-05,
+      "loss": 0.062,
+      "step": 2978
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.963597277347297e-05,
+      "loss": 0.0008,
+      "step": 2979
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.963572933104902e-05,
+      "loss": 0.0525,
+      "step": 2980
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.963548580825202e-05,
+      "loss": 0.0086,
+      "step": 2981
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9635242205083314e-05,
+      "loss": 0.0001,
+      "step": 2982
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.963499852154423e-05,
+      "loss": 0.0515,
+      "step": 2983
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.963475475763611e-05,
+      "loss": 0.0092,
+      "step": 2984
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9634510913360292e-05,
+      "loss": 0.0,
+      "step": 2985
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.963426698871812e-05,
+      "loss": 0.2891,
+      "step": 2986
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9634022983710927e-05,
+      "loss": 0.1807,
+      "step": 2987
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9633778898340054e-05,
+      "loss": 0.0009,
+      "step": 2988
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9633534732606843e-05,
+      "loss": 0.0005,
+      "step": 2989
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.963329048651263e-05,
+      "loss": 0.0017,
+      "step": 2990
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.963304616005877e-05,
+      "loss": 0.0002,
+      "step": 2991
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9632801753246588e-05,
+      "loss": 0.0002,
+      "step": 2992
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9632557266077437e-05,
+      "loss": 0.1406,
+      "step": 2993
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9632312698552656e-05,
+      "loss": 0.1206,
+      "step": 2994
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9632068050673586e-05,
+      "loss": 0.1543,
+      "step": 2995
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9631823322441576e-05,
+      "loss": 0.0001,
+      "step": 2996
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9631578513857968e-05,
+      "loss": 0.1738,
+      "step": 2997
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9631333624924102e-05,
+      "loss": 0.0005,
+      "step": 2998
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9631088655641325e-05,
+      "loss": 0.2432,
+      "step": 2999
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9630843606010984e-05,
+      "loss": 0.3047,
+      "step": 3000
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9630598476034426e-05,
+      "loss": 1.4219,
+      "step": 3001
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9630353265712997e-05,
+      "loss": 0.8477,
+      "step": 3002
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9630107975048037e-05,
+      "loss": 0.7773,
+      "step": 3003
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9629862604040897e-05,
+      "loss": 0.3984,
+      "step": 3004
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.962961715269293e-05,
+      "loss": 0.0108,
+      "step": 3005
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9629371621005477e-05,
+      "loss": 0.4961,
+      "step": 3006
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.962912600897989e-05,
+      "loss": 0.4492,
+      "step": 3007
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.962888031661751e-05,
+      "loss": 1.1328,
+      "step": 3008
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.96286345439197e-05,
+      "loss": 1.0547,
+      "step": 3009
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9628388690887797e-05,
+      "loss": 1.4922,
+      "step": 3010
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.962814275752316e-05,
+      "loss": 0.5039,
+      "step": 3011
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9627896743827136e-05,
+      "loss": 0.6289,
+      "step": 3012
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9627650649801072e-05,
+      "loss": 0.7695,
+      "step": 3013
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9627404475446323e-05,
+      "loss": 0.1318,
+      "step": 3014
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9627158220764245e-05,
+      "loss": 0.0747,
+      "step": 3015
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9626911885756186e-05,
+      "loss": 0.1147,
+      "step": 3016
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.96266654704235e-05,
+      "loss": 0.0009,
+      "step": 3017
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 2.9626418974767536e-05,
+      "loss": 0.3145,
+      "step": 3018
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9626172398789652e-05,
+      "loss": 0.003,
+      "step": 3019
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9625925742491208e-05,
+      "loss": 0.0101,
+      "step": 3020
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9625679005873542e-05,
+      "loss": 0.002,
+      "step": 3021
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9625432188938024e-05,
+      "loss": 0.4238,
+      "step": 3022
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9625185291686e-05,
+      "loss": 0.8281,
+      "step": 3023
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9624938314118833e-05,
+      "loss": 0.8164,
+      "step": 3024
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9624691256237878e-05,
+      "loss": 1.0625,
+      "step": 3025
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9624444118044488e-05,
+      "loss": 0.0032,
+      "step": 3026
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.962419689954002e-05,
+      "loss": 0.0003,
+      "step": 3027
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9623949600725836e-05,
+      "loss": 0.2354,
+      "step": 3028
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9623702221603295e-05,
+      "loss": 0.6758,
+      "step": 3029
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9623454762173747e-05,
+      "loss": 0.1226,
+      "step": 3030
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9623207222438557e-05,
+      "loss": 0.0001,
+      "step": 3031
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9622959602399084e-05,
+      "loss": 0.0001,
+      "step": 3032
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.962271190205669e-05,
+      "loss": 0.0002,
+      "step": 3033
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.962246412141273e-05,
+      "loss": 0.0004,
+      "step": 3034
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.962221626046857e-05,
+      "loss": 0.0009,
+      "step": 3035
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9621968319225567e-05,
+      "loss": 0.0004,
+      "step": 3036
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9621720297685087e-05,
+      "loss": 0.0001,
+      "step": 3037
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9621472195848486e-05,
+      "loss": 0.0015,
+      "step": 3038
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.962122401371713e-05,
+      "loss": 0.021,
+      "step": 3039
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.962097575129238e-05,
+      "loss": 0.0001,
+      "step": 3040
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.96207274085756e-05,
+      "loss": 0.0713,
+      "step": 3041
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9620478985568162e-05,
+      "loss": 0.0008,
+      "step": 3042
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.962023048227142e-05,
+      "loss": 0.1523,
+      "step": 3043
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9619981898686737e-05,
+      "loss": 0.1367,
+      "step": 3044
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9619733234815487e-05,
+      "loss": 0.0923,
+      "step": 3045
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9619484490659027e-05,
+      "loss": 0.1201,
+      "step": 3046
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9619235666218733e-05,
+      "loss": 0.1289,
+      "step": 3047
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9618986761495963e-05,
+      "loss": 0.2217,
+      "step": 3048
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9618737776492087e-05,
+      "loss": 0.1377,
+      "step": 3049
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.961848871120847e-05,
+      "loss": 0.4609,
+      "step": 3050
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.961823956564648e-05,
+      "loss": 0.375,
+      "step": 3051
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9617990339807493e-05,
+      "loss": 0.4688,
+      "step": 3052
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.961774103369287e-05,
+      "loss": 1.1172,
+      "step": 3053
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.961749164730398e-05,
+      "loss": 0.1494,
+      "step": 3054
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9617242180642193e-05,
+      "loss": 0.0167,
+      "step": 3055
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9616992633708878e-05,
+      "loss": 0.9609,
+      "step": 3056
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9616743006505414e-05,
+      "loss": 0.2812,
+      "step": 3057
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9616493299033162e-05,
+      "loss": 0.2656,
+      "step": 3058
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9616243511293495e-05,
+      "loss": 0.6445,
+      "step": 3059
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.961599364328779e-05,
+      "loss": 0.0034,
+      "step": 3060
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.961574369501741e-05,
+      "loss": 0.025,
+      "step": 3061
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9615493666483743e-05,
+      "loss": 0.4043,
+      "step": 3062
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.961524355768814e-05,
+      "loss": 1.6328,
+      "step": 3063
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9614993368631996e-05,
+      "loss": 0.3223,
+      "step": 3064
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.961474309931667e-05,
+      "loss": 0.1553,
+      "step": 3065
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9614492749743543e-05,
+      "loss": 1.1484,
+      "step": 3066
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.961424231991399e-05,
+      "loss": 0.0014,
+      "step": 3067
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9613991809829383e-05,
+      "loss": 1.4453,
+      "step": 3068
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.96137412194911e-05,
+      "loss": 0.0008,
+      "step": 3069
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9613490548900524e-05,
+      "loss": 1.3359,
+      "step": 3070
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9613239798059016e-05,
+      "loss": 0.0571,
+      "step": 3071
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.961298896696796e-05,
+      "loss": 0.2773,
+      "step": 3072
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9612738055628742e-05,
+      "loss": 0.0167,
+      "step": 3073
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9612487064042727e-05,
+      "loss": 0.0003,
+      "step": 3074
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9612235992211304e-05,
+      "loss": 1.4219,
+      "step": 3075
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9611984840135846e-05,
+      "loss": 0.0004,
+      "step": 3076
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.961173360781773e-05,
+      "loss": 0.3652,
+      "step": 3077
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.961148229525834e-05,
+      "loss": 0.625,
+      "step": 3078
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9611230902459057e-05,
+      "loss": 0.0003,
+      "step": 3079
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9610979429421256e-05,
+      "loss": 0.0981,
+      "step": 3080
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.961072787614632e-05,
+      "loss": 0.0029,
+      "step": 3081
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.961047624263564e-05,
+      "loss": 0.3945,
+      "step": 3082
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9610224528890583e-05,
+      "loss": 0.0004,
+      "step": 3083
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9609972734912537e-05,
+      "loss": 0.0005,
+      "step": 3084
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9609720860702887e-05,
+      "loss": 0.0215,
+      "step": 3085
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 2.9609468906263018e-05,
+      "loss": 0.0041,
+      "step": 3086
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9609216871594306e-05,
+      "loss": 0.3203,
+      "step": 3087
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9608964756698143e-05,
+      "loss": 0.0003,
+      "step": 3088
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9608712561575908e-05,
+      "loss": 0.009,
+      "step": 3089
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9608460286228987e-05,
+      "loss": 0.0003,
+      "step": 3090
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.960820793065877e-05,
+      "loss": 0.0003,
+      "step": 3091
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9607955494866638e-05,
+      "loss": 0.0391,
+      "step": 3092
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9607702978853978e-05,
+      "loss": 0.0003,
+      "step": 3093
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.960745038262218e-05,
+      "loss": 0.0962,
+      "step": 3094
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9607197706172624e-05,
+      "loss": 0.1289,
+      "step": 3095
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.960694494950671e-05,
+      "loss": 0.1006,
+      "step": 3096
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9606692112625807e-05,
+      "loss": 0.1396,
+      "step": 3097
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9606439195531317e-05,
+      "loss": 0.125,
+      "step": 3098
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.960618619822463e-05,
+      "loss": 0.1953,
+      "step": 3099
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9605933120707126e-05,
+      "loss": 0.2617,
+      "step": 3100
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9605679962980207e-05,
+      "loss": 1.4219,
+      "step": 3101
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.960542672504526e-05,
+      "loss": 0.1416,
+      "step": 3102
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.960517340690366e-05,
+      "loss": 0.2812,
+      "step": 3103
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9604920008556818e-05,
+      "loss": 1.8594,
+      "step": 3104
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9604666530006118e-05,
+      "loss": 0.6641,
+      "step": 3105
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.960441297125295e-05,
+      "loss": 0.3477,
+      "step": 3106
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.960415933229871e-05,
+      "loss": 0.2051,
+      "step": 3107
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9603905613144792e-05,
+      "loss": 0.8281,
+      "step": 3108
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9603651813792584e-05,
+      "loss": 0.2988,
+      "step": 3109
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.960339793424348e-05,
+      "loss": 0.5078,
+      "step": 3110
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9603143974498882e-05,
+      "loss": 0.0991,
+      "step": 3111
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9602889934560176e-05,
+      "loss": 0.5781,
+      "step": 3112
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9602635814428758e-05,
+      "loss": 0.5312,
+      "step": 3113
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9602381614106027e-05,
+      "loss": 1.0938,
+      "step": 3114
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9602127333593382e-05,
+      "loss": 1.6328,
+      "step": 3115
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9601872972892213e-05,
+      "loss": 0.3242,
+      "step": 3116
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9601618532003915e-05,
+      "loss": 0.0591,
+      "step": 3117
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9601364010929895e-05,
+      "loss": 0.9102,
+      "step": 3118
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9601109409671542e-05,
+      "loss": 0.6445,
+      "step": 3119
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9600854728230258e-05,
+      "loss": 0.0437,
+      "step": 3120
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.960059996660744e-05,
+      "loss": 0.6328,
+      "step": 3121
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9600345124804492e-05,
+      "loss": 0.0272,
+      "step": 3122
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9600090202822802e-05,
+      "loss": 0.3906,
+      "step": 3123
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9599835200663786e-05,
+      "loss": 2.0781,
+      "step": 3124
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9599580118328828e-05,
+      "loss": 0.293,
+      "step": 3125
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9599324955819342e-05,
+      "loss": 0.0003,
+      "step": 3126
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.959906971313672e-05,
+      "loss": 0.0757,
+      "step": 3127
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.959881439028237e-05,
+      "loss": 0.0125,
+      "step": 3128
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9598558987257686e-05,
+      "loss": 0.5703,
+      "step": 3129
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.959830350406408e-05,
+      "loss": 0.0177,
+      "step": 3130
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9598047940702952e-05,
+      "loss": 0.0001,
+      "step": 3131
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9597792297175707e-05,
+      "loss": 0.5469,
+      "step": 3132
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.959753657348374e-05,
+      "loss": 0.127,
+      "step": 3133
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9597280769628465e-05,
+      "loss": 0.3027,
+      "step": 3134
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9597024885611283e-05,
+      "loss": 0.375,
+      "step": 3135
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9596768921433597e-05,
+      "loss": 0.0143,
+      "step": 3136
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9596512877096816e-05,
+      "loss": 0.0161,
+      "step": 3137
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.959625675260235e-05,
+      "loss": 0.0006,
+      "step": 3138
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.95960005479516e-05,
+      "loss": 0.0132,
+      "step": 3139
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9595744263145973e-05,
+      "loss": 0.003,
+      "step": 3140
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9595487898186872e-05,
+      "loss": 0.0737,
+      "step": 3141
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9595231453075717e-05,
+      "loss": 0.0006,
+      "step": 3142
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9594974927813906e-05,
+      "loss": 0.0002,
+      "step": 3143
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9594718322402853e-05,
+      "loss": 0.0004,
+      "step": 3144
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9594461636843964e-05,
+      "loss": 0.0002,
+      "step": 3145
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.959420487113865e-05,
+      "loss": 0.0007,
+      "step": 3146
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9593948025288317e-05,
+      "loss": 0.0002,
+      "step": 3147
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9593691099294386e-05,
+      "loss": 0.0001,
+      "step": 3148
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.959343409315826e-05,
+      "loss": 0.0001,
+      "step": 3149
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.959317700688135e-05,
+      "loss": 0.1523,
+      "step": 3150
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.959291984046507e-05,
+      "loss": 0.8906,
+      "step": 3151
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.959266259391083e-05,
+      "loss": 1.1719,
+      "step": 3152
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.959240526722005e-05,
+      "loss": 0.9375,
+      "step": 3153
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 2.9592147860394132e-05,
+      "loss": 0.6055,
+      "step": 3154
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.9591890373434495e-05,
+      "loss": 1.6172,
+      "step": 3155
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.9591632806342556e-05,
+      "loss": 0.2119,
+      "step": 3156
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.959137515911972e-05,
+      "loss": 1.6016,
+      "step": 3157
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.959111743176742e-05,
+      "loss": 0.6055,
+      "step": 3158
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.9590859624287053e-05,
+      "loss": 0.6523,
+      "step": 3159
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.959060173668004e-05,
+      "loss": 1.3906,
+      "step": 3160
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.95903437689478e-05,
+      "loss": 1.8047,
+      "step": 3161
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.959008572109175e-05,
+      "loss": 0.2871,
+      "step": 3162
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.9589827593113306e-05,
+      "loss": 0.019,
+      "step": 3163
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.9589569385013884e-05,
+      "loss": 0.083,
+      "step": 3164
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.9589311096794905e-05,
+      "loss": 0.0403,
+      "step": 3165
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.9589052728457784e-05,
+      "loss": 0.0471,
+      "step": 3166
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.9588794280003943e-05,
+      "loss": 0.1079,
+      "step": 3167
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.9588535751434795e-05,
+      "loss": 0.2871,
+      "step": 3168
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.9588277142751767e-05,
+      "loss": 0.8945,
+      "step": 3169
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.9588018453956278e-05,
+      "loss": 0.0001,
+      "step": 3170
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.958775968504974e-05,
+      "loss": 1.6797,
+      "step": 3171
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.958750083603359e-05,
+      "loss": 0.209,
+      "step": 3172
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.9587241906909238e-05,
+      "loss": 0.0001,
+      "step": 3173
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.9586982897678105e-05,
+      "loss": 0.103,
+      "step": 3174
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.958672380834162e-05,
+      "loss": 0.0535,
+      "step": 3175
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.95864646389012e-05,
+      "loss": 0.0557,
+      "step": 3176
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.9586205389358273e-05,
+      "loss": 0.0005,
+      "step": 3177
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.958594605971426e-05,
+      "loss": 0.0062,
+      "step": 3178
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.9585686649970585e-05,
+      "loss": 0.0026,
+      "step": 3179
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.9585427160128675e-05,
+      "loss": 1.4531,
+      "step": 3180
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.958516759018995e-05,
+      "loss": 0.6953,
+      "step": 3181
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.9584907940155845e-05,
+      "loss": 0.4043,
+      "step": 3182
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.9584648210027775e-05,
+      "loss": 0.0015,
+      "step": 3183
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.9584388399807172e-05,
+      "loss": 0.0001,
+      "step": 3184
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.958412850949546e-05,
+      "loss": 0.1357,
+      "step": 3185
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.958386853909407e-05,
+      "loss": 0.1055,
+      "step": 3186
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.958360848860443e-05,
+      "loss": 0.0723,
+      "step": 3187
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.9583348358027965e-05,
+      "loss": 0.6758,
+      "step": 3188
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.95830881473661e-05,
+      "loss": 0.0018,
+      "step": 3189
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.958282785662027e-05,
+      "loss": 0.0001,
+      "step": 3190
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.95825674857919e-05,
+      "loss": 0.5234,
+      "step": 3191
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.9582307034882425e-05,
+      "loss": 0.0031,
+      "step": 3192
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.9582046503893275e-05,
+      "loss": 0.0017,
+      "step": 3193
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.9581785892825877e-05,
+      "loss": 0.0,
+      "step": 3194
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.9581525201681658e-05,
+      "loss": 0.0002,
+      "step": 3195
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.958126443046206e-05,
+      "loss": 0.1118,
+      "step": 3196
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.958100357916851e-05,
+      "loss": 0.1416,
+      "step": 3197
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.958074264780244e-05,
+      "loss": 0.1953,
+      "step": 3198
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.9580481636365286e-05,
+      "loss": 0.1533,
+      "step": 3199
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.9580220544858476e-05,
+      "loss": 0.1504,
+      "step": 3200
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.9579959373283448e-05,
+      "loss": 1.2031,
+      "step": 3201
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.9579698121641638e-05,
+      "loss": 0.9375,
+      "step": 3202
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.9579436789934475e-05,
+      "loss": 0.8086,
+      "step": 3203
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.9579175378163397e-05,
+      "loss": 0.2539,
+      "step": 3204
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.957891388632984e-05,
+      "loss": 0.8828,
+      "step": 3205
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.9578652314435236e-05,
+      "loss": 0.6758,
+      "step": 3206
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.9578390662481026e-05,
+      "loss": 0.625,
+      "step": 3207
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.9578128930468648e-05,
+      "loss": 0.3398,
+      "step": 3208
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.9577867118399536e-05,
+      "loss": 0.7344,
+      "step": 3209
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.9577605226275128e-05,
+      "loss": 0.3672,
+      "step": 3210
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.957734325409686e-05,
+      "loss": 0.6758,
+      "step": 3211
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.957708120186618e-05,
+      "loss": 0.0079,
+      "step": 3212
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.957681906958452e-05,
+      "loss": 1.1953,
+      "step": 3213
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.9576556857253315e-05,
+      "loss": 0.7617,
+      "step": 3214
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.9576294564874015e-05,
+      "loss": 0.4902,
+      "step": 3215
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.957603219244805e-05,
+      "loss": 0.2363,
+      "step": 3216
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.957576973997687e-05,
+      "loss": 0.6758,
+      "step": 3217
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.957550720746191e-05,
+      "loss": 0.4746,
+      "step": 3218
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.9575244594904617e-05,
+      "loss": 0.8281,
+      "step": 3219
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.9574981902306433e-05,
+      "loss": 1.1719,
+      "step": 3220
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 2.9574719129668796e-05,
+      "loss": 0.4141,
+      "step": 3221
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.957445627699315e-05,
+      "loss": 0.0996,
+      "step": 3222
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.957419334428094e-05,
+      "loss": 0.0226,
+      "step": 3223
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.9573930331533612e-05,
+      "loss": 0.0228,
+      "step": 3224
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.9573667238752605e-05,
+      "loss": 0.0471,
+      "step": 3225
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.9573404065939365e-05,
+      "loss": 0.0036,
+      "step": 3226
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.957314081309534e-05,
+      "loss": 0.0004,
+      "step": 3227
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.957287748022198e-05,
+      "loss": 0.0003,
+      "step": 3228
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.957261406732072e-05,
+      "loss": 0.0095,
+      "step": 3229
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.957235057439301e-05,
+      "loss": 0.2695,
+      "step": 3230
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.9572087001440307e-05,
+      "loss": 0.3242,
+      "step": 3231
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.9571823348464042e-05,
+      "loss": 0.0056,
+      "step": 3232
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.9571559615465677e-05,
+      "loss": 0.7148,
+      "step": 3233
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.9571295802446656e-05,
+      "loss": 0.1104,
+      "step": 3234
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.9571031909408423e-05,
+      "loss": 0.0003,
+      "step": 3235
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.9570767936352434e-05,
+      "loss": 0.5117,
+      "step": 3236
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.9570503883280132e-05,
+      "loss": 0.0091,
+      "step": 3237
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.957023975019297e-05,
+      "loss": 0.0322,
+      "step": 3238
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.9569975537092402e-05,
+      "loss": 0.0006,
+      "step": 3239
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.9569711243979874e-05,
+      "loss": 0.0008,
+      "step": 3240
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.956944687085684e-05,
+      "loss": 0.0771,
+      "step": 3241
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.9569182417724753e-05,
+      "loss": 0.0002,
+      "step": 3242
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.956891788458506e-05,
+      "loss": 0.0018,
+      "step": 3243
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.9568653271439218e-05,
+      "loss": 0.1104,
+      "step": 3244
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.9568388578288682e-05,
+      "loss": 0.1436,
+      "step": 3245
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.9568123805134898e-05,
+      "loss": 0.0128,
+      "step": 3246
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.9567858951979324e-05,
+      "loss": 0.0002,
+      "step": 3247
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.956759401882342e-05,
+      "loss": 0.0004,
+      "step": 3248
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.9567329005668636e-05,
+      "loss": 0.1235,
+      "step": 3249
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.9567063912516426e-05,
+      "loss": 0.1787,
+      "step": 3250
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.9566798739368247e-05,
+      "loss": 1.2422,
+      "step": 3251
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.9566533486225558e-05,
+      "loss": 0.8008,
+      "step": 3252
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.956626815308981e-05,
+      "loss": 0.8906,
+      "step": 3253
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.9566002739962463e-05,
+      "loss": 1.3594,
+      "step": 3254
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.9565737246844976e-05,
+      "loss": 0.7773,
+      "step": 3255
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.9565471673738806e-05,
+      "loss": 1.5312,
+      "step": 3256
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.9565206020645407e-05,
+      "loss": 0.2041,
+      "step": 3257
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.9564940287566247e-05,
+      "loss": 1.2578,
+      "step": 3258
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.9564674474502782e-05,
+      "loss": 0.0286,
+      "step": 3259
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.9564408581456463e-05,
+      "loss": 0.6016,
+      "step": 3260
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.9564142608428762e-05,
+      "loss": 1.3359,
+      "step": 3261
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.9563876555421138e-05,
+      "loss": 0.2422,
+      "step": 3262
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.956361042243504e-05,
+      "loss": 2.2812,
+      "step": 3263
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.9563344209471945e-05,
+      "loss": 0.291,
+      "step": 3264
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.9563077916533307e-05,
+      "loss": 0.0776,
+      "step": 3265
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.9562811543620586e-05,
+      "loss": 0.2129,
+      "step": 3266
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.956254509073525e-05,
+      "loss": 0.9375,
+      "step": 3267
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.9562278557878765e-05,
+      "loss": 0.0522,
+      "step": 3268
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.956201194505259e-05,
+      "loss": 3.3906,
+      "step": 3269
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.9561745252258184e-05,
+      "loss": 0.001,
+      "step": 3270
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.9561478479497018e-05,
+      "loss": 0.4727,
+      "step": 3271
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.956121162677056e-05,
+      "loss": 0.053,
+      "step": 3272
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.9560944694080268e-05,
+      "loss": 0.0173,
+      "step": 3273
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.9560677681427615e-05,
+      "loss": 0.0007,
+      "step": 3274
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.956041058881406e-05,
+      "loss": 0.0044,
+      "step": 3275
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.9560143416241073e-05,
+      "loss": 0.6016,
+      "step": 3276
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.9559876163710122e-05,
+      "loss": 0.0019,
+      "step": 3277
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.955960883122268e-05,
+      "loss": 0.0004,
+      "step": 3278
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.9559341418780203e-05,
+      "loss": 0.0001,
+      "step": 3279
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.955907392638417e-05,
+      "loss": 0.0004,
+      "step": 3280
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.955880635403604e-05,
+      "loss": 0.5703,
+      "step": 3281
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.9558538701737292e-05,
+      "loss": 0.0713,
+      "step": 3282
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.955827096948939e-05,
+      "loss": 0.0166,
+      "step": 3283
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.955800315729381e-05,
+      "loss": 0.0153,
+      "step": 3284
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.9557735265152015e-05,
+      "loss": 0.6875,
+      "step": 3285
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.955746729306548e-05,
+      "loss": 0.0003,
+      "step": 3286
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.955719924103568e-05,
+      "loss": 0.0005,
+      "step": 3287
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.9556931109064084e-05,
+      "loss": 0.014,
+      "step": 3288
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 2.955666289715216e-05,
+      "loss": 0.0001,
+      "step": 3289
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.955639460530139e-05,
+      "loss": 0.6289,
+      "step": 3290
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.955612623351324e-05,
+      "loss": 0.0,
+      "step": 3291
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.9555857781789186e-05,
+      "loss": 0.1895,
+      "step": 3292
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.95555892501307e-05,
+      "loss": 0.0021,
+      "step": 3293
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.9555320638539263e-05,
+      "loss": 0.0605,
+      "step": 3294
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.9555051947016344e-05,
+      "loss": 0.0001,
+      "step": 3295
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.9554783175563425e-05,
+      "loss": 0.0,
+      "step": 3296
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.9554514324181974e-05,
+      "loss": 0.0,
+      "step": 3297
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.955424539287347e-05,
+      "loss": 0.1465,
+      "step": 3298
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.955397638163939e-05,
+      "loss": 0.1357,
+      "step": 3299
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.955370729048122e-05,
+      "loss": 0.1367,
+      "step": 3300
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.9553438119400422e-05,
+      "loss": 1.1875,
+      "step": 3301
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.955316886839849e-05,
+      "loss": 1.3438,
+      "step": 3302
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.9552899537476886e-05,
+      "loss": 1.2266,
+      "step": 3303
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.95526301266371e-05,
+      "loss": 1.4062,
+      "step": 3304
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.9552360635880612e-05,
+      "loss": 0.3223,
+      "step": 3305
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.9552091065208896e-05,
+      "loss": 0.1533,
+      "step": 3306
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.955182141462344e-05,
+      "loss": 1.1094,
+      "step": 3307
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.955155168412572e-05,
+      "loss": 1.0234,
+      "step": 3308
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.9551281873717213e-05,
+      "loss": 0.2256,
+      "step": 3309
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.955101198339941e-05,
+      "loss": 0.5977,
+      "step": 3310
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.9550742013173788e-05,
+      "loss": 0.8906,
+      "step": 3311
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.9550471963041826e-05,
+      "loss": 2.2969,
+      "step": 3312
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.9550201833005014e-05,
+      "loss": 0.1104,
+      "step": 3313
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.9549931623064833e-05,
+      "loss": 0.0442,
+      "step": 3314
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.9549661333222764e-05,
+      "loss": 0.832,
+      "step": 3315
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.9549390963480294e-05,
+      "loss": 1.9766,
+      "step": 3316
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.9549120513838913e-05,
+      "loss": 0.0532,
+      "step": 3317
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.9548849984300096e-05,
+      "loss": 0.3574,
+      "step": 3318
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.954857937486533e-05,
+      "loss": 0.5742,
+      "step": 3319
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.954830868553611e-05,
+      "loss": 0.4492,
+      "step": 3320
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.9548037916313914e-05,
+      "loss": 2.2969,
+      "step": 3321
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.9547767067200228e-05,
+      "loss": 2.4531,
+      "step": 3322
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.954749613819655e-05,
+      "loss": 0.5078,
+      "step": 3323
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.9547225129304356e-05,
+      "loss": 0.0136,
+      "step": 3324
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.9546954040525144e-05,
+      "loss": 0.2393,
+      "step": 3325
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.9546682871860395e-05,
+      "loss": 0.0393,
+      "step": 3326
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.95464116233116e-05,
+      "loss": 0.0001,
+      "step": 3327
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.954614029488025e-05,
+      "loss": 0.3086,
+      "step": 3328
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.9545868886567838e-05,
+      "loss": 0.0669,
+      "step": 3329
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.9545597398375847e-05,
+      "loss": 0.0022,
+      "step": 3330
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.9545325830305773e-05,
+      "loss": 0.0003,
+      "step": 3331
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.954505418235911e-05,
+      "loss": 0.0002,
+      "step": 3332
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.954478245453734e-05,
+      "loss": 1.5391,
+      "step": 3333
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.9544510646841967e-05,
+      "loss": 0.4922,
+      "step": 3334
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.9544238759274477e-05,
+      "loss": 0.1846,
+      "step": 3335
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.9543966791836358e-05,
+      "loss": 0.1689,
+      "step": 3336
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.9543694744529116e-05,
+      "loss": 0.0001,
+      "step": 3337
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.9543422617354237e-05,
+      "loss": 0.1387,
+      "step": 3338
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.9543150410313214e-05,
+      "loss": 0.0008,
+      "step": 3339
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.9542878123407545e-05,
+      "loss": 0.0002,
+      "step": 3340
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.954260575663873e-05,
+      "loss": 0.2559,
+      "step": 3341
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.954233331000826e-05,
+      "loss": 0.002,
+      "step": 3342
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.9542060783517623e-05,
+      "loss": 0.0001,
+      "step": 3343
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.9541788177168328e-05,
+      "loss": 0.0001,
+      "step": 3344
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.954151549096187e-05,
+      "loss": 0.0001,
+      "step": 3345
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.9541242724899744e-05,
+      "loss": 0.0001,
+      "step": 3346
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.9540969878983442e-05,
+      "loss": 0.0002,
+      "step": 3347
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.954069695321448e-05,
+      "loss": 0.1572,
+      "step": 3348
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.9540423947594334e-05,
+      "loss": 0.0825,
+      "step": 3349
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.954015086212452e-05,
+      "loss": 0.1758,
+      "step": 3350
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.953987769680653e-05,
+      "loss": 0.9219,
+      "step": 3351
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.9539604451641862e-05,
+      "loss": 0.9531,
+      "step": 3352
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.9539331126632027e-05,
+      "loss": 1.5312,
+      "step": 3353
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.9539057721778515e-05,
+      "loss": 0.8477,
+      "step": 3354
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.953878423708284e-05,
+      "loss": 0.6875,
+      "step": 3355
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.9538510672546487e-05,
+      "loss": 0.8164,
+      "step": 3356
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 2.953823702817097e-05,
+      "loss": 0.6641,
+      "step": 3357
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.953796330395779e-05,
+      "loss": 0.5156,
+      "step": 3358
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9537689499908447e-05,
+      "loss": 1.5234,
+      "step": 3359
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9537415616024447e-05,
+      "loss": 0.5234,
+      "step": 3360
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9537141652307297e-05,
+      "loss": 0.9648,
+      "step": 3361
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9536867608758495e-05,
+      "loss": 0.0854,
+      "step": 3362
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9536593485379554e-05,
+      "loss": 0.75,
+      "step": 3363
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.953631928217197e-05,
+      "loss": 0.8516,
+      "step": 3364
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9536044999137256e-05,
+      "loss": 0.3574,
+      "step": 3365
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.953577063627691e-05,
+      "loss": 0.5156,
+      "step": 3366
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9535496193592454e-05,
+      "loss": 1.6172,
+      "step": 3367
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9535221671085375e-05,
+      "loss": 0.0693,
+      "step": 3368
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9534947068757197e-05,
+      "loss": 0.0583,
+      "step": 3369
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9534672386609422e-05,
+      "loss": 0.0559,
+      "step": 3370
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9534397624643557e-05,
+      "loss": 0.1621,
+      "step": 3371
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.953412278286111e-05,
+      "loss": 0.2676,
+      "step": 3372
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9533847861263596e-05,
+      "loss": 0.0099,
+      "step": 3373
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.953357285985252e-05,
+      "loss": 0.1504,
+      "step": 3374
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9533297778629395e-05,
+      "loss": 0.0008,
+      "step": 3375
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9533022617595727e-05,
+      "loss": 0.0002,
+      "step": 3376
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9532747376753033e-05,
+      "loss": 0.4648,
+      "step": 3377
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9532472056102823e-05,
+      "loss": 0.1216,
+      "step": 3378
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9532196655646606e-05,
+      "loss": 0.0002,
+      "step": 3379
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9531921175385895e-05,
+      "loss": 0.4355,
+      "step": 3380
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9531645615322202e-05,
+      "loss": 0.6523,
+      "step": 3381
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9531369975457045e-05,
+      "loss": 0.3203,
+      "step": 3382
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.953109425579194e-05,
+      "loss": 0.0002,
+      "step": 3383
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9530818456328386e-05,
+      "loss": 0.0002,
+      "step": 3384
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9530542577067914e-05,
+      "loss": 0.0011,
+      "step": 3385
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9530266618012028e-05,
+      "loss": 0.0019,
+      "step": 3386
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.952999057916225e-05,
+      "loss": 0.3652,
+      "step": 3387
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9529714460520095e-05,
+      "loss": 0.0001,
+      "step": 3388
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9529438262087077e-05,
+      "loss": 0.004,
+      "step": 3389
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9529161983864715e-05,
+      "loss": 0.0001,
+      "step": 3390
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9528885625854525e-05,
+      "loss": 0.2002,
+      "step": 3391
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9528609188058023e-05,
+      "loss": 0.2969,
+      "step": 3392
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9528332670476733e-05,
+      "loss": 0.105,
+      "step": 3393
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9528056073112163e-05,
+      "loss": 0.0444,
+      "step": 3394
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9527779395965844e-05,
+      "loss": 0.5508,
+      "step": 3395
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9527502639039288e-05,
+      "loss": 0.1128,
+      "step": 3396
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9527225802334015e-05,
+      "loss": 0.2188,
+      "step": 3397
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9526948885851547e-05,
+      "loss": 0.208,
+      "step": 3398
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9526671889593406e-05,
+      "loss": 0.1885,
+      "step": 3399
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.952639481356111e-05,
+      "loss": 0.1279,
+      "step": 3400
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9526117657756183e-05,
+      "loss": 0.0159,
+      "step": 3401
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9525840422180148e-05,
+      "loss": 0.7188,
+      "step": 3402
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.952556310683452e-05,
+      "loss": 0.9102,
+      "step": 3403
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9525285711720836e-05,
+      "loss": 0.4004,
+      "step": 3404
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.95250082368406e-05,
+      "loss": 0.3711,
+      "step": 3405
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9524730682195356e-05,
+      "loss": 0.457,
+      "step": 3406
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.952445304778662e-05,
+      "loss": 0.5898,
+      "step": 3407
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.952417533361591e-05,
+      "loss": 1.7969,
+      "step": 3408
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9523897539684758e-05,
+      "loss": 0.3086,
+      "step": 3409
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.952361966599469e-05,
+      "loss": 1.2031,
+      "step": 3410
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9523341712547227e-05,
+      "loss": 0.4902,
+      "step": 3411
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.95230636793439e-05,
+      "loss": 1.6797,
+      "step": 3412
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9522785566386236e-05,
+      "loss": 0.5625,
+      "step": 3413
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.952250737367576e-05,
+      "loss": 0.5117,
+      "step": 3414
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9522229101214e-05,
+      "loss": 0.6602,
+      "step": 3415
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.952195074900248e-05,
+      "loss": 1.0547,
+      "step": 3416
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9521672317042737e-05,
+      "loss": 0.0006,
+      "step": 3417
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.95213938053363e-05,
+      "loss": 0.2236,
+      "step": 3418
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9521115213884688e-05,
+      "loss": 0.1719,
+      "step": 3419
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.952083654268944e-05,
+      "loss": 0.127,
+      "step": 3420
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9520557791752086e-05,
+      "loss": 0.6367,
+      "step": 3421
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9520278961074156e-05,
+      "loss": 0.0001,
+      "step": 3422
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9520000050657173e-05,
+      "loss": 0.6758,
+      "step": 3423
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.951972106050268e-05,
+      "loss": 0.1475,
+      "step": 3424
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9519441990612204e-05,
+      "loss": 0.0913,
+      "step": 3425
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.9519162840987276e-05,
+      "loss": 0.1416,
+      "step": 3426
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.9518883611629433e-05,
+      "loss": 1.9609,
+      "step": 3427
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.951860430254021e-05,
+      "loss": 0.4727,
+      "step": 3428
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.951832491372113e-05,
+      "loss": 0.0,
+      "step": 3429
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.9518045445173738e-05,
+      "loss": 0.3691,
+      "step": 3430
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.9517765896899567e-05,
+      "loss": 0.0013,
+      "step": 3431
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.9517486268900148e-05,
+      "loss": 0.0004,
+      "step": 3432
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.951720656117702e-05,
+      "loss": 0.1138,
+      "step": 3433
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.951692677373172e-05,
+      "loss": 0.0002,
+      "step": 3434
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.951664690656578e-05,
+      "loss": 0.4258,
+      "step": 3435
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.9516366959680744e-05,
+      "loss": 0.0022,
+      "step": 3436
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.9516086933078137e-05,
+      "loss": 0.0001,
+      "step": 3437
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.951580682675951e-05,
+      "loss": 0.0884,
+      "step": 3438
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.9515526640726394e-05,
+      "loss": 0.0006,
+      "step": 3439
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.951524637498033e-05,
+      "loss": 0.0001,
+      "step": 3440
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.9514966029522857e-05,
+      "loss": 0.0002,
+      "step": 3441
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.9514685604355514e-05,
+      "loss": 0.0015,
+      "step": 3442
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.9514405099479842e-05,
+      "loss": 0.0,
+      "step": 3443
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.9514124514897378e-05,
+      "loss": 0.0005,
+      "step": 3444
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.9513843850609665e-05,
+      "loss": 0.0,
+      "step": 3445
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.951356310661825e-05,
+      "loss": 0.0002,
+      "step": 3446
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.9513282282924663e-05,
+      "loss": 0.0002,
+      "step": 3447
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.951300137953046e-05,
+      "loss": 0.1611,
+      "step": 3448
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.9512720396437167e-05,
+      "loss": 0.1455,
+      "step": 3449
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.951243933364634e-05,
+      "loss": 0.1553,
+      "step": 3450
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.951215819115952e-05,
+      "loss": 1.4453,
+      "step": 3451
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.951187696897825e-05,
+      "loss": 0.2832,
+      "step": 3452
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.951159566710407e-05,
+      "loss": 0.832,
+      "step": 3453
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.9511314285538535e-05,
+      "loss": 0.5508,
+      "step": 3454
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.951103282428318e-05,
+      "loss": 0.7383,
+      "step": 3455
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.9510751283339553e-05,
+      "loss": 1.2734,
+      "step": 3456
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.9510469662709206e-05,
+      "loss": 0.7422,
+      "step": 3457
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.9510187962393676e-05,
+      "loss": 1.3516,
+      "step": 3458
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.9509906182394518e-05,
+      "loss": 0.5586,
+      "step": 3459
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.9509624322713277e-05,
+      "loss": 0.4746,
+      "step": 3460
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.95093423833515e-05,
+      "loss": 1.9922,
+      "step": 3461
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.950906036431074e-05,
+      "loss": 0.418,
+      "step": 3462
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.9508778265592535e-05,
+      "loss": 0.5273,
+      "step": 3463
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.9508496087198443e-05,
+      "loss": 0.2988,
+      "step": 3464
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.950821382913001e-05,
+      "loss": 1.3281,
+      "step": 3465
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.9507931491388787e-05,
+      "loss": 0.377,
+      "step": 3466
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.950764907397633e-05,
+      "loss": 0.1416,
+      "step": 3467
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.950736657689418e-05,
+      "loss": 0.0002,
+      "step": 3468
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.950708400014389e-05,
+      "loss": 0.7734,
+      "step": 3469
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.950680134372702e-05,
+      "loss": 0.5,
+      "step": 3470
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.9506518607645118e-05,
+      "loss": 0.3926,
+      "step": 3471
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.9506235791899733e-05,
+      "loss": 0.3848,
+      "step": 3472
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.950595289649242e-05,
+      "loss": 0.3164,
+      "step": 3473
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.9505669921424737e-05,
+      "loss": 0.0825,
+      "step": 3474
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.950538686669823e-05,
+      "loss": 0.167,
+      "step": 3475
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.950510373231446e-05,
+      "loss": 0.0605,
+      "step": 3476
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.950482051827498e-05,
+      "loss": 0.0003,
+      "step": 3477
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.9504537224581343e-05,
+      "loss": 0.025,
+      "step": 3478
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.9504253851235108e-05,
+      "loss": 0.0972,
+      "step": 3479
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.9503970398237828e-05,
+      "loss": 0.1172,
+      "step": 3480
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.9503686865591058e-05,
+      "loss": 0.0001,
+      "step": 3481
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.9503403253296368e-05,
+      "loss": 0.0164,
+      "step": 3482
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.9503119561355298e-05,
+      "loss": 0.4082,
+      "step": 3483
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.9502835789769418e-05,
+      "loss": 0.0581,
+      "step": 3484
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.9502551938540282e-05,
+      "loss": 0.2129,
+      "step": 3485
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.9502268007669446e-05,
+      "loss": 0.0186,
+      "step": 3486
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.9501983997158475e-05,
+      "loss": 0.0167,
+      "step": 3487
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.9501699907008924e-05,
+      "loss": 0.0728,
+      "step": 3488
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.9501415737222355e-05,
+      "loss": 0.0025,
+      "step": 3489
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.950113148780033e-05,
+      "loss": 0.0005,
+      "step": 3490
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.950084715874441e-05,
+      "loss": 0.0019,
+      "step": 3491
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.950056275005615e-05,
+      "loss": 0.0645,
+      "step": 3492
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.9500278261737117e-05,
+      "loss": 0.001,
+      "step": 3493
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.9499993693788874e-05,
+      "loss": 0.0002,
+      "step": 3494
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.9499709046212985e-05,
+      "loss": 0.0004,
+      "step": 3495
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.949942431901101e-05,
+      "loss": 0.1187,
+      "step": 3496
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.949913951218451e-05,
+      "loss": 0.1416,
+      "step": 3497
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.9498854625735057e-05,
+      "loss": 0.1357,
+      "step": 3498
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.949856965966421e-05,
+      "loss": 0.063,
+      "step": 3499
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.949828461397353e-05,
+      "loss": 0.0957,
+      "step": 3500
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.9497999488664595e-05,
+      "loss": 0.4258,
+      "step": 3501
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.949771428373896e-05,
+      "loss": 0.1621,
+      "step": 3502
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.949742899919819e-05,
+      "loss": 0.6523,
+      "step": 3503
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.949714363504386e-05,
+      "loss": 0.5625,
+      "step": 3504
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.9496858191277533e-05,
+      "loss": 0.6055,
+      "step": 3505
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.9496572667900778e-05,
+      "loss": 0.498,
+      "step": 3506
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.9496287064915154e-05,
+      "loss": 0.4746,
+      "step": 3507
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.9496001382322246e-05,
+      "loss": 0.4199,
+      "step": 3508
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.949571562012361e-05,
+      "loss": 0.6914,
+      "step": 3509
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.9495429778320816e-05,
+      "loss": 0.6211,
+      "step": 3510
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.949514385691544e-05,
+      "loss": 2.2188,
+      "step": 3511
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.949485785590905e-05,
+      "loss": 2.2031,
+      "step": 3512
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.9494571775303214e-05,
+      "loss": 1.3047,
+      "step": 3513
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.9494285615099502e-05,
+      "loss": 0.7383,
+      "step": 3514
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.9493999375299492e-05,
+      "loss": 0.8906,
+      "step": 3515
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.9493713055904748e-05,
+      "loss": 0.1357,
+      "step": 3516
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.9493426656916845e-05,
+      "loss": 0.8086,
+      "step": 3517
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.9493140178337366e-05,
+      "loss": 0.0229,
+      "step": 3518
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.949285362016787e-05,
+      "loss": 0.0015,
+      "step": 3519
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.9492566982409934e-05,
+      "loss": 0.0039,
+      "step": 3520
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.949228026506513e-05,
+      "loss": 0.2002,
+      "step": 3521
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.9491993468135044e-05,
+      "loss": 0.2734,
+      "step": 3522
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.9491706591621245e-05,
+      "loss": 0.0605,
+      "step": 3523
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.94914196355253e-05,
+      "loss": 0.1807,
+      "step": 3524
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.9491132599848795e-05,
+      "loss": 0.4492,
+      "step": 3525
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.94908454845933e-05,
+      "loss": 0.0001,
+      "step": 3526
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.94905582897604e-05,
+      "loss": 0.0306,
+      "step": 3527
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.9490271015351664e-05,
+      "loss": 0.002,
+      "step": 3528
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.948998366136867e-05,
+      "loss": 0.003,
+      "step": 3529
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.9489696227813003e-05,
+      "loss": 0.3926,
+      "step": 3530
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.9489408714686232e-05,
+      "loss": 0.4902,
+      "step": 3531
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.9489121121989945e-05,
+      "loss": 0.0435,
+      "step": 3532
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.9488833449725714e-05,
+      "loss": 0.5273,
+      "step": 3533
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.9488545697895125e-05,
+      "loss": 0.4023,
+      "step": 3534
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.9488257866499752e-05,
+      "loss": 0.0001,
+      "step": 3535
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.9487969955541183e-05,
+      "loss": 1.4062,
+      "step": 3536
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.948768196502099e-05,
+      "loss": 0.0004,
+      "step": 3537
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.948739389494076e-05,
+      "loss": 0.0,
+      "step": 3538
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.948710574530208e-05,
+      "loss": 0.0006,
+      "step": 3539
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.9486817516106523e-05,
+      "loss": 0.0002,
+      "step": 3540
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.948652920735567e-05,
+      "loss": 0.0001,
+      "step": 3541
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.9486240819051117e-05,
+      "loss": 0.0001,
+      "step": 3542
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.9485952351194436e-05,
+      "loss": 0.1982,
+      "step": 3543
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.948566380378722e-05,
+      "loss": 0.0291,
+      "step": 3544
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.9485375176831046e-05,
+      "loss": 0.2754,
+      "step": 3545
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.9485086470327504e-05,
+      "loss": 0.0281,
+      "step": 3546
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.9484797684278176e-05,
+      "loss": 0.1992,
+      "step": 3547
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.948450881868465e-05,
+      "loss": 0.0884,
+      "step": 3548
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.9484219873548514e-05,
+      "loss": 0.1484,
+      "step": 3549
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.9483930848871347e-05,
+      "loss": 0.208,
+      "step": 3550
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.9483641744654747e-05,
+      "loss": 1.1719,
+      "step": 3551
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.9483352560900293e-05,
+      "loss": 1.1953,
+      "step": 3552
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.9483063297609577e-05,
+      "loss": 0.0063,
+      "step": 3553
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.9482773954784193e-05,
+      "loss": 1.2578,
+      "step": 3554
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.948248453242572e-05,
+      "loss": 0.543,
+      "step": 3555
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.9482195030535748e-05,
+      "loss": 0.6523,
+      "step": 3556
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.9481905449115876e-05,
+      "loss": 1.6562,
+      "step": 3557
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.948161578816768e-05,
+      "loss": 0.4727,
+      "step": 3558
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.948132604769277e-05,
+      "loss": 0.9375,
+      "step": 3559
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.948103622769272e-05,
+      "loss": 0.4668,
+      "step": 3560
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.9480746328169125e-05,
+      "loss": 1.2656,
+      "step": 3561
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.9480456349123585e-05,
+      "loss": 0.051,
+      "step": 3562
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.9480166290557685e-05,
+      "loss": 1.2188,
+      "step": 3563
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.9479876152473025e-05,
+      "loss": 0.7383,
+      "step": 3564
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.9479585934871186e-05,
+      "loss": 0.0579,
+      "step": 3565
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.9479295637753773e-05,
+      "loss": 0.3125,
+      "step": 3566
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.9479005261122376e-05,
+      "loss": 0.6641,
+      "step": 3567
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.947871480497859e-05,
+      "loss": 0.6602,
+      "step": 3568
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.947842426932401e-05,
+      "loss": 0.0152,
+      "step": 3569
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.947813365416023e-05,
+      "loss": 0.3125,
+      "step": 3570
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.947784295948885e-05,
+      "loss": 0.0002,
+      "step": 3571
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.9477552185311462e-05,
+      "loss": 1.8828,
+      "step": 3572
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.947726133162967e-05,
+      "loss": 0.4082,
+      "step": 3573
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.9476970398445058e-05,
+      "loss": 0.3184,
+      "step": 3574
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.9476679385759236e-05,
+      "loss": 0.0732,
+      "step": 3575
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.9476388293573795e-05,
+      "loss": 0.1187,
+      "step": 3576
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.9476097121890337e-05,
+      "loss": 0.0933,
+      "step": 3577
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.9475805870710463e-05,
+      "loss": 0.0498,
+      "step": 3578
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.9475514540035767e-05,
+      "loss": 0.0001,
+      "step": 3579
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.9475223129867855e-05,
+      "loss": 0.0001,
+      "step": 3580
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.9474931640208318e-05,
+      "loss": 0.2891,
+      "step": 3581
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.9474640071058766e-05,
+      "loss": 0.0017,
+      "step": 3582
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.9474348422420797e-05,
+      "loss": 0.3535,
+      "step": 3583
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.947405669429601e-05,
+      "loss": 0.6523,
+      "step": 3584
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.9473764886686016e-05,
+      "loss": 1.0078,
+      "step": 3585
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.9473472999592402e-05,
+      "loss": 0.2109,
+      "step": 3586
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.9473181033016784e-05,
+      "loss": 0.0001,
+      "step": 3587
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.9472888986960765e-05,
+      "loss": 0.0018,
+      "step": 3588
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.9472596861425936e-05,
+      "loss": 0.0098,
+      "step": 3589
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.9472304656413915e-05,
+      "loss": 0.0002,
+      "step": 3590
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.9472012371926304e-05,
+      "loss": 0.0006,
+      "step": 3591
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.94717200079647e-05,
+      "loss": 0.002,
+      "step": 3592
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.947142756453072e-05,
+      "loss": 0.0002,
+      "step": 3593
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.9471135041625963e-05,
+      "loss": 0.0947,
+      "step": 3594
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.9470842439252036e-05,
+      "loss": 0.0002,
+      "step": 3595
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.947054975741055e-05,
+      "loss": 0.0001,
+      "step": 3596
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.9470256996103103e-05,
+      "loss": 0.0003,
+      "step": 3597
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.9469964155331313e-05,
+      "loss": 0.1387,
+      "step": 3598
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.9469671235096784e-05,
+      "loss": 0.0,
+      "step": 3599
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.9469378235401118e-05,
+      "loss": 0.1279,
+      "step": 3600
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.946908515624594e-05,
+      "loss": 0.5977,
+      "step": 3601
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.9468791997632844e-05,
+      "loss": 1.1719,
+      "step": 3602
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.9468498759563448e-05,
+      "loss": 0.7773,
+      "step": 3603
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.946820544203936e-05,
+      "loss": 0.033,
+      "step": 3604
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.9467912045062186e-05,
+      "loss": 0.7734,
+      "step": 3605
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.9467618568633548e-05,
+      "loss": 0.6328,
+      "step": 3606
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.946732501275505e-05,
+      "loss": 0.8398,
+      "step": 3607
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.9467031377428305e-05,
+      "loss": 1.9062,
+      "step": 3608
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.9466737662654927e-05,
+      "loss": 0.0275,
+      "step": 3609
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.946644386843653e-05,
+      "loss": 0.8164,
+      "step": 3610
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.9466149994774724e-05,
+      "loss": 1.8281,
+      "step": 3611
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.9465856041671122e-05,
+      "loss": 0.918,
+      "step": 3612
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.9465562009127346e-05,
+      "loss": 0.0737,
+      "step": 3613
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.9465267897145e-05,
+      "loss": 0.0388,
+      "step": 3614
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.946497370572571e-05,
+      "loss": 0.0156,
+      "step": 3615
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.9464679434871084e-05,
+      "loss": 0.0469,
+      "step": 3616
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.9464385084582743e-05,
+      "loss": 0.0013,
+      "step": 3617
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.9464090654862294e-05,
+      "loss": 1.6172,
+      "step": 3618
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.9463796145711364e-05,
+      "loss": 0.0007,
+      "step": 3619
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.9463501557131567e-05,
+      "loss": 0.6328,
+      "step": 3620
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.9463206889124517e-05,
+      "loss": 0.0022,
+      "step": 3621
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.9462912141691846e-05,
+      "loss": 0.0078,
+      "step": 3622
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.946261731483515e-05,
+      "loss": 0.1226,
+      "step": 3623
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.946232240855607e-05,
+      "loss": 0.0708,
+      "step": 3624
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.9462027422856212e-05,
+      "loss": 0.1504,
+      "step": 3625
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.94617323577372e-05,
+      "loss": 0.0015,
+      "step": 3626
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.9461437213200653e-05,
+      "loss": 0.0001,
+      "step": 3627
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.9461141989248194e-05,
+      "loss": 0.0104,
+      "step": 3628
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.9460846685881446e-05,
+      "loss": 0.0049,
+      "step": 3629
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.9460551303102022e-05,
+      "loss": 0.3535,
+      "step": 3630
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.9460255840911555e-05,
+      "loss": 0.0947,
+      "step": 3631
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.9459960299311665e-05,
+      "loss": 0.0002,
+      "step": 3632
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.9459664678303964e-05,
+      "loss": 0.0315,
+      "step": 3633
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.945936897789009e-05,
+      "loss": 0.3906,
+      "step": 3634
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.945907319807166e-05,
+      "loss": 0.0001,
+      "step": 3635
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.9458777338850303e-05,
+      "loss": 0.0,
+      "step": 3636
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.9458481400227635e-05,
+      "loss": 0.0001,
+      "step": 3637
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.945818538220529e-05,
+      "loss": 0.0001,
+      "step": 3638
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.9457889284784885e-05,
+      "loss": 0.0,
+      "step": 3639
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.9457593107968052e-05,
+      "loss": 0.0001,
+      "step": 3640
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.9457296851756415e-05,
+      "loss": 0.0002,
+      "step": 3641
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.94570005161516e-05,
+      "loss": 0.0,
+      "step": 3642
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.9456704101155242e-05,
+      "loss": 0.9922,
+      "step": 3643
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.9456407606768966e-05,
+      "loss": 0.1826,
+      "step": 3644
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.9456111032994394e-05,
+      "loss": 0.0,
+      "step": 3645
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.9455814379833157e-05,
+      "loss": 0.1982,
+      "step": 3646
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.9455517647286884e-05,
+      "loss": 0.208,
+      "step": 3647
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.9455220835357208e-05,
+      "loss": 0.0476,
+      "step": 3648
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.9454923944045753e-05,
+      "loss": 0.1689,
+      "step": 3649
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.9454626973354158e-05,
+      "loss": 0.2012,
+      "step": 3650
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.9454329923284045e-05,
+      "loss": 0.5195,
+      "step": 3651
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.9454032793837053e-05,
+      "loss": 0.5117,
+      "step": 3652
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.9453735585014806e-05,
+      "loss": 0.7148,
+      "step": 3653
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.9453438296818948e-05,
+      "loss": 0.0405,
+      "step": 3654
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.9453140929251095e-05,
+      "loss": 0.7109,
+      "step": 3655
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.945284348231289e-05,
+      "loss": 1.8516,
+      "step": 3656
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.9452545956005968e-05,
+      "loss": 0.7852,
+      "step": 3657
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.945224835033196e-05,
+      "loss": 0.4102,
+      "step": 3658
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.94519506652925e-05,
+      "loss": 0.5391,
+      "step": 3659
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.945165290088922e-05,
+      "loss": 0.543,
+      "step": 3660
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.945135505712376e-05,
+      "loss": 0.3086,
+      "step": 3661
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.9451057133997755e-05,
+      "loss": 0.9141,
+      "step": 3662
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.945075913151284e-05,
+      "loss": 0.5078,
+      "step": 3663
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.945046104967065e-05,
+      "loss": 2.5469,
+      "step": 3664
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.9450162888472825e-05,
+      "loss": 0.1973,
+      "step": 3665
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.9449864647920996e-05,
+      "loss": 0.582,
+      "step": 3666
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.944956632801681e-05,
+      "loss": 0.0859,
+      "step": 3667
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.9449267928761902e-05,
+      "loss": 0.125,
+      "step": 3668
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.9448969450157907e-05,
+      "loss": 0.0001,
+      "step": 3669
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.9448670892206467e-05,
+      "loss": 0.0122,
+      "step": 3670
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.944837225490922e-05,
+      "loss": 1.7734,
+      "step": 3671
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.944807353826781e-05,
+      "loss": 0.0003,
+      "step": 3672
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.944777474228387e-05,
+      "loss": 0.6875,
+      "step": 3673
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.944747586695905e-05,
+      "loss": 0.3965,
+      "step": 3674
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.9447176912294985e-05,
+      "loss": 0.0012,
+      "step": 3675
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.944687787829332e-05,
+      "loss": 0.0,
+      "step": 3676
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.9446578764955692e-05,
+      "loss": 0.064,
+      "step": 3677
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.9446279572283752e-05,
+      "loss": 0.0,
+      "step": 3678
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.9445980300279134e-05,
+      "loss": 0.0189,
+      "step": 3679
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.9445680948943486e-05,
+      "loss": 0.4902,
+      "step": 3680
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.9445381518278453e-05,
+      "loss": 0.0008,
+      "step": 3681
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.9445082008285677e-05,
+      "loss": 0.0001,
+      "step": 3682
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.9444782418966806e-05,
+      "loss": 0.0,
+      "step": 3683
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.944448275032348e-05,
+      "loss": 0.6797,
+      "step": 3684
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.944418300235735e-05,
+      "loss": 0.0,
+      "step": 3685
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.944388317507006e-05,
+      "loss": 0.0003,
+      "step": 3686
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.9443583268463255e-05,
+      "loss": 0.0,
+      "step": 3687
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.9443283282538583e-05,
+      "loss": 0.0,
+      "step": 3688
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.9442983217297686e-05,
+      "loss": 0.0,
+      "step": 3689
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.9442683072742226e-05,
+      "loss": 0.0,
+      "step": 3690
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.944238284887384e-05,
+      "loss": 0.0003,
+      "step": 3691
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.9442082545694178e-05,
+      "loss": 0.0967,
+      "step": 3692
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.9441782163204893e-05,
+      "loss": 0.0903,
+      "step": 3693
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.9441481701407632e-05,
+      "loss": 0.1016,
+      "step": 3694
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.944118116030404e-05,
+      "loss": 0.0933,
+      "step": 3695
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.9440880539895778e-05,
+      "loss": 0.1084,
+      "step": 3696
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.944057984018449e-05,
+      "loss": 0.8594,
+      "step": 3697
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.9440279061171828e-05,
+      "loss": 0.165,
+      "step": 3698
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.9439978202859443e-05,
+      "loss": 0.1191,
+      "step": 3699
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.943967726524899e-05,
+      "loss": 0.2041,
+      "step": 3700
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.9439376248342117e-05,
+      "loss": 0.8789,
+      "step": 3701
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.9439075152140486e-05,
+      "loss": 1.0859,
+      "step": 3702
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.9438773976645742e-05,
+      "loss": 1.4453,
+      "step": 3703
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.9438472721859543e-05,
+      "loss": 0.2109,
+      "step": 3704
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.943817138778354e-05,
+      "loss": 0.7148,
+      "step": 3705
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.943786997441939e-05,
+      "loss": 0.3398,
+      "step": 3706
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.943756848176875e-05,
+      "loss": 0.6133,
+      "step": 3707
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.943726690983327e-05,
+      "loss": 0.3496,
+      "step": 3708
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.9436965258614616e-05,
+      "loss": 0.4375,
+      "step": 3709
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.9436663528114435e-05,
+      "loss": 1.1172,
+      "step": 3710
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.9436361718334387e-05,
+      "loss": 0.5469,
+      "step": 3711
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.943605982927613e-05,
+      "loss": 0.4258,
+      "step": 3712
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.943575786094132e-05,
+      "loss": 0.0991,
+      "step": 3713
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.943545581333162e-05,
+      "loss": 0.7031,
+      "step": 3714
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.9435153686448684e-05,
+      "loss": 0.2676,
+      "step": 3715
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.9434851480294174e-05,
+      "loss": 1.0078,
+      "step": 3716
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.9434549194869743e-05,
+      "loss": 0.0152,
+      "step": 3717
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.9434246830177062e-05,
+      "loss": 0.0503,
+      "step": 3718
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.9433944386217784e-05,
+      "loss": 0.0095,
+      "step": 3719
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.9433641862993573e-05,
+      "loss": 0.0027,
+      "step": 3720
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.943333926050609e-05,
+      "loss": 0.7266,
+      "step": 3721
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.9433036578756992e-05,
+      "loss": 0.8086,
+      "step": 3722
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.9432733817747946e-05,
+      "loss": 0.0006,
+      "step": 3723
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.9432430977480614e-05,
+      "loss": 0.2617,
+      "step": 3724
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.9432128057956663e-05,
+      "loss": 0.3027,
+      "step": 3725
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.9431825059177743e-05,
+      "loss": 0.0,
+      "step": 3726
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.9431521981145536e-05,
+      "loss": 0.1621,
+      "step": 3727
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.9431218823861693e-05,
+      "loss": 0.1157,
+      "step": 3728
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.9430915587327886e-05,
+      "loss": 0.0,
+      "step": 3729
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.9430612271545775e-05,
+      "loss": 0.0444,
+      "step": 3730
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.9430308876517026e-05,
+      "loss": 0.0001,
+      "step": 3731
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.943000540224331e-05,
+      "loss": 0.1543,
+      "step": 3732
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.9429701848726295e-05,
+      "loss": 0.0002,
+      "step": 3733
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.942939821596764e-05,
+      "loss": 0.0,
+      "step": 3734
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.9429094503969014e-05,
+      "loss": 0.0001,
+      "step": 3735
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.9428790712732087e-05,
+      "loss": 0.0105,
+      "step": 3736
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.942848684225853e-05,
+      "loss": 0.0107,
+      "step": 3737
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.9428182892550012e-05,
+      "loss": 0.0001,
+      "step": 3738
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.9427878863608196e-05,
+      "loss": 0.0,
+      "step": 3739
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.9427574755434755e-05,
+      "loss": 0.0,
+      "step": 3740
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.9427270568031357e-05,
+      "loss": 0.0,
+      "step": 3741
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.942696630139968e-05,
+      "loss": 0.2598,
+      "step": 3742
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.9426661955541385e-05,
+      "loss": 0.0,
+      "step": 3743
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.942635753045815e-05,
+      "loss": 0.1865,
+      "step": 3744
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.942605302615164e-05,
+      "loss": 0.1553,
+      "step": 3745
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.9425748442623534e-05,
+      "loss": 0.1836,
+      "step": 3746
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.9425443779875505e-05,
+      "loss": 0.1719,
+      "step": 3747
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.942513903790922e-05,
+      "loss": 0.2676,
+      "step": 3748
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.9424834216726356e-05,
+      "loss": 0.1895,
+      "step": 3749
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.942452931632859e-05,
+      "loss": 0.1631,
+      "step": 3750
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.942422433671759e-05,
+      "loss": 0.5664,
+      "step": 3751
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.9423919277895033e-05,
+      "loss": 0.6914,
+      "step": 3752
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.9423614139862598e-05,
+      "loss": 0.4043,
+      "step": 3753
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.9423308922621956e-05,
+      "loss": 0.3145,
+      "step": 3754
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.942300362617479e-05,
+      "loss": 1.2891,
+      "step": 3755
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.9422698250522767e-05,
+      "loss": 0.4395,
+      "step": 3756
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.9422392795667568e-05,
+      "loss": 0.7344,
+      "step": 3757
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.942208726161087e-05,
+      "loss": 1.4062,
+      "step": 3758
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.9421781648354354e-05,
+      "loss": 0.332,
+      "step": 3759
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.9421475955899698e-05,
+      "loss": 0.1128,
+      "step": 3760
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.942117018424858e-05,
+      "loss": 0.75,
+      "step": 3761
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.942086433340267e-05,
+      "loss": 0.7266,
+      "step": 3762
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.942055840336366e-05,
+      "loss": 0.4473,
+      "step": 3763
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.9420252394133228e-05,
+      "loss": 0.5508,
+      "step": 3764
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.941994630571305e-05,
+      "loss": 0.0894,
+      "step": 3765
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.941964013810481e-05,
+      "loss": 0.4707,
+      "step": 3766
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.9419333891310187e-05,
+      "loss": 0.0481,
+      "step": 3767
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.9419027565330868e-05,
+      "loss": 0.2891,
+      "step": 3768
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.9418721160168527e-05,
+      "loss": 0.0175,
+      "step": 3769
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.9418414675824852e-05,
+      "loss": 0.0129,
+      "step": 3770
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.9418108112301526e-05,
+      "loss": 0.0035,
+      "step": 3771
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.941780146960023e-05,
+      "loss": 0.2383,
+      "step": 3772
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.9417494747722655e-05,
+      "loss": 0.0029,
+      "step": 3773
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.941718794667047e-05,
+      "loss": 2.0312,
+      "step": 3774
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.9416881066445382e-05,
+      "loss": 0.2539,
+      "step": 3775
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.9416574107049055e-05,
+      "loss": 0.5117,
+      "step": 3776
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.9416267068483188e-05,
+      "loss": 0.0002,
+      "step": 3777
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.941595995074946e-05,
+      "loss": 0.0046,
+      "step": 3778
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.941565275384956e-05,
+      "loss": 0.0001,
+      "step": 3779
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.941534547778518e-05,
+      "loss": 0.0002,
+      "step": 3780
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.9415038122557996e-05,
+      "loss": 0.0012,
+      "step": 3781
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.941473068816971e-05,
+      "loss": 0.2578,
+      "step": 3782
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.9414423174621995e-05,
+      "loss": 0.0005,
+      "step": 3783
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.9414115581916556e-05,
+      "loss": 0.668,
+      "step": 3784
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.9413807910055067e-05,
+      "loss": 0.167,
+      "step": 3785
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.9413500159039227e-05,
+      "loss": 1.8516,
+      "step": 3786
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.9413192328870726e-05,
+      "loss": 0.0,
+      "step": 3787
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.941288441955125e-05,
+      "loss": 0.0001,
+      "step": 3788
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.941257643108249e-05,
+      "loss": 0.6445,
+      "step": 3789
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.9412268363466143e-05,
+      "loss": 0.0105,
+      "step": 3790
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.9411960216703896e-05,
+      "loss": 0.3398,
+      "step": 3791
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.9411651990797438e-05,
+      "loss": 0.0,
+      "step": 3792
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.9411343685748467e-05,
+      "loss": 0.0,
+      "step": 3793
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.941103530155868e-05,
+      "loss": 1.75,
+      "step": 3794
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.941072683822976e-05,
+      "loss": 0.0182,
+      "step": 3795
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.9410418295763405e-05,
+      "loss": 1.6406,
+      "step": 3796
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.9410109674161318e-05,
+      "loss": 0.1465,
+      "step": 3797
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.9409800973425182e-05,
+      "loss": 0.1719,
+      "step": 3798
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.9409492193556696e-05,
+      "loss": 0.2207,
+      "step": 3799
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.940918333455756e-05,
+      "loss": 0.1699,
+      "step": 3800
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.9408874396429462e-05,
+      "loss": 1.3438,
+      "step": 3801
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.9408565379174108e-05,
+      "loss": 1.3984,
+      "step": 3802
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.9408256282793185e-05,
+      "loss": 0.582,
+      "step": 3803
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.9407947107288404e-05,
+      "loss": 1.125,
+      "step": 3804
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.9407637852661445e-05,
+      "loss": 1.0312,
+      "step": 3805
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.940732851891402e-05,
+      "loss": 1.4375,
+      "step": 3806
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.9407019106047824e-05,
+      "loss": 0.6172,
+      "step": 3807
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.9406709614064554e-05,
+      "loss": 0.4453,
+      "step": 3808
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.9406400042965912e-05,
+      "loss": 0.8516,
+      "step": 3809
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.9406090392753594e-05,
+      "loss": 0.3125,
+      "step": 3810
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.940578066342931e-05,
+      "loss": 0.6289,
+      "step": 3811
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.9405470854994755e-05,
+      "loss": 0.3555,
+      "step": 3812
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.940516096745162e-05,
+      "loss": 0.6797,
+      "step": 3813
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.9404851000801628e-05,
+      "loss": 0.0001,
+      "step": 3814
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.9404540955046466e-05,
+      "loss": 0.3711,
+      "step": 3815
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.9404230830187838e-05,
+      "loss": 1.1953,
+      "step": 3816
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.940392062622745e-05,
+      "loss": 0.4824,
+      "step": 3817
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.9403610343167008e-05,
+      "loss": 0.0039,
+      "step": 3818
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.9403299981008207e-05,
+      "loss": 0.3848,
+      "step": 3819
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.9402989539752765e-05,
+      "loss": 0.5312,
+      "step": 3820
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.9402679019402372e-05,
+      "loss": 0.0393,
+      "step": 3821
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.9402368419958746e-05,
+      "loss": 1.1953,
+      "step": 3822
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.9402057741423585e-05,
+      "loss": 0.0001,
+      "step": 3823
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.9401746983798596e-05,
+      "loss": 0.5195,
+      "step": 3824
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.940143614708549e-05,
+      "loss": 0.0425,
+      "step": 3825
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.940112523128597e-05,
+      "loss": 0.0498,
+      "step": 3826
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.9400814236401742e-05,
+      "loss": 0.0021,
+      "step": 3827
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.9400503162434518e-05,
+      "loss": 0.1895,
+      "step": 3828
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.9400192009386004e-05,
+      "loss": 1.0391,
+      "step": 3829
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.939988077725791e-05,
+      "loss": 0.0001,
+      "step": 3830
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.939956946605194e-05,
+      "loss": 0.0001,
+      "step": 3831
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.9399258075769814e-05,
+      "loss": 0.0007,
+      "step": 3832
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.939894660641323e-05,
+      "loss": 0.0002,
+      "step": 3833
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.9398635057983907e-05,
+      "loss": 0.0001,
+      "step": 3834
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.9398323430483556e-05,
+      "loss": 0.0308,
+      "step": 3835
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.939801172391388e-05,
+      "loss": 0.8672,
+      "step": 3836
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.93976999382766e-05,
+      "loss": 0.0011,
+      "step": 3837
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.9397388073573422e-05,
+      "loss": 0.0001,
+      "step": 3838
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.9397076129806062e-05,
+      "loss": 0.0006,
+      "step": 3839
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.9396764106976234e-05,
+      "loss": 0.0019,
+      "step": 3840
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.9396452005085648e-05,
+      "loss": 0.0002,
+      "step": 3841
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.9396139824136022e-05,
+      "loss": 0.3066,
+      "step": 3842
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.9395827564129064e-05,
+      "loss": 0.7344,
+      "step": 3843
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.939551522506649e-05,
+      "loss": 0.0001,
+      "step": 3844
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.9395202806950025e-05,
+      "loss": 0.2363,
+      "step": 3845
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.9394890309781373e-05,
+      "loss": 0.0001,
+      "step": 3846
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.9394577733562257e-05,
+      "loss": 0.208,
+      "step": 3847
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.9394265078294388e-05,
+      "loss": 0.1133,
+      "step": 3848
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.9393952343979493e-05,
+      "loss": 0.1865,
+      "step": 3849
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.9393639530619276e-05,
+      "loss": 0.2236,
+      "step": 3850
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.9393326638215463e-05,
+      "loss": 1.7031,
+      "step": 3851
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.939301366676977e-05,
+      "loss": 0.9805,
+      "step": 3852
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.939270061628392e-05,
+      "loss": 0.1533,
+      "step": 3853
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.9392387486759627e-05,
+      "loss": 0.9609,
+      "step": 3854
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.939207427819861e-05,
+      "loss": 0.2715,
+      "step": 3855
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.939176099060259e-05,
+      "loss": 0.4707,
+      "step": 3856
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.939144762397329e-05,
+      "loss": 0.8047,
+      "step": 3857
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.9391134178312433e-05,
+      "loss": 0.126,
+      "step": 3858
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.9390820653621733e-05,
+      "loss": 0.6172,
+      "step": 3859
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.9390507049902917e-05,
+      "loss": 1.2344,
+      "step": 3860
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.9390193367157706e-05,
+      "loss": 2.0781,
+      "step": 3861
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.938987960538782e-05,
+      "loss": 0.418,
+      "step": 3862
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.9389565764594986e-05,
+      "loss": 0.5938,
+      "step": 3863
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.9389251844780923e-05,
+      "loss": 0.4629,
+      "step": 3864
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.9388937845947363e-05,
+      "loss": 0.5859,
+      "step": 3865
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.938862376809602e-05,
+      "loss": 0.1553,
+      "step": 3866
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.9388309611228627e-05,
+      "loss": 1.0625,
+      "step": 3867
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.9387995375346903e-05,
+      "loss": 1.2969,
+      "step": 3868
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.938768106045258e-05,
+      "loss": 0.5586,
+      "step": 3869
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.938736666654738e-05,
+      "loss": 0.5195,
+      "step": 3870
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.9387052193633027e-05,
+      "loss": 1.9062,
+      "step": 3871
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.9386737641711255e-05,
+      "loss": 0.165,
+      "step": 3872
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.9386423010783783e-05,
+      "loss": 0.3574,
+      "step": 3873
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.938610830085235e-05,
+      "loss": 0.3887,
+      "step": 3874
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.9385793511918673e-05,
+      "loss": 1.2734,
+      "step": 3875
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.9385478643984484e-05,
+      "loss": 0.0043,
+      "step": 3876
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.9385163697051517e-05,
+      "loss": 0.01,
+      "step": 3877
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.9384848671121495e-05,
+      "loss": 0.0198,
+      "step": 3878
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.9384533566196155e-05,
+      "loss": 0.5391,
+      "step": 3879
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.938421838227722e-05,
+      "loss": 0.1689,
+      "step": 3880
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.9383903119366427e-05,
+      "loss": 0.0029,
+      "step": 3881
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.93835877774655e-05,
+      "loss": 0.0004,
+      "step": 3882
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.9383272356576182e-05,
+      "loss": 0.0092,
+      "step": 3883
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.938295685670019e-05,
+      "loss": 0.0047,
+      "step": 3884
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.9382641277839273e-05,
+      "loss": 0.0005,
+      "step": 3885
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.9382325619995154e-05,
+      "loss": 0.0018,
+      "step": 3886
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.9382009883169564e-05,
+      "loss": 0.2158,
+      "step": 3887
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.9381694067364243e-05,
+      "loss": 0.0003,
+      "step": 3888
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.9381378172580926e-05,
+      "loss": 0.0005,
+      "step": 3889
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.938106219882135e-05,
+      "loss": 0.0001,
+      "step": 3890
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.9380746146087238e-05,
+      "loss": 0.0001,
+      "step": 3891
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.9380430014380332e-05,
+      "loss": 0.0006,
+      "step": 3892
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.9380113803702376e-05,
+      "loss": 0.0,
+      "step": 3893
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.9379797514055098e-05,
+      "loss": 0.3652,
+      "step": 3894
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.9379481145440233e-05,
+      "loss": 0.0967,
+      "step": 3895
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.9379164697859525e-05,
+      "loss": 0.1924,
+      "step": 3896
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.937884817131471e-05,
+      "loss": 0.0,
+      "step": 3897
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.937853156580752e-05,
+      "loss": 0.127,
+      "step": 3898
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.9378214881339704e-05,
+      "loss": 0.1172,
+      "step": 3899
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.9377898117912997e-05,
+      "loss": 0.1396,
+      "step": 3900
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.9377581275529134e-05,
+      "loss": 0.6562,
+      "step": 3901
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.9377264354189855e-05,
+      "loss": 1.4297,
+      "step": 3902
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.9376947353896906e-05,
+      "loss": 0.2178,
+      "step": 3903
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.937663027465203e-05,
+      "loss": 0.7344,
+      "step": 3904
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.9376313116456958e-05,
+      "loss": 1.2031,
+      "step": 3905
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.9375995879313445e-05,
+      "loss": 0.1582,
+      "step": 3906
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.9375678563223222e-05,
+      "loss": 0.6523,
+      "step": 3907
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.9375361168188034e-05,
+      "loss": 1.0703,
+      "step": 3908
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.937504369420962e-05,
+      "loss": 0.5312,
+      "step": 3909
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.9374726141289737e-05,
+      "loss": 0.7266,
+      "step": 3910
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.9374408509430117e-05,
+      "loss": 0.7461,
+      "step": 3911
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.937409079863251e-05,
+      "loss": 0.0383,
+      "step": 3912
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.9373773008898653e-05,
+      "loss": 0.6289,
+      "step": 3913
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.9373455140230302e-05,
+      "loss": 0.5508,
+      "step": 3914
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.9373137192629197e-05,
+      "loss": 2.2812,
+      "step": 3915
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.9372819166097086e-05,
+      "loss": 0.5977,
+      "step": 3916
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.9372501060635707e-05,
+      "loss": 0.0466,
+      "step": 3917
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.937218287624682e-05,
+      "loss": 0.9336,
+      "step": 3918
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.937186461293216e-05,
+      "loss": 2.2656,
+      "step": 3919
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.9371546270693486e-05,
+      "loss": 0.5156,
+      "step": 3920
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.937122784953254e-05,
+      "loss": 0.1436,
+      "step": 3921
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.937090934945107e-05,
+      "loss": 0.2275,
+      "step": 3922
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.937059077045083e-05,
+      "loss": 0.9844,
+      "step": 3923
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.9370272112533564e-05,
+      "loss": 0.2832,
+      "step": 3924
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.9369953375701023e-05,
+      "loss": 1.2656,
+      "step": 3925
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.936963455995496e-05,
+      "loss": 0.2656,
+      "step": 3926
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.936931566529712e-05,
+      "loss": 0.0056,
+      "step": 3927
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.9368996691729265e-05,
+      "loss": 0.0262,
+      "step": 3928
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.9368677639253135e-05,
+      "loss": 0.1426,
+      "step": 3929
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.936835850787049e-05,
+      "loss": 0.1914,
+      "step": 3930
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.936803929758308e-05,
+      "loss": 0.1543,
+      "step": 3931
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.9367720008392658e-05,
+      "loss": 0.5781,
+      "step": 3932
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.9367400640300976e-05,
+      "loss": 0.0004,
+      "step": 3933
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.9367081193309785e-05,
+      "loss": 0.6055,
+      "step": 3934
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.9366761667420852e-05,
+      "loss": 0.0005,
+      "step": 3935
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.9366442062635913e-05,
+      "loss": 0.0001,
+      "step": 3936
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.9366122378956744e-05,
+      "loss": 0.0,
+      "step": 3937
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.936580261638508e-05,
+      "loss": 0.0001,
+      "step": 3938
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.9365482774922694e-05,
+      "loss": 0.0205,
+      "step": 3939
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.9365162854571328e-05,
+      "loss": 0.0004,
+      "step": 3940
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.936484285533275e-05,
+      "loss": 0.0001,
+      "step": 3941
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.9364522777208715e-05,
+      "loss": 0.0028,
+      "step": 3942
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.9364202620200974e-05,
+      "loss": 0.7891,
+      "step": 3943
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.9363882384311295e-05,
+      "loss": 0.0,
+      "step": 3944
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.9363562069541426e-05,
+      "loss": 0.0001,
+      "step": 3945
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.9363241675893134e-05,
+      "loss": 0.7109,
+      "step": 3946
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.936292120336818e-05,
+      "loss": 0.1177,
+      "step": 3947
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.9362600651968316e-05,
+      "loss": 0.0996,
+      "step": 3948
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.936228002169531e-05,
+      "loss": 0.4531,
+      "step": 3949
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.9361959312550915e-05,
+      "loss": 0.165,
+      "step": 3950
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.93616385245369e-05,
+      "loss": 0.5352,
+      "step": 3951
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.9361317657655018e-05,
+      "loss": 1.6562,
+      "step": 3952
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.936099671190704e-05,
+      "loss": 1.625,
+      "step": 3953
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.9360675687294727e-05,
+      "loss": 0.2734,
+      "step": 3954
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.9360354583819832e-05,
+      "loss": 0.7227,
+      "step": 3955
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.9360033401484134e-05,
+      "loss": 0.9219,
+      "step": 3956
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.9359712140289384e-05,
+      "loss": 2.0469,
+      "step": 3957
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.935939080023735e-05,
+      "loss": 0.3477,
+      "step": 3958
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.9359069381329798e-05,
+      "loss": 0.5742,
+      "step": 3959
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.9358747883568496e-05,
+      "loss": 0.832,
+      "step": 3960
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.93584263069552e-05,
+      "loss": 0.2314,
+      "step": 3961
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.9358104651491692e-05,
+      "loss": 0.6523,
+      "step": 3962
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.935778291717972e-05,
+      "loss": 0.0972,
+      "step": 3963
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.935746110402106e-05,
+      "loss": 0.1572,
+      "step": 3964
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.9357139212017484e-05,
+      "loss": 0.1455,
+      "step": 3965
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.9356817241170747e-05,
+      "loss": 0.3379,
+      "step": 3966
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.9356495191482632e-05,
+      "loss": 1.4297,
+      "step": 3967
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.9356173062954892e-05,
+      "loss": 0.5703,
+      "step": 3968
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.935585085558931e-05,
+      "loss": 0.0101,
+      "step": 3969
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.9355528569387643e-05,
+      "loss": 0.8164,
+      "step": 3970
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.9355206204351672e-05,
+      "loss": 1.6797,
+      "step": 3971
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.9354883760483164e-05,
+      "loss": 1.0234,
+      "step": 3972
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.9354561237783885e-05,
+      "loss": 1.1562,
+      "step": 3973
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.9354238636255607e-05,
+      "loss": 0.0258,
+      "step": 3974
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.935391595590011e-05,
+      "loss": 0.0171,
+      "step": 3975
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.9353593196719157e-05,
+      "loss": 1.0625,
+      "step": 3976
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.9353270358714523e-05,
+      "loss": 0.0013,
+      "step": 3977
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.935294744188798e-05,
+      "loss": 0.0004,
+      "step": 3978
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.9352624446241307e-05,
+      "loss": 0.5703,
+      "step": 3979
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.9352301371776272e-05,
+      "loss": 0.0002,
+      "step": 3980
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.935197821849465e-05,
+      "loss": 0.0002,
+      "step": 3981
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.935165498639822e-05,
+      "loss": 0.3105,
+      "step": 3982
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.9351331675488748e-05,
+      "loss": 0.0003,
+      "step": 3983
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.935100828576801e-05,
+      "loss": 0.0001,
+      "step": 3984
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.9350684817237797e-05,
+      "loss": 0.0001,
+      "step": 3985
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.935036126989987e-05,
+      "loss": 0.0033,
+      "step": 3986
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.9350037643756007e-05,
+      "loss": 0.0008,
+      "step": 3987
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.9349713938807995e-05,
+      "loss": 0.0138,
+      "step": 3988
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.9349390155057604e-05,
+      "loss": 0.1406,
+      "step": 3989
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.9349066292506613e-05,
+      "loss": 0.0503,
+      "step": 3990
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.93487423511568e-05,
+      "loss": 0.0003,
+      "step": 3991
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.934841833100995e-05,
+      "loss": 0.0002,
+      "step": 3992
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.9348094232067835e-05,
+      "loss": 0.0003,
+      "step": 3993
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.9347770054332232e-05,
+      "loss": 0.0001,
+      "step": 3994
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.934744579780493e-05,
+      "loss": 0.0018,
+      "step": 3995
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.934712146248771e-05,
+      "loss": 0.1797,
+      "step": 3996
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.934679704838235e-05,
+      "loss": 0.1523,
+      "step": 3997
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.9346472555490625e-05,
+      "loss": 0.1226,
+      "step": 3998
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.9346147983814327e-05,
+      "loss": 0.1177,
+      "step": 3999
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.9345823333355233e-05,
+      "loss": 0.2217,
+      "step": 4000
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.9345498604115126e-05,
+      "loss": 1.2656,
+      "step": 4001
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.9345173796095792e-05,
+      "loss": 0.4277,
+      "step": 4002
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.934484890929901e-05,
+      "loss": 0.668,
+      "step": 4003
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.9344523943726572e-05,
+      "loss": 0.8828,
+      "step": 4004
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.934419889938025e-05,
+      "loss": 0.375,
+      "step": 4005
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.9343873776261845e-05,
+      "loss": 0.5742,
+      "step": 4006
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.9343548574373132e-05,
+      "loss": 1.5312,
+      "step": 4007
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.93432232937159e-05,
+      "loss": 0.0149,
+      "step": 4008
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.9342897934291935e-05,
+      "loss": 0.1934,
+      "step": 4009
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.934257249610302e-05,
+      "loss": 0.4062,
+      "step": 4010
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.9342246979150945e-05,
+      "loss": 0.0649,
+      "step": 4011
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.93419213834375e-05,
+      "loss": 0.2363,
+      "step": 4012
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.934159570896447e-05,
+      "loss": 0.1748,
+      "step": 4013
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.9341269955733646e-05,
+      "loss": 0.7617,
+      "step": 4014
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.9340944123746817e-05,
+      "loss": 0.875,
+      "step": 4015
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.9340618213005765e-05,
+      "loss": 1.1172,
+      "step": 4016
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.934029222351229e-05,
+      "loss": 0.165,
+      "step": 4017
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.9339966155268174e-05,
+      "loss": 0.0527,
+      "step": 4018
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.933964000827521e-05,
+      "loss": 0.0679,
+      "step": 4019
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.9339313782535194e-05,
+      "loss": 1.0234,
+      "step": 4020
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.9338987478049914e-05,
+      "loss": 1.3672,
+      "step": 4021
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.933866109482116e-05,
+      "loss": 0.0854,
+      "step": 4022
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.933833463285073e-05,
+      "loss": 2.0156,
+      "step": 4023
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.9338008092140404e-05,
+      "loss": 0.3184,
+      "step": 4024
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.933768147269199e-05,
+      "loss": 0.003,
+      "step": 4025
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.9337354774507278e-05,
+      "loss": 0.373,
+      "step": 4026
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.9337027997588054e-05,
+      "loss": 0.0002,
+      "step": 4027
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.9336701141936124e-05,
+      "loss": 0.1396,
+      "step": 4028
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.9336374207553272e-05,
+      "loss": 0.0064,
+      "step": 4029
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.9336047194441302e-05,
+      "loss": 0.0001,
+      "step": 4030
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.9335720102602008e-05,
+      "loss": 0.0001,
+      "step": 4031
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.933539293203718e-05,
+      "loss": 0.0,
+      "step": 4032
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.933506568274862e-05,
+      "loss": 0.2197,
+      "step": 4033
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.9334738354738126e-05,
+      "loss": 0.1553,
+      "step": 4034
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.9334410948007496e-05,
+      "loss": 0.4434,
+      "step": 4035
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.9334083462558524e-05,
+      "loss": 0.0089,
+      "step": 4036
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.933375589839301e-05,
+      "loss": 0.0001,
+      "step": 4037
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.9333428255512753e-05,
+      "loss": 0.0,
+      "step": 4038
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.9333100533919553e-05,
+      "loss": 0.0001,
+      "step": 4039
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.933277273361521e-05,
+      "loss": 0.0001,
+      "step": 4040
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.9332444854601523e-05,
+      "loss": 0.0025,
+      "step": 4041
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.933211689688029e-05,
+      "loss": 0.4531,
+      "step": 4042
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.9331788860453317e-05,
+      "loss": 0.0001,
+      "step": 4043
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.9331460745322406e-05,
+      "loss": 0.0001,
+      "step": 4044
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.9331132551489354e-05,
+      "loss": 0.1885,
+      "step": 4045
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.9330804278955964e-05,
+      "loss": 0.1416,
+      "step": 4046
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.933047592772404e-05,
+      "loss": 0.0002,
+      "step": 4047
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.933014749779539e-05,
+      "loss": 0.0003,
+      "step": 4048
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.932981898917181e-05,
+      "loss": 0.1108,
+      "step": 4049
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.9329490401855107e-05,
+      "loss": 0.103,
+      "step": 4050
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.9329161735847083e-05,
+      "loss": 0.6797,
+      "step": 4051
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.932883299114955e-05,
+      "loss": 0.8047,
+      "step": 4052
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.9328504167764304e-05,
+      "loss": 0.6992,
+      "step": 4053
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.932817526569316e-05,
+      "loss": 1.5859,
+      "step": 4054
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.9327846284937916e-05,
+      "loss": 0.1133,
+      "step": 4055
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.9327517225500384e-05,
+      "loss": 1.5859,
+      "step": 4056
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.9327188087382372e-05,
+      "loss": 0.5742,
+      "step": 4057
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.932685887058568e-05,
+      "loss": 0.3633,
+      "step": 4058
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.9326529575112126e-05,
+      "loss": 1.5,
+      "step": 4059
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.9326200200963507e-05,
+      "loss": 0.0962,
+      "step": 4060
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.932587074814164e-05,
+      "loss": 0.4512,
+      "step": 4061
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.9325541216648332e-05,
+      "loss": 0.4629,
+      "step": 4062
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.9325211606485397e-05,
+      "loss": 1.6797,
+      "step": 4063
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.9324881917654633e-05,
+      "loss": 0.832,
+      "step": 4064
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.9324552150157863e-05,
+      "loss": 0.0131,
+      "step": 4065
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.9324222303996892e-05,
+      "loss": 0.0391,
+      "step": 4066
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.932389237917353e-05,
+      "loss": 0.0094,
+      "step": 4067
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.93235623756896e-05,
+      "loss": 0.3066,
+      "step": 4068
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.93232322935469e-05,
+      "loss": 2.1406,
+      "step": 4069
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.9322902132747246e-05,
+      "loss": 0.4238,
+      "step": 4070
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.9322571893292456e-05,
+      "loss": 1.5703,
+      "step": 4071
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.9322241575184337e-05,
+      "loss": 0.0001,
+      "step": 4072
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.9321911178424713e-05,
+      "loss": 0.0,
+      "step": 4073
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.932158070301539e-05,
+      "loss": 0.0,
+      "step": 4074
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.9321250148958184e-05,
+      "loss": 0.0908,
+      "step": 4075
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.9320919516254914e-05,
+      "loss": 0.0001,
+      "step": 4076
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.932058880490739e-05,
+      "loss": 0.0352,
+      "step": 4077
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.9320258014917432e-05,
+      "loss": 0.125,
+      "step": 4078
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.9319927146286856e-05,
+      "loss": 0.0043,
+      "step": 4079
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.9319596199017478e-05,
+      "loss": 0.0001,
+      "step": 4080
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.931926517311112e-05,
+      "loss": 0.0001,
+      "step": 4081
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.931893406856959e-05,
+      "loss": 0.0292,
+      "step": 4082
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.9318602885394717e-05,
+      "loss": 0.0001,
+      "step": 4083
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.931827162358831e-05,
+      "loss": 0.0298,
+      "step": 4084
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.9317940283152196e-05,
+      "loss": 0.0094,
+      "step": 4085
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.9317608864088193e-05,
+      "loss": 0.0003,
+      "step": 4086
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.931727736639812e-05,
+      "loss": 2.4531,
+      "step": 4087
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.9316945790083792e-05,
+      "loss": 0.0001,
+      "step": 4088
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.931661413514704e-05,
+      "loss": 0.0,
+      "step": 4089
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.9316282401589684e-05,
+      "loss": 0.0,
+      "step": 4090
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.9315950589413533e-05,
+      "loss": 0.0002,
+      "step": 4091
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.9315618698620425e-05,
+      "loss": 1.3203,
+      "step": 4092
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.931528672921217e-05,
+      "loss": 0.0001,
+      "step": 4093
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.9314954681190602e-05,
+      "loss": 0.0003,
+      "step": 4094
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.9314622554557535e-05,
+      "loss": 0.0184,
+      "step": 4095
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.93142903493148e-05,
+      "loss": 0.0,
+      "step": 4096
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.9313958065464218e-05,
+      "loss": 0.1206,
+      "step": 4097
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.9313625703007617e-05,
+      "loss": 0.1699,
+      "step": 4098
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.9313293261946816e-05,
+      "loss": 0.2051,
+      "step": 4099
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.9312960742283645e-05,
+      "loss": 0.1787,
+      "step": 4100
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.931262814401993e-05,
+      "loss": 0.5234,
+      "step": 4101
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.9312295467157495e-05,
+      "loss": 0.2441,
+      "step": 4102
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.931196271169817e-05,
+      "loss": 1.5312,
+      "step": 4103
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.931162987764378e-05,
+      "loss": 0.8789,
+      "step": 4104
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.931129696499615e-05,
+      "loss": 1.0312,
+      "step": 4105
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.931096397375712e-05,
+      "loss": 0.3145,
+      "step": 4106
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.9310630903928505e-05,
+      "loss": 0.5117,
+      "step": 4107
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.9310297755512143e-05,
+      "loss": 0.2227,
+      "step": 4108
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.9309964528509857e-05,
+      "loss": 0.4238,
+      "step": 4109
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.9309631222923482e-05,
+      "loss": 0.2041,
+      "step": 4110
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.9309297838754843e-05,
+      "loss": 1.2812,
+      "step": 4111
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.9308964376005777e-05,
+      "loss": 0.5312,
+      "step": 4112
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.9308630834678112e-05,
+      "loss": 0.6133,
+      "step": 4113
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.930829721477368e-05,
+      "loss": 0.4551,
+      "step": 4114
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.9307963516294313e-05,
+      "loss": 0.7578,
+      "step": 4115
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.9307629739241848e-05,
+      "loss": 0.0427,
+      "step": 4116
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.930729588361811e-05,
+      "loss": 0.0007,
+      "step": 4117
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.9306961949424932e-05,
+      "loss": 0.0349,
+      "step": 4118
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.9306627936664156e-05,
+      "loss": 0.0466,
+      "step": 4119
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.9306293845337616e-05,
+      "loss": 0.1162,
+      "step": 4120
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.930595967544714e-05,
+      "loss": 0.2598,
+      "step": 4121
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.9305625426994565e-05,
+      "loss": 0.0288,
+      "step": 4122
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.9305291099981727e-05,
+      "loss": 0.1108,
+      "step": 4123
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.9304956694410462e-05,
+      "loss": 0.0,
+      "step": 4124
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.9304622210282608e-05,
+      "loss": 0.2031,
+      "step": 4125
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.9304287647600003e-05,
+      "loss": 0.3887,
+      "step": 4126
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.9303953006364478e-05,
+      "loss": 0.5195,
+      "step": 4127
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.9303618286577878e-05,
+      "loss": 0.0,
+      "step": 4128
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.9303283488242037e-05,
+      "loss": 0.0001,
+      "step": 4129
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.930294861135879e-05,
+      "loss": 0.0001,
+      "step": 4130
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.9302613655929987e-05,
+      "loss": 0.0003,
+      "step": 4131
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.9302278621957457e-05,
+      "loss": 0.5938,
+      "step": 4132
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.9301943509443046e-05,
+      "loss": 0.377,
+      "step": 4133
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.9301608318388588e-05,
+      "loss": 0.0256,
+      "step": 4134
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.930127304879593e-05,
+      "loss": 0.0001,
+      "step": 4135
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.930093770066691e-05,
+      "loss": 0.0003,
+      "step": 4136
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.9300602274003375e-05,
+      "loss": 0.1025,
+      "step": 4137
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.9300266768807162e-05,
+      "loss": 0.0,
+      "step": 4138
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.929993118508011e-05,
+      "loss": 0.3906,
+      "step": 4139
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.9299595522824066e-05,
+      "loss": 0.0452,
+      "step": 4140
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.9299259782040875e-05,
+      "loss": 0.0004,
+      "step": 4141
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.9298923962732377e-05,
+      "loss": 0.0001,
+      "step": 4142
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.929858806490042e-05,
+      "loss": 0.0002,
+      "step": 4143
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.9298252088546845e-05,
+      "loss": 0.1689,
+      "step": 4144
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.92979160336735e-05,
+      "loss": 0.0767,
+      "step": 4145
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.9297579900282227e-05,
+      "loss": 0.1836,
+      "step": 4146
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.9297243688374876e-05,
+      "loss": 0.1787,
+      "step": 4147
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.9296907397953294e-05,
+      "loss": 0.1426,
+      "step": 4148
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.9296571029019322e-05,
+      "loss": 0.1865,
+      "step": 4149
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.929623458157481e-05,
+      "loss": 0.2158,
+      "step": 4150
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.9295898055621607e-05,
+      "loss": 0.668,
+      "step": 4151
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.9295561451161557e-05,
+      "loss": 1.4688,
+      "step": 4152
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.9295224768196514e-05,
+      "loss": 0.1299,
+      "step": 4153
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.9294888006728325e-05,
+      "loss": 0.1416,
+      "step": 4154
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.9294551166758837e-05,
+      "loss": 0.2988,
+      "step": 4155
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.9294214248289904e-05,
+      "loss": 0.5078,
+      "step": 4156
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.9293877251323374e-05,
+      "loss": 0.0571,
+      "step": 4157
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.9293540175861092e-05,
+      "loss": 0.7852,
+      "step": 4158
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.929320302190492e-05,
+      "loss": 0.3438,
+      "step": 4159
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.9292865789456702e-05,
+      "loss": 0.2471,
+      "step": 4160
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.929252847851829e-05,
+      "loss": 0.1484,
+      "step": 4161
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.929219108909154e-05,
+      "loss": 0.373,
+      "step": 4162
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.9291853621178308e-05,
+      "loss": 0.2715,
+      "step": 4163
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.9291516074780435e-05,
+      "loss": 0.0254,
+      "step": 4164
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.9291178449899786e-05,
+      "loss": 1.0156,
+      "step": 4165
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.9290840746538207e-05,
+      "loss": 1.5859,
+      "step": 4166
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.9290502964697562e-05,
+      "loss": 1.0703,
+      "step": 4167
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.9290165104379698e-05,
+      "loss": 0.0049,
+      "step": 4168
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.928982716558647e-05,
+      "loss": 0.1699,
+      "step": 4169
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.9289489148319743e-05,
+      "loss": 0.7305,
+      "step": 4170
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.9289151052581363e-05,
+      "loss": 0.3086,
+      "step": 4171
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.928881287837319e-05,
+      "loss": 0.5547,
+      "step": 4172
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.9288474625697083e-05,
+      "loss": 0.0083,
+      "step": 4173
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.9288136294554894e-05,
+      "loss": 0.0361,
+      "step": 4174
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.928779788494849e-05,
+      "loss": 0.0005,
+      "step": 4175
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.9287459396879722e-05,
+      "loss": 0.4121,
+      "step": 4176
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.9287120830350458e-05,
+      "loss": 1.3438,
+      "step": 4177
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.9286782185362542e-05,
+      "loss": 0.0352,
+      "step": 4178
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.9286443461917846e-05,
+      "loss": 0.0001,
+      "step": 4179
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.928610466001823e-05,
+      "loss": 0.0001,
+      "step": 4180
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.9285765779665545e-05,
+      "loss": 1.0859,
+      "step": 4181
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.928542682086166e-05,
+      "loss": 0.0001,
+      "step": 4182
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.9285087783608434e-05,
+      "loss": 0.3535,
+      "step": 4183
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.9284748667907735e-05,
+      "loss": 0.1914,
+      "step": 4184
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.9284409473761413e-05,
+      "loss": 0.1592,
+      "step": 4185
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.928407020117134e-05,
+      "loss": 0.5664,
+      "step": 4186
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.9283730850139374e-05,
+      "loss": 0.0,
+      "step": 4187
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.9283391420667384e-05,
+      "loss": 0.0,
+      "step": 4188
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.928305191275723e-05,
+      "loss": 0.0011,
+      "step": 4189
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.9282712326410776e-05,
+      "loss": 0.0001,
+      "step": 4190
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.928237266162989e-05,
+      "loss": 2.25,
+      "step": 4191
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.928203291841643e-05,
+      "loss": 0.0001,
+      "step": 4192
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.9281693096772276e-05,
+      "loss": 0.1084,
+      "step": 4193
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.928135319669928e-05,
+      "loss": 0.0251,
+      "step": 4194
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.9281013218199318e-05,
+      "loss": 0.0001,
+      "step": 4195
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.928067316127425e-05,
+      "loss": 0.0,
+      "step": 4196
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.928033302592595e-05,
+      "loss": 0.1299,
+      "step": 4197
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.927999281215628e-05,
+      "loss": 0.1138,
+      "step": 4198
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.9279652519967105e-05,
+      "loss": 0.1494,
+      "step": 4199
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.9279312149360307e-05,
+      "loss": 0.0,
+      "step": 4200
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.9278971700337747e-05,
+      "loss": 0.4316,
+      "step": 4201
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.927863117290129e-05,
+      "loss": 1.75,
+      "step": 4202
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.927829056705282e-05,
+      "loss": 1.4219,
+      "step": 4203
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.927794988279419e-05,
+      "loss": 0.9766,
+      "step": 4204
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.9277609120127284e-05,
+      "loss": 0.9922,
+      "step": 4205
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.9277268279053965e-05,
+      "loss": 1.1094,
+      "step": 4206
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.9276927359576116e-05,
+      "loss": 1.4688,
+      "step": 4207
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.9276586361695596e-05,
+      "loss": 0.0188,
+      "step": 4208
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.9276245285414283e-05,
+      "loss": 0.4277,
+      "step": 4209
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.9275904130734053e-05,
+      "loss": 1.5703,
+      "step": 4210
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.9275562897656777e-05,
+      "loss": 0.6875,
+      "step": 4211
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.9275221586184328e-05,
+      "loss": 0.25,
+      "step": 4212
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.927488019631858e-05,
+      "loss": 0.6484,
+      "step": 4213
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.927453872806141e-05,
+      "loss": 1.7969,
+      "step": 4214
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.927419718141469e-05,
+      "loss": 1.2109,
+      "step": 4215
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.92738555563803e-05,
+      "loss": 1.6641,
+      "step": 4216
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.9273513852960117e-05,
+      "loss": 0.1914,
+      "step": 4217
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.9273172071156012e-05,
+      "loss": 1.1484,
+      "step": 4218
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.927283021096986e-05,
+      "loss": 1.4766,
+      "step": 4219
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.9272488272403548e-05,
+      "loss": 1.7266,
+      "step": 4220
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.9272146255458946e-05,
+      "loss": 0.3164,
+      "step": 4221
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.9271804160137932e-05,
+      "loss": 0.0086,
+      "step": 4222
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.927146198644239e-05,
+      "loss": 0.0007,
+      "step": 4223
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.9271119734374198e-05,
+      "loss": 0.3926,
+      "step": 4224
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.927077740393523e-05,
+      "loss": 0.1162,
+      "step": 4225
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.9270434995127374e-05,
+      "loss": 0.6406,
+      "step": 4226
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.92700925079525e-05,
+      "loss": 0.4902,
+      "step": 4227
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.92697499424125e-05,
+      "loss": 0.0366,
+      "step": 4228
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.9269407298509248e-05,
+      "loss": 0.0001,
+      "step": 4229
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.9269064576244627e-05,
+      "loss": 0.0001,
+      "step": 4230
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.9268721775620518e-05,
+      "loss": 0.4492,
+      "step": 4231
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.926837889663881e-05,
+      "loss": 0.6758,
+      "step": 4232
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.9268035939301382e-05,
+      "loss": 0.3008,
+      "step": 4233
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.926769290361011e-05,
+      "loss": 0.0001,
+      "step": 4234
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.9267349789566892e-05,
+      "loss": 0.6836,
+      "step": 4235
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.9267006597173602e-05,
+      "loss": 0.0,
+      "step": 4236
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.9266663326432123e-05,
+      "loss": 0.1348,
+      "step": 4237
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.9266319977344348e-05,
+      "loss": 0.0002,
+      "step": 4238
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.926597654991216e-05,
+      "loss": 0.0002,
+      "step": 4239
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.9265633044137444e-05,
+      "loss": 0.3945,
+      "step": 4240
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.926528946002208e-05,
+      "loss": 0.0547,
+      "step": 4241
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.926494579756797e-05,
+      "loss": 0.0001,
+      "step": 4242
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.9264602056776988e-05,
+      "loss": 0.0003,
+      "step": 4243
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.9264258237651026e-05,
+      "loss": 0.0003,
+      "step": 4244
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.926391434019197e-05,
+      "loss": 0.0004,
+      "step": 4245
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.9263570364401712e-05,
+      "loss": 0.0001,
+      "step": 4246
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.9263226310282144e-05,
+      "loss": 0.0001,
+      "step": 4247
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.9262882177835145e-05,
+      "loss": 0.3027,
+      "step": 4248
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.9262537967062613e-05,
+      "loss": 0.126,
+      "step": 4249
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.926219367796644e-05,
+      "loss": 0.1816,
+      "step": 4250
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.9261849310548505e-05,
+      "loss": 0.6445,
+      "step": 4251
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.9261504864810714e-05,
+      "loss": 0.9102,
+      "step": 4252
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.926116034075495e-05,
+      "loss": 0.4727,
+      "step": 4253
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.9260815738383103e-05,
+      "loss": 0.6016,
+      "step": 4254
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.926047105769707e-05,
+      "loss": 0.5898,
+      "step": 4255
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.926012629869874e-05,
+      "loss": 0.9102,
+      "step": 4256
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.9259781461390012e-05,
+      "loss": 0.1982,
+      "step": 4257
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.925943654577278e-05,
+      "loss": 0.5742,
+      "step": 4258
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.9259091551848932e-05,
+      "loss": 0.1309,
+      "step": 4259
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.9258746479620366e-05,
+      "loss": 1.0156,
+      "step": 4260
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.9258401329088973e-05,
+      "loss": 0.7852,
+      "step": 4261
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.9258056100256656e-05,
+      "loss": 0.0009,
+      "step": 4262
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.92577107931253e-05,
+      "loss": 0.2891,
+      "step": 4263
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.9257365407696815e-05,
+      "loss": 0.1465,
+      "step": 4264
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.925701994397309e-05,
+      "loss": 0.0291,
+      "step": 4265
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.9256674401956015e-05,
+      "loss": 1.5,
+      "step": 4266
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.9256328781647502e-05,
+      "loss": 0.5938,
+      "step": 4267
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.9255983083049442e-05,
+      "loss": 0.0012,
+      "step": 4268
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.925563730616373e-05,
+      "loss": 0.0242,
+      "step": 4269
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.9255291450992274e-05,
+      "loss": 0.25,
+      "step": 4270
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.925494551753696e-05,
+      "loss": 0.0045,
+      "step": 4271
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.9254599505799707e-05,
+      "loss": 0.3691,
+      "step": 4272
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.9254253415782396e-05,
+      "loss": 0.0089,
+      "step": 4273
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.9253907247486934e-05,
+      "loss": 0.9609,
+      "step": 4274
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.9253561000915226e-05,
+      "loss": 0.4629,
+      "step": 4275
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.925321467606917e-05,
+      "loss": 0.0767,
+      "step": 4276
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.9252868272950667e-05,
+      "loss": 0.0005,
+      "step": 4277
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.9252521791561622e-05,
+      "loss": 0.0,
+      "step": 4278
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.9252175231903935e-05,
+      "loss": 0.25,
+      "step": 4279
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.9251828593979516e-05,
+      "loss": 0.0001,
+      "step": 4280
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.925148187779026e-05,
+      "loss": 0.5234,
+      "step": 4281
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.925113508333807e-05,
+      "loss": 0.3125,
+      "step": 4282
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.925078821062486e-05,
+      "loss": 0.1138,
+      "step": 4283
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.925044125965253e-05,
+      "loss": 0.0143,
+      "step": 4284
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.9250094230422984e-05,
+      "loss": 0.0017,
+      "step": 4285
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.9249747122938126e-05,
+      "loss": 0.0801,
+      "step": 4286
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.9249399937199868e-05,
+      "loss": 0.0002,
+      "step": 4287
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.9249052673210114e-05,
+      "loss": 0.0076,
+      "step": 4288
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.9248705330970765e-05,
+      "loss": 0.0002,
+      "step": 4289
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.924835791048374e-05,
+      "loss": 0.0004,
+      "step": 4290
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.9248010411750938e-05,
+      "loss": 0.0005,
+      "step": 4291
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.924766283477427e-05,
+      "loss": 1.6641,
+      "step": 4292
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.9247315179555645e-05,
+      "loss": 0.0001,
+      "step": 4293
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.9246967446096975e-05,
+      "loss": 0.0004,
+      "step": 4294
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.9246619634400163e-05,
+      "loss": 0.0001,
+      "step": 4295
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.9246271744467126e-05,
+      "loss": 0.1167,
+      "step": 4296
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.9245923776299767e-05,
+      "loss": 0.1445,
+      "step": 4297
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.92455757299e-05,
+      "loss": 0.1006,
+      "step": 4298
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.9245227605269744e-05,
+      "loss": 0.1177,
+      "step": 4299
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.9244879402410897e-05,
+      "loss": 0.1084,
+      "step": 4300
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.9244531121325383e-05,
+      "loss": 1.125,
+      "step": 4301
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.9244182762015106e-05,
+      "loss": 1.1953,
+      "step": 4302
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.9243834324481987e-05,
+      "loss": 0.1836,
+      "step": 4303
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.924348580872793e-05,
+      "loss": 1.6484,
+      "step": 4304
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.924313721475486e-05,
+      "loss": 0.5703,
+      "step": 4305
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.9242788542564682e-05,
+      "loss": 1.4297,
+      "step": 4306
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.9242439792159313e-05,
+      "loss": 1.3594,
+      "step": 4307
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9242090963540674e-05,
+      "loss": 0.4609,
+      "step": 4308
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9241742056710675e-05,
+      "loss": 0.3125,
+      "step": 4309
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.924139307167123e-05,
+      "loss": 1.3438,
+      "step": 4310
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.924104400842426e-05,
+      "loss": 0.6914,
+      "step": 4311
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9240694866971682e-05,
+      "loss": 0.0439,
+      "step": 4312
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9240345647315414e-05,
+      "loss": 0.0488,
+      "step": 4313
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9239996349457368e-05,
+      "loss": 0.6445,
+      "step": 4314
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9239646973399465e-05,
+      "loss": 0.25,
+      "step": 4315
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9239297519143625e-05,
+      "loss": 0.0135,
+      "step": 4316
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.923894798669177e-05,
+      "loss": 0.3008,
+      "step": 4317
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9238598376045813e-05,
+      "loss": 0.053,
+      "step": 4318
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9238248687207678e-05,
+      "loss": 0.0156,
+      "step": 4319
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.923789892017928e-05,
+      "loss": 0.0444,
+      "step": 4320
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9237549074962548e-05,
+      "loss": 0.0047,
+      "step": 4321
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.92371991515594e-05,
+      "loss": 0.0073,
+      "step": 4322
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9236849149971756e-05,
+      "loss": 0.1133,
+      "step": 4323
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9236499070201536e-05,
+      "loss": 0.1914,
+      "step": 4324
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.923614891225067e-05,
+      "loss": 0.0117,
+      "step": 4325
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9235798676121075e-05,
+      "loss": 0.0008,
+      "step": 4326
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9235448361814673e-05,
+      "loss": 0.0,
+      "step": 4327
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.923509796933339e-05,
+      "loss": 0.0004,
+      "step": 4328
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9234747498679154e-05,
+      "loss": 0.4004,
+      "step": 4329
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9234396949853885e-05,
+      "loss": 0.0,
+      "step": 4330
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9234046322859507e-05,
+      "loss": 0.0491,
+      "step": 4331
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.923369561769795e-05,
+      "loss": 0.0322,
+      "step": 4332
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.923334483437114e-05,
+      "loss": 0.001,
+      "step": 4333
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9232993972880996e-05,
+      "loss": 0.0072,
+      "step": 4334
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9232643033229454e-05,
+      "loss": 0.0001,
+      "step": 4335
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.923229201541844e-05,
+      "loss": 0.0004,
+      "step": 4336
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.923194091944987e-05,
+      "loss": 0.0002,
+      "step": 4337
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9231589745325686e-05,
+      "loss": 0.0001,
+      "step": 4338
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.923123849304781e-05,
+      "loss": 0.2715,
+      "step": 4339
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9230887162618172e-05,
+      "loss": 0.2949,
+      "step": 4340
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9230535754038697e-05,
+      "loss": 0.0005,
+      "step": 4341
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.923018426731133e-05,
+      "loss": 0.0001,
+      "step": 4342
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9229832702437982e-05,
+      "loss": 0.0024,
+      "step": 4343
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9229481059420594e-05,
+      "loss": 0.0,
+      "step": 4344
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9229129338261097e-05,
+      "loss": 0.0,
+      "step": 4345
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9228777538961417e-05,
+      "loss": 0.1758,
+      "step": 4346
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9228425661523494e-05,
+      "loss": 0.0001,
+      "step": 4347
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9228073705949257e-05,
+      "loss": 0.1641,
+      "step": 4348
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9227721672240633e-05,
+      "loss": 0.2129,
+      "step": 4349
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9227369560399562e-05,
+      "loss": 0.2207,
+      "step": 4350
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9227017370427975e-05,
+      "loss": 0.9766,
+      "step": 4351
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9226665102327804e-05,
+      "loss": 0.6055,
+      "step": 4352
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.922631275610099e-05,
+      "loss": 0.8008,
+      "step": 4353
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9225960331749468e-05,
+      "loss": 0.4785,
+      "step": 4354
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9225607829275165e-05,
+      "loss": 0.5391,
+      "step": 4355
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.922525524868002e-05,
+      "loss": 0.6055,
+      "step": 4356
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9224902589965974e-05,
+      "loss": 0.4766,
+      "step": 4357
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9224549853134958e-05,
+      "loss": 0.7227,
+      "step": 4358
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.922419703818891e-05,
+      "loss": 1.9609,
+      "step": 4359
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.922384414512977e-05,
+      "loss": 0.0908,
+      "step": 4360
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9223491173959474e-05,
+      "loss": 0.8516,
+      "step": 4361
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9223138124679962e-05,
+      "loss": 1.4453,
+      "step": 4362
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9222784997293174e-05,
+      "loss": 0.5703,
+      "step": 4363
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9222431791801044e-05,
+      "loss": 1.8281,
+      "step": 4364
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9222078508205517e-05,
+      "loss": 1.4219,
+      "step": 4365
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9221725146508528e-05,
+      "loss": 0.0359,
+      "step": 4366
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9221371706712024e-05,
+      "loss": 1.0703,
+      "step": 4367
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9221018188817938e-05,
+      "loss": 0.9766,
+      "step": 4368
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9220664592828217e-05,
+      "loss": 1.4609,
+      "step": 4369
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9220310918744802e-05,
+      "loss": 0.1162,
+      "step": 4370
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9219957166569633e-05,
+      "loss": 0.2178,
+      "step": 4371
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9219603336304656e-05,
+      "loss": 0.0002,
+      "step": 4372
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9219249427951813e-05,
+      "loss": 0.4062,
+      "step": 4373
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9218895441513046e-05,
+      "loss": 0.0001,
+      "step": 4374
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.92185413769903e-05,
+      "loss": 0.0001,
+      "step": 4375
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9218187234385518e-05,
+      "loss": 0.1089,
+      "step": 4376
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.921783301370065e-05,
+      "loss": 0.0011,
+      "step": 4377
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.921747871493763e-05,
+      "loss": 0.0,
+      "step": 4378
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9217124338098416e-05,
+      "loss": 0.002,
+      "step": 4379
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.921676988318495e-05,
+      "loss": 0.0,
+      "step": 4380
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9216415350199173e-05,
+      "loss": 0.0001,
+      "step": 4381
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.921606073914304e-05,
+      "loss": 0.1328,
+      "step": 4382
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9215706050018493e-05,
+      "loss": 0.0,
+      "step": 4383
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9215351282827487e-05,
+      "loss": 0.0003,
+      "step": 4384
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.921499643757196e-05,
+      "loss": 0.0001,
+      "step": 4385
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9214641514253864e-05,
+      "loss": 0.0001,
+      "step": 4386
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9214286512875154e-05,
+      "loss": 0.2373,
+      "step": 4387
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9213931433437767e-05,
+      "loss": 0.0003,
+      "step": 4388
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9213576275943665e-05,
+      "loss": 0.0,
+      "step": 4389
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9213221040394795e-05,
+      "loss": 0.9258,
+      "step": 4390
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.921286572679311e-05,
+      "loss": 0.0018,
+      "step": 4391
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9212510335140555e-05,
+      "loss": 0.0001,
+      "step": 4392
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9212154865439087e-05,
+      "loss": 0.0,
+      "step": 4393
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.921179931769066e-05,
+      "loss": 0.0,
+      "step": 4394
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9211443691897213e-05,
+      "loss": 0.0001,
+      "step": 4395
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.921108798806071e-05,
+      "loss": 0.0001,
+      "step": 4396
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.921073220618311e-05,
+      "loss": 0.1504,
+      "step": 4397
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9210376346266356e-05,
+      "loss": 0.0791,
+      "step": 4398
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9210020408312406e-05,
+      "loss": 0.0266,
+      "step": 4399
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9209664392323212e-05,
+      "loss": 0.2451,
+      "step": 4400
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9209308298300735e-05,
+      "loss": 0.6562,
+      "step": 4401
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9208952126246927e-05,
+      "loss": 0.9414,
+      "step": 4402
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9208595876163745e-05,
+      "loss": 0.6328,
+      "step": 4403
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9208239548053142e-05,
+      "loss": 0.416,
+      "step": 4404
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.920788314191708e-05,
+      "loss": 0.0815,
+      "step": 4405
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.920752665775751e-05,
+      "loss": 0.9375,
+      "step": 4406
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9207170095576398e-05,
+      "loss": 0.2207,
+      "step": 4407
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9206813455375695e-05,
+      "loss": 2.0312,
+      "step": 4408
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.920645673715736e-05,
+      "loss": 0.6641,
+      "step": 4409
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9206099940923357e-05,
+      "loss": 0.1396,
+      "step": 4410
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.920574306667564e-05,
+      "loss": 0.2266,
+      "step": 4411
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.920538611441617e-05,
+      "loss": 1.1719,
+      "step": 4412
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9205029084146908e-05,
+      "loss": 0.4199,
+      "step": 4413
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9204671975869815e-05,
+      "loss": 0.8047,
+      "step": 4414
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9204314789586857e-05,
+      "loss": 0.0003,
+      "step": 4415
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9203957525299984e-05,
+      "loss": 0.0217,
+      "step": 4416
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9203600183011168e-05,
+      "loss": 0.2021,
+      "step": 4417
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9203242762722368e-05,
+      "loss": 0.2773,
+      "step": 4418
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9202885264435545e-05,
+      "loss": 0.5273,
+      "step": 4419
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9202527688152664e-05,
+      "loss": 0.5625,
+      "step": 4420
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9202170033875688e-05,
+      "loss": 0.2695,
+      "step": 4421
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.920181230160658e-05,
+      "loss": 1.0703,
+      "step": 4422
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9201454491347305e-05,
+      "loss": 0.2266,
+      "step": 4423
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9201096603099835e-05,
+      "loss": 0.6992,
+      "step": 4424
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9200738636866127e-05,
+      "loss": 0.0776,
+      "step": 4425
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9200380592648144e-05,
+      "loss": 0.0,
+      "step": 4426
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9200022470447864e-05,
+      "loss": 1.1875,
+      "step": 4427
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9199664270267243e-05,
+      "loss": 0.0,
+      "step": 4428
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9199305992108255e-05,
+      "loss": 0.0001,
+      "step": 4429
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.919894763597286e-05,
+      "loss": 0.0001,
+      "step": 4430
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9198589201863034e-05,
+      "loss": 0.0894,
+      "step": 4431
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.919823068978074e-05,
+      "loss": 0.0004,
+      "step": 4432
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.919787209972795e-05,
+      "loss": 0.0002,
+      "step": 4433
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9197513431706634e-05,
+      "loss": 0.3535,
+      "step": 4434
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9197154685718758e-05,
+      "loss": 0.0007,
+      "step": 4435
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9196795861766294e-05,
+      "loss": 0.0001,
+      "step": 4436
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.919643695985121e-05,
+      "loss": 0.0079,
+      "step": 4437
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9196077979975477e-05,
+      "loss": 0.0004,
+      "step": 4438
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.919571892214107e-05,
+      "loss": 0.0005,
+      "step": 4439
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.919535978634996e-05,
+      "loss": 0.0005,
+      "step": 4440
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.919500057260412e-05,
+      "loss": 0.0004,
+      "step": 4441
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.919464128090552e-05,
+      "loss": 0.0008,
+      "step": 4442
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9194281911256135e-05,
+      "loss": 0.0031,
+      "step": 4443
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9193922463657936e-05,
+      "loss": 0.0006,
+      "step": 4444
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9193562938112896e-05,
+      "loss": 0.0001,
+      "step": 4445
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9193203334622997e-05,
+      "loss": 0.0002,
+      "step": 4446
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9192843653190203e-05,
+      "loss": 0.1855,
+      "step": 4447
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.91924838938165e-05,
+      "loss": 0.1777,
+      "step": 4448
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.919212405650386e-05,
+      "loss": 0.106,
+      "step": 4449
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9191764141254252e-05,
+      "loss": 0.1631,
+      "step": 4450
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.919140414806966e-05,
+      "loss": 0.9062,
+      "step": 4451
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9191044076952058e-05,
+      "loss": 0.4551,
+      "step": 4452
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9190683927903426e-05,
+      "loss": 0.4609,
+      "step": 4453
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.919032370092574e-05,
+      "loss": 1.2734,
+      "step": 4454
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9189963396020983e-05,
+      "loss": 0.2715,
+      "step": 4455
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9189603013191123e-05,
+      "loss": 1.0938,
+      "step": 4456
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9189242552438144e-05,
+      "loss": 0.5547,
+      "step": 4457
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.918888201376403e-05,
+      "loss": 0.5586,
+      "step": 4458
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9188521397170755e-05,
+      "loss": 0.0097,
+      "step": 4459
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.91881607026603e-05,
+      "loss": 0.4492,
+      "step": 4460
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9187799930234655e-05,
+      "loss": 0.4766,
+      "step": 4461
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.918743907989579e-05,
+      "loss": 1.9531,
+      "step": 4462
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.918707815164569e-05,
+      "loss": 0.5625,
+      "step": 4463
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9186717145486338e-05,
+      "loss": 0.5781,
+      "step": 4464
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9186356061419716e-05,
+      "loss": 0.0022,
+      "step": 4465
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9185994899447805e-05,
+      "loss": 0.416,
+      "step": 4466
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9185633659572593e-05,
+      "loss": 0.0282,
+      "step": 4467
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.918527234179606e-05,
+      "loss": 0.0317,
+      "step": 4468
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9184910946120193e-05,
+      "loss": 0.2324,
+      "step": 4469
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9184549472546975e-05,
+      "loss": 0.0037,
+      "step": 4470
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9184187921078384e-05,
+      "loss": 0.0262,
+      "step": 4471
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.918382629171642e-05,
+      "loss": 0.2812,
+      "step": 4472
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9183464584463058e-05,
+      "loss": 0.5664,
+      "step": 4473
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9183102799320287e-05,
+      "loss": 0.1074,
+      "step": 4474
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9182740936290098e-05,
+      "loss": 0.0132,
+      "step": 4475
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.918237899537447e-05,
+      "loss": 2.4531,
+      "step": 4476
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9182016976575398e-05,
+      "loss": 1.4219,
+      "step": 4477
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9181654879894867e-05,
+      "loss": 1.6016,
+      "step": 4478
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9181292705334868e-05,
+      "loss": 0.1934,
+      "step": 4479
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.918093045289738e-05,
+      "loss": 0.0439,
+      "step": 4480
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.918056812258441e-05,
+      "loss": 0.0001,
+      "step": 4481
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.918020571439793e-05,
+      "loss": 0.0006,
+      "step": 4482
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9179843228339943e-05,
+      "loss": 0.0767,
+      "step": 4483
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.917948066441243e-05,
+      "loss": 0.0,
+      "step": 4484
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.917911802261739e-05,
+      "loss": 0.0002,
+      "step": 4485
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.917875530295681e-05,
+      "loss": 0.0349,
+      "step": 4486
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9178392505432684e-05,
+      "loss": 0.001,
+      "step": 4487
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9178029630047e-05,
+      "loss": 0.0001,
+      "step": 4488
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9177666676801758e-05,
+      "loss": 0.0003,
+      "step": 4489
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9177303645698946e-05,
+      "loss": 0.0001,
+      "step": 4490
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9176940536740558e-05,
+      "loss": 0.0796,
+      "step": 4491
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.917657734992859e-05,
+      "loss": 0.0019,
+      "step": 4492
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9176214085265035e-05,
+      "loss": 0.0,
+      "step": 4493
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9175850742751888e-05,
+      "loss": 0.3086,
+      "step": 4494
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9175487322391145e-05,
+      "loss": 0.1084,
+      "step": 4495
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9175123824184797e-05,
+      "loss": 0.0008,
+      "step": 4496
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.917476024813485e-05,
+      "loss": 0.1875,
+      "step": 4497
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.917439659424329e-05,
+      "loss": 0.1738,
+      "step": 4498
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9174032862512125e-05,
+      "loss": 0.1201,
+      "step": 4499
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9173669052943347e-05,
+      "loss": 0.1797,
+      "step": 4500
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9173305165538947e-05,
+      "loss": 1.0469,
+      "step": 4501
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9172941200300936e-05,
+      "loss": 0.6094,
+      "step": 4502
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.91725771572313e-05,
+      "loss": 0.5977,
+      "step": 4503
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.917221303633205e-05,
+      "loss": 0.2871,
+      "step": 4504
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9171848837605178e-05,
+      "loss": 0.8867,
+      "step": 4505
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.917148456105269e-05,
+      "loss": 2.2656,
+      "step": 4506
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.917112020667658e-05,
+      "loss": 0.8672,
+      "step": 4507
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.9170755774478845e-05,
+      "loss": 0.6289,
+      "step": 4508
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.91703912644615e-05,
+      "loss": 0.5391,
+      "step": 4509
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.917002667662654e-05,
+      "loss": 0.6836,
+      "step": 4510
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.9169662010975964e-05,
+      "loss": 0.2695,
+      "step": 4511
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.916929726751178e-05,
+      "loss": 0.1191,
+      "step": 4512
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.916893244623599e-05,
+      "loss": 0.582,
+      "step": 4513
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.916856754715059e-05,
+      "loss": 1.0156,
+      "step": 4514
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.9168202570257595e-05,
+      "loss": 0.0239,
+      "step": 4515
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.9167837515558997e-05,
+      "loss": 0.6836,
+      "step": 4516
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.9167472383056818e-05,
+      "loss": 0.0134,
+      "step": 4517
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.9167107172753044e-05,
+      "loss": 0.5898,
+      "step": 4518
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.9166741884649693e-05,
+      "loss": 0.0006,
+      "step": 4519
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.9166376518748764e-05,
+      "loss": 0.832,
+      "step": 4520
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.9166011075052274e-05,
+      "loss": 0.0012,
+      "step": 4521
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.916564555356222e-05,
+      "loss": 3.2188,
+      "step": 4522
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.9165279954280606e-05,
+      "loss": 0.0001,
+      "step": 4523
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.9164914277209453e-05,
+      "loss": 1.0391,
+      "step": 4524
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.9164548522350758e-05,
+      "loss": 0.0601,
+      "step": 4525
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.9164182689706537e-05,
+      "loss": 0.25,
+      "step": 4526
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.9163816779278793e-05,
+      "loss": 0.0718,
+      "step": 4527
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.9163450791069534e-05,
+      "loss": 0.0,
+      "step": 4528
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.9163084725080777e-05,
+      "loss": 0.1309,
+      "step": 4529
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.916271858131453e-05,
+      "loss": 0.0205,
+      "step": 4530
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.91623523597728e-05,
+      "loss": 0.0016,
+      "step": 4531
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.9161986060457605e-05,
+      "loss": 0.0005,
+      "step": 4532
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.916161968337095e-05,
+      "loss": 0.0,
+      "step": 4533
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.9161253228514852e-05,
+      "loss": 0.0356,
+      "step": 4534
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.9160886695891318e-05,
+      "loss": 0.0044,
+      "step": 4535
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.916052008550236e-05,
+      "loss": 0.3633,
+      "step": 4536
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.9160153397350004e-05,
+      "loss": 0.0002,
+      "step": 4537
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.9159786631436245e-05,
+      "loss": 0.0,
+      "step": 4538
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.915941978776311e-05,
+      "loss": 0.0,
+      "step": 4539
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.9159052866332612e-05,
+      "loss": 0.0,
+      "step": 4540
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.9158685867146763e-05,
+      "loss": 0.5234,
+      "step": 4541
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.9158318790207578e-05,
+      "loss": 0.0004,
+      "step": 4542
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.9157951635517077e-05,
+      "loss": 0.0001,
+      "step": 4543
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.9157584403077274e-05,
+      "loss": 0.0002,
+      "step": 4544
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.9157217092890186e-05,
+      "loss": 0.0,
+      "step": 4545
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.9156849704957827e-05,
+      "loss": 0.0001,
+      "step": 4546
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.915648223928222e-05,
+      "loss": 0.0004,
+      "step": 4547
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.915611469586537e-05,
+      "loss": 0.103,
+      "step": 4548
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.9155747074709317e-05,
+      "loss": 0.0032,
+      "step": 4549
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.9155379375816062e-05,
+      "loss": 0.4297,
+      "step": 4550
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.9155011599187635e-05,
+      "loss": 1.2344,
+      "step": 4551
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.9154643744826047e-05,
+      "loss": 0.2314,
+      "step": 4552
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.9154275812733322e-05,
+      "loss": 0.4883,
+      "step": 4553
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.9153907802911483e-05,
+      "loss": 0.7383,
+      "step": 4554
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.9153539715362548e-05,
+      "loss": 0.0427,
+      "step": 4555
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.9153171550088537e-05,
+      "loss": 1.8672,
+      "step": 4556
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.9152803307091478e-05,
+      "loss": 0.7383,
+      "step": 4557
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.9152434986373386e-05,
+      "loss": 0.6367,
+      "step": 4558
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.9152066587936287e-05,
+      "loss": 0.0835,
+      "step": 4559
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.9151698111782205e-05,
+      "loss": 0.0043,
+      "step": 4560
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.915132955791316e-05,
+      "loss": 0.4355,
+      "step": 4561
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.915096092633118e-05,
+      "loss": 0.2217,
+      "step": 4562
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.9150592217038288e-05,
+      "loss": 0.6094,
+      "step": 4563
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.9150223430036507e-05,
+      "loss": 0.1816,
+      "step": 4564
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.9149854565327862e-05,
+      "loss": 1.9453,
+      "step": 4565
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.9149485622914384e-05,
+      "loss": 0.1934,
+      "step": 4566
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.914911660279809e-05,
+      "loss": 1.9297,
+      "step": 4567
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.914874750498102e-05,
+      "loss": 0.0006,
+      "step": 4568
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.9148378329465185e-05,
+      "loss": 0.0093,
+      "step": 4569
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.9148009076252624e-05,
+      "loss": 0.4883,
+      "step": 4570
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.914763974534536e-05,
+      "loss": 0.0001,
+      "step": 4571
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.9147270336745425e-05,
+      "loss": 0.0757,
+      "step": 4572
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.9146900850454842e-05,
+      "loss": 0.1768,
+      "step": 4573
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.9146531286475638e-05,
+      "loss": 0.0029,
+      "step": 4574
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.914616164480985e-05,
+      "loss": 0.0005,
+      "step": 4575
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.914579192545951e-05,
+      "loss": 1.1562,
+      "step": 4576
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.9145422128426638e-05,
+      "loss": 0.0098,
+      "step": 4577
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.914505225371327e-05,
+      "loss": 0.1318,
+      "step": 4578
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.9144682301321444e-05,
+      "loss": 0.4785,
+      "step": 4579
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.914431227125318e-05,
+      "loss": 0.013,
+      "step": 4580
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.9143942163510517e-05,
+      "loss": 0.0393,
+      "step": 4581
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.914357197809548e-05,
+      "loss": 1.0781,
+      "step": 4582
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.9143201715010118e-05,
+      "loss": 0.1914,
+      "step": 4583
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.9142831374256446e-05,
+      "loss": 0.0015,
+      "step": 4584
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.914246095583651e-05,
+      "loss": 0.0001,
+      "step": 4585
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.9142090459752336e-05,
+      "loss": 0.2891,
+      "step": 4586
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.914171988600596e-05,
+      "loss": 0.1055,
+      "step": 4587
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.9141349234599425e-05,
+      "loss": 0.1621,
+      "step": 4588
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.9140978505534756e-05,
+      "loss": 0.3125,
+      "step": 4589
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.9140607698814e-05,
+      "loss": 0.0001,
+      "step": 4590
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.914023681443918e-05,
+      "loss": 0.0002,
+      "step": 4591
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.9139865852412345e-05,
+      "loss": 0.2852,
+      "step": 4592
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.9139494812735524e-05,
+      "loss": 1.4609,
+      "step": 4593
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.9139123695410756e-05,
+      "loss": 0.0001,
+      "step": 4594
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.9138752500440084e-05,
+      "loss": 0.0001,
+      "step": 4595
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.9138381227825544e-05,
+      "loss": 0.2188,
+      "step": 4596
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.9138009877569172e-05,
+      "loss": 0.1523,
+      "step": 4597
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.9137638449673004e-05,
+      "loss": 0.1426,
+      "step": 4598
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.913726694413909e-05,
+      "loss": 0.1807,
+      "step": 4599
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.913689536096947e-05,
+      "loss": 0.1602,
+      "step": 4600
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.913652370016617e-05,
+      "loss": 1.5859,
+      "step": 4601
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.9136151961731247e-05,
+      "loss": 0.8125,
+      "step": 4602
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.913578014566673e-05,
+      "loss": 0.4961,
+      "step": 4603
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.913540825197467e-05,
+      "loss": 1.5703,
+      "step": 4604
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.913503628065711e-05,
+      "loss": 1.0703,
+      "step": 4605
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.9134664231716087e-05,
+      "loss": 0.7422,
+      "step": 4606
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.9134292105153643e-05,
+      "loss": 0.8164,
+      "step": 4607
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.9133919900971823e-05,
+      "loss": 0.3281,
+      "step": 4608
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.9133547619172675e-05,
+      "loss": 1.9219,
+      "step": 4609
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.913317525975824e-05,
+      "loss": 0.5664,
+      "step": 4610
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.9132802822730564e-05,
+      "loss": 0.5117,
+      "step": 4611
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.9132430308091695e-05,
+      "loss": 1.2578,
+      "step": 4612
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.9132057715843675e-05,
+      "loss": 0.084,
+      "step": 4613
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.9131685045988546e-05,
+      "loss": 2.1719,
+      "step": 4614
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.913131229852836e-05,
+      "loss": 0.0002,
+      "step": 4615
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.9130939473465165e-05,
+      "loss": 0.8867,
+      "step": 4616
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.913056657080101e-05,
+      "loss": 0.0762,
+      "step": 4617
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.9130193590537935e-05,
+      "loss": 0.0221,
+      "step": 4618
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.9129820532677992e-05,
+      "loss": 0.7891,
+      "step": 4619
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.9129447397223237e-05,
+      "loss": 0.5508,
+      "step": 4620
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.9129074184175704e-05,
+      "loss": 0.011,
+      "step": 4621
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.912870089353746e-05,
+      "loss": 0.0002,
+      "step": 4622
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.9128327525310543e-05,
+      "loss": 0.0001,
+      "step": 4623
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.9127954079497004e-05,
+      "loss": 0.0364,
+      "step": 4624
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.9127580556098896e-05,
+      "loss": 0.1001,
+      "step": 4625
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.912720695511827e-05,
+      "loss": 1.7656,
+      "step": 4626
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.912683327655718e-05,
+      "loss": 0.2852,
+      "step": 4627
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.9126459520417674e-05,
+      "loss": 0.1035,
+      "step": 4628
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.9126085686701812e-05,
+      "loss": 0.0586,
+      "step": 4629
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.9125711775411634e-05,
+      "loss": 0.0967,
+      "step": 4630
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.9125337786549206e-05,
+      "loss": 0.0001,
+      "step": 4631
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.912496372011657e-05,
+      "loss": 0.1963,
+      "step": 4632
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.9124589576115794e-05,
+      "loss": 0.1455,
+      "step": 4633
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.9124215354548926e-05,
+      "loss": 0.001,
+      "step": 4634
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.9123841055418018e-05,
+      "loss": 0.0001,
+      "step": 4635
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.912346667872513e-05,
+      "loss": 0.0752,
+      "step": 4636
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.9123092224472314e-05,
+      "loss": 0.0004,
+      "step": 4637
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.9122717692661624e-05,
+      "loss": 0.0003,
+      "step": 4638
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.912234308329513e-05,
+      "loss": 0.0001,
+      "step": 4639
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.9121968396374874e-05,
+      "loss": 0.0002,
+      "step": 4640
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.9121593631902923e-05,
+      "loss": 0.5938,
+      "step": 4641
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.912121878988133e-05,
+      "loss": 0.0001,
+      "step": 4642
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.912084387031216e-05,
+      "loss": 0.0003,
+      "step": 4643
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.912046887319746e-05,
+      "loss": 0.0001,
+      "step": 4644
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.91200937985393e-05,
+      "loss": 0.1235,
+      "step": 4645
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.9119718646339737e-05,
+      "loss": 0.0001,
+      "step": 4646
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.911934341660083e-05,
+      "loss": 0.0002,
+      "step": 4647
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.911896810932464e-05,
+      "loss": 0.0001,
+      "step": 4648
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.911859272451323e-05,
+      "loss": 0.1768,
+      "step": 4649
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.911821726216866e-05,
+      "loss": 0.0771,
+      "step": 4650
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.9117841722292987e-05,
+      "loss": 0.5781,
+      "step": 4651
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.911746610488828e-05,
+      "loss": 1.1172,
+      "step": 4652
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.9117090409956603e-05,
+      "loss": 0.793,
+      "step": 4653
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.9116714637500012e-05,
+      "loss": 1.0312,
+      "step": 4654
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.9116338787520577e-05,
+      "loss": 1.5859,
+      "step": 4655
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.9115962860020354e-05,
+      "loss": 0.8789,
+      "step": 4656
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.9115586855001418e-05,
+      "loss": 2.1562,
+      "step": 4657
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.9115210772465824e-05,
+      "loss": 0.875,
+      "step": 4658
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.9114834612415643e-05,
+      "loss": 0.6523,
+      "step": 4659
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.9114458374852946e-05,
+      "loss": 0.2559,
+      "step": 4660
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.9114082059779788e-05,
+      "loss": 0.2578,
+      "step": 4661
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.911370566719824e-05,
+      "loss": 0.6523,
+      "step": 4662
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.9113329197110368e-05,
+      "loss": 0.104,
+      "step": 4663
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.9112952649518243e-05,
+      "loss": 0.2656,
+      "step": 4664
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.9112576024423925e-05,
+      "loss": 1.6016,
+      "step": 4665
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.9112199321829494e-05,
+      "loss": 0.5117,
+      "step": 4666
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.911182254173701e-05,
+      "loss": 0.083,
+      "step": 4667
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.9111445684148546e-05,
+      "loss": 0.0884,
+      "step": 4668
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.911106874906617e-05,
+      "loss": 0.0012,
+      "step": 4669
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.9110691736491954e-05,
+      "loss": 0.0449,
+      "step": 4670
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.911031464642796e-05,
+      "loss": 0.0084,
+      "step": 4671
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.910993747887627e-05,
+      "loss": 0.0001,
+      "step": 4672
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.9109560233838946e-05,
+      "loss": 0.0002,
+      "step": 4673
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.9109182911318072e-05,
+      "loss": 1.8672,
+      "step": 4674
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.9108805511315706e-05,
+      "loss": 0.0299,
+      "step": 4675
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.9108428033833925e-05,
+      "loss": 0.001,
+      "step": 4676
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.910805047887481e-05,
+      "loss": 0.0,
+      "step": 4677
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.9107672846440423e-05,
+      "loss": 0.0001,
+      "step": 4678
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.9107295136532846e-05,
+      "loss": 0.0001,
+      "step": 4679
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.9106917349154147e-05,
+      "loss": 0.0007,
+      "step": 4680
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.910653948430641e-05,
+      "loss": 0.3301,
+      "step": 4681
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.9106161541991696e-05,
+      "loss": 0.0,
+      "step": 4682
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.910578352221209e-05,
+      "loss": 0.0001,
+      "step": 4683
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.910540542496967e-05,
+      "loss": 0.0,
+      "step": 4684
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.910502725026651e-05,
+      "loss": 0.0,
+      "step": 4685
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.9104648998104676e-05,
+      "loss": 0.0012,
+      "step": 4686
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.9104270668486262e-05,
+      "loss": 0.6289,
+      "step": 4687
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.9103892261413337e-05,
+      "loss": 0.1533,
+      "step": 4688
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.910351377688798e-05,
+      "loss": 1.2031,
+      "step": 4689
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.910313521491227e-05,
+      "loss": 0.0028,
+      "step": 4690
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.9102756575488284e-05,
+      "loss": 0.1328,
+      "step": 4691
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.9102377858618103e-05,
+      "loss": 0.0001,
+      "step": 4692
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.9101999064303808e-05,
+      "loss": 0.0002,
+      "step": 4693
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.910162019254748e-05,
+      "loss": 0.3496,
+      "step": 4694
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.910124124335119e-05,
+      "loss": 0.0002,
+      "step": 4695
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.9100862216717033e-05,
+      "loss": 0.0,
+      "step": 4696
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.910048311264708e-05,
+      "loss": 0.1299,
+      "step": 4697
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.9100103931143418e-05,
+      "loss": 0.0991,
+      "step": 4698
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.909972467220813e-05,
+      "loss": 0.1143,
+      "step": 4699
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.9099345335843296e-05,
+      "loss": 0.1768,
+      "step": 4700
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.9098965922050998e-05,
+      "loss": 0.6328,
+      "step": 4701
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.9098586430833325e-05,
+      "loss": 0.3789,
+      "step": 4702
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.9098206862192355e-05,
+      "loss": 0.3965,
+      "step": 4703
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.9097827216130175e-05,
+      "loss": 0.2129,
+      "step": 4704
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.909744749264887e-05,
+      "loss": 0.4336,
+      "step": 4705
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.9097067691750527e-05,
+      "loss": 0.1445,
+      "step": 4706
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.9096687813437225e-05,
+      "loss": 0.7422,
+      "step": 4707
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.909630785771106e-05,
+      "loss": 0.4941,
+      "step": 4708
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.9095927824574107e-05,
+      "loss": 0.7188,
+      "step": 4709
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.9095547714028464e-05,
+      "loss": 0.3613,
+      "step": 4710
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.909516752607621e-05,
+      "loss": 0.9102,
+      "step": 4711
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.909478726071944e-05,
+      "loss": 0.8867,
+      "step": 4712
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.909440691796024e-05,
+      "loss": 0.1504,
+      "step": 4713
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.9094026497800698e-05,
+      "loss": 0.4238,
+      "step": 4714
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.90936460002429e-05,
+      "loss": 0.0649,
+      "step": 4715
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.9093265425288943e-05,
+      "loss": 1.0859,
+      "step": 4716
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.9092884772940905e-05,
+      "loss": 0.582,
+      "step": 4717
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.9092504043200883e-05,
+      "loss": 0.2344,
+      "step": 4718
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.9092123236070974e-05,
+      "loss": 0.2275,
+      "step": 4719
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.9091742351553262e-05,
+      "loss": 0.1904,
+      "step": 4720
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.909136138964984e-05,
+      "loss": 0.0008,
+      "step": 4721
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.90909803503628e-05,
+      "loss": 0.0012,
+      "step": 4722
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.9090599233694235e-05,
+      "loss": 0.125,
+      "step": 4723
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.909021803964624e-05,
+      "loss": 0.5742,
+      "step": 4724
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.90898367682209e-05,
+      "loss": 0.0017,
+      "step": 4725
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.908945541942032e-05,
+      "loss": 0.0001,
+      "step": 4726
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.9089073993246587e-05,
+      "loss": 0.0026,
+      "step": 4727
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.9088692489701795e-05,
+      "loss": 0.0002,
+      "step": 4728
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.908831090878805e-05,
+      "loss": 0.0001,
+      "step": 4729
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.908792925050743e-05,
+      "loss": 0.3906,
+      "step": 4730
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.9087547514862042e-05,
+      "loss": 0.1377,
+      "step": 4731
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.9087165701853986e-05,
+      "loss": 0.4844,
+      "step": 4732
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.908678381148535e-05,
+      "loss": 0.0408,
+      "step": 4733
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.908640184375823e-05,
+      "loss": 0.2344,
+      "step": 4734
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.9086019798674736e-05,
+      "loss": 0.0002,
+      "step": 4735
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.9085637676236952e-05,
+      "loss": 0.1514,
+      "step": 4736
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.9085255476446984e-05,
+      "loss": 0.0,
+      "step": 4737
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.908487319930693e-05,
+      "loss": 0.0139,
+      "step": 4738
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.9084490844818886e-05,
+      "loss": 0.0008,
+      "step": 4739
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.9084108412984957e-05,
+      "loss": 0.0002,
+      "step": 4740
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.908372590380724e-05,
+      "loss": 0.0012,
+      "step": 4741
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.9083343317287833e-05,
+      "loss": 0.418,
+      "step": 4742
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.908296065342885e-05,
+      "loss": 0.0002,
+      "step": 4743
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.9082577912232372e-05,
+      "loss": 0.0003,
+      "step": 4744
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.9082195093700513e-05,
+      "loss": 0.0001,
+      "step": 4745
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.9081812197835376e-05,
+      "loss": 0.1543,
+      "step": 4746
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.908142922463906e-05,
+      "loss": 0.1895,
+      "step": 4747
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.9081046174113667e-05,
+      "loss": 0.168,
+      "step": 4748
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.9080663046261307e-05,
+      "loss": 0.0001,
+      "step": 4749
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.908027984108408e-05,
+      "loss": 0.0001,
+      "step": 4750
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.9079896558584088e-05,
+      "loss": 0.7773,
+      "step": 4751
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.907951319876344e-05,
+      "loss": 0.5859,
+      "step": 4752
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.9079129761624237e-05,
+      "loss": 1.3203,
+      "step": 4753
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.9078746247168587e-05,
+      "loss": 0.8438,
+      "step": 4754
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.90783626553986e-05,
+      "loss": 0.0062,
+      "step": 4755
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.9077978986316377e-05,
+      "loss": 0.3008,
+      "step": 4756
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.9077595239924028e-05,
+      "loss": 0.9648,
+      "step": 4757
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.9077211416223652e-05,
+      "loss": 0.7109,
+      "step": 4758
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.9076827515217368e-05,
+      "loss": 1.1875,
+      "step": 4759
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.9076443536907286e-05,
+      "loss": 0.2354,
+      "step": 4760
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.9076059481295497e-05,
+      "loss": 0.3926,
+      "step": 4761
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.9075675348384128e-05,
+      "loss": 1.2734,
+      "step": 4762
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.9075291138175283e-05,
+      "loss": 1.0938,
+      "step": 4763
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.9074906850671074e-05,
+      "loss": 0.1211,
+      "step": 4764
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.9074522485873604e-05,
+      "loss": 0.5078,
+      "step": 4765
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.9074138043784987e-05,
+      "loss": 0.0684,
+      "step": 4766
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.9073753524407335e-05,
+      "loss": 0.3281,
+      "step": 4767
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.9073368927742764e-05,
+      "loss": 0.4805,
+      "step": 4768
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.907298425379338e-05,
+      "loss": 0.0004,
+      "step": 4769
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.9072599502561298e-05,
+      "loss": 0.3906,
+      "step": 4770
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.9072214674048627e-05,
+      "loss": 0.0074,
+      "step": 4771
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.907182976825749e-05,
+      "loss": 0.2852,
+      "step": 4772
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.907144478518999e-05,
+      "loss": 0.0474,
+      "step": 4773
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.9071059724848243e-05,
+      "loss": 0.7266,
+      "step": 4774
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.9070674587234368e-05,
+      "loss": 0.4766,
+      "step": 4775
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.9070289372350485e-05,
+      "loss": 2.1094,
+      "step": 4776
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.9069904080198694e-05,
+      "loss": 0.0002,
+      "step": 4777
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.9069518710781124e-05,
+      "loss": 0.0001,
+      "step": 4778
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.9069133264099885e-05,
+      "loss": 0.0483,
+      "step": 4779
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.9068747740157095e-05,
+      "loss": 0.0001,
+      "step": 4780
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.9068362138954875e-05,
+      "loss": 0.0002,
+      "step": 4781
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.9067976460495338e-05,
+      "loss": 0.0032,
+      "step": 4782
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.90675907047806e-05,
+      "loss": 0.0002,
+      "step": 4783
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.9067204871812784e-05,
+      "loss": 0.0002,
+      "step": 4784
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.906681896159401e-05,
+      "loss": 0.0003,
+      "step": 4785
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.906643297412639e-05,
+      "loss": 0.0001,
+      "step": 4786
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.9066046909412053e-05,
+      "loss": 0.0004,
+      "step": 4787
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.906566076745311e-05,
+      "loss": 0.0003,
+      "step": 4788
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.906527454825169e-05,
+      "loss": 0.0001,
+      "step": 4789
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.906488825180991e-05,
+      "loss": 0.2949,
+      "step": 4790
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.9064501878129887e-05,
+      "loss": 1.1953,
+      "step": 4791
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.9064115427213748e-05,
+      "loss": 0.0001,
+      "step": 4792
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.9063728899063614e-05,
+      "loss": 0.0001,
+      "step": 4793
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.906334229368161e-05,
+      "loss": 0.0001,
+      "step": 4794
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.9062955611069856e-05,
+      "loss": 0.0004,
+      "step": 4795
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.9062568851230478e-05,
+      "loss": 0.0001,
+      "step": 4796
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.9062182014165598e-05,
+      "loss": 0.1318,
+      "step": 4797
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.9061795099877334e-05,
+      "loss": 0.1113,
+      "step": 4798
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.9061408108367824e-05,
+      "loss": 0.1982,
+      "step": 4799
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.906102103963919e-05,
+      "loss": 0.1904,
+      "step": 4800
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.9060633893693547e-05,
+      "loss": 0.5117,
+      "step": 4801
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.9060246670533032e-05,
+      "loss": 0.1455,
+      "step": 4802
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.9059859370159765e-05,
+      "loss": 0.0471,
+      "step": 4803
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.9059471992575877e-05,
+      "loss": 0.8516,
+      "step": 4804
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.9059084537783496e-05,
+      "loss": 1.6875,
+      "step": 4805
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.9058697005784746e-05,
+      "loss": 0.5586,
+      "step": 4806
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.9058309396581755e-05,
+      "loss": 0.0287,
+      "step": 4807
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.9057921710176654e-05,
+      "loss": 0.4043,
+      "step": 4808
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.9057533946571575e-05,
+      "loss": 0.7266,
+      "step": 4809
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.905714610576864e-05,
+      "loss": 0.6406,
+      "step": 4810
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.9056758187769984e-05,
+      "loss": 0.6211,
+      "step": 4811
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.9056370192577738e-05,
+      "loss": 0.1553,
+      "step": 4812
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.9055982120194028e-05,
+      "loss": 0.6016,
+      "step": 4813
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.9055593970620984e-05,
+      "loss": 0.3809,
+      "step": 4814
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.905520574386075e-05,
+      "loss": 0.5352,
+      "step": 4815
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.9054817439915446e-05,
+      "loss": 0.5586,
+      "step": 4816
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.9054429058787207e-05,
+      "loss": 1.3047,
+      "step": 4817
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.9054040600478166e-05,
+      "loss": 0.2256,
+      "step": 4818
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.9053652064990458e-05,
+      "loss": 0.2412,
+      "step": 4819
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.9053263452326213e-05,
+      "loss": 0.5742,
+      "step": 4820
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.905287476248757e-05,
+      "loss": 0.3105,
+      "step": 4821
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.905248599547666e-05,
+      "loss": 0.1914,
+      "step": 4822
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.905209715129562e-05,
+      "loss": 0.0085,
+      "step": 4823
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.9051708229946586e-05,
+      "loss": 0.0288,
+      "step": 4824
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.9051319231431688e-05,
+      "loss": 0.9219,
+      "step": 4825
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.905093015575307e-05,
+      "loss": 0.0186,
+      "step": 4826
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.905054100291286e-05,
+      "loss": 0.053,
+      "step": 4827
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.90501517729132e-05,
+      "loss": 0.0001,
+      "step": 4828
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.904976246575624e-05,
+      "loss": 0.4102,
+      "step": 4829
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.904937308144409e-05,
+      "loss": 0.0003,
+      "step": 4830
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.9048983619978914e-05,
+      "loss": 0.0002,
+      "step": 4831
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.9048594081362836e-05,
+      "loss": 0.0001,
+      "step": 4832
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.9048204465597997e-05,
+      "loss": 0.105,
+      "step": 4833
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.9047814772686546e-05,
+      "loss": 0.0001,
+      "step": 4834
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.904742500263061e-05,
+      "loss": 0.0003,
+      "step": 4835
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.904703515543234e-05,
+      "loss": 0.0996,
+      "step": 4836
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.9046645231093872e-05,
+      "loss": 0.0014,
+      "step": 4837
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.9046255229617347e-05,
+      "loss": 0.0001,
+      "step": 4838
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.904586515100491e-05,
+      "loss": 0.0003,
+      "step": 4839
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.9045474995258696e-05,
+      "loss": 0.0004,
+      "step": 4840
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.9045084762380854e-05,
+      "loss": 0.0006,
+      "step": 4841
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.9044694452373524e-05,
+      "loss": 0.0001,
+      "step": 4842
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.9044304065238854e-05,
+      "loss": 0.0114,
+      "step": 4843
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.9043913600978984e-05,
+      "loss": 0.0001,
+      "step": 4844
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.904352305959606e-05,
+      "loss": 0.1064,
+      "step": 4845
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.904313244109222e-05,
+      "loss": 0.1475,
+      "step": 4846
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.9042741745469622e-05,
+      "loss": 0.1602,
+      "step": 4847
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.90423509727304e-05,
+      "loss": 0.0004,
+      "step": 4848
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.9041960122876703e-05,
+      "loss": 0.1582,
+      "step": 4849
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.9041569195910686e-05,
+      "loss": 0.1953,
+      "step": 4850
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.9041178191834483e-05,
+      "loss": 1.5,
+      "step": 4851
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.9040787110650244e-05,
+      "loss": 0.4375,
+      "step": 4852
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.904039595236012e-05,
+      "loss": 0.7227,
+      "step": 4853
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.9040004716966258e-05,
+      "loss": 1.3047,
+      "step": 4854
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.903961340447081e-05,
+      "loss": 1.6094,
+      "step": 4855
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.903922201487592e-05,
+      "loss": 0.0047,
+      "step": 4856
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.9038830548183738e-05,
+      "loss": 0.4629,
+      "step": 4857
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.9038439004396416e-05,
+      "loss": 1.0312,
+      "step": 4858
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.9038047383516106e-05,
+      "loss": 0.2852,
+      "step": 4859
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.903765568554495e-05,
+      "loss": 0.9336,
+      "step": 4860
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.9037263910485106e-05,
+      "loss": 0.2617,
+      "step": 4861
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.9036872058338723e-05,
+      "loss": 0.0082,
+      "step": 4862
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.9036480129107958e-05,
+      "loss": 0.3672,
+      "step": 4863
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.9036088122794958e-05,
+      "loss": 0.2539,
+      "step": 4864
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.9035696039401875e-05,
+      "loss": 0.5586,
+      "step": 4865
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.9035303878930864e-05,
+      "loss": 0.0019,
+      "step": 4866
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.9034911641384084e-05,
+      "loss": 0.0016,
+      "step": 4867
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.9034519326763678e-05,
+      "loss": 0.6406,
+      "step": 4868
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.903412693507181e-05,
+      "loss": 0.0576,
+      "step": 4869
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.9033734466310624e-05,
+      "loss": 0.0693,
+      "step": 4870
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.903334192048229e-05,
+      "loss": 0.1426,
+      "step": 4871
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.903294929758895e-05,
+      "loss": 0.3164,
+      "step": 4872
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.903255659763277e-05,
+      "loss": 0.1187,
+      "step": 4873
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.90321638206159e-05,
+      "loss": 0.0001,
+      "step": 4874
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.9031770966540504e-05,
+      "loss": 0.1738,
+      "step": 4875
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.903137803540873e-05,
+      "loss": 0.0001,
+      "step": 4876
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.9030985027222744e-05,
+      "loss": 0.0003,
+      "step": 4877
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.9030591941984695e-05,
+      "loss": 0.0001,
+      "step": 4878
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.903019877969676e-05,
+      "loss": 0.4062,
+      "step": 4879
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.9029805540361072e-05,
+      "loss": 0.0043,
+      "step": 4880
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.902941222397981e-05,
+      "loss": 0.0,
+      "step": 4881
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.902901883055513e-05,
+      "loss": 0.2773,
+      "step": 4882
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.9028625360089187e-05,
+      "loss": 0.0,
+      "step": 4883
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.9028231812584147e-05,
+      "loss": 0.1318,
+      "step": 4884
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.902783818804217e-05,
+      "loss": 0.0003,
+      "step": 4885
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.9027444486465414e-05,
+      "loss": 0.2471,
+      "step": 4886
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.9027050707856047e-05,
+      "loss": 0.0099,
+      "step": 4887
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.9026656852216227e-05,
+      "loss": 0.0001,
+      "step": 4888
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.9026262919548122e-05,
+      "loss": 0.0,
+      "step": 4889
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.9025868909853886e-05,
+      "loss": 0.0005,
+      "step": 4890
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.9025474823135692e-05,
+      "loss": 0.0013,
+      "step": 4891
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.9025080659395698e-05,
+      "loss": 0.0002,
+      "step": 4892
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.902468641863607e-05,
+      "loss": 0.7891,
+      "step": 4893
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.9024292100858982e-05,
+      "loss": 0.0996,
+      "step": 4894
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.9023897706066584e-05,
+      "loss": 0.5117,
+      "step": 4895
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.9023503234261055e-05,
+      "loss": 0.2715,
+      "step": 4896
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.9023108685444552e-05,
+      "loss": 0.0,
+      "step": 4897
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.9022714059619245e-05,
+      "loss": 0.2197,
+      "step": 4898
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.9022319356787303e-05,
+      "loss": 0.1992,
+      "step": 4899
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.902192457695089e-05,
+      "loss": 0.1299,
+      "step": 4900
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.902152972011218e-05,
+      "loss": 1.0703,
+      "step": 4901
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.9021134786273335e-05,
+      "loss": 0.4316,
+      "step": 4902
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.9020739775436523e-05,
+      "loss": 0.1689,
+      "step": 4903
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.9020344687603922e-05,
+      "loss": 0.5625,
+      "step": 4904
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.9019949522777694e-05,
+      "loss": 2.0938,
+      "step": 4905
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.9019554280960012e-05,
+      "loss": 1.2109,
+      "step": 4906
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.9019158962153043e-05,
+      "loss": 0.8711,
+      "step": 4907
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.9018763566358964e-05,
+      "loss": 0.1973,
+      "step": 4908
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.9018368093579938e-05,
+      "loss": 0.625,
+      "step": 4909
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.9017972543818143e-05,
+      "loss": 0.8086,
+      "step": 4910
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.9017576917075754e-05,
+      "loss": 0.918,
+      "step": 4911
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.9017181213354937e-05,
+      "loss": 0.0027,
+      "step": 4912
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.901678543265787e-05,
+      "loss": 1.6875,
+      "step": 4913
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.901638957498672e-05,
+      "loss": 0.0069,
+      "step": 4914
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.901599364034366e-05,
+      "loss": 0.0002,
+      "step": 4915
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.901559762873088e-05,
+      "loss": 0.0078,
+      "step": 4916
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.9015201540150538e-05,
+      "loss": 0.7461,
+      "step": 4917
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.9014805374604816e-05,
+      "loss": 0.3945,
+      "step": 4918
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.9014409132095885e-05,
+      "loss": 0.3438,
+      "step": 4919
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.901401281262593e-05,
+      "loss": 0.0001,
+      "step": 4920
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.9013616416197123e-05,
+      "loss": 0.0,
+      "step": 4921
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.9013219942811632e-05,
+      "loss": 0.5,
+      "step": 4922
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.901282339247165e-05,
+      "loss": 0.6484,
+      "step": 4923
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.901242676517934e-05,
+      "loss": 0.2139,
+      "step": 4924
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.9012030060936893e-05,
+      "loss": 0.0001,
+      "step": 4925
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.901163327974648e-05,
+      "loss": 0.0586,
+      "step": 4926
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.901123642161028e-05,
+      "loss": 1.9062,
+      "step": 4927
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.9010839486530473e-05,
+      "loss": 0.0547,
+      "step": 4928
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.901044247450924e-05,
+      "loss": 0.0742,
+      "step": 4929
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.901004538554876e-05,
+      "loss": 0.0001,
+      "step": 4930
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.9009648219651214e-05,
+      "loss": 0.5625,
+      "step": 4931
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.9009250976818785e-05,
+      "loss": 0.0026,
+      "step": 4932
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.9008853657053653e-05,
+      "loss": 0.0054,
+      "step": 4933
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.9008456260357997e-05,
+      "loss": 0.0461,
+      "step": 4934
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.9008058786734003e-05,
+      "loss": 0.022,
+      "step": 4935
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.900766123618386e-05,
+      "loss": 0.0003,
+      "step": 4936
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.900726360870974e-05,
+      "loss": 0.0064,
+      "step": 4937
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.9006865904313824e-05,
+      "loss": 0.0004,
+      "step": 4938
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.9006468122998308e-05,
+      "loss": 0.0006,
+      "step": 4939
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.900607026476537e-05,
+      "loss": 0.0074,
+      "step": 4940
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.9005672329617195e-05,
+      "loss": 0.0001,
+      "step": 4941
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.9005274317555973e-05,
+      "loss": 0.3164,
+      "step": 4942
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.9004876228583884e-05,
+      "loss": 0.1777,
+      "step": 4943
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.9004478062703115e-05,
+      "loss": 0.0051,
+      "step": 4944
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.9004079819915853e-05,
+      "loss": 0.0001,
+      "step": 4945
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.9003681500224292e-05,
+      "loss": 0.1245,
+      "step": 4946
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.9003283103630606e-05,
+      "loss": 0.0947,
+      "step": 4947
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.900288463013699e-05,
+      "loss": 0.1118,
+      "step": 4948
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.900248607974564e-05,
+      "loss": 0.1152,
+      "step": 4949
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.9002087452458725e-05,
+      "loss": 0.1641,
+      "step": 4950
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.9001688748278454e-05,
+      "loss": 0.9336,
+      "step": 4951
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.9001289967207007e-05,
+      "loss": 0.6602,
+      "step": 4952
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.9000891109246573e-05,
+      "loss": 0.7188,
+      "step": 4953
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.9000492174399347e-05,
+      "loss": 1.2812,
+      "step": 4954
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.9000093162667515e-05,
+      "loss": 0.3906,
+      "step": 4955
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.899969407405327e-05,
+      "loss": 1.2656,
+      "step": 4956
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.899929490855881e-05,
+      "loss": 0.7383,
+      "step": 4957
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.8998895666186315e-05,
+      "loss": 0.5938,
+      "step": 4958
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.899849634693799e-05,
+      "loss": 0.4902,
+      "step": 4959
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.8998096950816013e-05,
+      "loss": 0.6758,
+      "step": 4960
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.8997697477822592e-05,
+      "loss": 0.7617,
+      "step": 4961
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.8997297927959914e-05,
+      "loss": 0.8008,
+      "step": 4962
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.8996898301230173e-05,
+      "loss": 0.2129,
+      "step": 4963
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.8996498597635562e-05,
+      "loss": 0.4785,
+      "step": 4964
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.8996098817178285e-05,
+      "loss": 0.2773,
+      "step": 4965
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.899569895986053e-05,
+      "loss": 0.0986,
+      "step": 4966
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.8995299025684486e-05,
+      "loss": 0.0002,
+      "step": 4967
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.8994899014652362e-05,
+      "loss": 0.0518,
+      "step": 4968
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.899449892676635e-05,
+      "loss": 0.0081,
+      "step": 4969
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.8994098762028646e-05,
+      "loss": 0.0757,
+      "step": 4970
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.899369852044145e-05,
+      "loss": 0.1582,
+      "step": 4971
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.8993298202006953e-05,
+      "loss": 0.7031,
+      "step": 4972
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.8992897806727362e-05,
+      "loss": 0.0013,
+      "step": 4973
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.899249733460487e-05,
+      "loss": 0.4473,
+      "step": 4974
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.899209678564168e-05,
+      "loss": 0.7461,
+      "step": 4975
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.8991696159839994e-05,
+      "loss": 0.0,
+      "step": 4976
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.8991295457202003e-05,
+      "loss": 0.0002,
+      "step": 4977
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.8990894677729917e-05,
+      "loss": 0.1475,
+      "step": 4978
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.899049382142593e-05,
+      "loss": 0.0001,
+      "step": 4979
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.8990092888292248e-05,
+      "loss": 0.0001,
+      "step": 4980
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.8989691878331065e-05,
+      "loss": 0.1963,
+      "step": 4981
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.8989290791544594e-05,
+      "loss": 0.0001,
+      "step": 4982
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.8988889627935034e-05,
+      "loss": 0.0,
+      "step": 4983
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.8988488387504582e-05,
+      "loss": 0.0,
+      "step": 4984
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.8988087070255453e-05,
+      "loss": 0.0052,
+      "step": 4985
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.898768567618984e-05,
+      "loss": 0.0168,
+      "step": 4986
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.8987284205309947e-05,
+      "loss": 0.0,
+      "step": 4987
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.898688265761799e-05,
+      "loss": 0.1582,
+      "step": 4988
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.8986481033116158e-05,
+      "loss": 0.0001,
+      "step": 4989
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.8986079331806674e-05,
+      "loss": 0.0005,
+      "step": 4990
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.8985677553691732e-05,
+      "loss": 0.007,
+      "step": 4991
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.898527569877354e-05,
+      "loss": 0.0664,
+      "step": 4992
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.898487376705431e-05,
+      "loss": 0.0001,
+      "step": 4993
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.8984471758536243e-05,
+      "loss": 0.0002,
+      "step": 4994
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.898406967322155e-05,
+      "loss": 0.1689,
+      "step": 4995
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.898366751111244e-05,
+      "loss": 0.0001,
+      "step": 4996
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.898326527221112e-05,
+      "loss": 0.126,
+      "step": 4997
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.89828629565198e-05,
+      "loss": 0.1709,
+      "step": 4998
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.8982460564040686e-05,
+      "loss": 0.0913,
+      "step": 4999
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.898205809477599e-05,
+      "loss": 0.1357,
+      "step": 5000
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.8981655548727924e-05,
+      "loss": 0.4648,
+      "step": 5001
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.89812529258987e-05,
+      "loss": 0.252,
+      "step": 5002
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.898085022629052e-05,
+      "loss": 0.0728,
+      "step": 5003
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.8980447449905605e-05,
+      "loss": 0.5078,
+      "step": 5004
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.8980044596746164e-05,
+      "loss": 0.0352,
+      "step": 5005
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.8979641666814404e-05,
+      "loss": 0.8984,
+      "step": 5006
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.8979238660112545e-05,
+      "loss": 0.8125,
+      "step": 5007
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.89788355766428e-05,
+      "loss": 0.6484,
+      "step": 5008
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.8978432416407375e-05,
+      "loss": 0.0078,
+      "step": 5009
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.8978029179408492e-05,
+      "loss": 0.0422,
+      "step": 5010
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.897762586564836e-05,
+      "loss": 0.4141,
+      "step": 5011
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.8977222475129198e-05,
+      "loss": 0.3301,
+      "step": 5012
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.8976819007853224e-05,
+      "loss": 0.0957,
+      "step": 5013
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.8976415463822645e-05,
+      "loss": 0.0,
+      "step": 5014
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.8976011843039683e-05,
+      "loss": 0.1074,
+      "step": 5015
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.8975608145506548e-05,
+      "loss": 0.0073,
+      "step": 5016
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.8975204371225465e-05,
+      "loss": 0.0625,
+      "step": 5017
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.8974800520198646e-05,
+      "loss": 0.0005,
+      "step": 5018
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.8974396592428312e-05,
+      "loss": 0.1377,
+      "step": 5019
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.897399258791668e-05,
+      "loss": 0.106,
+      "step": 5020
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.897358850666597e-05,
+      "loss": 0.0283,
+      "step": 5021
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.8973184348678397e-05,
+      "loss": 0.1021,
+      "step": 5022
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.8972780113956188e-05,
+      "loss": 0.4395,
+      "step": 5023
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.897237580250155e-05,
+      "loss": 0.2559,
+      "step": 5024
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.897197141431672e-05,
+      "loss": 0.0002,
+      "step": 5025
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.8971566949403905e-05,
+      "loss": 0.083,
+      "step": 5026
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.897116240776533e-05,
+      "loss": 0.0007,
+      "step": 5027
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.897075778940322e-05,
+      "loss": 0.5078,
+      "step": 5028
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.8970353094319796e-05,
+      "loss": 0.104,
+      "step": 5029
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.8969948322517275e-05,
+      "loss": 0.6016,
+      "step": 5030
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.8969543473997884e-05,
+      "loss": 0.4629,
+      "step": 5031
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.896913854876385e-05,
+      "loss": 0.6641,
+      "step": 5032
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.8968733546817392e-05,
+      "loss": 0.0952,
+      "step": 5033
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.8968328468160736e-05,
+      "loss": 0.0005,
+      "step": 5034
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.8967923312796105e-05,
+      "loss": 0.0,
+      "step": 5035
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.8967518080725722e-05,
+      "loss": 0.0003,
+      "step": 5036
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.8967112771951817e-05,
+      "loss": 0.0002,
+      "step": 5037
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.8966707386476608e-05,
+      "loss": 0.1514,
+      "step": 5038
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.8966301924302335e-05,
+      "loss": 0.0008,
+      "step": 5039
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.896589638543121e-05,
+      "loss": 0.0,
+      "step": 5040
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.8965490769865473e-05,
+      "loss": 0.0002,
+      "step": 5041
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.896508507760734e-05,
+      "loss": 0.0002,
+      "step": 5042
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.8964679308659046e-05,
+      "loss": 0.1157,
+      "step": 5043
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.8964273463022815e-05,
+      "loss": 0.0,
+      "step": 5044
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.896386754070088e-05,
+      "loss": 0.0001,
+      "step": 5045
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.8963461541695464e-05,
+      "loss": 0.1289,
+      "step": 5046
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.89630554660088e-05,
+      "loss": 0.0796,
+      "step": 5047
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.8962649313643124e-05,
+      "loss": 0.1089,
+      "step": 5048
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.8962243084600658e-05,
+      "loss": 0.2393,
+      "step": 5049
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.8961836778883637e-05,
+      "loss": 0.2051,
+      "step": 5050
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.896143039649429e-05,
+      "loss": 1.0625,
+      "step": 5051
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.8961023937434848e-05,
+      "loss": 1.1406,
+      "step": 5052
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.896061740170755e-05,
+      "loss": 0.3945,
+      "step": 5053
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.896021078931462e-05,
+      "loss": 0.6758,
+      "step": 5054
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8959804100258294e-05,
+      "loss": 0.8555,
+      "step": 5055
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8959397334540803e-05,
+      "loss": 0.4258,
+      "step": 5056
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8958990492164385e-05,
+      "loss": 0.021,
+      "step": 5057
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8958583573131275e-05,
+      "loss": 0.1465,
+      "step": 5058
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8958176577443703e-05,
+      "loss": 1.4297,
+      "step": 5059
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8957769505103904e-05,
+      "loss": 2.5469,
+      "step": 5060
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8957362356114123e-05,
+      "loss": 0.0386,
+      "step": 5061
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8956955130476587e-05,
+      "loss": 0.332,
+      "step": 5062
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8956547828193527e-05,
+      "loss": 0.0014,
+      "step": 5063
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.895614044926719e-05,
+      "loss": 1.1406,
+      "step": 5064
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.895573299369981e-05,
+      "loss": 0.1748,
+      "step": 5065
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8955325461493628e-05,
+      "loss": 0.0713,
+      "step": 5066
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8954917852650872e-05,
+      "loss": 0.9336,
+      "step": 5067
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8954510167173793e-05,
+      "loss": 0.9922,
+      "step": 5068
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.895410240506462e-05,
+      "loss": 0.1055,
+      "step": 5069
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8953694566325596e-05,
+      "loss": 0.8125,
+      "step": 5070
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.895328665095896e-05,
+      "loss": 1.0703,
+      "step": 5071
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8952878658966956e-05,
+      "loss": 0.6719,
+      "step": 5072
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8952470590351816e-05,
+      "loss": 2.8281,
+      "step": 5073
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8952062445115786e-05,
+      "loss": 0.0013,
+      "step": 5074
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8951654223261114e-05,
+      "loss": 0.0302,
+      "step": 5075
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8951245924790028e-05,
+      "loss": 1.6328,
+      "step": 5076
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.895083754970478e-05,
+      "loss": 0.3672,
+      "step": 5077
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8950429098007613e-05,
+      "loss": 0.0001,
+      "step": 5078
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8950020569700767e-05,
+      "loss": 0.2598,
+      "step": 5079
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8949611964786483e-05,
+      "loss": 0.0889,
+      "step": 5080
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8949203283267003e-05,
+      "loss": 0.0001,
+      "step": 5081
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.894879452514458e-05,
+      "loss": 0.7852,
+      "step": 5082
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8948385690421456e-05,
+      "loss": 0.0012,
+      "step": 5083
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8947976779099874e-05,
+      "loss": 0.3457,
+      "step": 5084
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8947567791182075e-05,
+      "loss": 0.1445,
+      "step": 5085
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.894715872667032e-05,
+      "loss": 0.0001,
+      "step": 5086
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8946749585566835e-05,
+      "loss": 0.0,
+      "step": 5087
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8946340367873884e-05,
+      "loss": 0.0452,
+      "step": 5088
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.89459310735937e-05,
+      "loss": 0.8477,
+      "step": 5089
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.894552170272855e-05,
+      "loss": 0.2715,
+      "step": 5090
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.894511225528066e-05,
+      "loss": 0.0001,
+      "step": 5091
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.894470273125229e-05,
+      "loss": 0.0,
+      "step": 5092
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8944293130645695e-05,
+      "loss": 0.4629,
+      "step": 5093
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.894388345346311e-05,
+      "loss": 0.7227,
+      "step": 5094
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8943473699706793e-05,
+      "loss": 0.0001,
+      "step": 5095
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8943063869378997e-05,
+      "loss": 0.1729,
+      "step": 5096
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8942653962481964e-05,
+      "loss": 0.1309,
+      "step": 5097
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8942243979017953e-05,
+      "loss": 0.165,
+      "step": 5098
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8941833918989207e-05,
+      "loss": 0.1885,
+      "step": 5099
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8941423782397987e-05,
+      "loss": 0.1133,
+      "step": 5100
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8941013569246542e-05,
+      "loss": 0.4141,
+      "step": 5101
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8940603279537122e-05,
+      "loss": 1.1406,
+      "step": 5102
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.894019291327198e-05,
+      "loss": 1.3359,
+      "step": 5103
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.893978247045338e-05,
+      "loss": 0.7227,
+      "step": 5104
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8939371951083562e-05,
+      "loss": 0.2891,
+      "step": 5105
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8938961355164786e-05,
+      "loss": 2.1562,
+      "step": 5106
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8938550682699307e-05,
+      "loss": 0.3145,
+      "step": 5107
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.893813993368938e-05,
+      "loss": 1.5078,
+      "step": 5108
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.893772910813726e-05,
+      "loss": 0.0598,
+      "step": 5109
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8937318206045204e-05,
+      "loss": 0.3926,
+      "step": 5110
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.893690722741547e-05,
+      "loss": 0.0339,
+      "step": 5111
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8936496172250312e-05,
+      "loss": 0.4629,
+      "step": 5112
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8936085040551988e-05,
+      "loss": 0.0815,
+      "step": 5113
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8935673832322757e-05,
+      "loss": 0.5977,
+      "step": 5114
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.893526254756488e-05,
+      "loss": 1.2422,
+      "step": 5115
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8934851186280606e-05,
+      "loss": 0.0007,
+      "step": 5116
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.8934439748472206e-05,
+      "loss": 0.0112,
+      "step": 5117
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.893402823414193e-05,
+      "loss": 0.0454,
+      "step": 5118
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.893361664329204e-05,
+      "loss": 1.6641,
+      "step": 5119
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.89332049759248e-05,
+      "loss": 0.0204,
+      "step": 5120
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.893279323204247e-05,
+      "loss": 0.3887,
+      "step": 5121
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8932381411647305e-05,
+      "loss": 0.248,
+      "step": 5122
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8931969514741575e-05,
+      "loss": 0.027,
+      "step": 5123
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8931557541327537e-05,
+      "loss": 0.6094,
+      "step": 5124
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8931145491407453e-05,
+      "loss": 0.3086,
+      "step": 5125
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8930733364983588e-05,
+      "loss": 0.6016,
+      "step": 5126
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8930321162058207e-05,
+      "loss": 0.0298,
+      "step": 5127
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8929908882633567e-05,
+      "loss": 0.0383,
+      "step": 5128
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8929496526711938e-05,
+      "loss": 0.0008,
+      "step": 5129
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.892908409429558e-05,
+      "loss": 0.2637,
+      "step": 5130
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8928671585386764e-05,
+      "loss": 0.0006,
+      "step": 5131
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.892825899998775e-05,
+      "loss": 0.0005,
+      "step": 5132
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8927846338100808e-05,
+      "loss": 0.0002,
+      "step": 5133
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8927433599728197e-05,
+      "loss": 0.3984,
+      "step": 5134
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8927020784872192e-05,
+      "loss": 0.0003,
+      "step": 5135
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8926607893535054e-05,
+      "loss": 0.0103,
+      "step": 5136
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8926194925719057e-05,
+      "loss": 0.0654,
+      "step": 5137
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.892578188142646e-05,
+      "loss": 0.168,
+      "step": 5138
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.892536876065954e-05,
+      "loss": 0.1816,
+      "step": 5139
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8924955563420555e-05,
+      "loss": 0.4043,
+      "step": 5140
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8924542289711786e-05,
+      "loss": 0.0649,
+      "step": 5141
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8924128939535494e-05,
+      "loss": 0.0003,
+      "step": 5142
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8923715512893954e-05,
+      "loss": 0.0549,
+      "step": 5143
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8923302009789438e-05,
+      "loss": 0.1494,
+      "step": 5144
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8922888430224203e-05,
+      "loss": 0.0002,
+      "step": 5145
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.892247477420054e-05,
+      "loss": 0.8555,
+      "step": 5146
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8922061041720705e-05,
+      "loss": 0.8164,
+      "step": 5147
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8921647232786978e-05,
+      "loss": 0.167,
+      "step": 5148
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8921233347401633e-05,
+      "loss": 0.1152,
+      "step": 5149
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8920819385566936e-05,
+      "loss": 0.0059,
+      "step": 5150
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8920405347285164e-05,
+      "loss": 1.0391,
+      "step": 5151
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8919991232558587e-05,
+      "loss": 0.6523,
+      "step": 5152
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8919577041389483e-05,
+      "loss": 0.2891,
+      "step": 5153
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8919162773780134e-05,
+      "loss": 2.4688,
+      "step": 5154
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.89187484297328e-05,
+      "loss": 0.0977,
+      "step": 5155
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8918334009249766e-05,
+      "loss": 0.6445,
+      "step": 5156
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.89179195123333e-05,
+      "loss": 1.0156,
+      "step": 5157
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8917504938985692e-05,
+      "loss": 0.3828,
+      "step": 5158
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8917090289209206e-05,
+      "loss": 0.032,
+      "step": 5159
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8916675563006122e-05,
+      "loss": 0.707,
+      "step": 5160
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8916260760378722e-05,
+      "loss": 0.9766,
+      "step": 5161
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.891584588132928e-05,
+      "loss": 2.3125,
+      "step": 5162
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8915430925860073e-05,
+      "loss": 0.0811,
+      "step": 5163
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.891501589397338e-05,
+      "loss": 0.4453,
+      "step": 5164
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8914600785671484e-05,
+      "loss": 0.0405,
+      "step": 5165
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8914185600956663e-05,
+      "loss": 0.6055,
+      "step": 5166
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.89137703398312e-05,
+      "loss": 0.3652,
+      "step": 5167
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8913355002297367e-05,
+      "loss": 1.5781,
+      "step": 5168
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8912939588357453e-05,
+      "loss": 0.1064,
+      "step": 5169
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.891252409801373e-05,
+      "loss": 0.2197,
+      "step": 5170
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8912108531268493e-05,
+      "loss": 0.0923,
+      "step": 5171
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.891169288812401e-05,
+      "loss": 0.5117,
+      "step": 5172
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8911277168582575e-05,
+      "loss": 0.1289,
+      "step": 5173
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8910861372646468e-05,
+      "loss": 0.1484,
+      "step": 5174
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8910445500317967e-05,
+      "loss": 0.0003,
+      "step": 5175
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8910029551599363e-05,
+      "loss": 1.1953,
+      "step": 5176
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.890961352649293e-05,
+      "loss": 0.2598,
+      "step": 5177
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.890919742500097e-05,
+      "loss": 0.0432,
+      "step": 5178
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8908781247125747e-05,
+      "loss": 0.0603,
+      "step": 5179
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.890836499286956e-05,
+      "loss": 0.0031,
+      "step": 5180
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8907948662234695e-05,
+      "loss": 0.0002,
+      "step": 5181
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.890753225522343e-05,
+      "loss": 0.4844,
+      "step": 5182
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8907115771838066e-05,
+      "loss": 0.5039,
+      "step": 5183
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8906699212080868e-05,
+      "loss": 0.2129,
+      "step": 5184
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8906282575954143e-05,
+      "loss": 0.1484,
+      "step": 5185
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8905865863460177e-05,
+      "loss": 0.0004,
+      "step": 5186
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8905449074601248e-05,
+      "loss": 0.7695,
+      "step": 5187
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8905032209379653e-05,
+      "loss": 0.0003,
+      "step": 5188
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8904615267797678e-05,
+      "loss": 0.0535,
+      "step": 5189
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.890419824985761e-05,
+      "loss": 0.0096,
+      "step": 5190
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8903781155561752e-05,
+      "loss": 0.0002,
+      "step": 5191
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8903363984912378e-05,
+      "loss": 0.0153,
+      "step": 5192
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.890294673791179e-05,
+      "loss": 0.082,
+      "step": 5193
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.890252941456227e-05,
+      "loss": 0.1167,
+      "step": 5194
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8902112014866118e-05,
+      "loss": 0.1357,
+      "step": 5195
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8901694538825625e-05,
+      "loss": 0.0967,
+      "step": 5196
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.890127698644308e-05,
+      "loss": 0.0009,
+      "step": 5197
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8900859357720778e-05,
+      "loss": 0.0121,
+      "step": 5198
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.890044165266101e-05,
+      "loss": 0.1396,
+      "step": 5199
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.890002387126608e-05,
+      "loss": 0.0005,
+      "step": 5200
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8899606013538267e-05,
+      "loss": 1.1406,
+      "step": 5201
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.889918807947988e-05,
+      "loss": 0.3984,
+      "step": 5202
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8898770069093202e-05,
+      "loss": 0.7344,
+      "step": 5203
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.889835198238054e-05,
+      "loss": 0.0366,
+      "step": 5204
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8897933819344172e-05,
+      "loss": 0.0486,
+      "step": 5205
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8897515579986417e-05,
+      "loss": 0.9766,
+      "step": 5206
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8897097264309557e-05,
+      "loss": 1.0234,
+      "step": 5207
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8896678872315892e-05,
+      "loss": 0.5586,
+      "step": 5208
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8896260404007726e-05,
+      "loss": 0.2021,
+      "step": 5209
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8895841859387347e-05,
+      "loss": 0.5195,
+      "step": 5210
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8895423238457058e-05,
+      "loss": 1.6016,
+      "step": 5211
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8895004541219164e-05,
+      "loss": 1.2031,
+      "step": 5212
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8894585767675953e-05,
+      "loss": 1.1406,
+      "step": 5213
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8894166917829732e-05,
+      "loss": 0.8359,
+      "step": 5214
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.88937479916828e-05,
+      "loss": 0.2637,
+      "step": 5215
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8893328989237456e-05,
+      "loss": 0.5781,
+      "step": 5216
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8892909910496e-05,
+      "loss": 0.0062,
+      "step": 5217
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8892490755460738e-05,
+      "loss": 0.0058,
+      "step": 5218
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8892071524133967e-05,
+      "loss": 0.0016,
+      "step": 5219
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8891652216517995e-05,
+      "loss": 0.0669,
+      "step": 5220
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8891232832615118e-05,
+      "loss": 0.1011,
+      "step": 5221
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8890813372427644e-05,
+      "loss": 0.5742,
+      "step": 5222
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8890393835957874e-05,
+      "loss": 0.0018,
+      "step": 5223
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.888997422320811e-05,
+      "loss": 0.0,
+      "step": 5224
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8889554534180664e-05,
+      "loss": 0.2676,
+      "step": 5225
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8889134768877834e-05,
+      "loss": 0.0603,
+      "step": 5226
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.888871492730192e-05,
+      "loss": 0.0145,
+      "step": 5227
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.888829500945524e-05,
+      "loss": 0.0001,
+      "step": 5228
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8887875015340095e-05,
+      "loss": 0.0,
+      "step": 5229
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.888745494495879e-05,
+      "loss": 0.0007,
+      "step": 5230
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8887034798313634e-05,
+      "loss": 0.3867,
+      "step": 5231
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8886614575406934e-05,
+      "loss": 0.0002,
+      "step": 5232
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.888619427624099e-05,
+      "loss": 0.0211,
+      "step": 5233
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8885773900818123e-05,
+      "loss": 0.0001,
+      "step": 5234
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.888535344914063e-05,
+      "loss": 0.001,
+      "step": 5235
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.888493292121083e-05,
+      "loss": 0.0,
+      "step": 5236
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.888451231703103e-05,
+      "loss": 0.0001,
+      "step": 5237
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8884091636603528e-05,
+      "loss": 0.4375,
+      "step": 5238
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.888367087993065e-05,
+      "loss": 0.0002,
+      "step": 5239
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8883250047014698e-05,
+      "loss": 0.0007,
+      "step": 5240
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8882829137857987e-05,
+      "loss": 0.165,
+      "step": 5241
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8882408152462825e-05,
+      "loss": 0.032,
+      "step": 5242
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8881987090831524e-05,
+      "loss": 0.0703,
+      "step": 5243
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.88815659529664e-05,
+      "loss": 0.2676,
+      "step": 5244
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.888114473886976e-05,
+      "loss": 0.5664,
+      "step": 5245
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.888072344854392e-05,
+      "loss": 0.0098,
+      "step": 5246
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8880302081991198e-05,
+      "loss": 0.0001,
+      "step": 5247
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8879880639213905e-05,
+      "loss": 0.0002,
+      "step": 5248
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.887945912021435e-05,
+      "loss": 0.1553,
+      "step": 5249
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8879037524994856e-05,
+      "loss": 0.1953,
+      "step": 5250
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8878615853557736e-05,
+      "loss": 0.4961,
+      "step": 5251
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.88781941059053e-05,
+      "loss": 1.7969,
+      "step": 5252
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.887777228203987e-05,
+      "loss": 0.2188,
+      "step": 5253
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.887735038196376e-05,
+      "loss": 0.918,
+      "step": 5254
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8876928405679286e-05,
+      "loss": 0.6211,
+      "step": 5255
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8876506353188768e-05,
+      "loss": 0.75,
+      "step": 5256
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.8876084224494525e-05,
+      "loss": 1.3672,
+      "step": 5257
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.887566201959887e-05,
+      "loss": 0.4258,
+      "step": 5258
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.8875239738504124e-05,
+      "loss": 0.4824,
+      "step": 5259
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.8874817381212606e-05,
+      "loss": 1.4609,
+      "step": 5260
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.8874394947726638e-05,
+      "loss": 1.7734,
+      "step": 5261
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.887397243804854e-05,
+      "loss": 0.418,
+      "step": 5262
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.887354985218062e-05,
+      "loss": 0.6719,
+      "step": 5263
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.8873127190125212e-05,
+      "loss": 2.1719,
+      "step": 5264
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.8872704451884637e-05,
+      "loss": 0.1406,
+      "step": 5265
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.887228163746121e-05,
+      "loss": 0.0018,
+      "step": 5266
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.887185874685726e-05,
+      "loss": 0.3594,
+      "step": 5267
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.88714357800751e-05,
+      "loss": 0.0007,
+      "step": 5268
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.887101273711706e-05,
+      "loss": 0.0138,
+      "step": 5269
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.887058961798546e-05,
+      "loss": 0.2314,
+      "step": 5270
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.8870166422682628e-05,
+      "loss": 0.0981,
+      "step": 5271
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.8869743151210882e-05,
+      "loss": 0.0002,
+      "step": 5272
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.8869319803572548e-05,
+      "loss": 0.0004,
+      "step": 5273
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.8868896379769956e-05,
+      "loss": 0.0259,
+      "step": 5274
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.8868472879805424e-05,
+      "loss": 0.0005,
+      "step": 5275
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.8868049303681284e-05,
+      "loss": 0.0009,
+      "step": 5276
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.8867625651399852e-05,
+      "loss": 0.0003,
+      "step": 5277
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.8867201922963467e-05,
+      "loss": 0.0006,
+      "step": 5278
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.8866778118374453e-05,
+      "loss": 1.1875,
+      "step": 5279
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.8866354237635127e-05,
+      "loss": 0.3535,
+      "step": 5280
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.886593028074783e-05,
+      "loss": 0.1143,
+      "step": 5281
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.886550624771488e-05,
+      "loss": 0.0001,
+      "step": 5282
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.8865082138538615e-05,
+      "loss": 0.0164,
+      "step": 5283
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.886465795322136e-05,
+      "loss": 0.0271,
+      "step": 5284
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.8864233691765443e-05,
+      "loss": 0.7461,
+      "step": 5285
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.8863809354173194e-05,
+      "loss": 0.0003,
+      "step": 5286
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.886338494044694e-05,
+      "loss": 0.0002,
+      "step": 5287
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.886296045058902e-05,
+      "loss": 0.0002,
+      "step": 5288
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.8862535884601758e-05,
+      "loss": 0.0005,
+      "step": 5289
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.886211124248749e-05,
+      "loss": 0.0002,
+      "step": 5290
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.886168652424855e-05,
+      "loss": 0.0026,
+      "step": 5291
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.8861261729887258e-05,
+      "loss": 0.0003,
+      "step": 5292
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.8860836859405962e-05,
+      "loss": 0.0002,
+      "step": 5293
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.8860411912806986e-05,
+      "loss": 0.0003,
+      "step": 5294
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.8859986890092667e-05,
+      "loss": 0.5625,
+      "step": 5295
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.885956179126534e-05,
+      "loss": 0.1523,
+      "step": 5296
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.8859136616327334e-05,
+      "loss": 0.0835,
+      "step": 5297
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.885871136528099e-05,
+      "loss": 0.0005,
+      "step": 5298
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.8858286038128635e-05,
+      "loss": 0.2188,
+      "step": 5299
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.8857860634872618e-05,
+      "loss": 0.1475,
+      "step": 5300
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.8857435155515268e-05,
+      "loss": 0.6836,
+      "step": 5301
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.8857009600058915e-05,
+      "loss": 0.2656,
+      "step": 5302
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.8856583968505906e-05,
+      "loss": 0.0069,
+      "step": 5303
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.8856158260858574e-05,
+      "loss": 0.793,
+      "step": 5304
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.885573247711926e-05,
+      "loss": 0.1309,
+      "step": 5305
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.8855306617290298e-05,
+      "loss": 0.6641,
+      "step": 5306
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.8854880681374027e-05,
+      "loss": 0.3516,
+      "step": 5307
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.8854454669372785e-05,
+      "loss": 0.5664,
+      "step": 5308
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.8854028581288915e-05,
+      "loss": 0.0295,
+      "step": 5309
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.885360241712476e-05,
+      "loss": 0.6484,
+      "step": 5310
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.8853176176882653e-05,
+      "loss": 0.4043,
+      "step": 5311
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.8852749860564935e-05,
+      "loss": 0.053,
+      "step": 5312
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.8852323468173952e-05,
+      "loss": 0.498,
+      "step": 5313
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.885189699971205e-05,
+      "loss": 0.0371,
+      "step": 5314
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.8851470455181555e-05,
+      "loss": 0.1602,
+      "step": 5315
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.8851043834584822e-05,
+      "loss": 0.793,
+      "step": 5316
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.8850617137924192e-05,
+      "loss": 0.0247,
+      "step": 5317
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.8850190365202002e-05,
+      "loss": 0.1846,
+      "step": 5318
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.8849763516420607e-05,
+      "loss": 0.3281,
+      "step": 5319
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.884933659158234e-05,
+      "loss": 0.2949,
+      "step": 5320
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.884890959068955e-05,
+      "loss": 1.5859,
+      "step": 5321
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.884848251374459e-05,
+      "loss": 0.0623,
+      "step": 5322
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.8848055360749787e-05,
+      "loss": 0.1689,
+      "step": 5323
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.8847628131707502e-05,
+      "loss": 2.4375,
+      "step": 5324
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.8847200826620074e-05,
+      "loss": 0.0007,
+      "step": 5325
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.884677344548985e-05,
+      "loss": 0.0189,
+      "step": 5326
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.8846345988319183e-05,
+      "loss": 0.1875,
+      "step": 5327
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.8845918455110413e-05,
+      "loss": 0.0845,
+      "step": 5328
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.8845490845865895e-05,
+      "loss": 0.0004,
+      "step": 5329
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.8845063160587968e-05,
+      "loss": 0.0073,
+      "step": 5330
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.8844635399278986e-05,
+      "loss": 0.0064,
+      "step": 5331
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.88442075619413e-05,
+      "loss": 0.0004,
+      "step": 5332
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.884377964857726e-05,
+      "loss": 0.0079,
+      "step": 5333
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.884335165918921e-05,
+      "loss": 0.0009,
+      "step": 5334
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.8842923593779506e-05,
+      "loss": 0.0004,
+      "step": 5335
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.8842495452350495e-05,
+      "loss": 0.0001,
+      "step": 5336
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.884206723490453e-05,
+      "loss": 0.0229,
+      "step": 5337
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.884163894144396e-05,
+      "loss": 0.0003,
+      "step": 5338
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.8841210571971142e-05,
+      "loss": 0.0005,
+      "step": 5339
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.8840782126488424e-05,
+      "loss": 0.0023,
+      "step": 5340
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.8840353604998162e-05,
+      "loss": 0.0015,
+      "step": 5341
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.8839925007502703e-05,
+      "loss": 0.0001,
+      "step": 5342
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.883949633400441e-05,
+      "loss": 0.0004,
+      "step": 5343
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.8839067584505635e-05,
+      "loss": 0.1099,
+      "step": 5344
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.8838638759008724e-05,
+      "loss": 0.1416,
+      "step": 5345
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.883820985751604e-05,
+      "loss": 0.0009,
+      "step": 5346
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.8837780880029938e-05,
+      "loss": 0.1553,
+      "step": 5347
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.8837351826552767e-05,
+      "loss": 0.1426,
+      "step": 5348
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.883692269708689e-05,
+      "loss": 0.1328,
+      "step": 5349
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.8836493491634665e-05,
+      "loss": 0.2656,
+      "step": 5350
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.8836064210198444e-05,
+      "loss": 0.375,
+      "step": 5351
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.883563485278059e-05,
+      "loss": 0.5234,
+      "step": 5352
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.883520541938345e-05,
+      "loss": 1.3359,
+      "step": 5353
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.883477591000939e-05,
+      "loss": 0.9609,
+      "step": 5354
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.883434632466077e-05,
+      "loss": 0.8789,
+      "step": 5355
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.883391666333995e-05,
+      "loss": 0.6758,
+      "step": 5356
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.8833486926049282e-05,
+      "loss": 1.5781,
+      "step": 5357
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.8833057112791133e-05,
+      "loss": 0.4395,
+      "step": 5358
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.883262722356786e-05,
+      "loss": 0.0237,
+      "step": 5359
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.8832197258381824e-05,
+      "loss": 2.2344,
+      "step": 5360
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.8831767217235386e-05,
+      "loss": 0.6133,
+      "step": 5361
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.8831337100130913e-05,
+      "loss": 0.0508,
+      "step": 5362
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.8830906907070756e-05,
+      "loss": 0.0266,
+      "step": 5363
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.8830476638057287e-05,
+      "loss": 0.0664,
+      "step": 5364
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.8830046293092864e-05,
+      "loss": 0.3477,
+      "step": 5365
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.8829615872179858e-05,
+      "loss": 1.4688,
+      "step": 5366
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.882918537532062e-05,
+      "loss": 0.0356,
+      "step": 5367
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.8828754802517522e-05,
+      "loss": 0.1621,
+      "step": 5368
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.8828324153772928e-05,
+      "loss": 0.0019,
+      "step": 5369
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.88278934290892e-05,
+      "loss": 1.1641,
+      "step": 5370
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.882746262846871e-05,
+      "loss": 0.5781,
+      "step": 5371
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.882703175191382e-05,
+      "loss": 2.6094,
+      "step": 5372
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.8826600799426893e-05,
+      "loss": 0.0129,
+      "step": 5373
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.8826169771010294e-05,
+      "loss": 0.0476,
+      "step": 5374
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.8825738666666397e-05,
+      "loss": 0.4492,
+      "step": 5375
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.8825307486397566e-05,
+      "loss": 0.0001,
+      "step": 5376
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.8824876230206174e-05,
+      "loss": 0.6172,
+      "step": 5377
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.8824444898094582e-05,
+      "loss": 0.0001,
+      "step": 5378
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.8824013490065158e-05,
+      "loss": 0.5234,
+      "step": 5379
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.8823582006120278e-05,
+      "loss": 0.0,
+      "step": 5380
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.8823150446262305e-05,
+      "loss": 0.0001,
+      "step": 5381
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.8822718810493617e-05,
+      "loss": 0.0001,
+      "step": 5382
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.8822287098816576e-05,
+      "loss": 0.1211,
+      "step": 5383
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.8821855311233555e-05,
+      "loss": 0.0962,
+      "step": 5384
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.8821423447746926e-05,
+      "loss": 1.5938,
+      "step": 5385
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.8820991508359064e-05,
+      "loss": 0.1807,
+      "step": 5386
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.8820559493072337e-05,
+      "loss": 0.1807,
+      "step": 5387
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.8820127401889116e-05,
+      "loss": 0.0001,
+      "step": 5388
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.8819695234811778e-05,
+      "loss": 0.832,
+      "step": 5389
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.8819262991842696e-05,
+      "loss": 0.0028,
+      "step": 5390
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.881883067298424e-05,
+      "loss": 0.0,
+      "step": 5391
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.8818398278238786e-05,
+      "loss": 0.0003,
+      "step": 5392
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.881796580760871e-05,
+      "loss": 0.0002,
+      "step": 5393
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.8817533261096385e-05,
+      "loss": 0.1572,
+      "step": 5394
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.8817100638704183e-05,
+      "loss": 0.0,
+      "step": 5395
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.881666794043449e-05,
+      "loss": 0.4941,
+      "step": 5396
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.8816235166289674e-05,
+      "loss": 0.1641,
+      "step": 5397
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.8815802316272113e-05,
+      "loss": 0.1475,
+      "step": 5398
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.8815369390384184e-05,
+      "loss": 0.1001,
+      "step": 5399
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.8814936388628267e-05,
+      "loss": 0.1846,
+      "step": 5400
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.8814503311006735e-05,
+      "loss": 0.0471,
+      "step": 5401
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.8814070157521976e-05,
+      "loss": 0.9297,
+      "step": 5402
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.8813636928176355e-05,
+      "loss": 0.5938,
+      "step": 5403
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.881320362297226e-05,
+      "loss": 1.9375,
+      "step": 5404
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.881277024191207e-05,
+      "loss": 0.5352,
+      "step": 5405
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.8812336784998162e-05,
+      "loss": 1.4688,
+      "step": 5406
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.8811903252232916e-05,
+      "loss": 0.1973,
+      "step": 5407
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.881146964361872e-05,
+      "loss": 0.5938,
+      "step": 5408
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.8811035959157942e-05,
+      "loss": 1.6016,
+      "step": 5409
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.881060219885298e-05,
+      "loss": 1.8125,
+      "step": 5410
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.8810168362706203e-05,
+      "loss": 1.0312,
+      "step": 5411
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.8809734450719997e-05,
+      "loss": 0.8516,
+      "step": 5412
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.880930046289675e-05,
+      "loss": 0.1807,
+      "step": 5413
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.8808866399238834e-05,
+      "loss": 0.6016,
+      "step": 5414
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.8808432259748648e-05,
+      "loss": 1.2812,
+      "step": 5415
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.880799804442856e-05,
+      "loss": 0.0206,
+      "step": 5416
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.8807563753280966e-05,
+      "loss": 0.4375,
+      "step": 5417
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.8807129386308248e-05,
+      "loss": 0.0214,
+      "step": 5418
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.880669494351279e-05,
+      "loss": 0.207,
+      "step": 5419
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.8806260424896978e-05,
+      "loss": 0.1426,
+      "step": 5420
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.88058258304632e-05,
+      "loss": 0.2002,
+      "step": 5421
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.8805391160213838e-05,
+      "loss": 0.006,
+      "step": 5422
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.8804956414151283e-05,
+      "loss": 0.2598,
+      "step": 5423
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.880452159227792e-05,
+      "loss": 2.375,
+      "step": 5424
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.880408669459614e-05,
+      "loss": 0.0284,
+      "step": 5425
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.8803651721108333e-05,
+      "loss": 0.5547,
+      "step": 5426
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.880321667181688e-05,
+      "loss": 0.0327,
+      "step": 5427
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.8802781546724178e-05,
+      "loss": 0.0845,
+      "step": 5428
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.880234634583261e-05,
+      "loss": 0.8477,
+      "step": 5429
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.8801911069144573e-05,
+      "loss": 0.0796,
+      "step": 5430
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.8801475716662447e-05,
+      "loss": 0.0002,
+      "step": 5431
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.8801040288388637e-05,
+      "loss": 0.4238,
+      "step": 5432
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.880060478432552e-05,
+      "loss": 0.2305,
+      "step": 5433
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.8800169204475496e-05,
+      "loss": 0.0981,
+      "step": 5434
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.8799733548840956e-05,
+      "loss": 0.0003,
+      "step": 5435
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.8799297817424293e-05,
+      "loss": 0.0762,
+      "step": 5436
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.8798862010227895e-05,
+      "loss": 0.0001,
+      "step": 5437
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.8798426127254157e-05,
+      "loss": 0.0001,
+      "step": 5438
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.8797990168505476e-05,
+      "loss": 0.1895,
+      "step": 5439
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.8797554133984245e-05,
+      "loss": 0.0002,
+      "step": 5440
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.879711802369286e-05,
+      "loss": 0.0167,
+      "step": 5441
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.8796681837633707e-05,
+      "loss": 0.0001,
+      "step": 5442
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.8796245575809196e-05,
+      "loss": 0.0,
+      "step": 5443
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.8795809238221705e-05,
+      "loss": 0.0001,
+      "step": 5444
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.8795372824873646e-05,
+      "loss": 0.1611,
+      "step": 5445
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.8794936335767407e-05,
+      "loss": 0.0005,
+      "step": 5446
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.8794499770905394e-05,
+      "loss": 0.1191,
+      "step": 5447
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.879406313028999e-05,
+      "loss": 0.0767,
+      "step": 5448
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.8793626413923605e-05,
+      "loss": 0.1289,
+      "step": 5449
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.8793189621808632e-05,
+      "loss": 0.1982,
+      "step": 5450
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.8792752753947467e-05,
+      "loss": 2.3438,
+      "step": 5451
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.8792315810342522e-05,
+      "loss": 0.1387,
+      "step": 5452
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.879187879099618e-05,
+      "loss": 1.1484,
+      "step": 5453
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.879144169591085e-05,
+      "loss": 0.7148,
+      "step": 5454
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.879100452508893e-05,
+      "loss": 0.1216,
+      "step": 5455
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.8790567278532824e-05,
+      "loss": 0.8867,
+      "step": 5456
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.8790129956244927e-05,
+      "loss": 0.498,
+      "step": 5457
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.8789692558227646e-05,
+      "loss": 0.8398,
+      "step": 5458
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.878925508448338e-05,
+      "loss": 0.8047,
+      "step": 5459
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.8788817535014532e-05,
+      "loss": 0.4355,
+      "step": 5460
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.878837990982351e-05,
+      "loss": 0.5547,
+      "step": 5461
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.8787942208912705e-05,
+      "loss": 0.6562,
+      "step": 5462
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.8787504432284535e-05,
+      "loss": 0.9453,
+      "step": 5463
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.8787066579941393e-05,
+      "loss": 0.498,
+      "step": 5464
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.878662865188569e-05,
+      "loss": 2.0625,
+      "step": 5465
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.8786190648119824e-05,
+      "loss": 0.1104,
+      "step": 5466
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.878575256864621e-05,
+      "loss": 0.0238,
+      "step": 5467
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.878531441346725e-05,
+      "loss": 0.0002,
+      "step": 5468
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.8784876182585345e-05,
+      "loss": 0.0087,
+      "step": 5469
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.878443787600291e-05,
+      "loss": 0.0009,
+      "step": 5470
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.878399949372234e-05,
+      "loss": 0.0016,
+      "step": 5471
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.8783561035746054e-05,
+      "loss": 0.0042,
+      "step": 5472
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.8783122502076458e-05,
+      "loss": 0.0003,
+      "step": 5473
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.878268389271596e-05,
+      "loss": 0.0,
+      "step": 5474
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.878224520766696e-05,
+      "loss": 0.0,
+      "step": 5475
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.8781806446931875e-05,
+      "loss": 0.0007,
+      "step": 5476
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.878136761051312e-05,
+      "loss": 0.4277,
+      "step": 5477
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.878092869841309e-05,
+      "loss": 1.5078,
+      "step": 5478
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.8780489710634205e-05,
+      "loss": 0.0001,
+      "step": 5479
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.8780050647178877e-05,
+      "loss": 0.0004,
+      "step": 5480
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.8779611508049513e-05,
+      "loss": 0.1924,
+      "step": 5481
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.8779172293248523e-05,
+      "loss": 0.0,
+      "step": 5482
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.8778733002778322e-05,
+      "loss": 0.0,
+      "step": 5483
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.8778293636641323e-05,
+      "loss": 0.2832,
+      "step": 5484
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.8777854194839945e-05,
+      "loss": 0.6641,
+      "step": 5485
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.8777414677376586e-05,
+      "loss": 0.0001,
+      "step": 5486
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.877697508425367e-05,
+      "loss": 0.0,
+      "step": 5487
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.8776535415473605e-05,
+      "loss": 0.0,
+      "step": 5488
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.8776095671038817e-05,
+      "loss": 0.0001,
+      "step": 5489
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.877565585095171e-05,
+      "loss": 0.0593,
+      "step": 5490
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.87752159552147e-05,
+      "loss": 0.0001,
+      "step": 5491
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.8774775983830207e-05,
+      "loss": 0.0283,
+      "step": 5492
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.8774335936800648e-05,
+      "loss": 0.0,
+      "step": 5493
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.8773895814128433e-05,
+      "loss": 0.0001,
+      "step": 5494
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.8773455615815984e-05,
+      "loss": 0.0,
+      "step": 5495
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.877301534186572e-05,
+      "loss": 2.4062,
+      "step": 5496
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.877257499228005e-05,
+      "loss": 0.167,
+      "step": 5497
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.8772134567061407e-05,
+      "loss": 0.0991,
+      "step": 5498
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.8771694066212194e-05,
+      "loss": 0.1206,
+      "step": 5499
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.8771253489734838e-05,
+      "loss": 0.1138,
+      "step": 5500
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.8770812837631758e-05,
+      "loss": 0.1943,
+      "step": 5501
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.877037210990538e-05,
+      "loss": 1.8984,
+      "step": 5502
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.876993130655811e-05,
+      "loss": 1.5625,
+      "step": 5503
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.8769490427592375e-05,
+      "loss": 0.9805,
+      "step": 5504
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.8769049473010597e-05,
+      "loss": 0.061,
+      "step": 5505
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.8768608442815204e-05,
+      "loss": 0.5,
+      "step": 5506
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.8768167337008606e-05,
+      "loss": 0.793,
+      "step": 5507
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.8767726155593237e-05,
+      "loss": 1.3828,
+      "step": 5508
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.876728489857151e-05,
+      "loss": 1.1016,
+      "step": 5509
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.876684356594585e-05,
+      "loss": 0.0023,
+      "step": 5510
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.876640215771869e-05,
+      "loss": 0.0581,
+      "step": 5511
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.876596067389244e-05,
+      "loss": 0.0021,
+      "step": 5512
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.8765519114469533e-05,
+      "loss": 0.04,
+      "step": 5513
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.8765077479452395e-05,
+      "loss": 0.0457,
+      "step": 5514
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.876463576884345e-05,
+      "loss": 0.127,
+      "step": 5515
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.8764193982645113e-05,
+      "loss": 0.1553,
+      "step": 5516
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.8763752120859824e-05,
+      "loss": 0.2314,
+      "step": 5517
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.8763310183490007e-05,
+      "loss": 0.4434,
+      "step": 5518
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.8762868170538084e-05,
+      "loss": 0.021,
+      "step": 5519
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.8762426082006486e-05,
+      "loss": 0.0047,
+      "step": 5520
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.876198391789764e-05,
+      "loss": 0.1162,
+      "step": 5521
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.8761541678213973e-05,
+      "loss": 0.8672,
+      "step": 5522
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.8761099362957913e-05,
+      "loss": 0.8438,
+      "step": 5523
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.8760656972131894e-05,
+      "loss": 0.0002,
+      "step": 5524
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.8760214505738336e-05,
+      "loss": 0.0017,
+      "step": 5525
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.8759771963779677e-05,
+      "loss": 0.0001,
+      "step": 5526
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.8759329346258348e-05,
+      "loss": 0.0027,
+      "step": 5527
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.8758886653176777e-05,
+      "loss": 0.0,
+      "step": 5528
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.875844388453739e-05,
+      "loss": 0.918,
+      "step": 5529
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.875800104034263e-05,
+      "loss": 0.0021,
+      "step": 5530
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.8757558120594916e-05,
+      "loss": 0.0,
+      "step": 5531
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.8757115125296685e-05,
+      "loss": 0.2598,
+      "step": 5532
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.8756672054450373e-05,
+      "loss": 0.2422,
+      "step": 5533
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.8756228908058416e-05,
+      "loss": 0.168,
+      "step": 5534
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.875578568612324e-05,
+      "loss": 0.0,
+      "step": 5535
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.8755342388647282e-05,
+      "loss": 0.0,
+      "step": 5536
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.8754899015632978e-05,
+      "loss": 0.0,
+      "step": 5537
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.8754455567082755e-05,
+      "loss": 0.0,
+      "step": 5538
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.8754012042999057e-05,
+      "loss": 0.0051,
+      "step": 5539
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.875356844338432e-05,
+      "loss": 0.0011,
+      "step": 5540
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.8753124768240973e-05,
+      "loss": 0.0,
+      "step": 5541
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.875268101757146e-05,
+      "loss": 0.0,
+      "step": 5542
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.875223719137821e-05,
+      "loss": 0.1914,
+      "step": 5543
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.875179328966367e-05,
+      "loss": 0.0,
+      "step": 5544
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.8751349312430265e-05,
+      "loss": 0.8359,
+      "step": 5545
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.875090525968045e-05,
+      "loss": 0.1167,
+      "step": 5546
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.8750461131416645e-05,
+      "loss": 0.1201,
+      "step": 5547
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.8750016927641303e-05,
+      "loss": 0.1113,
+      "step": 5548
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.874957264835686e-05,
+      "loss": 0.1543,
+      "step": 5549
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.874912829356575e-05,
+      "loss": 0.1885,
+      "step": 5550
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.8748683863270418e-05,
+      "loss": 0.875,
+      "step": 5551
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.8748239357473308e-05,
+      "loss": 0.5469,
+      "step": 5552
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.874779477617685e-05,
+      "loss": 0.5078,
+      "step": 5553
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.87473501193835e-05,
+      "loss": 1.1016,
+      "step": 5554
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.8746905387095684e-05,
+      "loss": 1.0547,
+      "step": 5555
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.874646057931586e-05,
+      "loss": 0.0001,
+      "step": 5556
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.8746015696046458e-05,
+      "loss": 0.5078,
+      "step": 5557
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.8745570737289928e-05,
+      "loss": 0.0334,
+      "step": 5558
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.8745125703048712e-05,
+      "loss": 0.6016,
+      "step": 5559
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.8744680593325256e-05,
+      "loss": 0.4727,
+      "step": 5560
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.8744235408122003e-05,
+      "loss": 1.1328,
+      "step": 5561
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.874379014744139e-05,
+      "loss": 2.125,
+      "step": 5562
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.8743344811285873e-05,
+      "loss": 0.0007,
+      "step": 5563
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.8742899399657896e-05,
+      "loss": 0.0728,
+      "step": 5564
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.87424539125599e-05,
+      "loss": 1.9922,
+      "step": 5565
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.8742008349994334e-05,
+      "loss": 0.0359,
+      "step": 5566
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.8741562711963646e-05,
+      "loss": 0.543,
+      "step": 5567
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.874111699847028e-05,
+      "loss": 0.002,
+      "step": 5568
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.874067120951669e-05,
+      "loss": 0.2393,
+      "step": 5569
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.8740225345105317e-05,
+      "loss": 0.0004,
+      "step": 5570
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.8739779405238615e-05,
+      "loss": 0.9336,
+      "step": 5571
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.873933338991903e-05,
+      "loss": 0.6055,
+      "step": 5572
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.873888729914901e-05,
+      "loss": 0.9375,
+      "step": 5573
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.8738441132931004e-05,
+      "loss": 0.5898,
+      "step": 5574
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.873799489126747e-05,
+      "loss": 0.0532,
+      "step": 5575
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.873754857416085e-05,
+      "loss": 0.0002,
+      "step": 5576
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.8737102181613596e-05,
+      "loss": 0.021,
+      "step": 5577
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.8736655713628168e-05,
+      "loss": 0.0508,
+      "step": 5578
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.8736209170207004e-05,
+      "loss": 0.6758,
+      "step": 5579
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.873576255135257e-05,
+      "loss": 0.6797,
+      "step": 5580
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.873531585706731e-05,
+      "loss": 0.4297,
+      "step": 5581
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.8734869087353678e-05,
+      "loss": 0.0005,
+      "step": 5582
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.8734422242214127e-05,
+      "loss": 0.0,
+      "step": 5583
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.8733975321651117e-05,
+      "loss": 0.0003,
+      "step": 5584
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.8733528325667094e-05,
+      "loss": 0.3965,
+      "step": 5585
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.8733081254264518e-05,
+      "loss": 0.0001,
+      "step": 5586
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.873263410744584e-05,
+      "loss": 0.0422,
+      "step": 5587
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.8732186885213524e-05,
+      "loss": 0.0056,
+      "step": 5588
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.8731739587570013e-05,
+      "loss": 0.0001,
+      "step": 5589
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.8731292214517777e-05,
+      "loss": 0.0001,
+      "step": 5590
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.8730844766059267e-05,
+      "loss": 0.1631,
+      "step": 5591
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.8730397242196935e-05,
+      "loss": 0.0001,
+      "step": 5592
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.8729949642933245e-05,
+      "loss": 0.0016,
+      "step": 5593
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.872950196827065e-05,
+      "loss": 0.0001,
+      "step": 5594
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.8729054218211615e-05,
+      "loss": 0.0004,
+      "step": 5595
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.8728606392758594e-05,
+      "loss": 0.0121,
+      "step": 5596
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.8728158491914048e-05,
+      "loss": 0.1406,
+      "step": 5597
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.8727710515680438e-05,
+      "loss": 0.1167,
+      "step": 5598
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.872726246406022e-05,
+      "loss": 0.1465,
+      "step": 5599
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.8726814337055856e-05,
+      "loss": 0.2061,
+      "step": 5600
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.872636613466981e-05,
+      "loss": 0.6992,
+      "step": 5601
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.8725917856904542e-05,
+      "loss": 0.4629,
+      "step": 5602
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.872546950376251e-05,
+      "loss": 0.0471,
+      "step": 5603
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.8725021075246183e-05,
+      "loss": 1.2578,
+      "step": 5604
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.872457257135802e-05,
+      "loss": 0.1621,
+      "step": 5605
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.8724123992100483e-05,
+      "loss": 0.5664,
+      "step": 5606
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.8723675337476028e-05,
+      "loss": 0.793,
+      "step": 5607
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.8723226607487135e-05,
+      "loss": 0.4883,
+      "step": 5608
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.872277780213626e-05,
+      "loss": 0.106,
+      "step": 5609
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.8722328921425863e-05,
+      "loss": 0.4102,
+      "step": 5610
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.872187996535842e-05,
+      "loss": 0.0898,
+      "step": 5611
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.872143093393639e-05,
+      "loss": 0.6406,
+      "step": 5612
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.8720981827162235e-05,
+      "loss": 0.6719,
+      "step": 5613
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.8720532645038423e-05,
+      "loss": 0.4746,
+      "step": 5614
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.872008338756743e-05,
+      "loss": 1.0391,
+      "step": 5615
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.871963405475171e-05,
+      "loss": 0.0388,
+      "step": 5616
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.8719184646593745e-05,
+      "loss": 0.5312,
+      "step": 5617
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.871873516309599e-05,
+      "loss": 0.6094,
+      "step": 5618
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.8718285604260916e-05,
+      "loss": 0.2773,
+      "step": 5619
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.8717835970090995e-05,
+      "loss": 0.0,
+      "step": 5620
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.8717386260588696e-05,
+      "loss": 2.1094,
+      "step": 5621
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.871693647575649e-05,
+      "loss": 0.0199,
+      "step": 5622
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.8716486615596844e-05,
+      "loss": 2.0938,
+      "step": 5623
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.8716036680112226e-05,
+      "loss": 0.1406,
+      "step": 5624
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.871558666930511e-05,
+      "loss": 0.0048,
+      "step": 5625
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.8715136583177973e-05,
+      "loss": 0.019,
+      "step": 5626
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.8714686421733276e-05,
+      "loss": 0.0747,
+      "step": 5627
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.87142361849735e-05,
+      "loss": 0.0688,
+      "step": 5628
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.871378587290111e-05,
+      "loss": 0.1445,
+      "step": 5629
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.871333548551858e-05,
+      "loss": 0.0001,
+      "step": 5630
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.871288502282839e-05,
+      "loss": 0.1885,
+      "step": 5631
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.8712434484833012e-05,
+      "loss": 0.459,
+      "step": 5632
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.8711983871534912e-05,
+      "loss": 0.0157,
+      "step": 5633
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.8711533182936573e-05,
+      "loss": 0.0013,
+      "step": 5634
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.8711082419040468e-05,
+      "loss": 0.0001,
+      "step": 5635
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.8710631579849073e-05,
+      "loss": 0.0006,
+      "step": 5636
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.8710180665364858e-05,
+      "loss": 1.9688,
+      "step": 5637
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.8709729675590307e-05,
+      "loss": 0.8125,
+      "step": 5638
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.8709278610527894e-05,
+      "loss": 0.0004,
+      "step": 5639
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.8708827470180094e-05,
+      "loss": 0.0003,
+      "step": 5640
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.8708376254549388e-05,
+      "loss": 0.0002,
+      "step": 5641
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.8707924963638246e-05,
+      "loss": 0.0044,
+      "step": 5642
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.870747359744916e-05,
+      "loss": 0.5625,
+      "step": 5643
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.8707022155984595e-05,
+      "loss": 0.0002,
+      "step": 5644
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.870657063924704e-05,
+      "loss": 0.0898,
+      "step": 5645
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.8706119047238965e-05,
+      "loss": 0.0011,
+      "step": 5646
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.8705667379962858e-05,
+      "loss": 0.0723,
+      "step": 5647
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.8705215637421196e-05,
+      "loss": 0.3926,
+      "step": 5648
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.870476381961646e-05,
+      "loss": 0.1924,
+      "step": 5649
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.8704311926551136e-05,
+      "loss": 0.2246,
+      "step": 5650
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.8703859958227698e-05,
+      "loss": 0.7891,
+      "step": 5651
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.8703407914648632e-05,
+      "loss": 0.5117,
+      "step": 5652
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.870295579581642e-05,
+      "loss": 0.6719,
+      "step": 5653
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.8702503601733544e-05,
+      "loss": 1.625,
+      "step": 5654
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.8702051332402485e-05,
+      "loss": 0.7852,
+      "step": 5655
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.8701598987825738e-05,
+      "loss": 0.2949,
+      "step": 5656
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.870114656800577e-05,
+      "loss": 0.8906,
+      "step": 5657
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.8700694072945076e-05,
+      "loss": 0.1387,
+      "step": 5658
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.870024150264614e-05,
+      "loss": 0.3945,
+      "step": 5659
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.8699788857111446e-05,
+      "loss": 0.3242,
+      "step": 5660
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.8699336136343485e-05,
+      "loss": 1.7734,
+      "step": 5661
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.869888334034473e-05,
+      "loss": 0.7734,
+      "step": 5662
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.8698430469117683e-05,
+      "loss": 0.0238,
+      "step": 5663
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.8697977522664817e-05,
+      "loss": 0.5039,
+      "step": 5664
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.8697524500988628e-05,
+      "loss": 0.0255,
+      "step": 5665
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.8697071404091603e-05,
+      "loss": 0.0544,
+      "step": 5666
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.869661823197623e-05,
+      "loss": 0.0396,
+      "step": 5667
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.8696164984644996e-05,
+      "loss": 0.375,
+      "step": 5668
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.869571166210039e-05,
+      "loss": 0.0713,
+      "step": 5669
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.86952582643449e-05,
+      "loss": 0.1079,
+      "step": 5670
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.8694804791381025e-05,
+      "loss": 0.0008,
+      "step": 5671
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.869435124321124e-05,
+      "loss": 0.0008,
+      "step": 5672
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.8693897619838048e-05,
+      "loss": 0.0422,
+      "step": 5673
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.869344392126394e-05,
+      "loss": 0.05,
+      "step": 5674
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.8692990147491398e-05,
+      "loss": 0.5117,
+      "step": 5675
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.8692536298522924e-05,
+      "loss": 0.0002,
+      "step": 5676
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.8692082374361004e-05,
+      "loss": 0.0267,
+      "step": 5677
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.869162837500813e-05,
+      "loss": 0.0002,
+      "step": 5678
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.8691174300466798e-05,
+      "loss": 0.0674,
+      "step": 5679
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.8690720150739506e-05,
+      "loss": 0.1113,
+      "step": 5680
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.8690265925828742e-05,
+      "loss": 0.2031,
+      "step": 5681
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.8689811625737e-05,
+      "loss": 0.0295,
+      "step": 5682
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.8689357250466774e-05,
+      "loss": 0.0002,
+      "step": 5683
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.8688902800020567e-05,
+      "loss": 0.1445,
+      "step": 5684
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.868844827440087e-05,
+      "loss": 0.2793,
+      "step": 5685
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.8687993673610173e-05,
+      "loss": 0.3301,
+      "step": 5686
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.8687538997650984e-05,
+      "loss": 0.0002,
+      "step": 5687
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.8687084246525786e-05,
+      "loss": 0.0002,
+      "step": 5688
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.8686629420237094e-05,
+      "loss": 0.0002,
+      "step": 5689
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.868617451878739e-05,
+      "loss": 0.0005,
+      "step": 5690
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.868571954217918e-05,
+      "loss": 0.0001,
+      "step": 5691
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.868526449041496e-05,
+      "loss": 0.4277,
+      "step": 5692
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.8684809363497228e-05,
+      "loss": 0.0923,
+      "step": 5693
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.8684354161428488e-05,
+      "loss": 0.1455,
+      "step": 5694
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.8683898884211232e-05,
+      "loss": 0.1167,
+      "step": 5695
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.8683443531847967e-05,
+      "loss": 0.0002,
+      "step": 5696
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.8682988104341195e-05,
+      "loss": 0.1592,
+      "step": 5697
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.8682532601693408e-05,
+      "loss": 0.2598,
+      "step": 5698
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.8682077023907117e-05,
+      "loss": 0.1172,
+      "step": 5699
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.868162137098482e-05,
+      "loss": 0.1328,
+      "step": 5700
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.8681165642929012e-05,
+      "loss": 0.8594,
+      "step": 5701
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.8680709839742212e-05,
+      "loss": 0.3027,
+      "step": 5702
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.8680253961426907e-05,
+      "loss": 1.6406,
+      "step": 5703
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.867979800798561e-05,
+      "loss": 0.5938,
+      "step": 5704
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.8679341979420823e-05,
+      "loss": 0.5977,
+      "step": 5705
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.8678885875735048e-05,
+      "loss": 2.2656,
+      "step": 5706
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.867842969693079e-05,
+      "loss": 0.75,
+      "step": 5707
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.8677973443010552e-05,
+      "loss": 0.5039,
+      "step": 5708
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.8677517113976853e-05,
+      "loss": 0.6914,
+      "step": 5709
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.867706070983218e-05,
+      "loss": 0.8711,
+      "step": 5710
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.867660423057905e-05,
+      "loss": 0.0049,
+      "step": 5711
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.8676147676219968e-05,
+      "loss": 0.1074,
+      "step": 5712
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.8675691046757444e-05,
+      "loss": 0.2344,
+      "step": 5713
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.867523434219398e-05,
+      "loss": 0.1875,
+      "step": 5714
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.867477756253209e-05,
+      "loss": 0.0236,
+      "step": 5715
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.8674320707774272e-05,
+      "loss": 0.0002,
+      "step": 5716
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.8673863777923047e-05,
+      "loss": 0.0583,
+      "step": 5717
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.8673406772980918e-05,
+      "loss": 0.0004,
+      "step": 5718
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.86729496929504e-05,
+      "loss": 0.0009,
+      "step": 5719
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.8672492537833994e-05,
+      "loss": 0.2256,
+      "step": 5720
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.8672035307634216e-05,
+      "loss": 0.1543,
+      "step": 5721
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.8671578002353577e-05,
+      "loss": 0.0002,
+      "step": 5722
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.8671120621994588e-05,
+      "loss": 1.4062,
+      "step": 5723
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.867066316655976e-05,
+      "loss": 0.0042,
+      "step": 5724
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.8670205636051608e-05,
+      "loss": 0.0002,
+      "step": 5725
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.866974803047264e-05,
+      "loss": 0.1543,
+      "step": 5726
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.866929034982537e-05,
+      "loss": 0.0015,
+      "step": 5727
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.8668832594112314e-05,
+      "loss": 0.0002,
+      "step": 5728
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.8668374763335986e-05,
+      "loss": 2.6406,
+      "step": 5729
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.8667916857498897e-05,
+      "loss": 0.4883,
+      "step": 5730
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.8667458876603563e-05,
+      "loss": 0.126,
+      "step": 5731
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.86670008206525e-05,
+      "loss": 0.0001,
+      "step": 5732
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.866654268964822e-05,
+      "loss": 0.4961,
+      "step": 5733
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8666084483593247e-05,
+      "loss": 0.0403,
+      "step": 5734
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8665626202490087e-05,
+      "loss": 0.0598,
+      "step": 5735
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.866516784634126e-05,
+      "loss": 0.0137,
+      "step": 5736
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.866470941514929e-05,
+      "loss": 0.6484,
+      "step": 5737
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8664250908916686e-05,
+      "loss": 0.0344,
+      "step": 5738
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8663792327645967e-05,
+      "loss": 0.001,
+      "step": 5739
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.866333367133966e-05,
+      "loss": 0.5,
+      "step": 5740
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8662874940000273e-05,
+      "loss": 0.0004,
+      "step": 5741
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8662416133630326e-05,
+      "loss": 0.0033,
+      "step": 5742
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8661957252232344e-05,
+      "loss": 0.6953,
+      "step": 5743
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8661498295808846e-05,
+      "loss": 0.1147,
+      "step": 5744
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.866103926436235e-05,
+      "loss": 0.0001,
+      "step": 5745
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.866058015789538e-05,
+      "loss": 0.1191,
+      "step": 5746
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8660120976410452e-05,
+      "loss": 0.123,
+      "step": 5747
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8659661719910093e-05,
+      "loss": 0.1035,
+      "step": 5748
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.865920238839682e-05,
+      "loss": 0.1211,
+      "step": 5749
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.865874298187316e-05,
+      "loss": 0.1309,
+      "step": 5750
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8658283500341633e-05,
+      "loss": 1.4609,
+      "step": 5751
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8657823943804766e-05,
+      "loss": 1.1172,
+      "step": 5752
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.865736431226508e-05,
+      "loss": 0.5547,
+      "step": 5753
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8656904605725096e-05,
+      "loss": 0.3535,
+      "step": 5754
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8656444824187343e-05,
+      "loss": 0.062,
+      "step": 5755
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8655984967654346e-05,
+      "loss": 0.2754,
+      "step": 5756
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.865552503612863e-05,
+      "loss": 0.8789,
+      "step": 5757
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8655065029612716e-05,
+      "loss": 0.5938,
+      "step": 5758
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8654604948109134e-05,
+      "loss": 0.0026,
+      "step": 5759
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.865414479162041e-05,
+      "loss": 0.7227,
+      "step": 5760
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8653684560149077e-05,
+      "loss": 1.5391,
+      "step": 5761
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.865322425369765e-05,
+      "loss": 0.4023,
+      "step": 5762
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8652763872268668e-05,
+      "loss": 0.498,
+      "step": 5763
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.865230341586465e-05,
+      "loss": 0.0942,
+      "step": 5764
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8651842884488135e-05,
+      "loss": 0.6016,
+      "step": 5765
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8651382278141644e-05,
+      "loss": 0.6797,
+      "step": 5766
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8650921596827707e-05,
+      "loss": 0.8008,
+      "step": 5767
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.865046084054886e-05,
+      "loss": 0.0128,
+      "step": 5768
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8650000009307623e-05,
+      "loss": 0.0108,
+      "step": 5769
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.864953910310654e-05,
+      "loss": 0.9609,
+      "step": 5770
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8649078121948126e-05,
+      "loss": 0.0723,
+      "step": 5771
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8648617065834926e-05,
+      "loss": 0.0752,
+      "step": 5772
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8648155934769465e-05,
+      "loss": 0.3164,
+      "step": 5773
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.864769472875428e-05,
+      "loss": 0.0532,
+      "step": 5774
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.864723344779191e-05,
+      "loss": 0.7891,
+      "step": 5775
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8646772091884862e-05,
+      "loss": 0.0481,
+      "step": 5776
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.86463106610357e-05,
+      "loss": 0.1348,
+      "step": 5777
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8645849155246943e-05,
+      "loss": 0.0096,
+      "step": 5778
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.864538757452112e-05,
+      "loss": 0.031,
+      "step": 5779
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.864492591886078e-05,
+      "loss": 0.0664,
+      "step": 5780
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8644464188268446e-05,
+      "loss": 0.0001,
+      "step": 5781
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8644002382746665e-05,
+      "loss": 1.3516,
+      "step": 5782
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8643540502297965e-05,
+      "loss": 0.3691,
+      "step": 5783
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.864307854692488e-05,
+      "loss": 0.0001,
+      "step": 5784
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8642616516629958e-05,
+      "loss": 0.0004,
+      "step": 5785
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8642154411415726e-05,
+      "loss": 0.0051,
+      "step": 5786
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8641692231284727e-05,
+      "loss": 0.0398,
+      "step": 5787
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8641229976239494e-05,
+      "loss": 0.6641,
+      "step": 5788
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8640767646282576e-05,
+      "loss": 0.2031,
+      "step": 5789
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.86403052414165e-05,
+      "loss": 0.0452,
+      "step": 5790
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8639842761643808e-05,
+      "loss": 0.001,
+      "step": 5791
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8639380206967047e-05,
+      "loss": 0.2061,
+      "step": 5792
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8638917577388748e-05,
+      "loss": 0.0042,
+      "step": 5793
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8638454872911455e-05,
+      "loss": 0.0005,
+      "step": 5794
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8637992093537714e-05,
+      "loss": 0.001,
+      "step": 5795
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.863752923927006e-05,
+      "loss": 0.0835,
+      "step": 5796
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.863706631011103e-05,
+      "loss": 0.0806,
+      "step": 5797
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8636603306063186e-05,
+      "loss": 0.2451,
+      "step": 5798
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.8636140227129047e-05,
+      "loss": 0.2158,
+      "step": 5799
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.863567707331117e-05,
+      "loss": 0.1147,
+      "step": 5800
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8635213844612095e-05,
+      "loss": 0.6055,
+      "step": 5801
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8634750541034366e-05,
+      "loss": 1.3594,
+      "step": 5802
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8634287162580526e-05,
+      "loss": 0.9336,
+      "step": 5803
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8633823709253124e-05,
+      "loss": 0.707,
+      "step": 5804
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.86333601810547e-05,
+      "loss": 0.2012,
+      "step": 5805
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.86328965779878e-05,
+      "loss": 0.8047,
+      "step": 5806
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8632432900054975e-05,
+      "loss": 0.165,
+      "step": 5807
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8631969147258763e-05,
+      "loss": 0.5117,
+      "step": 5808
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.863150531960172e-05,
+      "loss": 0.5352,
+      "step": 5809
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8631041417086382e-05,
+      "loss": 0.2373,
+      "step": 5810
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8630577439715308e-05,
+      "loss": 0.8789,
+      "step": 5811
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8630113387491042e-05,
+      "loss": 0.0942,
+      "step": 5812
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8629649260416128e-05,
+      "loss": 0.1396,
+      "step": 5813
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8629185058493116e-05,
+      "loss": 0.2871,
+      "step": 5814
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.862872078172456e-05,
+      "loss": 0.7188,
+      "step": 5815
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8628256430113008e-05,
+      "loss": 0.7891,
+      "step": 5816
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.862779200366101e-05,
+      "loss": 0.0017,
+      "step": 5817
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8627327502371114e-05,
+      "loss": 0.0117,
+      "step": 5818
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.862686292624587e-05,
+      "loss": 0.0835,
+      "step": 5819
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8626398275287834e-05,
+      "loss": 1.9062,
+      "step": 5820
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8625933549499557e-05,
+      "loss": 0.0004,
+      "step": 5821
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.862546874888359e-05,
+      "loss": 0.6719,
+      "step": 5822
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8625003873442475e-05,
+      "loss": 0.127,
+      "step": 5823
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8624538923178784e-05,
+      "loss": 0.0442,
+      "step": 5824
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8624073898095057e-05,
+      "loss": 0.0021,
+      "step": 5825
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.862360879819385e-05,
+      "loss": 0.0002,
+      "step": 5826
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8623143623477725e-05,
+      "loss": 0.0007,
+      "step": 5827
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8622678373949226e-05,
+      "loss": 0.332,
+      "step": 5828
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8622213049610914e-05,
+      "loss": 0.0374,
+      "step": 5829
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8621747650465338e-05,
+      "loss": 0.0,
+      "step": 5830
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.862128217651506e-05,
+      "loss": 0.0001,
+      "step": 5831
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.862081662776264e-05,
+      "loss": 0.0776,
+      "step": 5832
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.862035100421062e-05,
+      "loss": 0.0006,
+      "step": 5833
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8619885305861574e-05,
+      "loss": 0.2617,
+      "step": 5834
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.861941953271805e-05,
+      "loss": 0.0002,
+      "step": 5835
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8618953684782603e-05,
+      "loss": 1.3359,
+      "step": 5836
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.86184877620578e-05,
+      "loss": 0.0002,
+      "step": 5837
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8618021764546196e-05,
+      "loss": 0.0322,
+      "step": 5838
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8617555692250347e-05,
+      "loss": 0.0067,
+      "step": 5839
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8617089545172815e-05,
+      "loss": 0.0004,
+      "step": 5840
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8616623323316158e-05,
+      "loss": 0.0002,
+      "step": 5841
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.861615702668294e-05,
+      "loss": 1.0547,
+      "step": 5842
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8615690655275717e-05,
+      "loss": 0.0006,
+      "step": 5843
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8615224209097055e-05,
+      "loss": 0.0001,
+      "step": 5844
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8614757688149514e-05,
+      "loss": 0.0007,
+      "step": 5845
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8614291092435655e-05,
+      "loss": 0.0009,
+      "step": 5846
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8613824421958037e-05,
+      "loss": 0.7227,
+      "step": 5847
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8613357676719228e-05,
+      "loss": 0.1748,
+      "step": 5848
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8612890856721792e-05,
+      "loss": 0.0352,
+      "step": 5849
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8612423961968288e-05,
+      "loss": 0.0571,
+      "step": 5850
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8611956992461284e-05,
+      "loss": 0.4023,
+      "step": 5851
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8611489948203344e-05,
+      "loss": 1.0781,
+      "step": 5852
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8611022829197023e-05,
+      "loss": 0.6523,
+      "step": 5853
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.86105556354449e-05,
+      "loss": 0.5273,
+      "step": 5854
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8610088366949533e-05,
+      "loss": 0.4844,
+      "step": 5855
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8609621023713492e-05,
+      "loss": 0.4141,
+      "step": 5856
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.860915360573934e-05,
+      "loss": 0.377,
+      "step": 5857
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8608686113029646e-05,
+      "loss": 2.0,
+      "step": 5858
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8608218545586973e-05,
+      "loss": 1.2344,
+      "step": 5859
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.86077509034139e-05,
+      "loss": 0.4492,
+      "step": 5860
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8607283186512982e-05,
+      "loss": 0.3867,
+      "step": 5861
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8606815394886794e-05,
+      "loss": 1.7734,
+      "step": 5862
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8606347528537906e-05,
+      "loss": 0.2734,
+      "step": 5863
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.860587958746888e-05,
+      "loss": 0.0591,
+      "step": 5864
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8605411571682295e-05,
+      "loss": 1.3984,
+      "step": 5865
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8604943481180717e-05,
+      "loss": 0.459,
+      "step": 5866
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.8604475315966713e-05,
+      "loss": 0.0243,
+      "step": 5867
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.860400707604286e-05,
+      "loss": 0.6523,
+      "step": 5868
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.860353876141173e-05,
+      "loss": 1.3359,
+      "step": 5869
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8603070372075887e-05,
+      "loss": 0.75,
+      "step": 5870
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8602601908037906e-05,
+      "loss": 0.0581,
+      "step": 5871
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8602133369300367e-05,
+      "loss": 0.5898,
+      "step": 5872
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8601664755865835e-05,
+      "loss": 0.0001,
+      "step": 5873
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8601196067736885e-05,
+      "loss": 0.208,
+      "step": 5874
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8600727304916092e-05,
+      "loss": 0.0059,
+      "step": 5875
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.860025846740603e-05,
+      "loss": 0.0034,
+      "step": 5876
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8599789555209274e-05,
+      "loss": 0.0006,
+      "step": 5877
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8599320568328398e-05,
+      "loss": 0.6562,
+      "step": 5878
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.859885150676598e-05,
+      "loss": 0.3555,
+      "step": 5879
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8598382370524593e-05,
+      "loss": 0.0001,
+      "step": 5880
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8597913159606813e-05,
+      "loss": 0.0002,
+      "step": 5881
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.859744387401522e-05,
+      "loss": 0.0583,
+      "step": 5882
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8596974513752384e-05,
+      "loss": 0.0002,
+      "step": 5883
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.859650507882089e-05,
+      "loss": 0.0035,
+      "step": 5884
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8596035569223316e-05,
+      "loss": 0.1348,
+      "step": 5885
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.859556598496224e-05,
+      "loss": 0.0001,
+      "step": 5886
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8595096326040233e-05,
+      "loss": 0.0972,
+      "step": 5887
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8594626592459882e-05,
+      "loss": 0.0398,
+      "step": 5888
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8594156784223765e-05,
+      "loss": 0.0017,
+      "step": 5889
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8593686901334463e-05,
+      "loss": 0.6758,
+      "step": 5890
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8593216943794552e-05,
+      "loss": 0.0001,
+      "step": 5891
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8592746911606613e-05,
+      "loss": 0.0001,
+      "step": 5892
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8592276804773232e-05,
+      "loss": 0.1475,
+      "step": 5893
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8591806623296993e-05,
+      "loss": 0.0002,
+      "step": 5894
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.859133636718047e-05,
+      "loss": 0.0001,
+      "step": 5895
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.859086603642625e-05,
+      "loss": 0.0003,
+      "step": 5896
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8590395631036906e-05,
+      "loss": 0.1338,
+      "step": 5897
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8589925151015038e-05,
+      "loss": 0.0435,
+      "step": 5898
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8589454596363222e-05,
+      "loss": 0.0586,
+      "step": 5899
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8588983967084042e-05,
+      "loss": 0.1631,
+      "step": 5900
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.858851326318008e-05,
+      "loss": 0.125,
+      "step": 5901
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8588042484653926e-05,
+      "loss": 0.3711,
+      "step": 5902
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.858757163150816e-05,
+      "loss": 0.4707,
+      "step": 5903
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.858710070374537e-05,
+      "loss": 0.2354,
+      "step": 5904
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8586629701368143e-05,
+      "loss": 1.0547,
+      "step": 5905
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.858615862437906e-05,
+      "loss": 0.7305,
+      "step": 5906
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.858568747278072e-05,
+      "loss": 1.0156,
+      "step": 5907
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.85852162465757e-05,
+      "loss": 0.8125,
+      "step": 5908
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8584744945766592e-05,
+      "loss": 2.3281,
+      "step": 5909
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8584273570355984e-05,
+      "loss": 0.1992,
+      "step": 5910
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.858380212034646e-05,
+      "loss": 0.3008,
+      "step": 5911
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8583330595740613e-05,
+      "loss": 0.0334,
+      "step": 5912
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8582858996541034e-05,
+      "loss": 0.0581,
+      "step": 5913
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8582387322750314e-05,
+      "loss": 0.2324,
+      "step": 5914
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8581915574371035e-05,
+      "loss": 0.0388,
+      "step": 5915
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8581443751405795e-05,
+      "loss": 0.0096,
+      "step": 5916
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.858097185385718e-05,
+      "loss": 0.5352,
+      "step": 5917
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.858049988172779e-05,
+      "loss": 0.0062,
+      "step": 5918
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8580027835020205e-05,
+      "loss": 0.9727,
+      "step": 5919
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8579555713737026e-05,
+      "loss": 0.3574,
+      "step": 5920
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8579083517880842e-05,
+      "loss": 0.2129,
+      "step": 5921
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.857861124745425e-05,
+      "loss": 0.0128,
+      "step": 5922
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.857813890245984e-05,
+      "loss": 0.75,
+      "step": 5923
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8577666482900205e-05,
+      "loss": 0.0016,
+      "step": 5924
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.857719398877794e-05,
+      "loss": 0.0001,
+      "step": 5925
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8576721420095644e-05,
+      "loss": 0.0014,
+      "step": 5926
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8576248776855906e-05,
+      "loss": 0.459,
+      "step": 5927
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8575776059061323e-05,
+      "loss": 0.0011,
+      "step": 5928
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8575303266714496e-05,
+      "loss": 1.4766,
+      "step": 5929
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.857483039981801e-05,
+      "loss": 0.7578,
+      "step": 5930
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.857435745837448e-05,
+      "loss": 0.1084,
+      "step": 5931
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8573884442386488e-05,
+      "loss": 0.1377,
+      "step": 5932
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8573411351856636e-05,
+      "loss": 0.2236,
+      "step": 5933
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.8572938186787522e-05,
+      "loss": 0.2578,
+      "step": 5934
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.857246494718175e-05,
+      "loss": 0.4258,
+      "step": 5935
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.8571991633041907e-05,
+      "loss": 0.0001,
+      "step": 5936
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.8571518244370596e-05,
+      "loss": 0.0018,
+      "step": 5937
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.8571044781170425e-05,
+      "loss": 0.0001,
+      "step": 5938
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.8570571243443986e-05,
+      "loss": 0.0016,
+      "step": 5939
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.857009763119388e-05,
+      "loss": 0.1641,
+      "step": 5940
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.8569623944422713e-05,
+      "loss": 0.0005,
+      "step": 5941
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.8569150183133088e-05,
+      "loss": 0.0005,
+      "step": 5942
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.8568676347327596e-05,
+      "loss": 0.0001,
+      "step": 5943
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.8568202437008842e-05,
+      "loss": 0.0003,
+      "step": 5944
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.8567728452179436e-05,
+      "loss": 0.1689,
+      "step": 5945
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.8567254392841973e-05,
+      "loss": 0.1709,
+      "step": 5946
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.8566780258999064e-05,
+      "loss": 0.0001,
+      "step": 5947
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.8566306050653306e-05,
+      "loss": 0.0752,
+      "step": 5948
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.85658317678073e-05,
+      "loss": 0.1973,
+      "step": 5949
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.8565357410463664e-05,
+      "loss": 0.1387,
+      "step": 5950
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.856488297862499e-05,
+      "loss": 1.1953,
+      "step": 5951
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.856440847229389e-05,
+      "loss": 0.0422,
+      "step": 5952
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.856393389147297e-05,
+      "loss": 0.5234,
+      "step": 5953
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.856345923616483e-05,
+      "loss": 0.3633,
+      "step": 5954
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.8562984506372085e-05,
+      "loss": 0.6133,
+      "step": 5955
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.8562509702097334e-05,
+      "loss": 0.0309,
+      "step": 5956
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.8562034823343196e-05,
+      "loss": 0.0032,
+      "step": 5957
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.8561559870112263e-05,
+      "loss": 0.5156,
+      "step": 5958
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.8561084842407154e-05,
+      "loss": 1.4453,
+      "step": 5959
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.856060974023048e-05,
+      "loss": 1.4375,
+      "step": 5960
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.8560134563584838e-05,
+      "loss": 0.8164,
+      "step": 5961
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.8559659312472844e-05,
+      "loss": 0.3848,
+      "step": 5962
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.8559183986897114e-05,
+      "loss": 1.3125,
+      "step": 5963
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.8558708586860252e-05,
+      "loss": 2.1562,
+      "step": 5964
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.8558233112364864e-05,
+      "loss": 0.0103,
+      "step": 5965
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.855775756341357e-05,
+      "loss": 0.0002,
+      "step": 5966
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.855728194000898e-05,
+      "loss": 0.5547,
+      "step": 5967
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.85568062421537e-05,
+      "loss": 0.0012,
+      "step": 5968
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.855633046985035e-05,
+      "loss": 0.0072,
+      "step": 5969
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.855585462310154e-05,
+      "loss": 0.0013,
+      "step": 5970
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.855537870190988e-05,
+      "loss": 0.015,
+      "step": 5971
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.855490270627798e-05,
+      "loss": 0.0947,
+      "step": 5972
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.855442663620847e-05,
+      "loss": 0.0601,
+      "step": 5973
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.8553950491703945e-05,
+      "loss": 0.0001,
+      "step": 5974
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.8553474272767034e-05,
+      "loss": 0.209,
+      "step": 5975
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.8552997979400345e-05,
+      "loss": 0.8008,
+      "step": 5976
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.8552521611606495e-05,
+      "loss": 0.0001,
+      "step": 5977
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.8552045169388104e-05,
+      "loss": 0.2197,
+      "step": 5978
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.855156865274778e-05,
+      "loss": 0.0001,
+      "step": 5979
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.8551092061688146e-05,
+      "loss": 0.6484,
+      "step": 5980
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.8550615396211823e-05,
+      "loss": 0.0128,
+      "step": 5981
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.855013865632142e-05,
+      "loss": 0.0002,
+      "step": 5982
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.8549661842019558e-05,
+      "loss": 0.3281,
+      "step": 5983
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.8549184953308857e-05,
+      "loss": 0.0004,
+      "step": 5984
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.8548707990191933e-05,
+      "loss": 0.0003,
+      "step": 5985
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.8548230952671408e-05,
+      "loss": 0.0043,
+      "step": 5986
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.8547753840749904e-05,
+      "loss": 0.0022,
+      "step": 5987
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.854727665443004e-05,
+      "loss": 0.4883,
+      "step": 5988
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.854679939371443e-05,
+      "loss": 0.0003,
+      "step": 5989
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.85463220586057e-05,
+      "loss": 0.0004,
+      "step": 5990
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.854584464910647e-05,
+      "loss": 0.7227,
+      "step": 5991
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.8545367165219363e-05,
+      "loss": 0.0003,
+      "step": 5992
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.8544889606947003e-05,
+      "loss": 0.1729,
+      "step": 5993
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.8544411974292008e-05,
+      "loss": 0.1182,
+      "step": 5994
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.8543934267257007e-05,
+      "loss": 0.0014,
+      "step": 5995
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.8543456485844612e-05,
+      "loss": 0.2402,
+      "step": 5996
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.8542978630057463e-05,
+      "loss": 0.1094,
+      "step": 5997
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.854250069989817e-05,
+      "loss": 0.1426,
+      "step": 5998
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.8542022695369366e-05,
+      "loss": 0.1377,
+      "step": 5999
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.854154461647367e-05,
+      "loss": 0.1084,
+      "step": 6000
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.8541066463213713e-05,
+      "loss": 1.7266,
+      "step": 6001
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.8540588235592117e-05,
+      "loss": 1.3672,
+      "step": 6002
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8540109933611508e-05,
+      "loss": 0.6758,
+      "step": 6003
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8539631557274514e-05,
+      "loss": 0.7461,
+      "step": 6004
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8539153106583766e-05,
+      "loss": 0.5547,
+      "step": 6005
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.853867458154189e-05,
+      "loss": 0.7344,
+      "step": 6006
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.85381959821515e-05,
+      "loss": 0.7617,
+      "step": 6007
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8537717308415246e-05,
+      "loss": 0.0012,
+      "step": 6008
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.853723856033574e-05,
+      "loss": 0.1777,
+      "step": 6009
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.853675973791562e-05,
+      "loss": 0.6758,
+      "step": 6010
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.853628084115752e-05,
+      "loss": 0.7031,
+      "step": 6011
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8535801870064052e-05,
+      "loss": 0.3047,
+      "step": 6012
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8535322824637865e-05,
+      "loss": 0.2949,
+      "step": 6013
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8534843704881575e-05,
+      "loss": 0.0728,
+      "step": 6014
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8534364510797822e-05,
+      "loss": 1.8281,
+      "step": 6015
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8533885242389235e-05,
+      "loss": 0.8906,
+      "step": 6016
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.853340589965845e-05,
+      "loss": 0.0664,
+      "step": 6017
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8532926482608094e-05,
+      "loss": 0.0244,
+      "step": 6018
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8532446991240802e-05,
+      "loss": 0.0001,
+      "step": 6019
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8531967425559206e-05,
+      "loss": 0.5273,
+      "step": 6020
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.853148778556594e-05,
+      "loss": 0.0194,
+      "step": 6021
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8531008071263637e-05,
+      "loss": 0.2148,
+      "step": 6022
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8530528282654932e-05,
+      "loss": 0.0293,
+      "step": 6023
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8530048419742465e-05,
+      "loss": 0.0444,
+      "step": 6024
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.852956848252886e-05,
+      "loss": 0.1768,
+      "step": 6025
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.852908847101676e-05,
+      "loss": 0.0003,
+      "step": 6026
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8528608385208807e-05,
+      "loss": 0.0002,
+      "step": 6027
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8528128225107627e-05,
+      "loss": 0.0182,
+      "step": 6028
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.852764799071586e-05,
+      "loss": 0.0008,
+      "step": 6029
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8527167682036147e-05,
+      "loss": 0.0022,
+      "step": 6030
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8526687299071115e-05,
+      "loss": 1.4922,
+      "step": 6031
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.852620684182341e-05,
+      "loss": 0.0002,
+      "step": 6032
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8525726310295678e-05,
+      "loss": 0.008,
+      "step": 6033
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8525245704490542e-05,
+      "loss": 0.0767,
+      "step": 6034
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8524765024410655e-05,
+      "loss": 0.0002,
+      "step": 6035
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8524284270058647e-05,
+      "loss": 0.1572,
+      "step": 6036
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.852380344143716e-05,
+      "loss": 0.0002,
+      "step": 6037
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.852332253854884e-05,
+      "loss": 1.1875,
+      "step": 6038
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.852284156139632e-05,
+      "loss": 0.0003,
+      "step": 6039
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.852236050998225e-05,
+      "loss": 0.0003,
+      "step": 6040
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8521879384309268e-05,
+      "loss": 0.0649,
+      "step": 6041
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8521398184380013e-05,
+      "loss": 0.0014,
+      "step": 6042
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8520916910197126e-05,
+      "loss": 0.0056,
+      "step": 6043
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.852043556176326e-05,
+      "loss": 0.1865,
+      "step": 6044
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8519954139081047e-05,
+      "loss": 0.0037,
+      "step": 6045
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.851947264215314e-05,
+      "loss": 0.0972,
+      "step": 6046
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.851899107098218e-05,
+      "loss": 0.001,
+      "step": 6047
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8518509425570806e-05,
+      "loss": 0.208,
+      "step": 6048
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8518027705921667e-05,
+      "loss": 0.0002,
+      "step": 6049
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8517545912037412e-05,
+      "loss": 0.1094,
+      "step": 6050
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8517064043920684e-05,
+      "loss": 1.6875,
+      "step": 6051
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8516582101574133e-05,
+      "loss": 0.8789,
+      "step": 6052
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.85161000850004e-05,
+      "loss": 0.0674,
+      "step": 6053
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.851561799420213e-05,
+      "loss": 0.3828,
+      "step": 6054
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8515135829181972e-05,
+      "loss": 0.0226,
+      "step": 6055
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8514653589942583e-05,
+      "loss": 1.9141,
+      "step": 6056
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8514171276486597e-05,
+      "loss": 0.4316,
+      "step": 6057
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8513688888816674e-05,
+      "loss": 0.498,
+      "step": 6058
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8513206426935464e-05,
+      "loss": 0.0811,
+      "step": 6059
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8512723890845603e-05,
+      "loss": 0.0603,
+      "step": 6060
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.851224128054975e-05,
+      "loss": 0.2598,
+      "step": 6061
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.851175859605056e-05,
+      "loss": 0.3789,
+      "step": 6062
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8511275837350674e-05,
+      "loss": 0.3906,
+      "step": 6063
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.851079300445275e-05,
+      "loss": 0.0001,
+      "step": 6064
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8510310097359435e-05,
+      "loss": 0.0093,
+      "step": 6065
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8509827116073383e-05,
+      "loss": 0.0001,
+      "step": 6066
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8509344060597247e-05,
+      "loss": 0.0,
+      "step": 6067
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8508860930933683e-05,
+      "loss": 0.2324,
+      "step": 6068
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8508377727085333e-05,
+      "loss": 0.9102,
+      "step": 6069
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.850789444905486e-05,
+      "loss": 0.0347,
+      "step": 6070
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8507411096844917e-05,
+      "loss": 0.0508,
+      "step": 6071
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8506927670458155e-05,
+      "loss": 1.9531,
+      "step": 6072
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.850644416989723e-05,
+      "loss": 0.0564,
+      "step": 6073
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.85059605951648e-05,
+      "loss": 0.0,
+      "step": 6074
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8505476946263515e-05,
+      "loss": 0.375,
+      "step": 6075
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.850499322319604e-05,
+      "loss": 0.7266,
+      "step": 6076
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.850450942596502e-05,
+      "loss": 0.0,
+      "step": 6077
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8504025554573123e-05,
+      "loss": 0.082,
+      "step": 6078
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8503541609022996e-05,
+      "loss": 0.1152,
+      "step": 6079
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8503057589317305e-05,
+      "loss": 0.0,
+      "step": 6080
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8502573495458702e-05,
+      "loss": 0.0349,
+      "step": 6081
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.850208932744985e-05,
+      "loss": 0.3281,
+      "step": 6082
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8501605085293404e-05,
+      "loss": 0.5508,
+      "step": 6083
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8501120768992026e-05,
+      "loss": 0.0001,
+      "step": 6084
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.850063637854837e-05,
+      "loss": 0.1182,
+      "step": 6085
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8500151913965106e-05,
+      "loss": 0.0206,
+      "step": 6086
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8499667375244887e-05,
+      "loss": 0.0005,
+      "step": 6087
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.849918276239038e-05,
+      "loss": 0.0002,
+      "step": 6088
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8498698075404235e-05,
+      "loss": 0.0354,
+      "step": 6089
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8498213314289125e-05,
+      "loss": 0.7656,
+      "step": 6090
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.849772847904771e-05,
+      "loss": 0.0003,
+      "step": 6091
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.849724356968265e-05,
+      "loss": 1.0391,
+      "step": 6092
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8496758586196605e-05,
+      "loss": 0.0,
+      "step": 6093
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8496273528592245e-05,
+      "loss": 0.1523,
+      "step": 6094
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.849578839687223e-05,
+      "loss": 0.168,
+      "step": 6095
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8495303191039224e-05,
+      "loss": 0.1738,
+      "step": 6096
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8494817911095892e-05,
+      "loss": 0.0007,
+      "step": 6097
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8494332557044902e-05,
+      "loss": 0.1543,
+      "step": 6098
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.849384712888891e-05,
+      "loss": 0.1738,
+      "step": 6099
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8493361626630594e-05,
+      "loss": 0.1807,
+      "step": 6100
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8492876050272612e-05,
+      "loss": 1.5781,
+      "step": 6101
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8492390399817635e-05,
+      "loss": 1.6172,
+      "step": 6102
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8491904675268325e-05,
+      "loss": 0.1406,
+      "step": 6103
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.849141887662736e-05,
+      "loss": 0.6836,
+      "step": 6104
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.849093300389739e-05,
+      "loss": 0.0962,
+      "step": 6105
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8490447057081102e-05,
+      "loss": 0.2988,
+      "step": 6106
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8489961036181148e-05,
+      "loss": 0.4336,
+      "step": 6107
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.848947494120021e-05,
+      "loss": 0.5039,
+      "step": 6108
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8488988772140952e-05,
+      "loss": 0.6875,
+      "step": 6109
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.848850252900604e-05,
+      "loss": 0.0417,
+      "step": 6110
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8488016211798155e-05,
+      "loss": 0.4766,
+      "step": 6111
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8487529820519956e-05,
+      "loss": 1.2969,
+      "step": 6112
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8487043355174123e-05,
+      "loss": 0.6641,
+      "step": 6113
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8486556815763322e-05,
+      "loss": 0.8359,
+      "step": 6114
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8486070202290224e-05,
+      "loss": 0.5977,
+      "step": 6115
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8485583514757508e-05,
+      "loss": 1.4141,
+      "step": 6116
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8485096753167838e-05,
+      "loss": 0.0289,
+      "step": 6117
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8484609917523896e-05,
+      "loss": 0.1465,
+      "step": 6118
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.848412300782835e-05,
+      "loss": 0.0,
+      "step": 6119
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8483636024083875e-05,
+      "loss": 0.0001,
+      "step": 6120
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8483148966293143e-05,
+      "loss": 0.3164,
+      "step": 6121
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8482661834458832e-05,
+      "loss": 0.334,
+      "step": 6122
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8482174628583616e-05,
+      "loss": 0.0383,
+      "step": 6123
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.848168734867017e-05,
+      "loss": 0.9062,
+      "step": 6124
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8481199994721173e-05,
+      "loss": 0.0002,
+      "step": 6125
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8480712566739295e-05,
+      "loss": 0.0,
+      "step": 6126
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.848022506472722e-05,
+      "loss": 0.1523,
+      "step": 6127
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.847973748868762e-05,
+      "loss": 0.0003,
+      "step": 6128
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8479249838623175e-05,
+      "loss": 0.0,
+      "step": 6129
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.847876211453656e-05,
+      "loss": 0.0002,
+      "step": 6130
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.847827431643046e-05,
+      "loss": 0.0001,
+      "step": 6131
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8477786444307547e-05,
+      "loss": 0.0,
+      "step": 6132
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8477298498170502e-05,
+      "loss": 0.0,
+      "step": 6133
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8476810478022002e-05,
+      "loss": 0.0001,
+      "step": 6134
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8476322383864738e-05,
+      "loss": 0.0004,
+      "step": 6135
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.847583421570138e-05,
+      "loss": 0.416,
+      "step": 6136
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8475345973534605e-05,
+      "loss": 0.9727,
+      "step": 6137
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8474857657367108e-05,
+      "loss": 0.0027,
+      "step": 6138
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.847436926720156e-05,
+      "loss": 0.2949,
+      "step": 6139
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.8473880803040645e-05,
+      "loss": 0.0001,
+      "step": 6140
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.847339226488705e-05,
+      "loss": 0.0001,
+      "step": 6141
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.8472903652743448e-05,
+      "loss": 0.8281,
+      "step": 6142
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.8472414966612536e-05,
+      "loss": 0.0,
+      "step": 6143
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.847192620649699e-05,
+      "loss": 0.0004,
+      "step": 6144
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.847143737239949e-05,
+      "loss": 0.0001,
+      "step": 6145
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.847094846432273e-05,
+      "loss": 0.0001,
+      "step": 6146
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.847045948226938e-05,
+      "loss": 0.1973,
+      "step": 6147
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.8469970426242146e-05,
+      "loss": 0.1348,
+      "step": 6148
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.84694812962437e-05,
+      "loss": 0.0796,
+      "step": 6149
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.8468992092276728e-05,
+      "loss": 0.1416,
+      "step": 6150
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.8468502814343922e-05,
+      "loss": 0.1069,
+      "step": 6151
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.8468013462447966e-05,
+      "loss": 0.3379,
+      "step": 6152
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.8467524036591543e-05,
+      "loss": 0.3496,
+      "step": 6153
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.846703453677735e-05,
+      "loss": 0.6992,
+      "step": 6154
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.846654496300807e-05,
+      "loss": 0.0057,
+      "step": 6155
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.8466055315286387e-05,
+      "loss": 0.4746,
+      "step": 6156
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.8465565593615e-05,
+      "loss": 0.03,
+      "step": 6157
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.846507579799659e-05,
+      "loss": 0.8086,
+      "step": 6158
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.8464585928433854e-05,
+      "loss": 0.0292,
+      "step": 6159
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.8464095984929474e-05,
+      "loss": 1.4219,
+      "step": 6160
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.846360596748615e-05,
+      "loss": 0.4941,
+      "step": 6161
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.8463115876106563e-05,
+      "loss": 1.0078,
+      "step": 6162
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.846262571079341e-05,
+      "loss": 0.1602,
+      "step": 6163
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.8462135471549383e-05,
+      "loss": 0.8398,
+      "step": 6164
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.8461645158377172e-05,
+      "loss": 0.5586,
+      "step": 6165
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.846115477127947e-05,
+      "loss": 0.0981,
+      "step": 6166
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.8460664310258965e-05,
+      "loss": 0.0002,
+      "step": 6167
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.8460173775318366e-05,
+      "loss": 0.7266,
+      "step": 6168
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.845968316646035e-05,
+      "loss": 0.1436,
+      "step": 6169
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.845919248368762e-05,
+      "loss": 0.0044,
+      "step": 6170
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.8458701727002868e-05,
+      "loss": 0.1235,
+      "step": 6171
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.8458210896408787e-05,
+      "loss": 0.126,
+      "step": 6172
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.845771999190808e-05,
+      "loss": 0.3242,
+      "step": 6173
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.8457229013503435e-05,
+      "loss": 0.4473,
+      "step": 6174
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.845673796119755e-05,
+      "loss": 0.0038,
+      "step": 6175
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.8456246834993126e-05,
+      "loss": 0.3066,
+      "step": 6176
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.8455755634892854e-05,
+      "loss": 0.0031,
+      "step": 6177
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.8455264360899434e-05,
+      "loss": 0.3203,
+      "step": 6178
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.8454773013015566e-05,
+      "loss": 0.1709,
+      "step": 6179
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.8454281591243947e-05,
+      "loss": 0.127,
+      "step": 6180
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.8453790095587273e-05,
+      "loss": 0.0,
+      "step": 6181
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.8453298526048245e-05,
+      "loss": 0.0005,
+      "step": 6182
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.8452806882629564e-05,
+      "loss": 0.8711,
+      "step": 6183
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.8452315165333926e-05,
+      "loss": 0.0014,
+      "step": 6184
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.8451823374164035e-05,
+      "loss": 0.0008,
+      "step": 6185
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.845133150912259e-05,
+      "loss": 0.0002,
+      "step": 6186
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.845083957021229e-05,
+      "loss": 0.0002,
+      "step": 6187
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.845034755743584e-05,
+      "loss": 0.6484,
+      "step": 6188
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.8449855470795946e-05,
+      "loss": 0.0,
+      "step": 6189
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.8449363310295302e-05,
+      "loss": 0.3262,
+      "step": 6190
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.8448871075936614e-05,
+      "loss": 0.0001,
+      "step": 6191
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.8448378767722586e-05,
+      "loss": 0.2617,
+      "step": 6192
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.8447886385655918e-05,
+      "loss": 0.123,
+      "step": 6193
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.844739392973932e-05,
+      "loss": 0.0006,
+      "step": 6194
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.8446901399975493e-05,
+      "loss": 0.0206,
+      "step": 6195
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.8446408796367144e-05,
+      "loss": 1.3125,
+      "step": 6196
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.844591611891697e-05,
+      "loss": 0.1104,
+      "step": 6197
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.8445423367627685e-05,
+      "loss": 0.2471,
+      "step": 6198
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.8444930542501992e-05,
+      "loss": 0.0028,
+      "step": 6199
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.84444376435426e-05,
+      "loss": 0.1641,
+      "step": 6200
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.844394467075221e-05,
+      "loss": 0.6758,
+      "step": 6201
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.8443451624133536e-05,
+      "loss": 0.9805,
+      "step": 6202
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.8442958503689277e-05,
+      "loss": 0.9922,
+      "step": 6203
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.844246530942215e-05,
+      "loss": 0.2217,
+      "step": 6204
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.8441972041334862e-05,
+      "loss": 0.3027,
+      "step": 6205
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.8441478699430123e-05,
+      "loss": 1.5469,
+      "step": 6206
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.8440985283710633e-05,
+      "loss": 0.7773,
+      "step": 6207
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.8440491794179106e-05,
+      "loss": 0.1982,
+      "step": 6208
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.8439998230838257e-05,
+      "loss": 1.0469,
+      "step": 6209
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.8439504593690792e-05,
+      "loss": 0.334,
+      "step": 6210
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.8439010882739422e-05,
+      "loss": 0.0898,
+      "step": 6211
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.8438517097986864e-05,
+      "loss": 2.1406,
+      "step": 6212
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.8438023239435815e-05,
+      "loss": 0.4551,
+      "step": 6213
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.8437529307089006e-05,
+      "loss": 1.3359,
+      "step": 6214
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.8437035300949137e-05,
+      "loss": 0.7578,
+      "step": 6215
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.8436541221018923e-05,
+      "loss": 0.1211,
+      "step": 6216
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.8436047067301076e-05,
+      "loss": 0.0508,
+      "step": 6217
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.8435552839798318e-05,
+      "loss": 1.7266,
+      "step": 6218
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.843505853851335e-05,
+      "loss": 0.024,
+      "step": 6219
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.8434564163448897e-05,
+      "loss": 0.3555,
+      "step": 6220
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.8434069714607672e-05,
+      "loss": 0.165,
+      "step": 6221
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.843357519199239e-05,
+      "loss": 0.012,
+      "step": 6222
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.843308059560576e-05,
+      "loss": 0.2617,
+      "step": 6223
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.843258592545051e-05,
+      "loss": 0.0001,
+      "step": 6224
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.8432091181529346e-05,
+      "loss": 0.0593,
+      "step": 6225
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.8431596363844994e-05,
+      "loss": 0.375,
+      "step": 6226
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.843110147240016e-05,
+      "loss": 0.9297,
+      "step": 6227
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.843060650719757e-05,
+      "loss": 0.0144,
+      "step": 6228
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.843011146823994e-05,
+      "loss": 0.0004,
+      "step": 6229
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.842961635552999e-05,
+      "loss": 0.0378,
+      "step": 6230
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.842912116907044e-05,
+      "loss": 0.0006,
+      "step": 6231
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.8428625908864004e-05,
+      "loss": 0.8516,
+      "step": 6232
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.8428130574913403e-05,
+      "loss": 0.2695,
+      "step": 6233
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.842763516722136e-05,
+      "loss": 0.0027,
+      "step": 6234
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.8427139685790598e-05,
+      "loss": 0.1914,
+      "step": 6235
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.8426644130623834e-05,
+      "loss": 0.2832,
+      "step": 6236
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.8426148501723785e-05,
+      "loss": 0.0024,
+      "step": 6237
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.8425652799093183e-05,
+      "loss": 0.0007,
+      "step": 6238
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.8425157022734743e-05,
+      "loss": 0.0376,
+      "step": 6239
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.8424661172651185e-05,
+      "loss": 0.1436,
+      "step": 6240
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.842416524884524e-05,
+      "loss": 0.6211,
+      "step": 6241
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.842366925131963e-05,
+      "loss": 0.0806,
+      "step": 6242
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.8423173180077072e-05,
+      "loss": 0.0579,
+      "step": 6243
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.8422677035120302e-05,
+      "loss": 0.0845,
+      "step": 6244
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.842218081645203e-05,
+      "loss": 0.0879,
+      "step": 6245
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.842168452407499e-05,
+      "loss": 0.0889,
+      "step": 6246
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.8421188157991908e-05,
+      "loss": 0.1494,
+      "step": 6247
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.842069171820551e-05,
+      "loss": 0.2041,
+      "step": 6248
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.8420195204718515e-05,
+      "loss": 0.0918,
+      "step": 6249
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.841969861753366e-05,
+      "loss": 0.1289,
+      "step": 6250
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.8419201956653657e-05,
+      "loss": 0.9102,
+      "step": 6251
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.841870522208125e-05,
+      "loss": 0.7422,
+      "step": 6252
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.8418208413819156e-05,
+      "loss": 1.2188,
+      "step": 6253
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.841771153187011e-05,
+      "loss": 0.5859,
+      "step": 6254
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.841721457623684e-05,
+      "loss": 0.6172,
+      "step": 6255
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.841671754692207e-05,
+      "loss": 0.9648,
+      "step": 6256
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.8416220443928533e-05,
+      "loss": 0.0222,
+      "step": 6257
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.841572326725896e-05,
+      "loss": 0.5781,
+      "step": 6258
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.8415226016916078e-05,
+      "loss": 0.6406,
+      "step": 6259
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.841472869290262e-05,
+      "loss": 1.1016,
+      "step": 6260
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.8414231295221314e-05,
+      "loss": 0.0864,
+      "step": 6261
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.8413733823874896e-05,
+      "loss": 0.5156,
+      "step": 6262
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.8413236278866095e-05,
+      "loss": 0.793,
+      "step": 6263
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.8412738660197642e-05,
+      "loss": 0.0173,
+      "step": 6264
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.841224096787227e-05,
+      "loss": 0.5234,
+      "step": 6265
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.841174320189272e-05,
+      "loss": 0.3086,
+      "step": 6266
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.8411245362261717e-05,
+      "loss": 0.0435,
+      "step": 6267
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.8410747448981998e-05,
+      "loss": 0.0566,
+      "step": 6268
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.841024946205629e-05,
+      "loss": 0.0005,
+      "step": 6269
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.8409751401487342e-05,
+      "loss": 0.0019,
+      "step": 6270
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.840925326727788e-05,
+      "loss": 0.8711,
+      "step": 6271
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.8408755059430635e-05,
+      "loss": 0.5039,
+      "step": 6272
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.8408256777948354e-05,
+      "loss": 0.3164,
+      "step": 6273
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.840775842283377e-05,
+      "loss": 0.0019,
+      "step": 6274
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8407259994089613e-05,
+      "loss": 0.625,
+      "step": 6275
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.840676149171863e-05,
+      "loss": 0.0141,
+      "step": 6276
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8406262915723547e-05,
+      "loss": 0.0259,
+      "step": 6277
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8405764266107116e-05,
+      "loss": 0.3379,
+      "step": 6278
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8405265542872064e-05,
+      "loss": 0.3086,
+      "step": 6279
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8404766746021134e-05,
+      "loss": 0.0003,
+      "step": 6280
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8404267875557064e-05,
+      "loss": 0.1143,
+      "step": 6281
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8403768931482596e-05,
+      "loss": 0.0004,
+      "step": 6282
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.840326991380047e-05,
+      "loss": 0.0364,
+      "step": 6283
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8402770822513424e-05,
+      "loss": 0.0003,
+      "step": 6284
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8402271657624198e-05,
+      "loss": 0.0013,
+      "step": 6285
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8401772419135538e-05,
+      "loss": 0.0001,
+      "step": 6286
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.840127310705018e-05,
+      "loss": 1.2031,
+      "step": 6287
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8400773721370867e-05,
+      "loss": 0.0081,
+      "step": 6288
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8400274262100344e-05,
+      "loss": 0.001,
+      "step": 6289
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8399774729241355e-05,
+      "loss": 0.0005,
+      "step": 6290
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8399275122796636e-05,
+      "loss": 0.0001,
+      "step": 6291
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.839877544276894e-05,
+      "loss": 0.0008,
+      "step": 6292
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8398275689161006e-05,
+      "loss": 0.0008,
+      "step": 6293
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8397775861975577e-05,
+      "loss": 0.0796,
+      "step": 6294
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8397275961215403e-05,
+      "loss": 0.1533,
+      "step": 6295
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.839677598688322e-05,
+      "loss": 0.064,
+      "step": 6296
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.839627593898179e-05,
+      "loss": 0.0001,
+      "step": 6297
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.839577581751384e-05,
+      "loss": 0.1328,
+      "step": 6298
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.839527562248213e-05,
+      "loss": 0.166,
+      "step": 6299
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8394775353889396e-05,
+      "loss": 0.0491,
+      "step": 6300
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8394275011738394e-05,
+      "loss": 0.7188,
+      "step": 6301
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8393774596031872e-05,
+      "loss": 0.8945,
+      "step": 6302
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8393274106772572e-05,
+      "loss": 0.1748,
+      "step": 6303
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8392773543963247e-05,
+      "loss": 0.2871,
+      "step": 6304
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8392272907606644e-05,
+      "loss": 0.5078,
+      "step": 6305
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.839177219770551e-05,
+      "loss": 0.373,
+      "step": 6306
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8391271414262602e-05,
+      "loss": 0.2871,
+      "step": 6307
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8390770557280663e-05,
+      "loss": 0.543,
+      "step": 6308
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.839026962676245e-05,
+      "loss": 0.0205,
+      "step": 6309
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8389768622710704e-05,
+      "loss": 0.4043,
+      "step": 6310
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8389267545128185e-05,
+      "loss": 1.8203,
+      "step": 6311
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.838876639401764e-05,
+      "loss": 0.6406,
+      "step": 6312
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8388265169381824e-05,
+      "loss": 0.9453,
+      "step": 6313
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8387763871223488e-05,
+      "loss": 1.875,
+      "step": 6314
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8387262499545386e-05,
+      "loss": 0.4609,
+      "step": 6315
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.838676105435027e-05,
+      "loss": 0.0559,
+      "step": 6316
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8386259535640892e-05,
+      "loss": 0.2812,
+      "step": 6317
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8385757943420012e-05,
+      "loss": 0.3086,
+      "step": 6318
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.838525627769038e-05,
+      "loss": 0.8242,
+      "step": 6319
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.838475453845475e-05,
+      "loss": 0.0001,
+      "step": 6320
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.838425272571588e-05,
+      "loss": 0.3086,
+      "step": 6321
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.838375083947653e-05,
+      "loss": 0.0066,
+      "step": 6322
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8383248879739445e-05,
+      "loss": 0.6797,
+      "step": 6323
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.838274684650739e-05,
+      "loss": 0.3281,
+      "step": 6324
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8382244739783117e-05,
+      "loss": 0.002,
+      "step": 6325
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8381742559569394e-05,
+      "loss": 0.0347,
+      "step": 6326
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8381240305868962e-05,
+      "loss": 0.3652,
+      "step": 6327
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.838073797868459e-05,
+      "loss": 0.0001,
+      "step": 6328
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.838023557801904e-05,
+      "loss": 0.0002,
+      "step": 6329
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8379733103875058e-05,
+      "loss": 0.4453,
+      "step": 6330
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8379230556255413e-05,
+      "loss": 0.0698,
+      "step": 6331
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8378727935162863e-05,
+      "loss": 0.5039,
+      "step": 6332
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.837822524060017e-05,
+      "loss": 0.0004,
+      "step": 6333
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8377722472570087e-05,
+      "loss": 0.0064,
+      "step": 6334
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8377219631075387e-05,
+      "loss": 0.0417,
+      "step": 6335
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.837671671611882e-05,
+      "loss": 0.0038,
+      "step": 6336
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8376213727703156e-05,
+      "loss": 0.3555,
+      "step": 6337
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.837571066583115e-05,
+      "loss": 0.0015,
+      "step": 6338
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8375207530505566e-05,
+      "loss": 0.0016,
+      "step": 6339
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8374704321729177e-05,
+      "loss": 0.0039,
+      "step": 6340
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8374201039504732e-05,
+      "loss": 0.0864,
+      "step": 6341
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8373697683835005e-05,
+      "loss": 0.0003,
+      "step": 6342
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.8373194254722755e-05,
+      "loss": 0.0006,
+      "step": 6343
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.837269075217075e-05,
+      "loss": 0.6406,
+      "step": 6344
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.8372187176181756e-05,
+      "loss": 0.0066,
+      "step": 6345
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.8371683526758526e-05,
+      "loss": 0.1475,
+      "step": 6346
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.8371179803903845e-05,
+      "loss": 0.0004,
+      "step": 6347
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.8370676007620464e-05,
+      "loss": 0.1416,
+      "step": 6348
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.837017213791116e-05,
+      "loss": 0.2168,
+      "step": 6349
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.8369668194778695e-05,
+      "loss": 0.2197,
+      "step": 6350
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.8369164178225836e-05,
+      "loss": 1.5781,
+      "step": 6351
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.836866008825535e-05,
+      "loss": 0.4004,
+      "step": 6352
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.8368155924870007e-05,
+      "loss": 0.3535,
+      "step": 6353
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.8367651688072573e-05,
+      "loss": 0.1328,
+      "step": 6354
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.8367147377865826e-05,
+      "loss": 0.4941,
+      "step": 6355
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.8366642994252524e-05,
+      "loss": 1.7812,
+      "step": 6356
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.8366138537235444e-05,
+      "loss": 0.7031,
+      "step": 6357
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.8365634006817353e-05,
+      "loss": 1.0391,
+      "step": 6358
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.8365129403001027e-05,
+      "loss": 1.5859,
+      "step": 6359
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.836462472578923e-05,
+      "loss": 0.7461,
+      "step": 6360
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.8364119975184736e-05,
+      "loss": 0.0432,
+      "step": 6361
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.836361515119032e-05,
+      "loss": 0.3613,
+      "step": 6362
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.836311025380875e-05,
+      "loss": 0.1709,
+      "step": 6363
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.8362605283042797e-05,
+      "loss": 0.3184,
+      "step": 6364
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.836210023889524e-05,
+      "loss": 0.3613,
+      "step": 6365
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.8361595121368853e-05,
+      "loss": 0.0261,
+      "step": 6366
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.8361089930466403e-05,
+      "loss": 0.4316,
+      "step": 6367
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.8360584666190668e-05,
+      "loss": 0.0786,
+      "step": 6368
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.836007932854442e-05,
+      "loss": 0.3281,
+      "step": 6369
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.835957391753044e-05,
+      "loss": 1.3047,
+      "step": 6370
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.83590684331515e-05,
+      "loss": 0.418,
+      "step": 6371
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.8358562875410376e-05,
+      "loss": 0.0618,
+      "step": 6372
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.8358057244309846e-05,
+      "loss": 0.0459,
+      "step": 6373
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.8357551539852685e-05,
+      "loss": 0.027,
+      "step": 6374
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.835704576204167e-05,
+      "loss": 0.0002,
+      "step": 6375
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.8356539910879576e-05,
+      "loss": 0.5859,
+      "step": 6376
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.8356033986369187e-05,
+      "loss": 0.291,
+      "step": 6377
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.8355527988513274e-05,
+      "loss": 0.5,
+      "step": 6378
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.8355021917314625e-05,
+      "loss": 0.3906,
+      "step": 6379
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.835451577277601e-05,
+      "loss": 0.0427,
+      "step": 6380
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.8354009554900213e-05,
+      "loss": 0.0002,
+      "step": 6381
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.8353503263690017e-05,
+      "loss": 0.026,
+      "step": 6382
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.8352996899148194e-05,
+      "loss": 0.0265,
+      "step": 6383
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.835249046127753e-05,
+      "loss": 1.7422,
+      "step": 6384
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.835198395008081e-05,
+      "loss": 0.0002,
+      "step": 6385
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.8351477365560804e-05,
+      "loss": 0.1348,
+      "step": 6386
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.8350970707720308e-05,
+      "loss": 0.1748,
+      "step": 6387
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.8350463976562094e-05,
+      "loss": 0.3301,
+      "step": 6388
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.8349957172088947e-05,
+      "loss": 0.0002,
+      "step": 6389
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.834945029430365e-05,
+      "loss": 0.0001,
+      "step": 6390
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.8348943343208992e-05,
+      "loss": 0.0052,
+      "step": 6391
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.8348436318807748e-05,
+      "loss": 0.0007,
+      "step": 6392
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.834792922110271e-05,
+      "loss": 0.0001,
+      "step": 6393
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.8347422050096662e-05,
+      "loss": 0.6016,
+      "step": 6394
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.834691480579238e-05,
+      "loss": 0.0002,
+      "step": 6395
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.8346407488192666e-05,
+      "loss": 0.1221,
+      "step": 6396
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.834590009730029e-05,
+      "loss": 0.0003,
+      "step": 6397
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.834539263311805e-05,
+      "loss": 0.1445,
+      "step": 6398
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.834488509564872e-05,
+      "loss": 0.1934,
+      "step": 6399
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.83443774848951e-05,
+      "loss": 0.1963,
+      "step": 6400
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.8343869800859972e-05,
+      "loss": 0.2988,
+      "step": 6401
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.8343362043546123e-05,
+      "loss": 0.6406,
+      "step": 6402
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.8342854212956345e-05,
+      "loss": 0.4941,
+      "step": 6403
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.8342346309093422e-05,
+      "loss": 0.5742,
+      "step": 6404
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.8341838331960148e-05,
+      "loss": 0.6172,
+      "step": 6405
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.8341330281559313e-05,
+      "loss": 0.0483,
+      "step": 6406
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.8340822157893696e-05,
+      "loss": 0.582,
+      "step": 6407
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.83403139609661e-05,
+      "loss": 0.4746,
+      "step": 6408
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.8339805690779312e-05,
+      "loss": 0.5195,
+      "step": 6409
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.8339297347336123e-05,
+      "loss": 0.0002,
+      "step": 6410
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.833878893063933e-05,
+      "loss": 0.4785,
+      "step": 6411
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.8338280440691708e-05,
+      "loss": 0.0481,
+      "step": 6412
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.833777187749607e-05,
+      "loss": 0.0054,
+      "step": 6413
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.8337263241055195e-05,
+      "loss": 1.5547,
+      "step": 6414
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.833675453137188e-05,
+      "loss": 0.2871,
+      "step": 6415
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.8336245748448922e-05,
+      "loss": 0.8086,
+      "step": 6416
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.8335736892289114e-05,
+      "loss": 0.1211,
+      "step": 6417
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.8335227962895243e-05,
+      "loss": 0.0022,
+      "step": 6418
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.8334718960270116e-05,
+      "loss": 0.0452,
+      "step": 6419
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.8334209884416518e-05,
+      "loss": 0.1709,
+      "step": 6420
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.8333700735337246e-05,
+      "loss": 1.8828,
+      "step": 6421
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.8333191513035104e-05,
+      "loss": 0.1182,
+      "step": 6422
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.833268221751288e-05,
+      "loss": 0.0017,
+      "step": 6423
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.8332172848773376e-05,
+      "loss": 0.6406,
+      "step": 6424
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.8331663406819385e-05,
+      "loss": 0.0002,
+      "step": 6425
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.8331153891653707e-05,
+      "loss": 0.004,
+      "step": 6426
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.833064430327914e-05,
+      "loss": 1.3594,
+      "step": 6427
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.833013464169848e-05,
+      "loss": 0.0388,
+      "step": 6428
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.8329624906914534e-05,
+      "loss": 0.1328,
+      "step": 6429
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.8329115098930088e-05,
+      "loss": 0.7109,
+      "step": 6430
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.8328605217747948e-05,
+      "loss": 0.0248,
+      "step": 6431
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.8328095263370923e-05,
+      "loss": 0.0003,
+      "step": 6432
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.8327585235801798e-05,
+      "loss": 0.0007,
+      "step": 6433
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.8327075135043385e-05,
+      "loss": 0.0001,
+      "step": 6434
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.8326564961098482e-05,
+      "loss": 0.3574,
+      "step": 6435
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.8326054713969886e-05,
+      "loss": 0.0439,
+      "step": 6436
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.832554439366041e-05,
+      "loss": 0.3789,
+      "step": 6437
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.8325034000172843e-05,
+      "loss": 0.0,
+      "step": 6438
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.8324523533509996e-05,
+      "loss": 0.0001,
+      "step": 6439
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.832401299367467e-05,
+      "loss": 0.0001,
+      "step": 6440
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.832350238066967e-05,
+      "loss": 0.1611,
+      "step": 6441
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.8322991694497802e-05,
+      "loss": 1.4844,
+      "step": 6442
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.8322480935161867e-05,
+      "loss": 0.0005,
+      "step": 6443
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.832197010266467e-05,
+      "loss": 0.0,
+      "step": 6444
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.8321459197009016e-05,
+      "loss": 0.0977,
+      "step": 6445
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.8320948218197713e-05,
+      "loss": 0.0001,
+      "step": 6446
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.8320437166233565e-05,
+      "loss": 0.1011,
+      "step": 6447
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.8319926041119378e-05,
+      "loss": 0.1289,
+      "step": 6448
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.8319414842857966e-05,
+      "loss": 0.1328,
+      "step": 6449
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.8318903571452122e-05,
+      "loss": 0.2539,
+      "step": 6450
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.8318392226904668e-05,
+      "loss": 1.1953,
+      "step": 6451
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.8317880809218402e-05,
+      "loss": 0.3555,
+      "step": 6452
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.8317369318396137e-05,
+      "loss": 1.9844,
+      "step": 6453
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.8316857754440685e-05,
+      "loss": 0.8828,
+      "step": 6454
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.831634611735485e-05,
+      "loss": 0.3828,
+      "step": 6455
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.8315834407141443e-05,
+      "loss": 2.0469,
+      "step": 6456
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.831532262380328e-05,
+      "loss": 0.4121,
+      "step": 6457
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.8314810767343155e-05,
+      "loss": 0.0825,
+      "step": 6458
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.8314298837763902e-05,
+      "loss": 0.2188,
+      "step": 6459
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.8313786835068314e-05,
+      "loss": 3.125,
+      "step": 6460
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.8313274759259207e-05,
+      "loss": 0.6602,
+      "step": 6461
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.83127626103394e-05,
+      "loss": 0.0723,
+      "step": 6462
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.8312250388311697e-05,
+      "loss": 0.7422,
+      "step": 6463
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.8311738093178916e-05,
+      "loss": 2.4531,
+      "step": 6464
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.8311225724943867e-05,
+      "loss": 0.2734,
+      "step": 6465
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.8310713283609367e-05,
+      "loss": 0.1602,
+      "step": 6466
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.8310200769178228e-05,
+      "loss": 1.1797,
+      "step": 6467
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.8309688181653264e-05,
+      "loss": 0.0001,
+      "step": 6468
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.8309175521037296e-05,
+      "loss": 1.3047,
+      "step": 6469
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.830866278733313e-05,
+      "loss": 0.0007,
+      "step": 6470
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.8308149980543586e-05,
+      "loss": 1.8047,
+      "step": 6471
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.830763710067148e-05,
+      "loss": 0.0026,
+      "step": 6472
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.8307124147719633e-05,
+      "loss": 0.1021,
+      "step": 6473
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.8306611121690853e-05,
+      "loss": 0.4023,
+      "step": 6474
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.830609802258796e-05,
+      "loss": 1.5312,
+      "step": 6475
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.8305584850413776e-05,
+      "loss": 0.2021,
+      "step": 6476
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.830507160517112e-05,
+      "loss": 0.0034,
+      "step": 6477
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.83045582868628e-05,
+      "loss": 2.5625,
+      "step": 6478
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.8304044895491647e-05,
+      "loss": 0.0,
+      "step": 6479
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.830353143106048e-05,
+      "loss": 1.4531,
+      "step": 6480
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.8303017893572105e-05,
+      "loss": 0.1406,
+      "step": 6481
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.8302504283029354e-05,
+      "loss": 0.0,
+      "step": 6482
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.830199059943505e-05,
+      "loss": 0.0001,
+      "step": 6483
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.8301476842792003e-05,
+      "loss": 0.0,
+      "step": 6484
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.8300963013103042e-05,
+      "loss": 0.0291,
+      "step": 6485
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.8300449110370987e-05,
+      "loss": 0.0,
+      "step": 6486
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.8299935134598657e-05,
+      "loss": 0.0,
+      "step": 6487
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.8299421085788886e-05,
+      "loss": 0.0002,
+      "step": 6488
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.829890696394448e-05,
+      "loss": 0.25,
+      "step": 6489
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.8298392769068272e-05,
+      "loss": 0.1641,
+      "step": 6490
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.8297878501163088e-05,
+      "loss": 0.2637,
+      "step": 6491
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.8297364160231745e-05,
+      "loss": 0.0001,
+      "step": 6492
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.829684974627707e-05,
+      "loss": 0.0008,
+      "step": 6493
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.8296335259301896e-05,
+      "loss": 0.0679,
+      "step": 6494
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.8295820699309035e-05,
+      "loss": 0.0002,
+      "step": 6495
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.829530606630132e-05,
+      "loss": 0.001,
+      "step": 6496
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.829479136028158e-05,
+      "loss": 0.1572,
+      "step": 6497
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.8294276581252635e-05,
+      "loss": 0.0002,
+      "step": 6498
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.8293761729217315e-05,
+      "loss": 0.1064,
+      "step": 6499
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.8293246804178445e-05,
+      "loss": 0.1641,
+      "step": 6500
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.829273180613886e-05,
+      "loss": 0.6367,
+      "step": 6501
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.8292216735101382e-05,
+      "loss": 0.6055,
+      "step": 6502
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.829170159106884e-05,
+      "loss": 0.6719,
+      "step": 6503
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.829118637404406e-05,
+      "loss": 0.1157,
+      "step": 6504
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.8290671084029877e-05,
+      "loss": 1.3125,
+      "step": 6505
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.8290155721029115e-05,
+      "loss": 0.8086,
+      "step": 6506
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.8289640285044612e-05,
+      "loss": 1.1094,
+      "step": 6507
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.8289124776079195e-05,
+      "loss": 0.6484,
+      "step": 6508
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.8288609194135693e-05,
+      "loss": 0.6602,
+      "step": 6509
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.8288093539216938e-05,
+      "loss": 0.2031,
+      "step": 6510
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.8287577811325762e-05,
+      "loss": 0.123,
+      "step": 6511
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.8287062010465e-05,
+      "loss": 0.793,
+      "step": 6512
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.828654613663748e-05,
+      "loss": 0.208,
+      "step": 6513
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.8286030189846035e-05,
+      "loss": 0.0315,
+      "step": 6514
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.8285514170093507e-05,
+      "loss": 1.2734,
+      "step": 6515
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.8284998077382717e-05,
+      "loss": 1.2344,
+      "step": 6516
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.8284481911716504e-05,
+      "loss": 0.0796,
+      "step": 6517
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.828396567309771e-05,
+      "loss": 0.4004,
+      "step": 6518
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.828344936152916e-05,
+      "loss": 0.0005,
+      "step": 6519
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.8282932977013694e-05,
+      "loss": 0.7812,
+      "step": 6520
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.8282416519554146e-05,
+      "loss": 0.0879,
+      "step": 6521
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.828189998915335e-05,
+      "loss": 0.0063,
+      "step": 6522
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.8281383385814153e-05,
+      "loss": 0.0003,
+      "step": 6523
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.828086670953938e-05,
+      "loss": 0.3145,
+      "step": 6524
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.8280349960331876e-05,
+      "loss": 0.3535,
+      "step": 6525
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.8279833138194476e-05,
+      "loss": 0.0002,
+      "step": 6526
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.8279316243130012e-05,
+      "loss": 0.0,
+      "step": 6527
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.8278799275141334e-05,
+      "loss": 0.0014,
+      "step": 6528
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.8278282234231272e-05,
+      "loss": 0.8633,
+      "step": 6529
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.8277765120402674e-05,
+      "loss": 0.0004,
+      "step": 6530
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.827724793365837e-05,
+      "loss": 0.9844,
+      "step": 6531
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.8276730674001204e-05,
+      "loss": 0.3574,
+      "step": 6532
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.8276213341434015e-05,
+      "loss": 0.1406,
+      "step": 6533
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.8275695935959654e-05,
+      "loss": 0.0674,
+      "step": 6534
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.827517845758095e-05,
+      "loss": 0.0286,
+      "step": 6535
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.827466090630075e-05,
+      "loss": 0.3027,
+      "step": 6536
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.82741432821219e-05,
+      "loss": 0.0957,
+      "step": 6537
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.827362558504723e-05,
+      "loss": 0.0001,
+      "step": 6538
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.8273107815079596e-05,
+      "loss": 0.543,
+      "step": 6539
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.8272589972221833e-05,
+      "loss": 0.1162,
+      "step": 6540
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.827207205647679e-05,
+      "loss": 0.1299,
+      "step": 6541
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.8271554067847308e-05,
+      "loss": 0.0006,
+      "step": 6542
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.8271036006336236e-05,
+      "loss": 0.1963,
+      "step": 6543
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.8270517871946415e-05,
+      "loss": 0.0977,
+      "step": 6544
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.826999966468069e-05,
+      "loss": 0.1055,
+      "step": 6545
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8269481384541905e-05,
+      "loss": 0.1035,
+      "step": 6546
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8268963031532918e-05,
+      "loss": 0.1797,
+      "step": 6547
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.826844460565656e-05,
+      "loss": 0.0962,
+      "step": 6548
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8267926106915685e-05,
+      "loss": 0.1494,
+      "step": 6549
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8267407535313143e-05,
+      "loss": 0.1504,
+      "step": 6550
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8266888890851777e-05,
+      "loss": 0.7812,
+      "step": 6551
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8266370173534437e-05,
+      "loss": 0.3047,
+      "step": 6552
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.826585138336397e-05,
+      "loss": 0.5898,
+      "step": 6553
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.826533252034323e-05,
+      "loss": 0.208,
+      "step": 6554
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.826481358447506e-05,
+      "loss": 0.5078,
+      "step": 6555
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8264294575762314e-05,
+      "loss": 1.375,
+      "step": 6556
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.826377549420784e-05,
+      "loss": 0.1123,
+      "step": 6557
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8263256339814494e-05,
+      "loss": 0.7344,
+      "step": 6558
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8262737112585116e-05,
+      "loss": 0.1406,
+      "step": 6559
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8262217812522567e-05,
+      "loss": 2.1094,
+      "step": 6560
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.82616984396297e-05,
+      "loss": 0.7188,
+      "step": 6561
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8261178993909354e-05,
+      "loss": 0.625,
+      "step": 6562
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8260659475364397e-05,
+      "loss": 1.2812,
+      "step": 6563
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8260139883997672e-05,
+      "loss": 0.6953,
+      "step": 6564
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8259620219812034e-05,
+      "loss": 0.0942,
+      "step": 6565
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8259100482810337e-05,
+      "loss": 0.9141,
+      "step": 6566
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8258580672995445e-05,
+      "loss": 0.1138,
+      "step": 6567
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8258060790370197e-05,
+      "loss": 0.0072,
+      "step": 6568
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.825754083493746e-05,
+      "loss": 0.0889,
+      "step": 6569
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8257020806700083e-05,
+      "loss": 0.375,
+      "step": 6570
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.825650070566092e-05,
+      "loss": 0.3477,
+      "step": 6571
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8255980531822833e-05,
+      "loss": 0.498,
+      "step": 6572
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8255460285188676e-05,
+      "loss": 0.5469,
+      "step": 6573
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8254939965761308e-05,
+      "loss": 0.252,
+      "step": 6574
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8254419573543583e-05,
+      "loss": 1.2031,
+      "step": 6575
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8253899108538356e-05,
+      "loss": 0.002,
+      "step": 6576
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.825337857074849e-05,
+      "loss": 0.0195,
+      "step": 6577
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8252857960176848e-05,
+      "loss": 1.8281,
+      "step": 6578
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.825233727682628e-05,
+      "loss": 0.5703,
+      "step": 6579
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.825181652069965e-05,
+      "loss": 0.0334,
+      "step": 6580
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.825129569179982e-05,
+      "loss": 0.0082,
+      "step": 6581
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.825077479012964e-05,
+      "loss": 0.0569,
+      "step": 6582
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8250253815691984e-05,
+      "loss": 0.8438,
+      "step": 6583
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8249732768489702e-05,
+      "loss": 0.0488,
+      "step": 6584
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8249211648525667e-05,
+      "loss": 0.0618,
+      "step": 6585
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8248690455802726e-05,
+      "loss": 0.0003,
+      "step": 6586
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8248169190323753e-05,
+      "loss": 0.0001,
+      "step": 6587
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8247647852091607e-05,
+      "loss": 0.0003,
+      "step": 6588
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8247126441109145e-05,
+      "loss": 0.0574,
+      "step": 6589
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8246604957379242e-05,
+      "loss": 0.543,
+      "step": 6590
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8246083400904754e-05,
+      "loss": 0.0,
+      "step": 6591
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8245561771688546e-05,
+      "loss": 0.0145,
+      "step": 6592
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8245040069733482e-05,
+      "loss": 0.1807,
+      "step": 6593
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8244518295042432e-05,
+      "loss": 0.0069,
+      "step": 6594
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8243996447618252e-05,
+      "loss": 0.0042,
+      "step": 6595
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.824347452746382e-05,
+      "loss": 0.0011,
+      "step": 6596
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.824295253458199e-05,
+      "loss": 0.0001,
+      "step": 6597
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8242430468975635e-05,
+      "loss": 0.0825,
+      "step": 6598
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8241908330647625e-05,
+      "loss": 0.1846,
+      "step": 6599
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.824138611960082e-05,
+      "loss": 0.1846,
+      "step": 6600
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.824086383583809e-05,
+      "loss": 0.1426,
+      "step": 6601
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8240341479362308e-05,
+      "loss": 0.2559,
+      "step": 6602
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8239819050176337e-05,
+      "loss": 0.8281,
+      "step": 6603
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.823929654828305e-05,
+      "loss": 0.0786,
+      "step": 6604
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.823877397368531e-05,
+      "loss": 1.625,
+      "step": 6605
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8238251326385992e-05,
+      "loss": 1.1328,
+      "step": 6606
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8237728606387965e-05,
+      "loss": 0.7266,
+      "step": 6607
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.82372058136941e-05,
+      "loss": 0.0469,
+      "step": 6608
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.823668294830727e-05,
+      "loss": 0.5781,
+      "step": 6609
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.823616001023034e-05,
+      "loss": 0.1084,
+      "step": 6610
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.823563699946619e-05,
+      "loss": 0.2002,
+      "step": 6611
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8235113916017685e-05,
+      "loss": 0.9492,
+      "step": 6612
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8234590759887702e-05,
+      "loss": 0.1069,
+      "step": 6613
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.823406753107911e-05,
+      "loss": 0.4238,
+      "step": 6614
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8233544229594786e-05,
+      "loss": 0.0072,
+      "step": 6615
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.82330208554376e-05,
+      "loss": 0.0435,
+      "step": 6616
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.823249740861043e-05,
+      "loss": 0.0037,
+      "step": 6617
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8231973889116152e-05,
+      "loss": 0.0845,
+      "step": 6618
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8231450296957638e-05,
+      "loss": 0.0007,
+      "step": 6619
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.823092663213776e-05,
+      "loss": 0.4902,
+      "step": 6620
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8230402894659396e-05,
+      "loss": 0.0,
+      "step": 6621
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.822987908452543e-05,
+      "loss": 0.0762,
+      "step": 6622
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8229355201738726e-05,
+      "loss": 0.123,
+      "step": 6623
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8228831246302166e-05,
+      "loss": 0.1475,
+      "step": 6624
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.822830721821863e-05,
+      "loss": 2.7031,
+      "step": 6625
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8227783117490993e-05,
+      "loss": 0.0159,
+      "step": 6626
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.822725894412214e-05,
+      "loss": 0.2695,
+      "step": 6627
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8226734698114935e-05,
+      "loss": 0.7188,
+      "step": 6628
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8226210379472266e-05,
+      "loss": 0.0002,
+      "step": 6629
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8225685988197018e-05,
+      "loss": 0.0835,
+      "step": 6630
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8225161524292058e-05,
+      "loss": 0.0002,
+      "step": 6631
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8224636987760274e-05,
+      "loss": 0.0278,
+      "step": 6632
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.822411237860455e-05,
+      "loss": 0.0003,
+      "step": 6633
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.822358769682776e-05,
+      "loss": 0.0297,
+      "step": 6634
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.822306294243278e-05,
+      "loss": 0.0001,
+      "step": 6635
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8222538115422507e-05,
+      "loss": 0.6172,
+      "step": 6636
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.822201321579981e-05,
+      "loss": 0.0114,
+      "step": 6637
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.822148824356758e-05,
+      "loss": 0.3184,
+      "step": 6638
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.82209631987287e-05,
+      "loss": 0.0967,
+      "step": 6639
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8220438081286044e-05,
+      "loss": 0.0001,
+      "step": 6640
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8219912891242504e-05,
+      "loss": 0.0347,
+      "step": 6641
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8219387628600962e-05,
+      "loss": 0.2197,
+      "step": 6642
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8218862293364303e-05,
+      "loss": 0.0002,
+      "step": 6643
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.821833688553541e-05,
+      "loss": 0.5,
+      "step": 6644
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.821781140511717e-05,
+      "loss": 0.1719,
+      "step": 6645
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.821728585211247e-05,
+      "loss": 0.0003,
+      "step": 6646
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.821676022652419e-05,
+      "loss": 0.1514,
+      "step": 6647
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.821623452835522e-05,
+      "loss": 0.0737,
+      "step": 6648
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8215708757608454e-05,
+      "loss": 0.0003,
+      "step": 6649
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8215182914286768e-05,
+      "loss": 0.1729,
+      "step": 6650
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.821465699839306e-05,
+      "loss": 1.3125,
+      "step": 6651
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8214131009930207e-05,
+      "loss": 1.0859,
+      "step": 6652
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8213604948901105e-05,
+      "loss": 0.2734,
+      "step": 6653
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8213078815308645e-05,
+      "loss": 0.2773,
+      "step": 6654
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8212552609155708e-05,
+      "loss": 0.8281,
+      "step": 6655
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.821202633044519e-05,
+      "loss": 1.4219,
+      "step": 6656
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8211499979179977e-05,
+      "loss": 0.4004,
+      "step": 6657
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8210973555362963e-05,
+      "loss": 1.0859,
+      "step": 6658
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.821044705899704e-05,
+      "loss": 0.012,
+      "step": 6659
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.820992049008509e-05,
+      "loss": 0.4043,
+      "step": 6660
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.820939384863002e-05,
+      "loss": 0.6133,
+      "step": 6661
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.820886713463471e-05,
+      "loss": 0.5391,
+      "step": 6662
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.820834034810206e-05,
+      "loss": 1.0625,
+      "step": 6663
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8207813489034953e-05,
+      "loss": 0.3789,
+      "step": 6664
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.820728655743629e-05,
+      "loss": 0.6562,
+      "step": 6665
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.820675955330897e-05,
+      "loss": 0.0688,
+      "step": 6666
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.820623247665587e-05,
+      "loss": 0.0221,
+      "step": 6667
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.82057053274799e-05,
+      "loss": 0.0051,
+      "step": 6668
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8205178105783955e-05,
+      "loss": 0.2148,
+      "step": 6669
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8204650811570915e-05,
+      "loss": 0.0002,
+      "step": 6670
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8204123444843692e-05,
+      "loss": 0.003,
+      "step": 6671
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.820359600560517e-05,
+      "loss": 1.0391,
+      "step": 6672
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8203068493858257e-05,
+      "loss": 0.0042,
+      "step": 6673
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.820254090960584e-05,
+      "loss": 0.0161,
+      "step": 6674
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8202013252850818e-05,
+      "loss": 0.0403,
+      "step": 6675
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8201485523596093e-05,
+      "loss": 0.0001,
+      "step": 6676
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8200957721844563e-05,
+      "loss": 0.0072,
+      "step": 6677
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8200429847599122e-05,
+      "loss": 0.0001,
+      "step": 6678
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8199901900862674e-05,
+      "loss": 0.0811,
+      "step": 6679
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.819937388163811e-05,
+      "loss": 0.0001,
+      "step": 6680
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.819884578992834e-05,
+      "loss": 0.0001,
+      "step": 6681
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.819831762573626e-05,
+      "loss": 0.0002,
+      "step": 6682
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8197789389064763e-05,
+      "loss": 0.1543,
+      "step": 6683
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.819726107991676e-05,
+      "loss": 0.0002,
+      "step": 6684
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.819673269829515e-05,
+      "loss": 0.6172,
+      "step": 6685
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8196204244202835e-05,
+      "loss": 0.0001,
+      "step": 6686
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.819567571764271e-05,
+      "loss": 0.0001,
+      "step": 6687
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8195147118617687e-05,
+      "loss": 0.0002,
+      "step": 6688
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.819461844713066e-05,
+      "loss": 0.2363,
+      "step": 6689
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8194089703184542e-05,
+      "loss": 0.0996,
+      "step": 6690
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.819356088678223e-05,
+      "loss": 0.0002,
+      "step": 6691
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.819303199792663e-05,
+      "loss": 0.0,
+      "step": 6692
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8192503036620646e-05,
+      "loss": 0.0039,
+      "step": 6693
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.819197400286718e-05,
+      "loss": 0.0002,
+      "step": 6694
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.819144489666914e-05,
+      "loss": 0.0001,
+      "step": 6695
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8190915718029432e-05,
+      "loss": 0.0281,
+      "step": 6696
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.819038646695096e-05,
+      "loss": 0.0017,
+      "step": 6697
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.818985714343664e-05,
+      "loss": 0.0649,
+      "step": 6698
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8189327747489363e-05,
+      "loss": 0.1104,
+      "step": 6699
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.818879827911205e-05,
+      "loss": 0.1245,
+      "step": 6700
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8188268738307593e-05,
+      "loss": 0.8984,
+      "step": 6701
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.818773912507892e-05,
+      "loss": 1.0391,
+      "step": 6702
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8187209439428924e-05,
+      "loss": 0.668,
+      "step": 6703
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.818667968136052e-05,
+      "loss": 0.75,
+      "step": 6704
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8186149850876615e-05,
+      "loss": 0.5,
+      "step": 6705
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.818561994798012e-05,
+      "loss": 0.7695,
+      "step": 6706
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8185089972673946e-05,
+      "loss": 0.4766,
+      "step": 6707
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8184559924961e-05,
+      "loss": 2.25,
+      "step": 6708
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8184029804844197e-05,
+      "loss": 0.0033,
+      "step": 6709
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8183499612326446e-05,
+      "loss": 0.0132,
+      "step": 6710
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8182969347410658e-05,
+      "loss": 0.5703,
+      "step": 6711
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8182439010099746e-05,
+      "loss": 2.8438,
+      "step": 6712
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.818190860039662e-05,
+      "loss": 0.0786,
+      "step": 6713
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8181378118304195e-05,
+      "loss": 0.0476,
+      "step": 6714
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8180847563825386e-05,
+      "loss": 0.0003,
+      "step": 6715
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8180316936963105e-05,
+      "loss": 0.0762,
+      "step": 6716
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8179786237720263e-05,
+      "loss": 0.0698,
+      "step": 6717
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8179255466099778e-05,
+      "loss": 1.6484,
+      "step": 6718
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8178724622104565e-05,
+      "loss": 0.7188,
+      "step": 6719
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8178193705737536e-05,
+      "loss": 0.2734,
+      "step": 6720
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8177662717001608e-05,
+      "loss": 0.0006,
+      "step": 6721
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8177131655899698e-05,
+      "loss": 0.7617,
+      "step": 6722
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.817660052243472e-05,
+      "loss": 0.0082,
+      "step": 6723
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8176069316609597e-05,
+      "loss": 0.0591,
+      "step": 6724
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.817553803842724e-05,
+      "loss": 0.0654,
+      "step": 6725
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8175006687890564e-05,
+      "loss": 0.0256,
+      "step": 6726
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8174475265002495e-05,
+      "loss": 0.5195,
+      "step": 6727
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8173943769765947e-05,
+      "loss": 0.0031,
+      "step": 6728
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8173412202183837e-05,
+      "loss": 0.0056,
+      "step": 6729
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.817288056225909e-05,
+      "loss": 0.2344,
+      "step": 6730
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.817234884999462e-05,
+      "loss": 0.0085,
+      "step": 6731
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8171817065393346e-05,
+      "loss": 0.0,
+      "step": 6732
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8171285208458195e-05,
+      "loss": 0.0074,
+      "step": 6733
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.817075327919208e-05,
+      "loss": 0.0001,
+      "step": 6734
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.817022127759793e-05,
+      "loss": 0.0,
+      "step": 6735
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.816968920367866e-05,
+      "loss": 0.0,
+      "step": 6736
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.81691570574372e-05,
+      "loss": 0.2158,
+      "step": 6737
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8168624838876456e-05,
+      "loss": 0.0,
+      "step": 6738
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8168092547999367e-05,
+      "loss": 0.0898,
+      "step": 6739
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8167560184808852e-05,
+      "loss": 0.0,
+      "step": 6740
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8167027749307836e-05,
+      "loss": 0.0,
+      "step": 6741
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8166495241499232e-05,
+      "loss": 0.0001,
+      "step": 6742
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8165962661385975e-05,
+      "loss": 0.0,
+      "step": 6743
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.816543000897099e-05,
+      "loss": 0.0,
+      "step": 6744
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8164897284257197e-05,
+      "loss": 0.0,
+      "step": 6745
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.816436448724752e-05,
+      "loss": 0.6367,
+      "step": 6746
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8163831617944892e-05,
+      "loss": 0.0,
+      "step": 6747
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8163298676352237e-05,
+      "loss": 0.1396,
+      "step": 6748
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.816276566247248e-05,
+      "loss": 0.1108,
+      "step": 6749
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.8162232576308546e-05,
+      "loss": 0.1885,
+      "step": 6750
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.8161699417863365e-05,
+      "loss": 0.75,
+      "step": 6751
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.8161166187139866e-05,
+      "loss": 1.7031,
+      "step": 6752
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.8160632884140976e-05,
+      "loss": 0.4863,
+      "step": 6753
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.8160099508869624e-05,
+      "loss": 0.4805,
+      "step": 6754
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.815956606132874e-05,
+      "loss": 0.4297,
+      "step": 6755
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.8159032541521248e-05,
+      "loss": 0.0903,
+      "step": 6756
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.8158498949450086e-05,
+      "loss": 1.0547,
+      "step": 6757
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.8157965285118183e-05,
+      "loss": 0.2441,
+      "step": 6758
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.815743154852846e-05,
+      "loss": 0.957,
+      "step": 6759
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.815689773968386e-05,
+      "loss": 0.0001,
+      "step": 6760
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.8156363858587302e-05,
+      "loss": 1.6172,
+      "step": 6761
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.8155829905241734e-05,
+      "loss": 0.5469,
+      "step": 6762
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.815529587965008e-05,
+      "loss": 0.4473,
+      "step": 6763
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.8154761781815266e-05,
+      "loss": 0.2676,
+      "step": 6764
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.8154227611740235e-05,
+      "loss": 0.0,
+      "step": 6765
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.8153693369427916e-05,
+      "loss": 0.2373,
+      "step": 6766
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.8153159054881245e-05,
+      "loss": 0.0077,
+      "step": 6767
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.8152624668103154e-05,
+      "loss": 0.1006,
+      "step": 6768
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.815209020909658e-05,
+      "loss": 0.106,
+      "step": 6769
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.8151555677864452e-05,
+      "loss": 0.0001,
+      "step": 6770
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.8151021074409713e-05,
+      "loss": 0.0806,
+      "step": 6771
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.8150486398735294e-05,
+      "loss": 0.7188,
+      "step": 6772
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.8149951650844134e-05,
+      "loss": 0.0,
+      "step": 6773
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.8149416830739166e-05,
+      "loss": 0.0,
+      "step": 6774
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.814888193842333e-05,
+      "loss": 0.0,
+      "step": 6775
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.8148346973899562e-05,
+      "loss": 0.8242,
+      "step": 6776
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.8147811937170804e-05,
+      "loss": 0.4258,
+      "step": 6777
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.8147276828239988e-05,
+      "loss": 0.0,
+      "step": 6778
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.814674164711006e-05,
+      "loss": 0.0001,
+      "step": 6779
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.814620639378395e-05,
+      "loss": 0.9609,
+      "step": 6780
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.8145671068264602e-05,
+      "loss": 0.0002,
+      "step": 6781
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.814513567055496e-05,
+      "loss": 0.6445,
+      "step": 6782
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.8144600200657953e-05,
+      "loss": 0.0723,
+      "step": 6783
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.814406465857653e-05,
+      "loss": 1.7422,
+      "step": 6784
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.8143529044313635e-05,
+      "loss": 0.1504,
+      "step": 6785
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.8142993357872202e-05,
+      "loss": 0.832,
+      "step": 6786
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.8142457599255172e-05,
+      "loss": 0.3418,
+      "step": 6787
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.8141921768465495e-05,
+      "loss": 0.5117,
+      "step": 6788
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.8141385865506106e-05,
+      "loss": 0.0165,
+      "step": 6789
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.814084989037995e-05,
+      "loss": 0.0155,
+      "step": 6790
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.8140313843089976e-05,
+      "loss": 0.0062,
+      "step": 6791
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.813977772363912e-05,
+      "loss": 0.7695,
+      "step": 6792
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.8139241532030327e-05,
+      "loss": 1.2188,
+      "step": 6793
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.8138705268266548e-05,
+      "loss": 0.6211,
+      "step": 6794
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.813816893235072e-05,
+      "loss": 0.0481,
+      "step": 6795
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.8137632524285795e-05,
+      "loss": 0.0013,
+      "step": 6796
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.8137096044074717e-05,
+      "loss": 0.0635,
+      "step": 6797
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.8136559491720426e-05,
+      "loss": 1.1094,
+      "step": 6798
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.813602286722588e-05,
+      "loss": 1.6484,
+      "step": 6799
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.813548617059401e-05,
+      "loss": 1.5156,
+      "step": 6800
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.813494940182778e-05,
+      "loss": 0.0767,
+      "step": 6801
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.8134412560930126e-05,
+      "loss": 0.0095,
+      "step": 6802
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.8133875647904002e-05,
+      "loss": 0.0039,
+      "step": 6803
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.8133338662752355e-05,
+      "loss": 0.0008,
+      "step": 6804
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.8132801605478137e-05,
+      "loss": 0.0203,
+      "step": 6805
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.8132264476084287e-05,
+      "loss": 0.0001,
+      "step": 6806
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.8131727274573764e-05,
+      "loss": 0.0004,
+      "step": 6807
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.813119000094952e-05,
+      "loss": 0.0703,
+      "step": 6808
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.81306526552145e-05,
+      "loss": 0.0427,
+      "step": 6809
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.813011523737165e-05,
+      "loss": 0.0025,
+      "step": 6810
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.8129577747423936e-05,
+      "loss": 0.0262,
+      "step": 6811
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.8129040185374296e-05,
+      "loss": 0.1602,
+      "step": 6812
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.8128502551225687e-05,
+      "loss": 0.0001,
+      "step": 6813
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.812796484498106e-05,
+      "loss": 0.0001,
+      "step": 6814
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.812742706664337e-05,
+      "loss": 0.0337,
+      "step": 6815
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.8126889216215575e-05,
+      "loss": 0.377,
+      "step": 6816
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.812635129370062e-05,
+      "loss": 0.0059,
+      "step": 6817
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.812581329910146e-05,
+      "loss": 0.2207,
+      "step": 6818
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8125275232421055e-05,
+      "loss": 0.083,
+      "step": 6819
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8124737093662355e-05,
+      "loss": 0.0,
+      "step": 6820
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8124198882828318e-05,
+      "loss": 0.0,
+      "step": 6821
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8123660599921897e-05,
+      "loss": 0.0,
+      "step": 6822
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8123122244946047e-05,
+      "loss": 0.0001,
+      "step": 6823
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.812258381790373e-05,
+      "loss": 0.2236,
+      "step": 6824
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.81220453187979e-05,
+      "loss": 0.0,
+      "step": 6825
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.812150674763151e-05,
+      "loss": 0.0001,
+      "step": 6826
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8120968104407527e-05,
+      "loss": 0.0001,
+      "step": 6827
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.81204293891289e-05,
+      "loss": 0.0001,
+      "step": 6828
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8119890601798594e-05,
+      "loss": 0.0001,
+      "step": 6829
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.811935174241956e-05,
+      "loss": 0.0001,
+      "step": 6830
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8118812810994764e-05,
+      "loss": 0.0,
+      "step": 6831
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8118273807527166e-05,
+      "loss": 0.126,
+      "step": 6832
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.811773473201972e-05,
+      "loss": 0.1582,
+      "step": 6833
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.811719558447539e-05,
+      "loss": 0.5625,
+      "step": 6834
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8116656364897136e-05,
+      "loss": 1.3672,
+      "step": 6835
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.811611707328792e-05,
+      "loss": 0.2314,
+      "step": 6836
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8115577709650704e-05,
+      "loss": 0.2383,
+      "step": 6837
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.811503827398845e-05,
+      "loss": 0.0021,
+      "step": 6838
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.811449876630412e-05,
+      "loss": 1.4766,
+      "step": 6839
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8113959186600674e-05,
+      "loss": 0.7539,
+      "step": 6840
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.811341953488108e-05,
+      "loss": 0.8398,
+      "step": 6841
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8112879811148294e-05,
+      "loss": 1.3125,
+      "step": 6842
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8112340015405288e-05,
+      "loss": 0.2266,
+      "step": 6843
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.811180014765502e-05,
+      "loss": 1.8203,
+      "step": 6844
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8111260207900463e-05,
+      "loss": 0.793,
+      "step": 6845
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8110720196144573e-05,
+      "loss": 0.0439,
+      "step": 6846
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8110180112390324e-05,
+      "loss": 0.0265,
+      "step": 6847
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8109639956640674e-05,
+      "loss": 0.5742,
+      "step": 6848
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8109099728898592e-05,
+      "loss": 0.0175,
+      "step": 6849
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8108559429167046e-05,
+      "loss": 0.0115,
+      "step": 6850
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8108019057449e-05,
+      "loss": 1.0234,
+      "step": 6851
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.810747861374743e-05,
+      "loss": 0.2812,
+      "step": 6852
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8106938098065295e-05,
+      "loss": 0.1504,
+      "step": 6853
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8106397510405565e-05,
+      "loss": 0.377,
+      "step": 6854
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8105856850771213e-05,
+      "loss": 0.0255,
+      "step": 6855
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.81053161191652e-05,
+      "loss": 0.0001,
+      "step": 6856
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8104775315590504e-05,
+      "loss": 0.0049,
+      "step": 6857
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.810423444005009e-05,
+      "loss": 0.0006,
+      "step": 6858
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.810369349254693e-05,
+      "loss": 0.0001,
+      "step": 6859
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8103152473083992e-05,
+      "loss": 0.0576,
+      "step": 6860
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.810261138166425e-05,
+      "loss": 0.0003,
+      "step": 6861
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8102070218290675e-05,
+      "loss": 0.0,
+      "step": 6862
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.810152898296624e-05,
+      "loss": 1.1406,
+      "step": 6863
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8100987675693915e-05,
+      "loss": 0.1914,
+      "step": 6864
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8100446296476672e-05,
+      "loss": 0.0056,
+      "step": 6865
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8099904845317485e-05,
+      "loss": 0.0,
+      "step": 6866
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.809936332221933e-05,
+      "loss": 0.0,
+      "step": 6867
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8098821727185175e-05,
+      "loss": 0.0957,
+      "step": 6868
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8098280060218002e-05,
+      "loss": 0.168,
+      "step": 6869
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8097738321320774e-05,
+      "loss": 0.1318,
+      "step": 6870
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.809719651049648e-05,
+      "loss": 0.0,
+      "step": 6871
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8096654627748086e-05,
+      "loss": 0.0002,
+      "step": 6872
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.809611267307857e-05,
+      "loss": 0.0,
+      "step": 6873
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.809557064649091e-05,
+      "loss": 0.0001,
+      "step": 6874
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.809502854798808e-05,
+      "loss": 0.0001,
+      "step": 6875
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.809448637757306e-05,
+      "loss": 0.0,
+      "step": 6876
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8093944135248826e-05,
+      "loss": 0.1582,
+      "step": 6877
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8093401821018353e-05,
+      "loss": 0.106,
+      "step": 6878
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8092859434884623e-05,
+      "loss": 0.1021,
+      "step": 6879
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8092316976850612e-05,
+      "loss": 0.0,
+      "step": 6880
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.80917744469193e-05,
+      "loss": 0.2002,
+      "step": 6881
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8091231845093666e-05,
+      "loss": 0.1768,
+      "step": 6882
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.809068917137669e-05,
+      "loss": 0.1807,
+      "step": 6883
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 2.8090146425771353e-05,
+      "loss": 1.1953,
+      "step": 6884
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8089603608280635e-05,
+      "loss": 0.3574,
+      "step": 6885
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8089060718907517e-05,
+      "loss": 1.2344,
+      "step": 6886
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8088517757654976e-05,
+      "loss": 0.2061,
+      "step": 6887
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8087974724526005e-05,
+      "loss": 0.5977,
+      "step": 6888
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8087431619523573e-05,
+      "loss": 0.75,
+      "step": 6889
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8086888442650668e-05,
+      "loss": 1.0469,
+      "step": 6890
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8086345193910276e-05,
+      "loss": 0.2793,
+      "step": 6891
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8085801873305376e-05,
+      "loss": 0.126,
+      "step": 6892
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8085258480838953e-05,
+      "loss": 0.8555,
+      "step": 6893
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8084715016513988e-05,
+      "loss": 1.5625,
+      "step": 6894
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.808417148033347e-05,
+      "loss": 0.0054,
+      "step": 6895
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8083627872300388e-05,
+      "loss": 0.5273,
+      "step": 6896
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8083084192417715e-05,
+      "loss": 0.0352,
+      "step": 6897
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8082540440688447e-05,
+      "loss": 0.0369,
+      "step": 6898
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8081996617115566e-05,
+      "loss": 0.2002,
+      "step": 6899
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.808145272170206e-05,
+      "loss": 0.1328,
+      "step": 6900
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.808090875445091e-05,
+      "loss": 0.5664,
+      "step": 6901
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8080364715365116e-05,
+      "loss": 0.3555,
+      "step": 6902
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8079820604447647e-05,
+      "loss": 0.0391,
+      "step": 6903
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8079276421701507e-05,
+      "loss": 1.4375,
+      "step": 6904
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8078732167129684e-05,
+      "loss": 0.0344,
+      "step": 6905
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.807818784073515e-05,
+      "loss": 0.0186,
+      "step": 6906
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8077643442520916e-05,
+      "loss": 0.1504,
+      "step": 6907
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.807709897248996e-05,
+      "loss": 0.2012,
+      "step": 6908
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8076554430645275e-05,
+      "loss": 0.3887,
+      "step": 6909
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8076009816989845e-05,
+      "loss": 0.8906,
+      "step": 6910
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8075465131526666e-05,
+      "loss": 1.5859,
+      "step": 6911
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8074920374258732e-05,
+      "loss": 0.0087,
+      "step": 6912
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8074375545189028e-05,
+      "loss": 0.0,
+      "step": 6913
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8073830644320552e-05,
+      "loss": 0.1416,
+      "step": 6914
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8073285671656297e-05,
+      "loss": 0.0032,
+      "step": 6915
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8072740627199247e-05,
+      "loss": 0.2832,
+      "step": 6916
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.80721955109524e-05,
+      "loss": 0.0211,
+      "step": 6917
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8071650322918756e-05,
+      "loss": 0.0001,
+      "step": 6918
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8071105063101298e-05,
+      "loss": 0.0,
+      "step": 6919
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8070559731503027e-05,
+      "loss": 0.0,
+      "step": 6920
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.807001432812694e-05,
+      "loss": 0.0178,
+      "step": 6921
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8069468852976024e-05,
+      "loss": 0.1475,
+      "step": 6922
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.806892330605328e-05,
+      "loss": 0.0845,
+      "step": 6923
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8068377687361705e-05,
+      "loss": 0.0,
+      "step": 6924
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.806783199690429e-05,
+      "loss": 0.2188,
+      "step": 6925
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8067286234684034e-05,
+      "loss": 0.0,
+      "step": 6926
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.806674040070394e-05,
+      "loss": 0.0001,
+      "step": 6927
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8066194494966995e-05,
+      "loss": 0.0001,
+      "step": 6928
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8065648517476208e-05,
+      "loss": 0.1123,
+      "step": 6929
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8065102468234564e-05,
+      "loss": 0.0014,
+      "step": 6930
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8064556347245075e-05,
+      "loss": 0.0001,
+      "step": 6931
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8064010154510734e-05,
+      "loss": 0.0001,
+      "step": 6932
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8063463890034535e-05,
+      "loss": 0.106,
+      "step": 6933
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.806291755381949e-05,
+      "loss": 1.4922,
+      "step": 6934
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.806237114586859e-05,
+      "loss": 1.6797,
+      "step": 6935
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.806182466618484e-05,
+      "loss": 0.1128,
+      "step": 6936
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8061278114771238e-05,
+      "loss": 0.3281,
+      "step": 6937
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8060731491630786e-05,
+      "loss": 0.3613,
+      "step": 6938
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8060184796766488e-05,
+      "loss": 0.7891,
+      "step": 6939
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.805963803018134e-05,
+      "loss": 0.9688,
+      "step": 6940
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8059091191878356e-05,
+      "loss": 0.1846,
+      "step": 6941
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.805854428186053e-05,
+      "loss": 0.0243,
+      "step": 6942
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.805799730013087e-05,
+      "loss": 0.05,
+      "step": 6943
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8057450246692376e-05,
+      "loss": 0.0261,
+      "step": 6944
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8056903121548053e-05,
+      "loss": 0.1836,
+      "step": 6945
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8056355924700906e-05,
+      "loss": 0.2031,
+      "step": 6946
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8055808656153942e-05,
+      "loss": 0.0527,
+      "step": 6947
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8055261315910166e-05,
+      "loss": 0.8359,
+      "step": 6948
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.805471390397258e-05,
+      "loss": 0.0057,
+      "step": 6949
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8054166420344193e-05,
+      "loss": 0.0,
+      "step": 6950
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.805361886502801e-05,
+      "loss": 0.0067,
+      "step": 6951
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 2.8053071238027044e-05,
+      "loss": 1.5703,
+      "step": 6952
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.805252353934429e-05,
+      "loss": 0.0092,
+      "step": 6953
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.8051975768982768e-05,
+      "loss": 0.4121,
+      "step": 6954
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.8051427926945482e-05,
+      "loss": 0.0002,
+      "step": 6955
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.8050880013235437e-05,
+      "loss": 0.026,
+      "step": 6956
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.8050332027855646e-05,
+      "loss": 2.1406,
+      "step": 6957
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.804978397080912e-05,
+      "loss": 0.04,
+      "step": 6958
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.804923584209886e-05,
+      "loss": 0.0016,
+      "step": 6959
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.8048687641727888e-05,
+      "loss": 0.0295,
+      "step": 6960
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.8048139369699205e-05,
+      "loss": 0.3281,
+      "step": 6961
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.8047591026015825e-05,
+      "loss": 0.0,
+      "step": 6962
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.8047042610680756e-05,
+      "loss": 0.0,
+      "step": 6963
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.8046494123697016e-05,
+      "loss": 0.0002,
+      "step": 6964
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.8045945565067614e-05,
+      "loss": 0.4199,
+      "step": 6965
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.8045396934795562e-05,
+      "loss": 0.1104,
+      "step": 6966
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.8044848232883876e-05,
+      "loss": 0.0645,
+      "step": 6967
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.8044299459335565e-05,
+      "loss": 0.0239,
+      "step": 6968
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.8043750614153643e-05,
+      "loss": 0.0,
+      "step": 6969
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.8043201697341126e-05,
+      "loss": 0.0,
+      "step": 6970
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.804265270890103e-05,
+      "loss": 0.0171,
+      "step": 6971
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.8042103648836364e-05,
+      "loss": 0.0,
+      "step": 6972
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.804155451715015e-05,
+      "loss": 0.0,
+      "step": 6973
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.8041005313845394e-05,
+      "loss": 0.0001,
+      "step": 6974
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.8040456038925128e-05,
+      "loss": 0.0,
+      "step": 6975
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.803990669239235e-05,
+      "loss": 0.0,
+      "step": 6976
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.803935727425009e-05,
+      "loss": 0.4688,
+      "step": 6977
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.803880778450136e-05,
+      "loss": 0.0002,
+      "step": 6978
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.8038258223149174e-05,
+      "loss": 0.1465,
+      "step": 6979
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.8037708590196555e-05,
+      "loss": 0.1562,
+      "step": 6980
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.8037158885646525e-05,
+      "loss": 0.1045,
+      "step": 6981
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.8036609109502095e-05,
+      "loss": 0.0,
+      "step": 6982
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.8036059261766288e-05,
+      "loss": 0.1533,
+      "step": 6983
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.8035509342442123e-05,
+      "loss": 0.7969,
+      "step": 6984
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.8034959351532618e-05,
+      "loss": 0.5859,
+      "step": 6985
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.8034409289040796e-05,
+      "loss": 0.5039,
+      "step": 6986
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.803385915496968e-05,
+      "loss": 0.2266,
+      "step": 6987
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.8033308949322282e-05,
+      "loss": 0.0013,
+      "step": 6988
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.8032758672101635e-05,
+      "loss": 0.6094,
+      "step": 6989
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.8032208323310752e-05,
+      "loss": 0.4121,
+      "step": 6990
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.803165790295266e-05,
+      "loss": 0.7344,
+      "step": 6991
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.803110741103038e-05,
+      "loss": 1.25,
+      "step": 6992
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.8030556847546936e-05,
+      "loss": 1.4453,
+      "step": 6993
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.8030006212505348e-05,
+      "loss": 1.125,
+      "step": 6994
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.8029455505908642e-05,
+      "loss": 0.0381,
+      "step": 6995
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.802890472775985e-05,
+      "loss": 0.1465,
+      "step": 6996
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.802835387806198e-05,
+      "loss": 0.5703,
+      "step": 6997
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.802780295681807e-05,
+      "loss": 1.3594,
+      "step": 6998
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.802725196403114e-05,
+      "loss": 0.668,
+      "step": 6999
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.802670089970422e-05,
+      "loss": 0.0004,
+      "step": 7000
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.8026149763840336e-05,
+      "loss": 0.0918,
+      "step": 7001
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.802559855644251e-05,
+      "loss": 0.0004,
+      "step": 7002
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.802504727751377e-05,
+      "loss": 0.6602,
+      "step": 7003
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.8024495927057146e-05,
+      "loss": 0.0019,
+      "step": 7004
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.802394450507566e-05,
+      "loss": 0.0001,
+      "step": 7005
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.8023393011572355e-05,
+      "loss": 0.2129,
+      "step": 7006
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.802284144655024e-05,
+      "loss": 0.0,
+      "step": 7007
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.802228981001236e-05,
+      "loss": 0.1514,
+      "step": 7008
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.8021738101961737e-05,
+      "loss": 0.0005,
+      "step": 7009
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.80211863224014e-05,
+      "loss": 0.0,
+      "step": 7010
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.8020634471334377e-05,
+      "loss": 0.0347,
+      "step": 7011
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.8020082548763702e-05,
+      "loss": 0.0,
+      "step": 7012
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.801953055469241e-05,
+      "loss": 0.0312,
+      "step": 7013
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.801897848912353e-05,
+      "loss": 0.0103,
+      "step": 7014
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.801842635206009e-05,
+      "loss": 0.4844,
+      "step": 7015
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.801787414350512e-05,
+      "loss": 0.2061,
+      "step": 7016
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.8017321863461664e-05,
+      "loss": 0.4727,
+      "step": 7017
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.8016769511932742e-05,
+      "loss": 0.0,
+      "step": 7018
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.801621708892139e-05,
+      "loss": 1.8906,
+      "step": 7019
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 2.8015664594430648e-05,
+      "loss": 0.0243,
+      "step": 7020
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.8015112028463544e-05,
+      "loss": 0.0008,
+      "step": 7021
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.8014559391023123e-05,
+      "loss": 0.0162,
+      "step": 7022
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.8014006682112403e-05,
+      "loss": 0.0001,
+      "step": 7023
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.801345390173443e-05,
+      "loss": 0.2578,
+      "step": 7024
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.801290104989224e-05,
+      "loss": 0.0001,
+      "step": 7025
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.8012348126588865e-05,
+      "loss": 0.0003,
+      "step": 7026
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.8011795131827344e-05,
+      "loss": 0.1055,
+      "step": 7027
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.8011242065610713e-05,
+      "loss": 0.1055,
+      "step": 7028
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.801068892794201e-05,
+      "loss": 0.9961,
+      "step": 7029
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.8010135718824265e-05,
+      "loss": 0.0996,
+      "step": 7030
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.8009582438260526e-05,
+      "loss": 0.0952,
+      "step": 7031
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.800902908625383e-05,
+      "loss": 0.1426,
+      "step": 7032
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.8008475662807214e-05,
+      "loss": 0.1699,
+      "step": 7033
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.8007922167923716e-05,
+      "loss": 0.707,
+      "step": 7034
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.800736860160638e-05,
+      "loss": 0.3242,
+      "step": 7035
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.8006814963858238e-05,
+      "loss": 1.1016,
+      "step": 7036
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.8006261254682336e-05,
+      "loss": 0.2207,
+      "step": 7037
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.8005707474081712e-05,
+      "loss": 0.2773,
+      "step": 7038
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.8005153622059406e-05,
+      "loss": 0.252,
+      "step": 7039
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.8004599698618467e-05,
+      "loss": 0.0549,
+      "step": 7040
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.8004045703761933e-05,
+      "loss": 0.9219,
+      "step": 7041
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.8003491637492845e-05,
+      "loss": 0.5742,
+      "step": 7042
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.800293749981424e-05,
+      "loss": 1.6875,
+      "step": 7043
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.8002383290729173e-05,
+      "loss": 1.2031,
+      "step": 7044
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.800182901024068e-05,
+      "loss": 0.0405,
+      "step": 7045
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.8001274658351803e-05,
+      "loss": 0.2969,
+      "step": 7046
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.8000720235065594e-05,
+      "loss": 0.1777,
+      "step": 7047
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.800016574038509e-05,
+      "loss": 0.6992,
+      "step": 7048
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.7999611174313345e-05,
+      "loss": 0.0011,
+      "step": 7049
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.7999056536853393e-05,
+      "loss": 0.0001,
+      "step": 7050
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.7998501828008288e-05,
+      "loss": 0.0311,
+      "step": 7051
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.7997947047781075e-05,
+      "loss": 0.2344,
+      "step": 7052
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.7997392196174798e-05,
+      "loss": 0.0461,
+      "step": 7053
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.7996837273192507e-05,
+      "loss": 0.0075,
+      "step": 7054
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.7996282278837245e-05,
+      "loss": 0.7422,
+      "step": 7055
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.7995727213112067e-05,
+      "loss": 0.0013,
+      "step": 7056
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.7995172076020014e-05,
+      "loss": 0.0483,
+      "step": 7057
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.7994616867564137e-05,
+      "loss": 0.0002,
+      "step": 7058
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.7994061587747485e-05,
+      "loss": 0.0064,
+      "step": 7059
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.799350623657311e-05,
+      "loss": 0.0001,
+      "step": 7060
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.799295081404406e-05,
+      "loss": 0.0,
+      "step": 7061
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.799239532016338e-05,
+      "loss": 0.0,
+      "step": 7062
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.799183975493413e-05,
+      "loss": 0.0194,
+      "step": 7063
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.7991284118359356e-05,
+      "loss": 0.0175,
+      "step": 7064
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.7990728410442106e-05,
+      "loss": 0.0,
+      "step": 7065
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.7990172631185437e-05,
+      "loss": 0.0,
+      "step": 7066
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.7989616780592395e-05,
+      "loss": 0.0001,
+      "step": 7067
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.7989060858666044e-05,
+      "loss": 0.0002,
+      "step": 7068
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.7988504865409424e-05,
+      "loss": 0.0,
+      "step": 7069
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.7987948800825597e-05,
+      "loss": 0.0,
+      "step": 7070
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.798739266491761e-05,
+      "loss": 0.0001,
+      "step": 7071
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.7986836457688523e-05,
+      "loss": 0.0,
+      "step": 7072
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.798628017914139e-05,
+      "loss": 0.0,
+      "step": 7073
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.798572382927926e-05,
+      "loss": 0.0,
+      "step": 7074
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.7985167408105194e-05,
+      "loss": 0.0001,
+      "step": 7075
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.7984610915622247e-05,
+      "loss": 0.0009,
+      "step": 7076
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.798405435183347e-05,
+      "loss": 0.1035,
+      "step": 7077
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.7983497716741925e-05,
+      "loss": 0.125,
+      "step": 7078
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.7982941010350666e-05,
+      "loss": 0.1846,
+      "step": 7079
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.7982384232662754e-05,
+      "loss": 0.1279,
+      "step": 7080
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.798182738368124e-05,
+      "loss": 0.1699,
+      "step": 7081
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.7981270463409188e-05,
+      "loss": 0.1602,
+      "step": 7082
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.7980713471849658e-05,
+      "loss": 0.1553,
+      "step": 7083
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.7980156409005698e-05,
+      "loss": 0.0081,
+      "step": 7084
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.7979599274880375e-05,
+      "loss": 1.2422,
+      "step": 7085
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.797904206947675e-05,
+      "loss": 0.0625,
+      "step": 7086
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.7978484792797874e-05,
+      "loss": 0.1914,
+      "step": 7087
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 2.7977927444846826e-05,
+      "loss": 1.8125,
+      "step": 7088
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.7977370025626642e-05,
+      "loss": 0.1348,
+      "step": 7089
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.7976812535140402e-05,
+      "loss": 0.0071,
+      "step": 7090
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.7976254973391157e-05,
+      "loss": 0.0286,
+      "step": 7091
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.7975697340381974e-05,
+      "loss": 0.4805,
+      "step": 7092
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.7975139636115922e-05,
+      "loss": 0.0297,
+      "step": 7093
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.7974581860596048e-05,
+      "loss": 0.0009,
+      "step": 7094
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.7974024013825427e-05,
+      "loss": 0.001,
+      "step": 7095
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.7973466095807113e-05,
+      "loss": 1.2656,
+      "step": 7096
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.7972908106544178e-05,
+      "loss": 0.0327,
+      "step": 7097
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.7972350046039686e-05,
+      "loss": 0.009,
+      "step": 7098
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.7971791914296695e-05,
+      "loss": 0.0869,
+      "step": 7099
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.7971233711318277e-05,
+      "loss": 0.3691,
+      "step": 7100
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.7970675437107488e-05,
+      "loss": 0.0001,
+      "step": 7101
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.7970117091667408e-05,
+      "loss": 0.8047,
+      "step": 7102
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.7969558675001093e-05,
+      "loss": 0.0001,
+      "step": 7103
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.796900018711161e-05,
+      "loss": 0.1099,
+      "step": 7104
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.796844162800203e-05,
+      "loss": 0.0025,
+      "step": 7105
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.7967882997675424e-05,
+      "loss": 0.0022,
+      "step": 7106
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.7967324296134842e-05,
+      "loss": 0.0002,
+      "step": 7107
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.7966765523383375e-05,
+      "loss": 0.3984,
+      "step": 7108
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.7966206679424073e-05,
+      "loss": 0.0002,
+      "step": 7109
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.796564776426002e-05,
+      "loss": 0.0123,
+      "step": 7110
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.7965088777894274e-05,
+      "loss": 0.0051,
+      "step": 7111
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.7964529720329906e-05,
+      "loss": 0.0,
+      "step": 7112
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.7963970591569996e-05,
+      "loss": 0.0141,
+      "step": 7113
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.7963411391617603e-05,
+      "loss": 0.0,
+      "step": 7114
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.7962852120475798e-05,
+      "loss": 0.0,
+      "step": 7115
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.7962292778147664e-05,
+      "loss": 0.0,
+      "step": 7116
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.7961733364636263e-05,
+      "loss": 0.0,
+      "step": 7117
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.796117387994467e-05,
+      "loss": 0.0,
+      "step": 7118
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.7960614324075954e-05,
+      "loss": 0.0004,
+      "step": 7119
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.7960054697033188e-05,
+      "loss": 0.0703,
+      "step": 7120
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.795949499881945e-05,
+      "loss": 0.0,
+      "step": 7121
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.795893522943782e-05,
+      "loss": 0.0,
+      "step": 7122
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.7958375388891354e-05,
+      "loss": 0.7812,
+      "step": 7123
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.795781547718314e-05,
+      "loss": 0.0,
+      "step": 7124
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.7957255494316242e-05,
+      "loss": 0.0,
+      "step": 7125
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.7956695440293753e-05,
+      "loss": 0.0,
+      "step": 7126
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.795613531511873e-05,
+      "loss": 0.0,
+      "step": 7127
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.7955575118794255e-05,
+      "loss": 0.0,
+      "step": 7128
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.7955014851323407e-05,
+      "loss": 0.0,
+      "step": 7129
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.7954454512709262e-05,
+      "loss": 0.0,
+      "step": 7130
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.7953894102954897e-05,
+      "loss": 0.1416,
+      "step": 7131
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.7953333622063388e-05,
+      "loss": 0.1113,
+      "step": 7132
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.7952773070037814e-05,
+      "loss": 0.1387,
+      "step": 7133
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.7952212446881253e-05,
+      "loss": 0.7344,
+      "step": 7134
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.7951651752596788e-05,
+      "loss": 0.0435,
+      "step": 7135
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.795109098718749e-05,
+      "loss": 0.5234,
+      "step": 7136
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.7950530150656443e-05,
+      "loss": 0.377,
+      "step": 7137
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.794996924300673e-05,
+      "loss": 0.6406,
+      "step": 7138
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.794940826424142e-05,
+      "loss": 0.7734,
+      "step": 7139
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.794884721436361e-05,
+      "loss": 0.1465,
+      "step": 7140
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.7948286093376366e-05,
+      "loss": 0.5312,
+      "step": 7141
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.794772490128278e-05,
+      "loss": 0.4102,
+      "step": 7142
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.794716363808593e-05,
+      "loss": 0.3887,
+      "step": 7143
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.794660230378889e-05,
+      "loss": 0.1089,
+      "step": 7144
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.7946040898394758e-05,
+      "loss": 0.2324,
+      "step": 7145
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.7945479421906607e-05,
+      "loss": 1.125,
+      "step": 7146
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.7944917874327524e-05,
+      "loss": 1.0312,
+      "step": 7147
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.7944356255660594e-05,
+      "loss": 0.0383,
+      "step": 7148
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.7943794565908894e-05,
+      "loss": 0.0,
+      "step": 7149
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.794323280507551e-05,
+      "loss": 0.0527,
+      "step": 7150
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.7942670973163536e-05,
+      "loss": 0.0938,
+      "step": 7151
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.7942109070176056e-05,
+      "loss": 0.0129,
+      "step": 7152
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.7941547096116144e-05,
+      "loss": 0.0806,
+      "step": 7153
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.7940985050986895e-05,
+      "loss": 0.0,
+      "step": 7154
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.7940422934791396e-05,
+      "loss": 0.625,
+      "step": 7155
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 2.793986074753273e-05,
+      "loss": 0.582,
+      "step": 7156
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.7939298489213985e-05,
+      "loss": 0.0,
+      "step": 7157
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.7938736159838254e-05,
+      "loss": 0.0579,
+      "step": 7158
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.7938173759408618e-05,
+      "loss": 0.0481,
+      "step": 7159
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.7937611287928168e-05,
+      "loss": 0.0322,
+      "step": 7160
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.793704874539999e-05,
+      "loss": 0.0972,
+      "step": 7161
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.793648613182718e-05,
+      "loss": 0.0303,
+      "step": 7162
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.7935923447212828e-05,
+      "loss": 0.3867,
+      "step": 7163
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.7935360691560013e-05,
+      "loss": 0.0,
+      "step": 7164
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.7934797864871838e-05,
+      "loss": 0.0,
+      "step": 7165
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.7934234967151388e-05,
+      "loss": 0.0,
+      "step": 7166
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.7933671998401754e-05,
+      "loss": 0.001,
+      "step": 7167
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.7933108958626025e-05,
+      "loss": 0.0,
+      "step": 7168
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.7932545847827295e-05,
+      "loss": 0.0001,
+      "step": 7169
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.793198266600866e-05,
+      "loss": 0.0,
+      "step": 7170
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.7931419413173212e-05,
+      "loss": 0.0,
+      "step": 7171
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.793085608932404e-05,
+      "loss": 0.2041,
+      "step": 7172
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.7930292694464242e-05,
+      "loss": 0.0977,
+      "step": 7173
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.7929729228596907e-05,
+      "loss": 0.0,
+      "step": 7174
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.792916569172513e-05,
+      "loss": 0.0,
+      "step": 7175
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.792860208385201e-05,
+      "loss": 0.0938,
+      "step": 7176
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.7928038404980643e-05,
+      "loss": 0.0771,
+      "step": 7177
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.7927474655114118e-05,
+      "loss": 0.127,
+      "step": 7178
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.7926910834255536e-05,
+      "loss": 0.166,
+      "step": 7179
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.7926346942407993e-05,
+      "loss": 0.0,
+      "step": 7180
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.792578297957458e-05,
+      "loss": 0.1885,
+      "step": 7181
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.7925218945758396e-05,
+      "loss": 0.1221,
+      "step": 7182
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.7924654840962545e-05,
+      "loss": 0.167,
+      "step": 7183
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.7924090665190117e-05,
+      "loss": 1.3438,
+      "step": 7184
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.792352641844422e-05,
+      "loss": 0.5391,
+      "step": 7185
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.7922962100727938e-05,
+      "loss": 0.5586,
+      "step": 7186
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.7922397712044384e-05,
+      "loss": 1.5859,
+      "step": 7187
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.7921833252396653e-05,
+      "loss": 1.2734,
+      "step": 7188
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.792126872178784e-05,
+      "loss": 0.1973,
+      "step": 7189
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.7920704120221046e-05,
+      "loss": 0.3809,
+      "step": 7190
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.7920139447699373e-05,
+      "loss": 1.0391,
+      "step": 7191
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.7919574704225928e-05,
+      "loss": 0.9297,
+      "step": 7192
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.7919009889803808e-05,
+      "loss": 0.6367,
+      "step": 7193
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.791844500443611e-05,
+      "loss": 0.0075,
+      "step": 7194
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.7917880048125943e-05,
+      "loss": 0.0496,
+      "step": 7195
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.7917315020876403e-05,
+      "loss": 0.0001,
+      "step": 7196
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.7916749922690595e-05,
+      "loss": 0.0282,
+      "step": 7197
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.7916184753571632e-05,
+      "loss": 1.0156,
+      "step": 7198
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.7915619513522604e-05,
+      "loss": 0.0013,
+      "step": 7199
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.7915054202546622e-05,
+      "loss": 2.2031,
+      "step": 7200
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.7914488820646786e-05,
+      "loss": 0.0,
+      "step": 7201
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.7913923367826206e-05,
+      "loss": 0.0256,
+      "step": 7202
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.7913357844087985e-05,
+      "loss": 0.0,
+      "step": 7203
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.7912792249435228e-05,
+      "loss": 0.0001,
+      "step": 7204
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.791222658387104e-05,
+      "loss": 1.625,
+      "step": 7205
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.791166084739853e-05,
+      "loss": 0.0002,
+      "step": 7206
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.7911095040020803e-05,
+      "loss": 0.0,
+      "step": 7207
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.791052916174097e-05,
+      "loss": 0.123,
+      "step": 7208
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.7909963212562136e-05,
+      "loss": 0.7188,
+      "step": 7209
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.7909397192487406e-05,
+      "loss": 0.0096,
+      "step": 7210
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.7908831101519888e-05,
+      "loss": 0.0001,
+      "step": 7211
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.790826493966269e-05,
+      "loss": 0.0,
+      "step": 7212
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.7907698706918934e-05,
+      "loss": 0.0001,
+      "step": 7213
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.790713240329172e-05,
+      "loss": 0.003,
+      "step": 7214
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.7906566028784157e-05,
+      "loss": 0.0132,
+      "step": 7215
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.790599958339935e-05,
+      "loss": 0.0,
+      "step": 7216
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.790543306714042e-05,
+      "loss": 0.0,
+      "step": 7217
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.7904866480010473e-05,
+      "loss": 0.0,
+      "step": 7218
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.790429982201262e-05,
+      "loss": 0.0091,
+      "step": 7219
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.7903733093149977e-05,
+      "loss": 0.0,
+      "step": 7220
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.7903166293425655e-05,
+      "loss": 0.208,
+      "step": 7221
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.790259942284276e-05,
+      "loss": 0.0001,
+      "step": 7222
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 2.7902032481404415e-05,
+      "loss": 0.0,
+      "step": 7223
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.7901465469113726e-05,
+      "loss": 0.1699,
+      "step": 7224
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.790089838597381e-05,
+      "loss": 0.5469,
+      "step": 7225
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.790033123198778e-05,
+      "loss": 0.0,
+      "step": 7226
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.789976400715875e-05,
+      "loss": 0.0001,
+      "step": 7227
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.7899196711489835e-05,
+      "loss": 0.0,
+      "step": 7228
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.7898629344984156e-05,
+      "loss": 0.0044,
+      "step": 7229
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.7898061907644822e-05,
+      "loss": 0.1494,
+      "step": 7230
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.7897494399474953e-05,
+      "loss": 0.3828,
+      "step": 7231
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.7896926820477656e-05,
+      "loss": 0.1562,
+      "step": 7232
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.7896359170656062e-05,
+      "loss": 0.1514,
+      "step": 7233
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.789579145001328e-05,
+      "loss": 0.8711,
+      "step": 7234
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.7895223658552433e-05,
+      "loss": 0.8086,
+      "step": 7235
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.789465579627663e-05,
+      "loss": 0.3008,
+      "step": 7236
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.7894087863189e-05,
+      "loss": 0.5469,
+      "step": 7237
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.7893519859292653e-05,
+      "loss": 0.1777,
+      "step": 7238
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.7892951784590713e-05,
+      "loss": 0.7617,
+      "step": 7239
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.78923836390863e-05,
+      "loss": 0.3145,
+      "step": 7240
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.789181542278253e-05,
+      "loss": 0.2402,
+      "step": 7241
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.7891247135682532e-05,
+      "loss": 0.0603,
+      "step": 7242
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.7890678777789417e-05,
+      "loss": 0.4551,
+      "step": 7243
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.7890110349106306e-05,
+      "loss": 0.4609,
+      "step": 7244
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.7889541849636334e-05,
+      "loss": 0.2432,
+      "step": 7245
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.788897327938261e-05,
+      "loss": 0.0366,
+      "step": 7246
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.7888404638348254e-05,
+      "loss": 0.0001,
+      "step": 7247
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.7887835926536398e-05,
+      "loss": 0.0002,
+      "step": 7248
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.7887267143950162e-05,
+      "loss": 0.2129,
+      "step": 7249
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.788669829059267e-05,
+      "loss": 0.1436,
+      "step": 7250
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.7886129366467046e-05,
+      "loss": 1.9141,
+      "step": 7251
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.7885560371576417e-05,
+      "loss": 0.0001,
+      "step": 7252
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.7884991305923896e-05,
+      "loss": 1.2578,
+      "step": 7253
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.788442216951262e-05,
+      "loss": 1.1016,
+      "step": 7254
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.7883852962345713e-05,
+      "loss": 0.0035,
+      "step": 7255
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.7883283684426296e-05,
+      "loss": 0.6172,
+      "step": 7256
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.7882714335757496e-05,
+      "loss": 0.2178,
+      "step": 7257
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.7882144916342444e-05,
+      "loss": 0.0,
+      "step": 7258
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.7881575426184264e-05,
+      "loss": 0.0,
+      "step": 7259
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.7881005865286084e-05,
+      "loss": 0.0317,
+      "step": 7260
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.788043623365103e-05,
+      "loss": 0.0,
+      "step": 7261
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.787986653128223e-05,
+      "loss": 0.0001,
+      "step": 7262
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.7879296758182818e-05,
+      "loss": 0.0,
+      "step": 7263
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.7878726914355918e-05,
+      "loss": 0.0216,
+      "step": 7264
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.787815699980466e-05,
+      "loss": 0.105,
+      "step": 7265
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.787758701453218e-05,
+      "loss": 0.0894,
+      "step": 7266
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.7877016958541594e-05,
+      "loss": 0.3066,
+      "step": 7267
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.787644683183604e-05,
+      "loss": 0.0,
+      "step": 7268
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.7875876634418658e-05,
+      "loss": 0.0052,
+      "step": 7269
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.7875306366292565e-05,
+      "loss": 0.0003,
+      "step": 7270
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.78747360274609e-05,
+      "loss": 0.0903,
+      "step": 7271
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.7874165617926794e-05,
+      "loss": 0.0,
+      "step": 7272
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.7873595137693378e-05,
+      "loss": 0.0,
+      "step": 7273
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.7873024586763786e-05,
+      "loss": 0.001,
+      "step": 7274
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.7872453965141154e-05,
+      "loss": 0.0001,
+      "step": 7275
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.7871883272828614e-05,
+      "loss": 0.0,
+      "step": 7276
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.787131250982929e-05,
+      "loss": 0.0,
+      "step": 7277
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.7870741676146333e-05,
+      "loss": 0.0005,
+      "step": 7278
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.787017077178287e-05,
+      "loss": 0.0,
+      "step": 7279
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.7869599796742035e-05,
+      "loss": 0.0,
+      "step": 7280
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.786902875102696e-05,
+      "loss": 0.0,
+      "step": 7281
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.7868457634640793e-05,
+      "loss": 0.1299,
+      "step": 7282
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.7867886447586656e-05,
+      "loss": 0.0,
+      "step": 7283
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.7867315189867698e-05,
+      "loss": 0.75,
+      "step": 7284
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.786674386148705e-05,
+      "loss": 1.1641,
+      "step": 7285
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.7866172462447847e-05,
+      "loss": 0.041,
+      "step": 7286
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.7865600992753232e-05,
+      "loss": 0.0679,
+      "step": 7287
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.7865029452406343e-05,
+      "loss": 0.7578,
+      "step": 7288
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.7864457841410315e-05,
+      "loss": 0.4824,
+      "step": 7289
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.7863886159768287e-05,
+      "loss": 0.5898,
+      "step": 7290
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 2.7863314407483403e-05,
+      "loss": 0.1553,
+      "step": 7291
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.78627425845588e-05,
+      "loss": 0.4805,
+      "step": 7292
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.786217069099762e-05,
+      "loss": 0.3789,
+      "step": 7293
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.7861598726803e-05,
+      "loss": 0.019,
+      "step": 7294
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.7861026691978085e-05,
+      "loss": 0.0166,
+      "step": 7295
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.7860454586526012e-05,
+      "loss": 0.0698,
+      "step": 7296
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.7859882410449925e-05,
+      "loss": 0.1514,
+      "step": 7297
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.785931016375297e-05,
+      "loss": 0.0977,
+      "step": 7298
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.7858737846438285e-05,
+      "loss": 0.0542,
+      "step": 7299
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.7858165458509012e-05,
+      "loss": 0.0007,
+      "step": 7300
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.7857592999968294e-05,
+      "loss": 0.2715,
+      "step": 7301
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.7857020470819282e-05,
+      "loss": 0.0001,
+      "step": 7302
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.785644787106511e-05,
+      "loss": 0.4512,
+      "step": 7303
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.785587520070893e-05,
+      "loss": 0.0,
+      "step": 7304
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.7855302459753884e-05,
+      "loss": 0.0001,
+      "step": 7305
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.7854729648203115e-05,
+      "loss": 0.7344,
+      "step": 7306
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.7854156766059774e-05,
+      "loss": 0.6367,
+      "step": 7307
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.7853583813327003e-05,
+      "loss": 1.4688,
+      "step": 7308
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.785301079000795e-05,
+      "loss": 0.0003,
+      "step": 7309
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.785243769610576e-05,
+      "loss": 0.2852,
+      "step": 7310
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.7851864531623582e-05,
+      "loss": 0.0,
+      "step": 7311
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.7851291296564565e-05,
+      "loss": 0.0435,
+      "step": 7312
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.7850717990931852e-05,
+      "loss": 0.0243,
+      "step": 7313
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.7850144614728598e-05,
+      "loss": 0.0,
+      "step": 7314
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.7849571167957942e-05,
+      "loss": 0.0,
+      "step": 7315
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.7848997650623045e-05,
+      "loss": 0.0,
+      "step": 7316
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.7848424062727043e-05,
+      "loss": 0.6445,
+      "step": 7317
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.7847850404273103e-05,
+      "loss": 0.0001,
+      "step": 7318
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.7847276675264364e-05,
+      "loss": 0.0,
+      "step": 7319
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.784670287570397e-05,
+      "loss": 0.0281,
+      "step": 7320
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.784612900559509e-05,
+      "loss": 0.0184,
+      "step": 7321
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.784555506494086e-05,
+      "loss": 0.0,
+      "step": 7322
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.784498105374444e-05,
+      "loss": 0.0635,
+      "step": 7323
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.7844406972008983e-05,
+      "loss": 0.0001,
+      "step": 7324
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.784383281973764e-05,
+      "loss": 0.0,
+      "step": 7325
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.7843258596933557e-05,
+      "loss": 0.0,
+      "step": 7326
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.7842684303599898e-05,
+      "loss": 0.1523,
+      "step": 7327
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.7842109939739802e-05,
+      "loss": 0.0,
+      "step": 7328
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.7841535505356445e-05,
+      "loss": 0.0001,
+      "step": 7329
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.7840961000452963e-05,
+      "loss": 0.0,
+      "step": 7330
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.7840386425032517e-05,
+      "loss": 0.1406,
+      "step": 7331
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.7839811779098266e-05,
+      "loss": 0.1592,
+      "step": 7332
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.7839237062653364e-05,
+      "loss": 0.0001,
+      "step": 7333
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.783866227570096e-05,
+      "loss": 1.2031,
+      "step": 7334
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.783808741824422e-05,
+      "loss": 0.0312,
+      "step": 7335
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.7837512490286294e-05,
+      "loss": 0.2773,
+      "step": 7336
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.7836937491830344e-05,
+      "loss": 0.0061,
+      "step": 7337
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.7836362422879528e-05,
+      "loss": 0.1299,
+      "step": 7338
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.7835787283436997e-05,
+      "loss": 0.2695,
+      "step": 7339
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.783521207350592e-05,
+      "loss": 0.0209,
+      "step": 7340
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.7834636793089445e-05,
+      "loss": 0.0251,
+      "step": 7341
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.783406144219074e-05,
+      "loss": 0.5547,
+      "step": 7342
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.7833486020812957e-05,
+      "loss": 0.377,
+      "step": 7343
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.7832910528959265e-05,
+      "loss": 0.4688,
+      "step": 7344
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.783233496663281e-05,
+      "loss": 0.4297,
+      "step": 7345
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.783175933383677e-05,
+      "loss": 0.0006,
+      "step": 7346
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.7831183630574297e-05,
+      "loss": 0.0,
+      "step": 7347
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.7830607856848548e-05,
+      "loss": 1.1562,
+      "step": 7348
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.78300320126627e-05,
+      "loss": 0.0393,
+      "step": 7349
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.7829456098019898e-05,
+      "loss": 0.0008,
+      "step": 7350
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.7828880112923315e-05,
+      "loss": 0.7891,
+      "step": 7351
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.782830405737611e-05,
+      "loss": 0.0757,
+      "step": 7352
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.7827727931381452e-05,
+      "loss": 0.5508,
+      "step": 7353
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.78271517349425e-05,
+      "loss": 0.0005,
+      "step": 7354
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.7826575468062415e-05,
+      "loss": 0.0001,
+      "step": 7355
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.7825999130744365e-05,
+      "loss": 0.2969,
+      "step": 7356
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.7825422722991518e-05,
+      "loss": 0.0001,
+      "step": 7357
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.7824846244807037e-05,
+      "loss": 0.0001,
+      "step": 7358
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 2.7824269696194085e-05,
+      "loss": 0.7617,
+      "step": 7359
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.7823693077155835e-05,
+      "loss": 0.0,
+      "step": 7360
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.782311638769545e-05,
+      "loss": 1.6641,
+      "step": 7361
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.7822539627816094e-05,
+      "loss": 0.0171,
+      "step": 7362
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.7821962797520932e-05,
+      "loss": 0.0,
+      "step": 7363
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.782138589681314e-05,
+      "loss": 0.0,
+      "step": 7364
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.782080892569589e-05,
+      "loss": 0.0476,
+      "step": 7365
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.782023188417233e-05,
+      "loss": 0.0,
+      "step": 7366
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.7819654772245652e-05,
+      "loss": 0.5039,
+      "step": 7367
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.7819077589919006e-05,
+      "loss": 0.0255,
+      "step": 7368
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.7818500337195574e-05,
+      "loss": 0.0139,
+      "step": 7369
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.7817923014078528e-05,
+      "loss": 0.0,
+      "step": 7370
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.7817345620571027e-05,
+      "loss": 0.0981,
+      "step": 7371
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.781676815667625e-05,
+      "loss": 0.0002,
+      "step": 7372
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.781619062239737e-05,
+      "loss": 0.0,
+      "step": 7373
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.7815613017737548e-05,
+      "loss": 0.0026,
+      "step": 7374
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.7815035342699962e-05,
+      "loss": 0.0,
+      "step": 7375
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.7814457597287792e-05,
+      "loss": 0.0,
+      "step": 7376
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.7813879781504196e-05,
+      "loss": 0.123,
+      "step": 7377
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.7813301895352358e-05,
+      "loss": 0.0522,
+      "step": 7378
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.781272393883545e-05,
+      "loss": 0.0001,
+      "step": 7379
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.781214591195664e-05,
+      "loss": 0.1064,
+      "step": 7380
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.781156781471911e-05,
+      "loss": 0.2129,
+      "step": 7381
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.781098964712603e-05,
+      "loss": 0.1514,
+      "step": 7382
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.7810411409180577e-05,
+      "loss": 0.1689,
+      "step": 7383
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.7809833100885926e-05,
+      "loss": 1.4141,
+      "step": 7384
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.7809254722245247e-05,
+      "loss": 0.0413,
+      "step": 7385
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.7808676273261727e-05,
+      "loss": 0.3281,
+      "step": 7386
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.780809775393853e-05,
+      "loss": 0.9023,
+      "step": 7387
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.7807519164278848e-05,
+      "loss": 0.4316,
+      "step": 7388
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.780694050428585e-05,
+      "loss": 0.0085,
+      "step": 7389
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.7806361773962714e-05,
+      "loss": 0.3652,
+      "step": 7390
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.7805782973312615e-05,
+      "loss": 0.1816,
+      "step": 7391
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.7805204102338734e-05,
+      "loss": 0.2461,
+      "step": 7392
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.7804625161044256e-05,
+      "loss": 0.2256,
+      "step": 7393
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.7804046149432352e-05,
+      "loss": 0.4355,
+      "step": 7394
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.7803467067506204e-05,
+      "loss": 0.6328,
+      "step": 7395
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.7802887915268992e-05,
+      "loss": 0.377,
+      "step": 7396
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.78023086927239e-05,
+      "loss": 0.2285,
+      "step": 7397
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.7801729399874105e-05,
+      "loss": 0.9375,
+      "step": 7398
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.780115003672279e-05,
+      "loss": 0.0063,
+      "step": 7399
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.7800570603273136e-05,
+      "loss": 0.0003,
+      "step": 7400
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.7799991099528323e-05,
+      "loss": 0.0,
+      "step": 7401
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.7799411525491536e-05,
+      "loss": 0.4395,
+      "step": 7402
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.7798831881165956e-05,
+      "loss": 0.012,
+      "step": 7403
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.779825216655477e-05,
+      "loss": 0.0571,
+      "step": 7404
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.779767238166116e-05,
+      "loss": 0.3711,
+      "step": 7405
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.7797092526488304e-05,
+      "loss": 0.1758,
+      "step": 7406
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.7796512601039394e-05,
+      "loss": 0.0674,
+      "step": 7407
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.7795932605317608e-05,
+      "loss": 0.1709,
+      "step": 7408
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.7795352539326142e-05,
+      "loss": 0.0,
+      "step": 7409
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.779477240306817e-05,
+      "loss": 0.0025,
+      "step": 7410
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.779419219654688e-05,
+      "loss": 0.0001,
+      "step": 7411
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.7793611919765462e-05,
+      "loss": 0.0064,
+      "step": 7412
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.7793031572727102e-05,
+      "loss": 0.008,
+      "step": 7413
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.7792451155434986e-05,
+      "loss": 0.0403,
+      "step": 7414
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.77918706678923e-05,
+      "loss": 0.1709,
+      "step": 7415
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.7791290110102234e-05,
+      "loss": 0.0018,
+      "step": 7416
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.7790709482067974e-05,
+      "loss": 0.0,
+      "step": 7417
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.7790128783792715e-05,
+      "loss": 0.3477,
+      "step": 7418
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.7789548015279634e-05,
+      "loss": 0.0,
+      "step": 7419
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.778896717653193e-05,
+      "loss": 0.0,
+      "step": 7420
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.7788386267552786e-05,
+      "loss": 0.0014,
+      "step": 7421
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.7787805288345406e-05,
+      "loss": 0.0001,
+      "step": 7422
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.7787224238912964e-05,
+      "loss": 0.0,
+      "step": 7423
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.7786643119258658e-05,
+      "loss": 0.0001,
+      "step": 7424
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.7786061929385676e-05,
+      "loss": 0.0002,
+      "step": 7425
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.7785480669297217e-05,
+      "loss": 0.0,
+      "step": 7426
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 2.7784899338996466e-05,
+      "loss": 0.104,
+      "step": 7427
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.7784317938486618e-05,
+      "loss": 0.0,
+      "step": 7428
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.7783736467770863e-05,
+      "loss": 0.0,
+      "step": 7429
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.77831549268524e-05,
+      "loss": 0.0,
+      "step": 7430
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.7782573315734418e-05,
+      "loss": 0.0894,
+      "step": 7431
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.778199163442011e-05,
+      "loss": 0.1445,
+      "step": 7432
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.7781409882912676e-05,
+      "loss": 0.1709,
+      "step": 7433
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.7780828061215305e-05,
+      "loss": 0.6992,
+      "step": 7434
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.7780246169331198e-05,
+      "loss": 0.1807,
+      "step": 7435
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.777966420726354e-05,
+      "loss": 0.0026,
+      "step": 7436
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.7779082175015537e-05,
+      "loss": 0.4336,
+      "step": 7437
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.7778500072590387e-05,
+      "loss": 0.4395,
+      "step": 7438
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.7777917899991274e-05,
+      "loss": 0.6328,
+      "step": 7439
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.777733565722141e-05,
+      "loss": 0.0112,
+      "step": 7440
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.777675334428398e-05,
+      "loss": 0.3867,
+      "step": 7441
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.777617096118219e-05,
+      "loss": 0.3535,
+      "step": 7442
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.777558850791923e-05,
+      "loss": 0.7422,
+      "step": 7443
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.777500598449831e-05,
+      "loss": 0.7734,
+      "step": 7444
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.777442339092262e-05,
+      "loss": 0.5156,
+      "step": 7445
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.777384072719536e-05,
+      "loss": 0.04,
+      "step": 7446
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.7773257993319735e-05,
+      "loss": 0.002,
+      "step": 7447
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.777267518929894e-05,
+      "loss": 1.2188,
+      "step": 7448
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.7772092315136175e-05,
+      "loss": 0.0014,
+      "step": 7449
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.7771509370834648e-05,
+      "loss": 0.1191,
+      "step": 7450
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.7770926356397552e-05,
+      "loss": 0.2598,
+      "step": 7451
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.7770343271828097e-05,
+      "loss": 0.0001,
+      "step": 7452
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.7769760117129475e-05,
+      "loss": 0.0001,
+      "step": 7453
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.7769176892304896e-05,
+      "loss": 0.0,
+      "step": 7454
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.776859359735756e-05,
+      "loss": 0.4414,
+      "step": 7455
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.776801023229067e-05,
+      "loss": 0.5,
+      "step": 7456
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.7767426797107432e-05,
+      "loss": 0.0,
+      "step": 7457
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.7766843291811047e-05,
+      "loss": 0.0,
+      "step": 7458
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.7766259716404722e-05,
+      "loss": 0.003,
+      "step": 7459
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.776567607089166e-05,
+      "loss": 0.0,
+      "step": 7460
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.776509235527507e-05,
+      "loss": 0.0,
+      "step": 7461
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.7764508569558147e-05,
+      "loss": 0.3926,
+      "step": 7462
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.7763924713744103e-05,
+      "loss": 0.0,
+      "step": 7463
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.776334078783615e-05,
+      "loss": 0.0001,
+      "step": 7464
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.7762756791837493e-05,
+      "loss": 0.0,
+      "step": 7465
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.7762172725751328e-05,
+      "loss": 0.0,
+      "step": 7466
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.7761588589580877e-05,
+      "loss": 0.0354,
+      "step": 7467
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.7761004383329336e-05,
+      "loss": 0.0133,
+      "step": 7468
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.7760420106999923e-05,
+      "loss": 0.0,
+      "step": 7469
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.7759835760595838e-05,
+      "loss": 0.0244,
+      "step": 7470
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.7759251344120297e-05,
+      "loss": 0.2207,
+      "step": 7471
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.7758666857576508e-05,
+      "loss": 0.0,
+      "step": 7472
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.775808230096767e-05,
+      "loss": 0.0001,
+      "step": 7473
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.7757497674297012e-05,
+      "loss": 0.0525,
+      "step": 7474
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.775691297756773e-05,
+      "loss": 0.0,
+      "step": 7475
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.7756328210783038e-05,
+      "loss": 0.0,
+      "step": 7476
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.7755743373946154e-05,
+      "loss": 0.1133,
+      "step": 7477
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.7755158467060283e-05,
+      "loss": 0.0,
+      "step": 7478
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.7754573490128635e-05,
+      "loss": 0.2178,
+      "step": 7479
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.7753988443154427e-05,
+      "loss": 0.1523,
+      "step": 7480
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.7753403326140874e-05,
+      "loss": 0.1787,
+      "step": 7481
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.7752818139091184e-05,
+      "loss": 0.0001,
+      "step": 7482
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.7752232882008573e-05,
+      "loss": 0.1201,
+      "step": 7483
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.7751647554896253e-05,
+      "loss": 1.5625,
+      "step": 7484
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.7751062157757442e-05,
+      "loss": 0.2773,
+      "step": 7485
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.7750476690595353e-05,
+      "loss": 0.2295,
+      "step": 7486
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.7749891153413197e-05,
+      "loss": 0.6055,
+      "step": 7487
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.77493055462142e-05,
+      "loss": 0.5195,
+      "step": 7488
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.7748719869001566e-05,
+      "loss": 0.8125,
+      "step": 7489
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.774813412177852e-05,
+      "loss": 1.0625,
+      "step": 7490
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.7747548304548273e-05,
+      "loss": 0.9336,
+      "step": 7491
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.7746962417314044e-05,
+      "loss": 0.4062,
+      "step": 7492
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.7746376460079054e-05,
+      "loss": 0.1177,
+      "step": 7493
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.774579043284651e-05,
+      "loss": 0.2197,
+      "step": 7494
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 2.774520433561965e-05,
+      "loss": 0.3125,
+      "step": 7495
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.774461816840167e-05,
+      "loss": 0.0005,
+      "step": 7496
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.7744031931195803e-05,
+      "loss": 1.1016,
+      "step": 7497
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.7743445624005265e-05,
+      "loss": 0.0359,
+      "step": 7498
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.7742859246833274e-05,
+      "loss": 0.0154,
+      "step": 7499
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.7742272799683054e-05,
+      "loss": 0.0017,
+      "step": 7500
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.774168628255782e-05,
+      "loss": 0.0001,
+      "step": 7501
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.7741099695460797e-05,
+      "loss": 0.1768,
+      "step": 7502
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.7740513038395203e-05,
+      "loss": 0.002,
+      "step": 7503
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.7739926311364263e-05,
+      "loss": 0.2676,
+      "step": 7504
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.77393395143712e-05,
+      "loss": 0.0,
+      "step": 7505
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.7738752647419237e-05,
+      "loss": 0.0001,
+      "step": 7506
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.7738165710511588e-05,
+      "loss": 0.3555,
+      "step": 7507
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.773757870365149e-05,
+      "loss": 0.3203,
+      "step": 7508
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.773699162684215e-05,
+      "loss": 0.3086,
+      "step": 7509
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.7736404480086808e-05,
+      "loss": 0.0181,
+      "step": 7510
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.773581726338868e-05,
+      "loss": 0.0,
+      "step": 7511
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.7735229976750986e-05,
+      "loss": 0.4766,
+      "step": 7512
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.7734642620176963e-05,
+      "loss": 0.0073,
+      "step": 7513
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.773405519366983e-05,
+      "loss": 0.3887,
+      "step": 7514
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.7733467697232813e-05,
+      "loss": 0.0,
+      "step": 7515
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.773288013086914e-05,
+      "loss": 0.0006,
+      "step": 7516
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.7732292494582033e-05,
+      "loss": 0.0101,
+      "step": 7517
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.7731704788374725e-05,
+      "loss": 0.0515,
+      "step": 7518
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.773111701225044e-05,
+      "loss": 0.0,
+      "step": 7519
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.773052916621241e-05,
+      "loss": 0.0,
+      "step": 7520
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.7729941250263856e-05,
+      "loss": 0.0,
+      "step": 7521
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.7729353264408013e-05,
+      "loss": 0.332,
+      "step": 7522
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.7728765208648107e-05,
+      "loss": 0.0,
+      "step": 7523
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.772817708298737e-05,
+      "loss": 0.0,
+      "step": 7524
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.7727588887429025e-05,
+      "loss": 0.0,
+      "step": 7525
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.7727000621976307e-05,
+      "loss": 0.0,
+      "step": 7526
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.772641228663245e-05,
+      "loss": 0.0,
+      "step": 7527
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.772582388140068e-05,
+      "loss": 0.0,
+      "step": 7528
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.7725235406284228e-05,
+      "loss": 0.1484,
+      "step": 7529
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.7724646861286328e-05,
+      "loss": 0.0884,
+      "step": 7530
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.772405824641021e-05,
+      "loss": 0.1484,
+      "step": 7531
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.7723469561659106e-05,
+      "loss": 0.1426,
+      "step": 7532
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.7722880807036252e-05,
+      "loss": 0.1455,
+      "step": 7533
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.772229198254488e-05,
+      "loss": 0.3926,
+      "step": 7534
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.7721703088188226e-05,
+      "loss": 0.7344,
+      "step": 7535
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.7721114123969516e-05,
+      "loss": 0.416,
+      "step": 7536
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.7720525089891993e-05,
+      "loss": 0.918,
+      "step": 7537
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.7719935985958884e-05,
+      "loss": 0.8711,
+      "step": 7538
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.771934681217343e-05,
+      "loss": 0.5898,
+      "step": 7539
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.7718757568538867e-05,
+      "loss": 0.4668,
+      "step": 7540
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.7718168255058426e-05,
+      "loss": 0.8984,
+      "step": 7541
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.7717578871735345e-05,
+      "loss": 0.5469,
+      "step": 7542
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.771698941857286e-05,
+      "loss": 0.3691,
+      "step": 7543
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.7716399895574215e-05,
+      "loss": 0.0189,
+      "step": 7544
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.7715810302742637e-05,
+      "loss": 0.9688,
+      "step": 7545
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.771522064008137e-05,
+      "loss": 0.207,
+      "step": 7546
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.7714630907593647e-05,
+      "loss": 1.8359,
+      "step": 7547
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.7714041105282717e-05,
+      "loss": 0.0001,
+      "step": 7548
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.771345123315181e-05,
+      "loss": 0.0014,
+      "step": 7549
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.7712861291204164e-05,
+      "loss": 0.0001,
+      "step": 7550
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.7712271279443026e-05,
+      "loss": 0.0,
+      "step": 7551
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.771168119787163e-05,
+      "loss": 0.0364,
+      "step": 7552
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.771109104649322e-05,
+      "loss": 0.0461,
+      "step": 7553
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.7710500825311036e-05,
+      "loss": 0.0,
+      "step": 7554
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.7709910534328323e-05,
+      "loss": 0.0289,
+      "step": 7555
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.7709320173548313e-05,
+      "loss": 0.02,
+      "step": 7556
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.7708729742974256e-05,
+      "loss": 0.3926,
+      "step": 7557
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.770813924260939e-05,
+      "loss": 0.0,
+      "step": 7558
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.770754867245696e-05,
+      "loss": 0.0,
+      "step": 7559
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.770695803252021e-05,
+      "loss": 0.0457,
+      "step": 7560
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.770636732280238e-05,
+      "loss": 0.0,
+      "step": 7561
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.770577654330672e-05,
+      "loss": 0.0001,
+      "step": 7562
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 2.7705185694036465e-05,
+      "loss": 0.4551,
+      "step": 7563
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.7704594774994868e-05,
+      "loss": 0.0,
+      "step": 7564
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.7704003786185173e-05,
+      "loss": 0.0124,
+      "step": 7565
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.770341272761062e-05,
+      "loss": 0.1348,
+      "step": 7566
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.7702821599274464e-05,
+      "loss": 0.0,
+      "step": 7567
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.7702230401179943e-05,
+      "loss": 0.0,
+      "step": 7568
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.7701639133330307e-05,
+      "loss": 0.0,
+      "step": 7569
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.7701047795728802e-05,
+      "loss": 0.0,
+      "step": 7570
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.7700456388378677e-05,
+      "loss": 0.0001,
+      "step": 7571
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.769986491128318e-05,
+      "loss": 0.0,
+      "step": 7572
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.7699273364445555e-05,
+      "loss": 0.0918,
+      "step": 7573
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.7698681747869053e-05,
+      "loss": 0.1436,
+      "step": 7574
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.769809006155692e-05,
+      "loss": 0.0,
+      "step": 7575
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.7697498305512416e-05,
+      "loss": 0.0,
+      "step": 7576
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.769690647973878e-05,
+      "loss": 0.0001,
+      "step": 7577
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.769631458423926e-05,
+      "loss": 0.1348,
+      "step": 7578
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.769572261901712e-05,
+      "loss": 0.0,
+      "step": 7579
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.7695130584075594e-05,
+      "loss": 0.1748,
+      "step": 7580
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.769453847941795e-05,
+      "loss": 0.1279,
+      "step": 7581
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.7693946305047425e-05,
+      "loss": 0.1631,
+      "step": 7582
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.769335406096728e-05,
+      "loss": 0.1738,
+      "step": 7583
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.769276174718076e-05,
+      "loss": 1.5938,
+      "step": 7584
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.7692169363691127e-05,
+      "loss": 0.7422,
+      "step": 7585
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.769157691050163e-05,
+      "loss": 0.5859,
+      "step": 7586
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.769098438761552e-05,
+      "loss": 0.1846,
+      "step": 7587
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.769039179503605e-05,
+      "loss": 0.1533,
+      "step": 7588
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.7689799132766483e-05,
+      "loss": 0.6875,
+      "step": 7589
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.7689206400810067e-05,
+      "loss": 0.7109,
+      "step": 7590
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.768861359917005e-05,
+      "loss": 0.1245,
+      "step": 7591
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.76880207278497e-05,
+      "loss": 0.1099,
+      "step": 7592
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.768742778685227e-05,
+      "loss": 0.0114,
+      "step": 7593
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.768683477618101e-05,
+      "loss": 0.0356,
+      "step": 7594
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.7686241695839177e-05,
+      "loss": 0.1797,
+      "step": 7595
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.7685648545830043e-05,
+      "loss": 0.0056,
+      "step": 7596
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.7685055326156844e-05,
+      "loss": 0.0356,
+      "step": 7597
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.768446203682285e-05,
+      "loss": 0.0001,
+      "step": 7598
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.768386867783132e-05,
+      "loss": 0.0001,
+      "step": 7599
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.7683275249185507e-05,
+      "loss": 0.0001,
+      "step": 7600
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.7682681750888673e-05,
+      "loss": 0.0703,
+      "step": 7601
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.768208818294407e-05,
+      "loss": 0.0067,
+      "step": 7602
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.768149454535497e-05,
+      "loss": 0.0063,
+      "step": 7603
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.7680900838124625e-05,
+      "loss": 0.0061,
+      "step": 7604
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.7680307061256297e-05,
+      "loss": 0.3242,
+      "step": 7605
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.7679713214753245e-05,
+      "loss": 0.2676,
+      "step": 7606
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.7679119298618734e-05,
+      "loss": 0.7266,
+      "step": 7607
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.7678525312856025e-05,
+      "loss": 0.0,
+      "step": 7608
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.7677931257468376e-05,
+      "loss": 0.0076,
+      "step": 7609
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.7677337132459054e-05,
+      "loss": 0.0082,
+      "step": 7610
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.767674293783132e-05,
+      "loss": 0.1338,
+      "step": 7611
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.7676148673588432e-05,
+      "loss": 0.0,
+      "step": 7612
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.767555433973366e-05,
+      "loss": 0.083,
+      "step": 7613
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.767495993627027e-05,
+      "loss": 0.2334,
+      "step": 7614
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.7674365463201517e-05,
+      "loss": 0.026,
+      "step": 7615
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.767377092053067e-05,
+      "loss": 0.0,
+      "step": 7616
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.7673176308260998e-05,
+      "loss": 0.1543,
+      "step": 7617
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.7672581626395763e-05,
+      "loss": 0.0,
+      "step": 7618
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.7671986874938228e-05,
+      "loss": 0.1807,
+      "step": 7619
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.7671392053891664e-05,
+      "loss": 0.0187,
+      "step": 7620
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.7670797163259334e-05,
+      "loss": 0.0,
+      "step": 7621
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.7670202203044504e-05,
+      "loss": 0.0,
+      "step": 7622
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.7669607173250448e-05,
+      "loss": 0.0,
+      "step": 7623
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.7669012073880428e-05,
+      "loss": 0.0,
+      "step": 7624
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.766841690493771e-05,
+      "loss": 0.0,
+      "step": 7625
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.7667821666425567e-05,
+      "loss": 0.0,
+      "step": 7626
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.7667226358347262e-05,
+      "loss": 0.0,
+      "step": 7627
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.7666630980706075e-05,
+      "loss": 0.1035,
+      "step": 7628
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.766603553350527e-05,
+      "loss": 0.0,
+      "step": 7629
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 2.766544001674811e-05,
+      "loss": 0.1465,
+      "step": 7630
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.7664844430437875e-05,
+      "loss": 0.0947,
+      "step": 7631
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.766424877457783e-05,
+      "loss": 0.1226,
+      "step": 7632
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.766365304917125e-05,
+      "loss": 0.1309,
+      "step": 7633
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.7663057254221403e-05,
+      "loss": 1.5391,
+      "step": 7634
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.7662461389731563e-05,
+      "loss": 1.2031,
+      "step": 7635
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.7661865455705e-05,
+      "loss": 0.3574,
+      "step": 7636
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.766126945214499e-05,
+      "loss": 0.2539,
+      "step": 7637
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.7660673379054807e-05,
+      "loss": 0.2715,
+      "step": 7638
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.7660077236437718e-05,
+      "loss": 0.4297,
+      "step": 7639
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.7659481024297e-05,
+      "loss": 1.1641,
+      "step": 7640
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.7658884742635926e-05,
+      "loss": 0.0006,
+      "step": 7641
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.7658288391457774e-05,
+      "loss": 1.4297,
+      "step": 7642
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.7657691970765816e-05,
+      "loss": 0.0344,
+      "step": 7643
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.765709548056333e-05,
+      "loss": 0.0195,
+      "step": 7644
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.7656498920853586e-05,
+      "loss": 0.6016,
+      "step": 7645
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.7655902291639867e-05,
+      "loss": 0.4375,
+      "step": 7646
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.7655305592925445e-05,
+      "loss": 1.1406,
+      "step": 7647
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.76547088247136e-05,
+      "loss": 0.2324,
+      "step": 7648
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.7654111987007605e-05,
+      "loss": 1.0391,
+      "step": 7649
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.7653515079810744e-05,
+      "loss": 0.0,
+      "step": 7650
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.7652918103126285e-05,
+      "loss": 0.0,
+      "step": 7651
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.765232105695751e-05,
+      "loss": 1.0938,
+      "step": 7652
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.765172394130771e-05,
+      "loss": 0.002,
+      "step": 7653
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.7651126756180144e-05,
+      "loss": 0.0038,
+      "step": 7654
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.765052950157811e-05,
+      "loss": 0.0014,
+      "step": 7655
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.7649932177504873e-05,
+      "loss": 0.0,
+      "step": 7656
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.7649334783963724e-05,
+      "loss": 0.0093,
+      "step": 7657
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.7648737320957934e-05,
+      "loss": 0.0,
+      "step": 7658
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.7648139788490798e-05,
+      "loss": 0.0001,
+      "step": 7659
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.7647542186565577e-05,
+      "loss": 0.6758,
+      "step": 7660
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.764694451518557e-05,
+      "loss": 0.0125,
+      "step": 7661
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.7646346774354053e-05,
+      "loss": 0.0001,
+      "step": 7662
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.764574896407431e-05,
+      "loss": 0.0,
+      "step": 7663
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.7645151084349625e-05,
+      "loss": 0.0,
+      "step": 7664
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.7644553135183278e-05,
+      "loss": 0.0,
+      "step": 7665
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.764395511657855e-05,
+      "loss": 0.1016,
+      "step": 7666
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.7643357028538728e-05,
+      "loss": 0.0001,
+      "step": 7667
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.7642758871067105e-05,
+      "loss": 0.022,
+      "step": 7668
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.764216064416695e-05,
+      "loss": 0.0,
+      "step": 7669
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.764156234784156e-05,
+      "loss": 0.0479,
+      "step": 7670
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.7640963982094215e-05,
+      "loss": 0.9375,
+      "step": 7671
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.7640365546928206e-05,
+      "loss": 0.0,
+      "step": 7672
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.7639767042346818e-05,
+      "loss": 0.0123,
+      "step": 7673
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.7639168468353336e-05,
+      "loss": 0.0654,
+      "step": 7674
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.763856982495104e-05,
+      "loss": 0.0,
+      "step": 7675
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.763797111214323e-05,
+      "loss": 0.0,
+      "step": 7676
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.7637372329933188e-05,
+      "loss": 0.0001,
+      "step": 7677
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.7636773478324204e-05,
+      "loss": 0.0001,
+      "step": 7678
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.7636174557319565e-05,
+      "loss": 0.0,
+      "step": 7679
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.763557556692256e-05,
+      "loss": 0.1128,
+      "step": 7680
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.7634976507136477e-05,
+      "loss": 0.1904,
+      "step": 7681
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.7634377377964605e-05,
+      "loss": 0.1245,
+      "step": 7682
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.7633778179410244e-05,
+      "loss": 0.0659,
+      "step": 7683
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.7633178911476672e-05,
+      "loss": 1.2422,
+      "step": 7684
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.763257957416719e-05,
+      "loss": 0.4121,
+      "step": 7685
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.763198016748508e-05,
+      "loss": 0.7852,
+      "step": 7686
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.7631380691433638e-05,
+      "loss": 0.3086,
+      "step": 7687
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.7630781146016155e-05,
+      "loss": 0.3164,
+      "step": 7688
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.7630181531235927e-05,
+      "loss": 0.8594,
+      "step": 7689
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.7629581847096246e-05,
+      "loss": 0.0,
+      "step": 7690
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.7628982093600403e-05,
+      "loss": 0.4434,
+      "step": 7691
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.762838227075169e-05,
+      "loss": 0.0203,
+      "step": 7692
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.7627782378553404e-05,
+      "loss": 0.1128,
+      "step": 7693
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.762718241700884e-05,
+      "loss": 0.3945,
+      "step": 7694
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.7626582386121292e-05,
+      "loss": 0.9531,
+      "step": 7695
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.7625982285894054e-05,
+      "loss": 0.0264,
+      "step": 7696
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.762538211633042e-05,
+      "loss": 0.0437,
+      "step": 7697
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 2.7624781877433686e-05,
+      "loss": 0.6094,
+      "step": 7698
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.7624181569207154e-05,
+      "loss": 0.3887,
+      "step": 7699
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.762358119165412e-05,
+      "loss": 0.0001,
+      "step": 7700
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.7622980744777868e-05,
+      "loss": 0.5859,
+      "step": 7701
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.7622380228581714e-05,
+      "loss": 0.3223,
+      "step": 7702
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.762177964306894e-05,
+      "loss": 0.6133,
+      "step": 7703
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.7621178988242856e-05,
+      "loss": 0.0022,
+      "step": 7704
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.7620578264106754e-05,
+      "loss": 0.0001,
+      "step": 7705
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.7619977470663934e-05,
+      "loss": 0.0,
+      "step": 7706
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.7619376607917694e-05,
+      "loss": 0.0001,
+      "step": 7707
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.7618775675871338e-05,
+      "loss": 0.0,
+      "step": 7708
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.761817467452816e-05,
+      "loss": 0.0,
+      "step": 7709
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.761757360389147e-05,
+      "loss": 0.0,
+      "step": 7710
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.7616972463964557e-05,
+      "loss": 0.0208,
+      "step": 7711
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.7616371254750732e-05,
+      "loss": 0.2461,
+      "step": 7712
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.761576997625329e-05,
+      "loss": 0.0004,
+      "step": 7713
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.7615168628475535e-05,
+      "loss": 0.5,
+      "step": 7714
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.761456721142077e-05,
+      "loss": 0.0908,
+      "step": 7715
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.76139657250923e-05,
+      "loss": 0.0002,
+      "step": 7716
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.761336416949342e-05,
+      "loss": 0.0069,
+      "step": 7717
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.7612762544627447e-05,
+      "loss": 0.0001,
+      "step": 7718
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.761216085049767e-05,
+      "loss": 0.0001,
+      "step": 7719
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.7611559087107404e-05,
+      "loss": 0.0001,
+      "step": 7720
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.761095725445995e-05,
+      "loss": 0.0001,
+      "step": 7721
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.761035535255861e-05,
+      "loss": 0.0002,
+      "step": 7722
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.7609753381406697e-05,
+      "loss": 0.005,
+      "step": 7723
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.760915134100751e-05,
+      "loss": 0.0,
+      "step": 7724
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.760854923136435e-05,
+      "loss": 0.0005,
+      "step": 7725
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.760794705248054e-05,
+      "loss": 0.0001,
+      "step": 7726
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.7607344804359376e-05,
+      "loss": 0.0003,
+      "step": 7727
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.7606742487004164e-05,
+      "loss": 0.105,
+      "step": 7728
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.7606140100418215e-05,
+      "loss": 0.0,
+      "step": 7729
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.7605537644604836e-05,
+      "loss": 0.3887,
+      "step": 7730
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.760493511956734e-05,
+      "loss": 0.1621,
+      "step": 7731
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.7604332525309028e-05,
+      "loss": 0.1553,
+      "step": 7732
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.7603729861833212e-05,
+      "loss": 0.1689,
+      "step": 7733
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.7603127129143204e-05,
+      "loss": 0.5469,
+      "step": 7734
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.7602524327242317e-05,
+      "loss": 0.5234,
+      "step": 7735
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.7601921456133847e-05,
+      "loss": 0.0086,
+      "step": 7736
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.7601318515821122e-05,
+      "loss": 0.8945,
+      "step": 7737
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.7600715506307445e-05,
+      "loss": 0.127,
+      "step": 7738
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.7600112427596126e-05,
+      "loss": 0.5625,
+      "step": 7739
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.759950927969048e-05,
+      "loss": 0.4492,
+      "step": 7740
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.7598906062593818e-05,
+      "loss": 0.5273,
+      "step": 7741
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.759830277630945e-05,
+      "loss": 1.0078,
+      "step": 7742
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.7597699420840696e-05,
+      "loss": 0.0008,
+      "step": 7743
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.7597095996190862e-05,
+      "loss": 0.6328,
+      "step": 7744
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.7596492502363265e-05,
+      "loss": 0.6602,
+      "step": 7745
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.759588893936122e-05,
+      "loss": 0.1221,
+      "step": 7746
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.759528530718804e-05,
+      "loss": 0.7344,
+      "step": 7747
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.7594681605847044e-05,
+      "loss": 0.126,
+      "step": 7748
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.759407783534154e-05,
+      "loss": 0.8359,
+      "step": 7749
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.7593473995674845e-05,
+      "loss": 0.4199,
+      "step": 7750
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.7592870086850282e-05,
+      "loss": 0.0036,
+      "step": 7751
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.7592266108871158e-05,
+      "loss": 2.5,
+      "step": 7752
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.7591662061740795e-05,
+      "loss": 0.7852,
+      "step": 7753
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.7591057945462516e-05,
+      "loss": 0.0635,
+      "step": 7754
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.7590453760039625e-05,
+      "loss": 0.0001,
+      "step": 7755
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.758984950547545e-05,
+      "loss": 0.0002,
+      "step": 7756
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.7589245181773308e-05,
+      "loss": 0.7266,
+      "step": 7757
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.758864078893651e-05,
+      "loss": 0.0,
+      "step": 7758
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.7588036326968388e-05,
+      "loss": 0.0547,
+      "step": 7759
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.758743179587225e-05,
+      "loss": 0.1631,
+      "step": 7760
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.7586827195651426e-05,
+      "loss": 0.0,
+      "step": 7761
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.7586222526309225e-05,
+      "loss": 0.0,
+      "step": 7762
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.758561778784898e-05,
+      "loss": 0.1309,
+      "step": 7763
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.7585012980273995e-05,
+      "loss": 0.0005,
+      "step": 7764
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.758440810358761e-05,
+      "loss": 0.0001,
+      "step": 7765
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 2.7583803157793134e-05,
+      "loss": 0.0013,
+      "step": 7766
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.7583198142893894e-05,
+      "loss": 0.0,
+      "step": 7767
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.758259305889321e-05,
+      "loss": 0.0005,
+      "step": 7768
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.758198790579441e-05,
+      "loss": 0.0923,
+      "step": 7769
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.758138268360081e-05,
+      "loss": 0.3242,
+      "step": 7770
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.7580777392315743e-05,
+      "loss": 0.0001,
+      "step": 7771
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.7580172031942525e-05,
+      "loss": 0.0,
+      "step": 7772
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.757956660248448e-05,
+      "loss": 0.1602,
+      "step": 7773
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.7578961103944937e-05,
+      "loss": 0.1055,
+      "step": 7774
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.7578355536327222e-05,
+      "loss": 0.0,
+      "step": 7775
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.7577749899634658e-05,
+      "loss": 0.0845,
+      "step": 7776
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.7577144193870572e-05,
+      "loss": 0.104,
+      "step": 7777
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.7576538419038285e-05,
+      "loss": 0.165,
+      "step": 7778
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.7575932575141133e-05,
+      "loss": 0.0359,
+      "step": 7779
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.757532666218243e-05,
+      "loss": 0.0001,
+      "step": 7780
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.757472068016552e-05,
+      "loss": 0.1206,
+      "step": 7781
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.7574114629093717e-05,
+      "loss": 0.1719,
+      "step": 7782
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.757350850897036e-05,
+      "loss": 0.1406,
+      "step": 7783
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.7572902319798767e-05,
+      "loss": 0.0791,
+      "step": 7784
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.757229606158228e-05,
+      "loss": 0.9883,
+      "step": 7785
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.757168973432421e-05,
+      "loss": 0.8086,
+      "step": 7786
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.75710833380279e-05,
+      "loss": 0.3672,
+      "step": 7787
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.757047687269668e-05,
+      "loss": 0.0027,
+      "step": 7788
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.7569870338333873e-05,
+      "loss": 1.0156,
+      "step": 7789
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.756926373494282e-05,
+      "loss": 0.668,
+      "step": 7790
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.7568657062526842e-05,
+      "loss": 1.3359,
+      "step": 7791
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.7568050321089276e-05,
+      "loss": 0.6484,
+      "step": 7792
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.7567443510633454e-05,
+      "loss": 0.5234,
+      "step": 7793
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.7566836631162705e-05,
+      "loss": 0.7188,
+      "step": 7794
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.7566229682680366e-05,
+      "loss": 0.0388,
+      "step": 7795
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.7565622665189772e-05,
+      "loss": 0.7031,
+      "step": 7796
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.7565015578694247e-05,
+      "loss": 0.3418,
+      "step": 7797
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.756440842319713e-05,
+      "loss": 0.0747,
+      "step": 7798
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.756380119870176e-05,
+      "loss": 1.5078,
+      "step": 7799
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.7563193905211464e-05,
+      "loss": 1.2188,
+      "step": 7800
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.7562586542729582e-05,
+      "loss": 0.9258,
+      "step": 7801
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.7561979111259445e-05,
+      "loss": 0.0005,
+      "step": 7802
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.7561371610804394e-05,
+      "loss": 0.0006,
+      "step": 7803
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.7560764041367765e-05,
+      "loss": 0.3906,
+      "step": 7804
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.756015640295289e-05,
+      "loss": 0.0035,
+      "step": 7805
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.7559548695563104e-05,
+      "loss": 0.2031,
+      "step": 7806
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.7558940919201754e-05,
+      "loss": 0.0001,
+      "step": 7807
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.7558333073872166e-05,
+      "loss": 0.0258,
+      "step": 7808
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.755772515957769e-05,
+      "loss": 0.2969,
+      "step": 7809
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.7557117176321655e-05,
+      "loss": 0.064,
+      "step": 7810
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.7556509124107404e-05,
+      "loss": 0.0,
+      "step": 7811
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.7555901002938274e-05,
+      "loss": 0.0,
+      "step": 7812
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.755529281281761e-05,
+      "loss": 0.1074,
+      "step": 7813
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.7554684553748743e-05,
+      "loss": 0.0001,
+      "step": 7814
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.7554076225735018e-05,
+      "loss": 0.4258,
+      "step": 7815
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.755346782877978e-05,
+      "loss": 0.0442,
+      "step": 7816
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.755285936288636e-05,
+      "loss": 0.0001,
+      "step": 7817
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.755225082805811e-05,
+      "loss": 0.0,
+      "step": 7818
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.7551642224298364e-05,
+      "loss": 0.104,
+      "step": 7819
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.7551033551610466e-05,
+      "loss": 0.0,
+      "step": 7820
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.7550424809997762e-05,
+      "loss": 0.0,
+      "step": 7821
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.754981599946359e-05,
+      "loss": 0.1826,
+      "step": 7822
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.75492071200113e-05,
+      "loss": 0.0007,
+      "step": 7823
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.7548598171644227e-05,
+      "loss": 0.0,
+      "step": 7824
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.7547989154365728e-05,
+      "loss": 0.0,
+      "step": 7825
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.7547380068179134e-05,
+      "loss": 0.0001,
+      "step": 7826
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.754677091308779e-05,
+      "loss": 0.0,
+      "step": 7827
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.7546161689095052e-05,
+      "loss": 0.126,
+      "step": 7828
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.754555239620426e-05,
+      "loss": 0.0003,
+      "step": 7829
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.754494303441876e-05,
+      "loss": 0.0,
+      "step": 7830
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.7544333603741897e-05,
+      "loss": 0.209,
+      "step": 7831
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.7543724104177017e-05,
+      "loss": 0.1426,
+      "step": 7832
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.7543114535727474e-05,
+      "loss": 0.165,
+      "step": 7833
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 2.7542504898396602e-05,
+      "loss": 0.6055,
+      "step": 7834
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.7541895192187762e-05,
+      "loss": 1.2734,
+      "step": 7835
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.75412854171043e-05,
+      "loss": 0.1816,
+      "step": 7836
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.754067557314956e-05,
+      "loss": 0.1592,
+      "step": 7837
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.7540065660326894e-05,
+      "loss": 0.6055,
+      "step": 7838
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.7539455678639653e-05,
+      "loss": 0.4102,
+      "step": 7839
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.753884562809118e-05,
+      "loss": 0.0205,
+      "step": 7840
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.7538235508684833e-05,
+      "loss": 0.1553,
+      "step": 7841
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.7537625320423955e-05,
+      "loss": 0.5156,
+      "step": 7842
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.7537015063311903e-05,
+      "loss": 1.0156,
+      "step": 7843
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.7536404737352027e-05,
+      "loss": 0.0718,
+      "step": 7844
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.7535794342547678e-05,
+      "loss": 0.3594,
+      "step": 7845
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.7535183878902203e-05,
+      "loss": 0.0,
+      "step": 7846
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.7534573346418964e-05,
+      "loss": 0.2617,
+      "step": 7847
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.7533962745101308e-05,
+      "loss": 0.0125,
+      "step": 7848
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.753335207495259e-05,
+      "loss": 0.0001,
+      "step": 7849
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.753274133597616e-05,
+      "loss": 0.6016,
+      "step": 7850
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.7532130528175376e-05,
+      "loss": 0.0002,
+      "step": 7851
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.753151965155359e-05,
+      "loss": 1.1641,
+      "step": 7852
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.753090870611416e-05,
+      "loss": 0.1245,
+      "step": 7853
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.7530297691860436e-05,
+      "loss": 1.1719,
+      "step": 7854
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.7529686608795776e-05,
+      "loss": 0.0004,
+      "step": 7855
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.7529075456923536e-05,
+      "loss": 0.0154,
+      "step": 7856
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.7528464236247075e-05,
+      "loss": 0.0099,
+      "step": 7857
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.7527852946769745e-05,
+      "loss": 0.0571,
+      "step": 7858
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.7527241588494902e-05,
+      "loss": 0.0684,
+      "step": 7859
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.7526630161425906e-05,
+      "loss": 0.0,
+      "step": 7860
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.752601866556612e-05,
+      "loss": 0.0413,
+      "step": 7861
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.7525407100918888e-05,
+      "loss": 0.0,
+      "step": 7862
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.7524795467487584e-05,
+      "loss": 0.0001,
+      "step": 7863
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.7524183765275555e-05,
+      "loss": 0.1099,
+      "step": 7864
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.752357199428617e-05,
+      "loss": 0.1001,
+      "step": 7865
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.7522960154522782e-05,
+      "loss": 0.084,
+      "step": 7866
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.752234824598875e-05,
+      "loss": 0.0015,
+      "step": 7867
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.752173626868744e-05,
+      "loss": 0.0005,
+      "step": 7868
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.7521124222622206e-05,
+      "loss": 0.0004,
+      "step": 7869
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.7520512107796416e-05,
+      "loss": 0.0001,
+      "step": 7870
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.7519899924213425e-05,
+      "loss": 0.0649,
+      "step": 7871
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.75192876718766e-05,
+      "loss": 0.0001,
+      "step": 7872
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.7518675350789303e-05,
+      "loss": 0.0162,
+      "step": 7873
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.7518062960954894e-05,
+      "loss": 0.0001,
+      "step": 7874
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.7517450502376737e-05,
+      "loss": 0.0002,
+      "step": 7875
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.751683797505819e-05,
+      "loss": 0.1846,
+      "step": 7876
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.751622537900263e-05,
+      "loss": 0.1279,
+      "step": 7877
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.7515612714213408e-05,
+      "loss": 0.001,
+      "step": 7878
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.7514999980693893e-05,
+      "loss": 0.0864,
+      "step": 7879
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.751438717844745e-05,
+      "loss": 0.1572,
+      "step": 7880
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.7513774307477452e-05,
+      "loss": 0.1426,
+      "step": 7881
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.7513161367787248e-05,
+      "loss": 0.106,
+      "step": 7882
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.751254835938022e-05,
+      "loss": 0.167,
+      "step": 7883
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.7511935282259722e-05,
+      "loss": 0.6367,
+      "step": 7884
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.7511322136429126e-05,
+      "loss": 0.8477,
+      "step": 7885
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.7510708921891804e-05,
+      "loss": 0.0014,
+      "step": 7886
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.7510095638651122e-05,
+      "loss": 0.7266,
+      "step": 7887
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.7509482286710436e-05,
+      "loss": 0.6484,
+      "step": 7888
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.750886886607313e-05,
+      "loss": 0.9453,
+      "step": 7889
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.750825537674256e-05,
+      "loss": 1.1562,
+      "step": 7890
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.7507641818722104e-05,
+      "loss": 0.3047,
+      "step": 7891
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.750702819201513e-05,
+      "loss": 1.2734,
+      "step": 7892
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.7506414496625005e-05,
+      "loss": 0.249,
+      "step": 7893
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.7505800732555096e-05,
+      "loss": 0.103,
+      "step": 7894
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.7505186899808786e-05,
+      "loss": 0.4941,
+      "step": 7895
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.7504572998389428e-05,
+      "loss": 0.105,
+      "step": 7896
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.750395902830041e-05,
+      "loss": 0.1846,
+      "step": 7897
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.7503344989545094e-05,
+      "loss": 0.0002,
+      "step": 7898
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.7502730882126853e-05,
+      "loss": 0.5508,
+      "step": 7899
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.750211670604906e-05,
+      "loss": 0.0,
+      "step": 7900
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.750150246131509e-05,
+      "loss": 0.0001,
+      "step": 7901
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 2.7500888147928317e-05,
+      "loss": 0.0291,
+      "step": 7902
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.750027376589211e-05,
+      "loss": 0.0293,
+      "step": 7903
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.749965931520985e-05,
+      "loss": 0.02,
+      "step": 7904
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.7499044795884903e-05,
+      "loss": 0.0115,
+      "step": 7905
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.7498430207920645e-05,
+      "loss": 0.0001,
+      "step": 7906
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.7497815551320454e-05,
+      "loss": 0.5234,
+      "step": 7907
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.7497200826087708e-05,
+      "loss": 0.0,
+      "step": 7908
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.7496586032225777e-05,
+      "loss": 0.1377,
+      "step": 7909
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.749597116973804e-05,
+      "loss": 0.0036,
+      "step": 7910
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.7495356238627878e-05,
+      "loss": 0.0002,
+      "step": 7911
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.7494741238898658e-05,
+      "loss": 0.0991,
+      "step": 7912
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.7494126170553763e-05,
+      "loss": 0.0001,
+      "step": 7913
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.749351103359657e-05,
+      "loss": 0.0001,
+      "step": 7914
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.749289582803046e-05,
+      "loss": 0.0,
+      "step": 7915
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.7492280553858808e-05,
+      "loss": 0.0,
+      "step": 7916
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.7491665211084992e-05,
+      "loss": 0.0103,
+      "step": 7917
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.749104979971239e-05,
+      "loss": 0.0272,
+      "step": 7918
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.749043431974439e-05,
+      "loss": 0.0004,
+      "step": 7919
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.7489818771184364e-05,
+      "loss": 0.0002,
+      "step": 7920
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.7489203154035695e-05,
+      "loss": 0.0613,
+      "step": 7921
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.7488587468301763e-05,
+      "loss": 0.1235,
+      "step": 7922
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.748797171398595e-05,
+      "loss": 0.0,
+      "step": 7923
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.7487355891091634e-05,
+      "loss": 0.0004,
+      "step": 7924
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.7486739999622205e-05,
+      "loss": 0.0002,
+      "step": 7925
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.7486124039581033e-05,
+      "loss": 0.0708,
+      "step": 7926
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.7485508010971514e-05,
+      "loss": 0.8516,
+      "step": 7927
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.748489191379702e-05,
+      "loss": 0.1484,
+      "step": 7928
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.7484275748060944e-05,
+      "loss": 0.0649,
+      "step": 7929
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.7483659513766655e-05,
+      "loss": 0.1416,
+      "step": 7930
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.7483043210917555e-05,
+      "loss": 0.126,
+      "step": 7931
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.7482426839517018e-05,
+      "loss": 0.1426,
+      "step": 7932
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.748181039956843e-05,
+      "loss": 0.0,
+      "step": 7933
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.7481193891075175e-05,
+      "loss": 0.6641,
+      "step": 7934
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.7480577314040646e-05,
+      "loss": 0.1021,
+      "step": 7935
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.747996066846822e-05,
+      "loss": 0.0141,
+      "step": 7936
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.7479343954361285e-05,
+      "loss": 0.7617,
+      "step": 7937
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.7478727171723234e-05,
+      "loss": 0.6797,
+      "step": 7938
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.7478110320557448e-05,
+      "loss": 0.0649,
+      "step": 7939
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.7477493400867314e-05,
+      "loss": 0.457,
+      "step": 7940
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.7476876412656226e-05,
+      "loss": 0.4219,
+      "step": 7941
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.7476259355927565e-05,
+      "loss": 0.249,
+      "step": 7942
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.7475642230684727e-05,
+      "loss": 1.3438,
+      "step": 7943
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.7475025036931096e-05,
+      "loss": 1.7656,
+      "step": 7944
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.7474407774670064e-05,
+      "loss": 0.4297,
+      "step": 7945
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.7473790443905014e-05,
+      "loss": 0.3066,
+      "step": 7946
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.7473173044639344e-05,
+      "loss": 1.9453,
+      "step": 7947
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.7472555576876443e-05,
+      "loss": 0.2578,
+      "step": 7948
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.74719380406197e-05,
+      "loss": 0.0006,
+      "step": 7949
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.7471320435872503e-05,
+      "loss": 0.0209,
+      "step": 7950
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.747070276263825e-05,
+      "loss": 0.0,
+      "step": 7951
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.747008502092033e-05,
+      "loss": 0.014,
+      "step": 7952
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.7469467210722138e-05,
+      "loss": 0.0005,
+      "step": 7953
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.7468849332047064e-05,
+      "loss": 0.0017,
+      "step": 7954
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.74682313848985e-05,
+      "loss": 1.2188,
+      "step": 7955
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.746761336927984e-05,
+      "loss": 0.0001,
+      "step": 7956
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.746699528519448e-05,
+      "loss": 0.0019,
+      "step": 7957
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.7466377132645812e-05,
+      "loss": 0.416,
+      "step": 7958
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.7465758911637237e-05,
+      "loss": 0.0112,
+      "step": 7959
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.746514062217214e-05,
+      "loss": 0.0535,
+      "step": 7960
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.7464522264253925e-05,
+      "loss": 0.418,
+      "step": 7961
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.746390383788598e-05,
+      "loss": 0.0001,
+      "step": 7962
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.7463285343071706e-05,
+      "loss": 0.127,
+      "step": 7963
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.74626667798145e-05,
+      "loss": 0.0256,
+      "step": 7964
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.7462048148117755e-05,
+      "loss": 0.0002,
+      "step": 7965
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.7461429447984874e-05,
+      "loss": 0.0109,
+      "step": 7966
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.746081067941925e-05,
+      "loss": 0.0001,
+      "step": 7967
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.7460191842424282e-05,
+      "loss": 0.0001,
+      "step": 7968
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.745957293700337e-05,
+      "loss": 0.0,
+      "step": 7969
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 2.7458953963159915e-05,
+      "loss": 0.0001,
+      "step": 7970
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.7458334920897304e-05,
+      "loss": 0.0001,
+      "step": 7971
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.745771581021895e-05,
+      "loss": 0.0008,
+      "step": 7972
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.7457096631128248e-05,
+      "loss": 0.0718,
+      "step": 7973
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.74564773836286e-05,
+      "loss": 0.0,
+      "step": 7974
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.7455858067723402e-05,
+      "loss": 0.0,
+      "step": 7975
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.745523868341606e-05,
+      "loss": 0.0,
+      "step": 7976
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.745461923070997e-05,
+      "loss": 0.0,
+      "step": 7977
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.7453999709608542e-05,
+      "loss": 0.0001,
+      "step": 7978
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.745338012011517e-05,
+      "loss": 0.0001,
+      "step": 7979
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.745276046223326e-05,
+      "loss": 0.127,
+      "step": 7980
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.7452140735966217e-05,
+      "loss": 0.1211,
+      "step": 7981
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.745152094131744e-05,
+      "loss": 0.1992,
+      "step": 7982
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.7450901078290335e-05,
+      "loss": 0.1768,
+      "step": 7983
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.7450281146888305e-05,
+      "loss": 0.6602,
+      "step": 7984
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.7449661147114758e-05,
+      "loss": 0.3691,
+      "step": 7985
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.7449041078973093e-05,
+      "loss": 0.3867,
+      "step": 7986
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.7448420942466714e-05,
+      "loss": 0.1631,
+      "step": 7987
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.7447800737599037e-05,
+      "loss": 0.0184,
+      "step": 7988
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.744718046437346e-05,
+      "loss": 1.5938,
+      "step": 7989
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.744656012279339e-05,
+      "loss": 0.6016,
+      "step": 7990
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.7445939712862235e-05,
+      "loss": 1.5625,
+      "step": 7991
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.7445319234583398e-05,
+      "loss": 1.0938,
+      "step": 7992
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.744469868796029e-05,
+      "loss": 0.8516,
+      "step": 7993
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.7444078072996324e-05,
+      "loss": 0.5234,
+      "step": 7994
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.7443457389694893e-05,
+      "loss": 0.3105,
+      "step": 7995
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.7442836638059425e-05,
+      "loss": 0.2314,
+      "step": 7996
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.7442215818093312e-05,
+      "loss": 1.0156,
+      "step": 7997
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.7441594929799974e-05,
+      "loss": 0.2793,
+      "step": 7998
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.7440973973182818e-05,
+      "loss": 0.373,
+      "step": 7999
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.7440352948245254e-05,
+      "loss": 0.0781,
+      "step": 8000
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.7439731854990685e-05,
+      "loss": 0.0143,
+      "step": 8001
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.743911069342253e-05,
+      "loss": 0.3672,
+      "step": 8002
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.7438489463544203e-05,
+      "loss": 1.1953,
+      "step": 8003
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.7437868165359106e-05,
+      "loss": 0.1318,
+      "step": 8004
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.7437246798870655e-05,
+      "loss": 0.015,
+      "step": 8005
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.7436625364082268e-05,
+      "loss": 0.0019,
+      "step": 8006
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.743600386099735e-05,
+      "loss": 0.3867,
+      "step": 8007
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.7435382289619317e-05,
+      "loss": 1.2109,
+      "step": 8008
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.7434760649951583e-05,
+      "loss": 0.0259,
+      "step": 8009
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.7434138941997558e-05,
+      "loss": 0.3398,
+      "step": 8010
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.743351716576066e-05,
+      "loss": 0.0211,
+      "step": 8011
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.7432895321244302e-05,
+      "loss": 0.0055,
+      "step": 8012
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.7432273408451902e-05,
+      "loss": 0.0,
+      "step": 8013
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.743165142738687e-05,
+      "loss": 0.0001,
+      "step": 8014
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.743102937805263e-05,
+      "loss": 0.0,
+      "step": 8015
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.7430407260452585e-05,
+      "loss": 0.0197,
+      "step": 8016
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.742978507459016e-05,
+      "loss": 0.0002,
+      "step": 8017
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.7429162820468772e-05,
+      "loss": 0.0002,
+      "step": 8018
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.7428540498091836e-05,
+      "loss": 0.0014,
+      "step": 8019
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.7427918107462773e-05,
+      "loss": 0.334,
+      "step": 8020
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.742729564858499e-05,
+      "loss": 0.0001,
+      "step": 8021
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.742667312146192e-05,
+      "loss": 0.0001,
+      "step": 8022
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.7426050526096976e-05,
+      "loss": 0.0,
+      "step": 8023
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.7425427862493573e-05,
+      "loss": 0.0,
+      "step": 8024
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.7424805130655136e-05,
+      "loss": 0.0003,
+      "step": 8025
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.7424182330585082e-05,
+      "loss": 0.1133,
+      "step": 8026
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.742355946228683e-05,
+      "loss": 0.0,
+      "step": 8027
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.7422936525763803e-05,
+      "loss": 0.0879,
+      "step": 8028
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.7422313521019418e-05,
+      "loss": 0.4824,
+      "step": 8029
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.74216904480571e-05,
+      "loss": 0.0004,
+      "step": 8030
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.742106730688027e-05,
+      "loss": 0.1416,
+      "step": 8031
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.7420444097492353e-05,
+      "loss": 0.1348,
+      "step": 8032
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.7419820819896762e-05,
+      "loss": 0.1226,
+      "step": 8033
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.7419197474096933e-05,
+      "loss": 0.832,
+      "step": 8034
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.7418574060096274e-05,
+      "loss": 0.0806,
+      "step": 8035
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.741795057789822e-05,
+      "loss": 0.0464,
+      "step": 8036
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 2.7417327027506195e-05,
+      "loss": 0.0664,
+      "step": 8037
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.7416703408923613e-05,
+      "loss": 1.1641,
+      "step": 8038
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.741607972215391e-05,
+      "loss": 0.5312,
+      "step": 8039
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.7415455967200504e-05,
+      "loss": 0.0231,
+      "step": 8040
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.7414832144066824e-05,
+      "loss": 1.3359,
+      "step": 8041
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.7414208252756296e-05,
+      "loss": 1.1328,
+      "step": 8042
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.7413584293272345e-05,
+      "loss": 0.0571,
+      "step": 8043
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.7412960265618393e-05,
+      "loss": 0.3906,
+      "step": 8044
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.7412336169797875e-05,
+      "loss": 1.3359,
+      "step": 8045
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.7411712005814214e-05,
+      "loss": 0.0444,
+      "step": 8046
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.7411087773670834e-05,
+      "loss": 0.0003,
+      "step": 8047
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.741046347337117e-05,
+      "loss": 0.001,
+      "step": 8048
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.7409839104918647e-05,
+      "loss": 1.6797,
+      "step": 8049
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.7409214668316697e-05,
+      "loss": 1.4453,
+      "step": 8050
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.740859016356874e-05,
+      "loss": 0.0006,
+      "step": 8051
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.7407965590678216e-05,
+      "loss": 0.0022,
+      "step": 8052
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.7407340949648546e-05,
+      "loss": 0.0674,
+      "step": 8053
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.740671624048317e-05,
+      "loss": 0.0124,
+      "step": 8054
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.740609146318551e-05,
+      "loss": 0.0002,
+      "step": 8055
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.7405466617759005e-05,
+      "loss": 1.0547,
+      "step": 8056
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.740484170420708e-05,
+      "loss": 0.0001,
+      "step": 8057
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.7404216722533166e-05,
+      "loss": 0.6484,
+      "step": 8058
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.74035916727407e-05,
+      "loss": 0.0037,
+      "step": 8059
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.740296655483311e-05,
+      "loss": 0.0,
+      "step": 8060
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.7402341368813835e-05,
+      "loss": 0.0009,
+      "step": 8061
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.7401716114686305e-05,
+      "loss": 2.3125,
+      "step": 8062
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.740109079245395e-05,
+      "loss": 0.0894,
+      "step": 8063
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.7400465402120202e-05,
+      "loss": 0.0,
+      "step": 8064
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.739983994368851e-05,
+      "loss": 0.0,
+      "step": 8065
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.7399214417162292e-05,
+      "loss": 0.0,
+      "step": 8066
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.7398588822544994e-05,
+      "loss": 0.008,
+      "step": 8067
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.7397963159840043e-05,
+      "loss": 0.0,
+      "step": 8068
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.739733742905089e-05,
+      "loss": 0.0,
+      "step": 8069
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.739671163018095e-05,
+      "loss": 0.0,
+      "step": 8070
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.7396085763233676e-05,
+      "loss": 0.918,
+      "step": 8071
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.73954598282125e-05,
+      "loss": 0.0,
+      "step": 8072
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.7394833825120853e-05,
+      "loss": 0.0002,
+      "step": 8073
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.7394207753962182e-05,
+      "loss": 0.8281,
+      "step": 8074
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.7393581614739924e-05,
+      "loss": 0.0,
+      "step": 8075
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.7392955407457512e-05,
+      "loss": 0.1162,
+      "step": 8076
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.739232913211839e-05,
+      "loss": 0.106,
+      "step": 8077
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.7391702788725997e-05,
+      "loss": 0.0057,
+      "step": 8078
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.7391076377283766e-05,
+      "loss": 0.1768,
+      "step": 8079
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.7390449897795142e-05,
+      "loss": 0.0,
+      "step": 8080
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.738982335026357e-05,
+      "loss": 0.1235,
+      "step": 8081
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.7389196734692483e-05,
+      "loss": 0.0996,
+      "step": 8082
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.7388570051085325e-05,
+      "loss": 0.1768,
+      "step": 8083
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.7387943299445537e-05,
+      "loss": 0.5352,
+      "step": 8084
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.738731647977656e-05,
+      "loss": 0.5273,
+      "step": 8085
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.738668959208184e-05,
+      "loss": 0.6641,
+      "step": 8086
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.7386062636364816e-05,
+      "loss": 0.1079,
+      "step": 8087
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.738543561262894e-05,
+      "loss": 0.9414,
+      "step": 8088
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.7384808520877637e-05,
+      "loss": 0.1108,
+      "step": 8089
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.7384181361114363e-05,
+      "loss": 0.8945,
+      "step": 8090
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.738355413334256e-05,
+      "loss": 0.7539,
+      "step": 8091
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.7382926837565676e-05,
+      "loss": 0.0977,
+      "step": 8092
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.738229947378715e-05,
+      "loss": 0.7539,
+      "step": 8093
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.738167204201043e-05,
+      "loss": 0.3457,
+      "step": 8094
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.7381044542238964e-05,
+      "loss": 0.416,
+      "step": 8095
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.738041697447619e-05,
+      "loss": 0.8594,
+      "step": 8096
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.7379789338725563e-05,
+      "loss": 0.8047,
+      "step": 8097
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.7379161634990528e-05,
+      "loss": 0.9609,
+      "step": 8098
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.7378533863274525e-05,
+      "loss": 0.5586,
+      "step": 8099
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.737790602358101e-05,
+      "loss": 0.0,
+      "step": 8100
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.737727811591343e-05,
+      "loss": 0.2266,
+      "step": 8101
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.737665014027523e-05,
+      "loss": 0.8242,
+      "step": 8102
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.7376022096669858e-05,
+      "loss": 0.6484,
+      "step": 8103
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.7375393985100764e-05,
+      "loss": 0.832,
+      "step": 8104
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 2.7374765805571398e-05,
+      "loss": 0.1924,
+      "step": 8105
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.737413755808521e-05,
+      "loss": 0.7344,
+      "step": 8106
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.737350924264565e-05,
+      "loss": 1.5781,
+      "step": 8107
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.7372880859256167e-05,
+      "loss": 0.0001,
+      "step": 8108
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.737225240792021e-05,
+      "loss": 0.0898,
+      "step": 8109
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.7371623888641232e-05,
+      "loss": 0.0,
+      "step": 8110
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.7370995301422694e-05,
+      "loss": 0.0015,
+      "step": 8111
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.737036664626803e-05,
+      "loss": 0.0002,
+      "step": 8112
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.736973792318071e-05,
+      "loss": 0.0889,
+      "step": 8113
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.736910913216417e-05,
+      "loss": 0.0,
+      "step": 8114
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.7368480273221877e-05,
+      "loss": 0.2188,
+      "step": 8115
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.7367851346357276e-05,
+      "loss": 0.0178,
+      "step": 8116
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.7367222351573823e-05,
+      "loss": 0.0,
+      "step": 8117
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.736659328887498e-05,
+      "loss": 0.0,
+      "step": 8118
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.7365964158264186e-05,
+      "loss": 0.4336,
+      "step": 8119
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.7365334959744904e-05,
+      "loss": 0.0278,
+      "step": 8120
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.7364705693320594e-05,
+      "loss": 0.0,
+      "step": 8121
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.7364076358994707e-05,
+      "loss": 0.0001,
+      "step": 8122
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.73634469567707e-05,
+      "loss": 0.0002,
+      "step": 8123
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.7362817486652024e-05,
+      "loss": 0.001,
+      "step": 8124
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.736218794864214e-05,
+      "loss": 0.0,
+      "step": 8125
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.736155834274451e-05,
+      "loss": 0.0002,
+      "step": 8126
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.736092866896258e-05,
+      "loss": 0.1113,
+      "step": 8127
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.7360298927299826e-05,
+      "loss": 0.0001,
+      "step": 8128
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.7359669117759684e-05,
+      "loss": 0.1934,
+      "step": 8129
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.7359039240345633e-05,
+      "loss": 0.2119,
+      "step": 8130
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.7358409295061115e-05,
+      "loss": 0.1982,
+      "step": 8131
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.73577792819096e-05,
+      "loss": 0.1572,
+      "step": 8132
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.7357149200894548e-05,
+      "loss": 0.1147,
+      "step": 8133
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.7356519052019414e-05,
+      "loss": 0.0493,
+      "step": 8134
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.735588883528766e-05,
+      "loss": 0.4805,
+      "step": 8135
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.7355258550702747e-05,
+      "loss": 0.053,
+      "step": 8136
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.7354628198268137e-05,
+      "loss": 0.2559,
+      "step": 8137
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.735399777798729e-05,
+      "loss": 2.0625,
+      "step": 8138
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.7353367289863675e-05,
+      "loss": 0.3965,
+      "step": 8139
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.7352736733900745e-05,
+      "loss": 0.4609,
+      "step": 8140
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.7352106110101966e-05,
+      "loss": 0.0166,
+      "step": 8141
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.7351475418470802e-05,
+      "loss": 0.4258,
+      "step": 8142
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.7350844659010714e-05,
+      "loss": 0.416,
+      "step": 8143
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.735021383172517e-05,
+      "loss": 0.3203,
+      "step": 8144
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.7349582936617632e-05,
+      "loss": 0.028,
+      "step": 8145
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.7348951973691564e-05,
+      "loss": 0.8672,
+      "step": 8146
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.7348320942950435e-05,
+      "loss": 0.9219,
+      "step": 8147
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.7347689844397705e-05,
+      "loss": 0.8047,
+      "step": 8148
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.734705867803684e-05,
+      "loss": 0.2393,
+      "step": 8149
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.734642744387131e-05,
+      "loss": 0.0003,
+      "step": 8150
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.7345796141904576e-05,
+      "loss": 0.0001,
+      "step": 8151
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.7345164772140112e-05,
+      "loss": 0.5039,
+      "step": 8152
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.7344533334581378e-05,
+      "loss": 0.0047,
+      "step": 8153
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.7343901829231845e-05,
+      "loss": 0.0001,
+      "step": 8154
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.7343270256094985e-05,
+      "loss": 0.0001,
+      "step": 8155
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.7342638615174255e-05,
+      "loss": 0.0001,
+      "step": 8156
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.7342006906473142e-05,
+      "loss": 0.0002,
+      "step": 8157
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.7341375129995096e-05,
+      "loss": 0.0,
+      "step": 8158
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.7340743285743595e-05,
+      "loss": 0.0,
+      "step": 8159
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.734011137372211e-05,
+      "loss": 0.3262,
+      "step": 8160
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.7339479393934106e-05,
+      "loss": 0.1406,
+      "step": 8161
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.733884734638306e-05,
+      "loss": 0.2871,
+      "step": 8162
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.7338215231072438e-05,
+      "loss": 0.0001,
+      "step": 8163
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.7337583048005715e-05,
+      "loss": 0.0,
+      "step": 8164
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.7336950797186362e-05,
+      "loss": 0.0095,
+      "step": 8165
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.7336318478617848e-05,
+      "loss": 0.0,
+      "step": 8166
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.733568609230365e-05,
+      "loss": 0.0,
+      "step": 8167
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.733505363824723e-05,
+      "loss": 0.0,
+      "step": 8168
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.7334421116452076e-05,
+      "loss": 0.0767,
+      "step": 8169
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.733378852692166e-05,
+      "loss": 0.0,
+      "step": 8170
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.733315586965944e-05,
+      "loss": 0.0,
+      "step": 8171
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.7332523144668907e-05,
+      "loss": 0.0091,
+      "step": 8172
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.7331890351953527e-05,
+      "loss": 0.0033,
+      "step": 8173
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.733125749151678e-05,
+      "loss": 0.0,
+      "step": 8174
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.7330624563362136e-05,
+      "loss": 0.1157,
+      "step": 8175
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.7329991567493077e-05,
+      "loss": 0.1226,
+      "step": 8176
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.7329358503913078e-05,
+      "loss": 0.0,
+      "step": 8177
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.7328725372625606e-05,
+      "loss": 0.1328,
+      "step": 8178
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.7328092173634153e-05,
+      "loss": 0.25,
+      "step": 8179
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.7327458906942182e-05,
+      "loss": 0.1309,
+      "step": 8180
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.7326825572553184e-05,
+      "loss": 0.0,
+      "step": 8181
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.7326192170470626e-05,
+      "loss": 0.1035,
+      "step": 8182
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.732555870069799e-05,
+      "loss": 0.1099,
+      "step": 8183
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.7324925163238758e-05,
+      "loss": 0.373,
+      "step": 8184
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.7324291558096406e-05,
+      "loss": 0.7188,
+      "step": 8185
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.7323657885274414e-05,
+      "loss": 0.0347,
+      "step": 8186
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.7323024144776262e-05,
+      "loss": 0.3945,
+      "step": 8187
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.732239033660543e-05,
+      "loss": 2.0938,
+      "step": 8188
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.7321756460765394e-05,
+      "loss": 1.5,
+      "step": 8189
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.7321122517259645e-05,
+      "loss": 1.5938,
+      "step": 8190
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.7320488506091653e-05,
+      "loss": 1.6406,
+      "step": 8191
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.731985442726491e-05,
+      "loss": 0.793,
+      "step": 8192
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.7319220280782896e-05,
+      "loss": 0.4844,
+      "step": 8193
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.7318586066649087e-05,
+      "loss": 1.3359,
+      "step": 8194
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.731795178486697e-05,
+      "loss": 1.125,
+      "step": 8195
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.731731743544003e-05,
+      "loss": 0.0,
+      "step": 8196
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.731668301837175e-05,
+      "loss": 0.8203,
+      "step": 8197
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.731604853366561e-05,
+      "loss": 0.3691,
+      "step": 8198
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.73154139813251e-05,
+      "loss": 0.3594,
+      "step": 8199
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.7314779361353702e-05,
+      "loss": 0.0054,
+      "step": 8200
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.73141446737549e-05,
+      "loss": 0.0001,
+      "step": 8201
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.7313509918532177e-05,
+      "loss": 0.0654,
+      "step": 8202
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.7312875095689023e-05,
+      "loss": 0.5156,
+      "step": 8203
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.7312240205228926e-05,
+      "loss": 0.2773,
+      "step": 8204
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.731160524715537e-05,
+      "loss": 0.0011,
+      "step": 8205
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.7310970221471836e-05,
+      "loss": 0.0311,
+      "step": 8206
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.7310335128181822e-05,
+      "loss": 0.0304,
+      "step": 8207
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.7309699967288808e-05,
+      "loss": 0.0,
+      "step": 8208
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.7309064738796286e-05,
+      "loss": 0.0,
+      "step": 8209
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.7308429442707747e-05,
+      "loss": 0.0,
+      "step": 8210
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.7307794079026666e-05,
+      "loss": 0.0125,
+      "step": 8211
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.730715864775655e-05,
+      "loss": 0.0,
+      "step": 8212
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.7306523148900877e-05,
+      "loss": 0.0,
+      "step": 8213
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.730588758246314e-05,
+      "loss": 0.0,
+      "step": 8214
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.7305251948446833e-05,
+      "loss": 0.0001,
+      "step": 8215
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.7304616246855443e-05,
+      "loss": 0.0,
+      "step": 8216
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.7303980477692454e-05,
+      "loss": 0.0,
+      "step": 8217
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.7303344640961375e-05,
+      "loss": 0.0,
+      "step": 8218
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.730270873666568e-05,
+      "loss": 0.0452,
+      "step": 8219
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.730207276480887e-05,
+      "loss": 0.0,
+      "step": 8220
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.7301436725394436e-05,
+      "loss": 0.0,
+      "step": 8221
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.730080061842587e-05,
+      "loss": 0.0,
+      "step": 8222
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.7300164443906666e-05,
+      "loss": 0.0,
+      "step": 8223
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.729952820184032e-05,
+      "loss": 0.0001,
+      "step": 8224
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.7298891892230323e-05,
+      "loss": 0.0,
+      "step": 8225
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.7298255515080165e-05,
+      "loss": 0.0,
+      "step": 8226
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.7297619070393354e-05,
+      "loss": 0.0,
+      "step": 8227
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.7296982558173367e-05,
+      "loss": 0.0013,
+      "step": 8228
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.7296345978423717e-05,
+      "loss": 0.165,
+      "step": 8229
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.729570933114789e-05,
+      "loss": 0.0001,
+      "step": 8230
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.729507261634938e-05,
+      "loss": 0.1631,
+      "step": 8231
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.7294435834031693e-05,
+      "loss": 0.1455,
+      "step": 8232
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.729379898419832e-05,
+      "loss": 0.1436,
+      "step": 8233
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.7293162066852757e-05,
+      "loss": 1.5859,
+      "step": 8234
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.72925250819985e-05,
+      "loss": 0.2051,
+      "step": 8235
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.7291888029639058e-05,
+      "loss": 0.2637,
+      "step": 8236
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.7291250909777916e-05,
+      "loss": 0.4355,
+      "step": 8237
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.7290613722418583e-05,
+      "loss": 0.0156,
+      "step": 8238
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.728997646756455e-05,
+      "loss": 1.5,
+      "step": 8239
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.728933914521932e-05,
+      "loss": 0.1152,
+      "step": 8240
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.72887017553864e-05,
+      "loss": 1.0312,
+      "step": 8241
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.7288064298069277e-05,
+      "loss": 0.6016,
+      "step": 8242
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.728742677327146e-05,
+      "loss": 0.0,
+      "step": 8243
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.728678918099645e-05,
+      "loss": 0.1836,
+      "step": 8244
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.7286151521247744e-05,
+      "loss": 0.5469,
+      "step": 8245
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.7285513794028848e-05,
+      "loss": 0.0001,
+      "step": 8246
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.7284875999343263e-05,
+      "loss": 0.2471,
+      "step": 8247
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.728423813719449e-05,
+      "loss": 0.0,
+      "step": 8248
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.7283600207586035e-05,
+      "loss": 0.0,
+      "step": 8249
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.72829622105214e-05,
+      "loss": 0.6758,
+      "step": 8250
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.7282324146004086e-05,
+      "loss": 0.0062,
+      "step": 8251
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.7281686014037598e-05,
+      "loss": 0.0,
+      "step": 8252
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.728104781462544e-05,
+      "loss": 0.0001,
+      "step": 8253
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.7280409547771117e-05,
+      "loss": 0.8047,
+      "step": 8254
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.7279771213478137e-05,
+      "loss": 0.0072,
+      "step": 8255
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.7279132811750005e-05,
+      "loss": 0.0,
+      "step": 8256
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.727849434259022e-05,
+      "loss": 0.0903,
+      "step": 8257
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.7277855806002294e-05,
+      "loss": 0.0001,
+      "step": 8258
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.7277217201989738e-05,
+      "loss": 0.0,
+      "step": 8259
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.727657853055605e-05,
+      "loss": 0.0,
+      "step": 8260
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.7275939791704746e-05,
+      "loss": 0.0001,
+      "step": 8261
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.727530098543932e-05,
+      "loss": 0.0,
+      "step": 8262
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.7274662111763295e-05,
+      "loss": 0.0001,
+      "step": 8263
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.7274023170680168e-05,
+      "loss": 0.9219,
+      "step": 8264
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.7273384162193462e-05,
+      "loss": 0.0,
+      "step": 8265
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.727274508630667e-05,
+      "loss": 0.0,
+      "step": 8266
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.727210594302331e-05,
+      "loss": 0.0,
+      "step": 8267
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.7271466732346894e-05,
+      "loss": 0.0,
+      "step": 8268
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.7270827454280924e-05,
+      "loss": 0.0481,
+      "step": 8269
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.727018810882892e-05,
+      "loss": 0.0,
+      "step": 8270
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.7269548695994384e-05,
+      "loss": 0.0,
+      "step": 8271
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.7268909215780834e-05,
+      "loss": 0.0,
+      "step": 8272
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.7268269668191783e-05,
+      "loss": 0.0,
+      "step": 8273
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.7267630053230737e-05,
+      "loss": 0.0057,
+      "step": 8274
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.726699037090121e-05,
+      "loss": 0.0001,
+      "step": 8275
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.7266350621206717e-05,
+      "loss": 0.0,
+      "step": 8276
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.726571080415077e-05,
+      "loss": 0.0,
+      "step": 8277
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.726507091973688e-05,
+      "loss": 0.5977,
+      "step": 8278
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.726443096796857e-05,
+      "loss": 0.0,
+      "step": 8279
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.7263790948849344e-05,
+      "loss": 0.0001,
+      "step": 8280
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.7263150862382722e-05,
+      "loss": 0.1416,
+      "step": 8281
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.7262510708572218e-05,
+      "loss": 0.1445,
+      "step": 8282
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.7261870487421347e-05,
+      "loss": 0.1523,
+      "step": 8283
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.7261230198933627e-05,
+      "loss": 1.7344,
+      "step": 8284
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.7260589843112568e-05,
+      "loss": 1.125,
+      "step": 8285
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.7259949419961697e-05,
+      "loss": 1.2188,
+      "step": 8286
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.7259308929484524e-05,
+      "loss": 0.2949,
+      "step": 8287
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.7258668371684562e-05,
+      "loss": 0.4453,
+      "step": 8288
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.7258027746565342e-05,
+      "loss": 1.2188,
+      "step": 8289
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.7257387054130368e-05,
+      "loss": 0.7852,
+      "step": 8290
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.7256746294383165e-05,
+      "loss": 0.6094,
+      "step": 8291
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.7256105467327252e-05,
+      "loss": 0.2676,
+      "step": 8292
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.7255464572966145e-05,
+      "loss": 0.3359,
+      "step": 8293
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.7254823611303368e-05,
+      "loss": 0.4727,
+      "step": 8294
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.7254182582342437e-05,
+      "loss": 0.7266,
+      "step": 8295
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.725354148608688e-05,
+      "loss": 0.0043,
+      "step": 8296
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.7252900322540203e-05,
+      "loss": 0.1309,
+      "step": 8297
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.725225909170594e-05,
+      "loss": 0.3047,
+      "step": 8298
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.7251617793587606e-05,
+      "loss": 0.001,
+      "step": 8299
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.7250976428188726e-05,
+      "loss": 0.373,
+      "step": 8300
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.725033499551282e-05,
+      "loss": 0.9688,
+      "step": 8301
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.7249693495563414e-05,
+      "loss": 0.0085,
+      "step": 8302
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.7249051928344026e-05,
+      "loss": 1.6328,
+      "step": 8303
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.7248410293858178e-05,
+      "loss": 0.0001,
+      "step": 8304
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.7247768592109403e-05,
+      "loss": 0.5039,
+      "step": 8305
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.724712682310121e-05,
+      "loss": 0.0,
+      "step": 8306
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.724648498683714e-05,
+      "loss": 0.1914,
+      "step": 8307
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.7245843083320707e-05,
+      "loss": 2.6875,
+      "step": 8308
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.724520111255544e-05,
+      "loss": 0.0,
+      "step": 8309
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.7244559074544864e-05,
+      "loss": 0.0002,
+      "step": 8310
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.7243916969292503e-05,
+      "loss": 0.2871,
+      "step": 8311
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.724327479680188e-05,
+      "loss": 0.2148,
+      "step": 8312
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.724263255707653e-05,
+      "loss": 1.4375,
+      "step": 8313
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.7241990250119973e-05,
+      "loss": 0.0162,
+      "step": 8314
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.7241347875935742e-05,
+      "loss": 0.0,
+      "step": 8315
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.724070543452736e-05,
+      "loss": 0.104,
+      "step": 8316
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.724006292589836e-05,
+      "loss": 0.0109,
+      "step": 8317
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.7239420350052256e-05,
+      "loss": 0.0002,
+      "step": 8318
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.72387777069926e-05,
+      "loss": 0.0009,
+      "step": 8319
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.7238134996722902e-05,
+      "loss": 0.0001,
+      "step": 8320
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.72374922192467e-05,
+      "loss": 0.1953,
+      "step": 8321
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.7236849374567524e-05,
+      "loss": 0.7344,
+      "step": 8322
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.7236206462688896e-05,
+      "loss": 0.0,
+      "step": 8323
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.7235563483614362e-05,
+      "loss": 0.0,
+      "step": 8324
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.7234920437347438e-05,
+      "loss": 0.0,
+      "step": 8325
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.7234277323891662e-05,
+      "loss": 0.0,
+      "step": 8326
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.7233634143250564e-05,
+      "loss": 0.0004,
+      "step": 8327
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.723299089542768e-05,
+      "loss": 0.1143,
+      "step": 8328
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.7232347580426537e-05,
+      "loss": 0.2871,
+      "step": 8329
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.723170419825067e-05,
+      "loss": 0.1123,
+      "step": 8330
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.7231060748903613e-05,
+      "loss": 0.0,
+      "step": 8331
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.7230417232388903e-05,
+      "loss": 0.1602,
+      "step": 8332
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.7229773648710063e-05,
+      "loss": 0.123,
+      "step": 8333
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.7229129997870642e-05,
+      "loss": 0.6289,
+      "step": 8334
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.7228486279874157e-05,
+      "loss": 0.2471,
+      "step": 8335
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.7227842494724157e-05,
+      "loss": 0.8633,
+      "step": 8336
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.7227198642424177e-05,
+      "loss": 0.8945,
+      "step": 8337
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.7226554722977747e-05,
+      "loss": 0.5547,
+      "step": 8338
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.7225910736388405e-05,
+      "loss": 0.0354,
+      "step": 8339
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.722526668265968e-05,
+      "loss": 1.0547,
+      "step": 8340
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.7224622561795124e-05,
+      "loss": 0.0287,
+      "step": 8341
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.7223978373798268e-05,
+      "loss": 0.8086,
+      "step": 8342
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.7223334118672647e-05,
+      "loss": 0.0527,
+      "step": 8343
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.7222689796421796e-05,
+      "loss": 0.3242,
+      "step": 8344
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.722204540704926e-05,
+      "loss": 0.8711,
+      "step": 8345
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.7221400950558578e-05,
+      "loss": 0.2324,
+      "step": 8346
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.722075642695328e-05,
+      "loss": 0.8672,
+      "step": 8347
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.722011183623692e-05,
+      "loss": 0.0009,
+      "step": 8348
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.7219467178413022e-05,
+      "loss": 0.0001,
+      "step": 8349
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.721882245348514e-05,
+      "loss": 0.0004,
+      "step": 8350
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.7218177661456807e-05,
+      "loss": 0.2412,
+      "step": 8351
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.7217532802331562e-05,
+      "loss": 0.25,
+      "step": 8352
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.7216887876112952e-05,
+      "loss": 0.6641,
+      "step": 8353
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.7216242882804518e-05,
+      "loss": 0.0008,
+      "step": 8354
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.7215597822409797e-05,
+      "loss": 0.0007,
+      "step": 8355
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.721495269493233e-05,
+      "loss": 0.0003,
+      "step": 8356
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.7214307500375675e-05,
+      "loss": 0.291,
+      "step": 8357
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.7213662238743357e-05,
+      "loss": 0.1445,
+      "step": 8358
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.721301691003893e-05,
+      "loss": 0.0188,
+      "step": 8359
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.721237151426593e-05,
+      "loss": 0.5703,
+      "step": 8360
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.7211726051427914e-05,
+      "loss": 0.25,
+      "step": 8361
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.7211080521528414e-05,
+      "loss": 0.0001,
+      "step": 8362
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.721043492457098e-05,
+      "loss": 0.0292,
+      "step": 8363
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.7209789260559157e-05,
+      "loss": 0.0398,
+      "step": 8364
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.7209143529496488e-05,
+      "loss": 0.0938,
+      "step": 8365
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.7208497731386527e-05,
+      "loss": 0.0017,
+      "step": 8366
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.7207851866232816e-05,
+      "loss": 0.0003,
+      "step": 8367
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.7207205934038897e-05,
+      "loss": 0.0001,
+      "step": 8368
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.720655993480832e-05,
+      "loss": 0.0003,
+      "step": 8369
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.7205913868544635e-05,
+      "loss": 0.0002,
+      "step": 8370
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.720526773525139e-05,
+      "loss": 0.0,
+      "step": 8371
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.720462153493213e-05,
+      "loss": 0.004,
+      "step": 8372
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.7203975267590407e-05,
+      "loss": 0.0,
+      "step": 8373
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.720332893322977e-05,
+      "loss": 0.0,
+      "step": 8374
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.7202682531853763e-05,
+      "loss": 0.0001,
+      "step": 8375
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.7202036063465942e-05,
+      "loss": 0.0,
+      "step": 8376
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.720138952806986e-05,
+      "loss": 0.0859,
+      "step": 8377
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.7200742925669057e-05,
+      "loss": 0.0,
+      "step": 8378
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.7200096256267087e-05,
+      "loss": 0.1426,
+      "step": 8379
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.7199449519867508e-05,
+      "loss": 0.1621,
+      "step": 8380
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.7198802716473864e-05,
+      "loss": 0.1729,
+      "step": 8381
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.719815584608971e-05,
+      "loss": 0.1533,
+      "step": 8382
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.7197508908718603e-05,
+      "loss": 0.1182,
+      "step": 8383
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.7196861904364085e-05,
+      "loss": 0.2637,
+      "step": 8384
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.7196214833029718e-05,
+      "loss": 0.8789,
+      "step": 8385
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.719556769471905e-05,
+      "loss": 0.3027,
+      "step": 8386
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.719492048943564e-05,
+      "loss": 1.0156,
+      "step": 8387
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.719427321718304e-05,
+      "loss": 0.0874,
+      "step": 8388
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.7193625877964807e-05,
+      "loss": 0.0635,
+      "step": 8389
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.7192978471784482e-05,
+      "loss": 0.124,
+      "step": 8390
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.7192330998645644e-05,
+      "loss": 0.6367,
+      "step": 8391
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.7191683458551822e-05,
+      "loss": 0.9844,
+      "step": 8392
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.7191035851506596e-05,
+      "loss": 0.5195,
+      "step": 8393
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.719038817751351e-05,
+      "loss": 0.033,
+      "step": 8394
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.7189740436576123e-05,
+      "loss": 0.1143,
+      "step": 8395
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.7189092628697996e-05,
+      "loss": 0.9531,
+      "step": 8396
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.7188444753882677e-05,
+      "loss": 0.6484,
+      "step": 8397
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.7187796812133733e-05,
+      "loss": 2.1719,
+      "step": 8398
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.7187148803454715e-05,
+      "loss": 0.2393,
+      "step": 8399
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.7186500727849186e-05,
+      "loss": 0.8906,
+      "step": 8400
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.7185852585320704e-05,
+      "loss": 0.0048,
+      "step": 8401
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.7185204375872828e-05,
+      "loss": 0.0447,
+      "step": 8402
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.718455609950912e-05,
+      "loss": 0.0016,
+      "step": 8403
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.7183907756233138e-05,
+      "loss": 0.0557,
+      "step": 8404
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.7183259346048442e-05,
+      "loss": 0.0,
+      "step": 8405
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.7182610868958595e-05,
+      "loss": 0.0007,
+      "step": 8406
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.718196232496716e-05,
+      "loss": 0.0069,
+      "step": 8407
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.7181313714077688e-05,
+      "loss": 0.6016,
+      "step": 8408
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.7180665036293745e-05,
+      "loss": 0.0001,
+      "step": 8409
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.7180016291618906e-05,
+      "loss": 0.2754,
+      "step": 8410
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.7179367480056715e-05,
+      "loss": 0.0058,
+      "step": 8411
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.7178718601610752e-05,
+      "loss": 0.1426,
+      "step": 8412
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.7178069656284565e-05,
+      "loss": 0.0,
+      "step": 8413
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.7177420644081732e-05,
+      "loss": 0.2139,
+      "step": 8414
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.7176771565005804e-05,
+      "loss": 0.1172,
+      "step": 8415
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.7176122419060354e-05,
+      "loss": 0.0,
+      "step": 8416
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.7175473206248945e-05,
+      "loss": 0.105,
+      "step": 8417
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.7174823926575137e-05,
+      "loss": 0.0,
+      "step": 8418
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.7174174580042508e-05,
+      "loss": 0.0001,
+      "step": 8419
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.717352516665461e-05,
+      "loss": 0.0,
+      "step": 8420
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.717287568641502e-05,
+      "loss": 0.0,
+      "step": 8421
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.7172226139327296e-05,
+      "loss": 0.0003,
+      "step": 8422
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.717157652539501e-05,
+      "loss": 0.0001,
+      "step": 8423
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.7170926844621728e-05,
+      "loss": 0.0001,
+      "step": 8424
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.7170277097011016e-05,
+      "loss": 0.1602,
+      "step": 8425
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.716962728256645e-05,
+      "loss": 0.0977,
+      "step": 8426
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.7168977401291587e-05,
+      "loss": 0.0004,
+      "step": 8427
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.7168327453190002e-05,
+      "loss": 0.0923,
+      "step": 8428
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.716767743826527e-05,
+      "loss": 0.1152,
+      "step": 8429
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.7167027356520948e-05,
+      "loss": 0.1318,
+      "step": 8430
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.7166377207960617e-05,
+      "loss": 0.1191,
+      "step": 8431
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.7165726992587842e-05,
+      "loss": 0.165,
+      "step": 8432
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.716507671040619e-05,
+      "loss": 0.1934,
+      "step": 8433
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.716442636141924e-05,
+      "loss": 0.2344,
+      "step": 8434
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.716377594563056e-05,
+      "loss": 1.3047,
+      "step": 8435
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.716312546304372e-05,
+      "loss": 0.8789,
+      "step": 8436
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.7162474913662295e-05,
+      "loss": 2.125,
+      "step": 8437
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.716182429748986e-05,
+      "loss": 1.7578,
+      "step": 8438
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.716117361452998e-05,
+      "loss": 0.0659,
+      "step": 8439
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.7160522864786235e-05,
+      "loss": 0.7695,
+      "step": 8440
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.7159872048262193e-05,
+      "loss": 0.2988,
+      "step": 8441
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.715922116496143e-05,
+      "loss": 1.0703,
+      "step": 8442
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.7158570214887528e-05,
+      "loss": 0.8711,
+      "step": 8443
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.715791919804405e-05,
+      "loss": 1.2031,
+      "step": 8444
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.7157268114434582e-05,
+      "loss": 0.0002,
+      "step": 8445
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.715661696406269e-05,
+      "loss": 1.0625,
+      "step": 8446
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.7155965746931952e-05,
+      "loss": 0.0037,
+      "step": 8447
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.7155314463045952e-05,
+      "loss": 0.0006,
+      "step": 8448
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.715466311240825e-05,
+      "loss": 1.0312,
+      "step": 8449
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.7154011695022444e-05,
+      "loss": 0.4102,
+      "step": 8450
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.7153360210892095e-05,
+      "loss": 0.0015,
+      "step": 8451
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.715270866002079e-05,
+      "loss": 0.0033,
+      "step": 8452
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.71520570424121e-05,
+      "loss": 0.0898,
+      "step": 8453
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.715140535806961e-05,
+      "loss": 0.0004,
+      "step": 8454
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.7150753606996894e-05,
+      "loss": 0.0001,
+      "step": 8455
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.7150101789197533e-05,
+      "loss": 0.085,
+      "step": 8456
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.714944990467511e-05,
+      "loss": 0.015,
+      "step": 8457
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.7148797953433196e-05,
+      "loss": 0.0001,
+      "step": 8458
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.714814593547538e-05,
+      "loss": 1.0312,
+      "step": 8459
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.7147493850805234e-05,
+      "loss": 0.1064,
+      "step": 8460
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.7146841699426344e-05,
+      "loss": 0.0001,
+      "step": 8461
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.71461894813423e-05,
+      "loss": 0.0129,
+      "step": 8462
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.7145537196556666e-05,
+      "loss": 0.084,
+      "step": 8463
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.7144884845073036e-05,
+      "loss": 0.0,
+      "step": 8464
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.714423242689499e-05,
+      "loss": 0.0591,
+      "step": 8465
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.714357994202611e-05,
+      "loss": 0.0649,
+      "step": 8466
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.714292739046998e-05,
+      "loss": 0.1611,
+      "step": 8467
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.714227477223018e-05,
+      "loss": 0.0,
+      "step": 8468
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.71416220873103e-05,
+      "loss": 0.0001,
+      "step": 8469
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.7140969335713918e-05,
+      "loss": 0.0,
+      "step": 8470
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.7140316517444623e-05,
+      "loss": 0.0,
+      "step": 8471
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.7139663632506e-05,
+      "loss": 0.0,
+      "step": 8472
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.7139010680901634e-05,
+      "loss": 0.0001,
+      "step": 8473
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.7138357662635108e-05,
+      "loss": 0.0,
+      "step": 8474
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.7137704577710013e-05,
+      "loss": 0.0,
+      "step": 8475
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.7137051426129927e-05,
+      "loss": 0.1191,
+      "step": 8476
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.7136398207898444e-05,
+      "loss": 0.0001,
+      "step": 8477
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.7135744923019154e-05,
+      "loss": 0.0,
+      "step": 8478
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.7135091571495632e-05,
+      "loss": 0.1484,
+      "step": 8479
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.713443815333148e-05,
+      "loss": 0.1484,
+      "step": 8480
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.7133784668530275e-05,
+      "loss": 0.0918,
+      "step": 8481
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.7133131117095614e-05,
+      "loss": 0.1504,
+      "step": 8482
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.7132477499031086e-05,
+      "loss": 0.0747,
+      "step": 8483
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.713182381434027e-05,
+      "loss": 0.4961,
+      "step": 8484
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.713117006302677e-05,
+      "loss": 1.1094,
+      "step": 8485
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.7130516245094165e-05,
+      "loss": 0.0289,
+      "step": 8486
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.712986236054605e-05,
+      "loss": 0.3848,
+      "step": 8487
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.7129208409386015e-05,
+      "loss": 0.9648,
+      "step": 8488
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.712855439161765e-05,
+      "loss": 0.4668,
+      "step": 8489
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.7127900307244554e-05,
+      "loss": 0.2559,
+      "step": 8490
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.712724615627031e-05,
+      "loss": 0.3887,
+      "step": 8491
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.712659193869851e-05,
+      "loss": 0.2656,
+      "step": 8492
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.7125937654532755e-05,
+      "loss": 0.2461,
+      "step": 8493
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.7125283303776633e-05,
+      "loss": 0.8828,
+      "step": 8494
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.7124628886433733e-05,
+      "loss": 0.1699,
+      "step": 8495
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.712397440250766e-05,
+      "loss": 0.016,
+      "step": 8496
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.7123319852001995e-05,
+      "loss": 0.0005,
+      "step": 8497
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.712266523492034e-05,
+      "loss": 0.0618,
+      "step": 8498
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.712201055126629e-05,
+      "loss": 0.2031,
+      "step": 8499
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.7121355801043444e-05,
+      "loss": 1.5859,
+      "step": 8500
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.7120700984255388e-05,
+      "loss": 0.0001,
+      "step": 8501
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.712004610090572e-05,
+      "loss": 0.0073,
+      "step": 8502
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.711939115099804e-05,
+      "loss": 0.8438,
+      "step": 8503
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.711873613453595e-05,
+      "loss": 0.416,
+      "step": 8504
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.7118081051523037e-05,
+      "loss": 0.0,
+      "step": 8505
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.7117425901962902e-05,
+      "loss": 0.0007,
+      "step": 8506
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.711677068585914e-05,
+      "loss": 0.4453,
+      "step": 8507
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.7116115403215354e-05,
+      "loss": 0.0,
+      "step": 8508
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.7115460054035148e-05,
+      "loss": 0.006,
+      "step": 8509
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.7114804638322102e-05,
+      "loss": 0.0,
+      "step": 8510
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.7114149156079833e-05,
+      "loss": 0.0001,
+      "step": 8511
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.7113493607311935e-05,
+      "loss": 0.0442,
+      "step": 8512
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.7112837992022006e-05,
+      "loss": 2.2812,
+      "step": 8513
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.7112182310213645e-05,
+      "loss": 0.0,
+      "step": 8514
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.711152656189046e-05,
+      "loss": 0.0003,
+      "step": 8515
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.7110870747056045e-05,
+      "loss": 0.0,
+      "step": 8516
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.7110214865714007e-05,
+      "loss": 0.0,
+      "step": 8517
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.710955891786794e-05,
+      "loss": 0.0052,
+      "step": 8518
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.7108902903521455e-05,
+      "loss": 0.0002,
+      "step": 8519
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.710824682267815e-05,
+      "loss": 0.3398,
+      "step": 8520
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.710759067534163e-05,
+      "loss": 0.0001,
+      "step": 8521
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.7106934461515492e-05,
+      "loss": 0.0,
+      "step": 8522
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.7106278181203344e-05,
+      "loss": 0.4473,
+      "step": 8523
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.71056218344088e-05,
+      "loss": 0.2236,
+      "step": 8524
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.7104965421135447e-05,
+      "loss": 0.1562,
+      "step": 8525
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.71043089413869e-05,
+      "loss": 0.0,
+      "step": 8526
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.710365239516676e-05,
+      "loss": 0.0,
+      "step": 8527
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.7102995782478636e-05,
+      "loss": 0.1719,
+      "step": 8528
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.710233910332613e-05,
+      "loss": 0.1416,
+      "step": 8529
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.7101682357712848e-05,
+      "loss": 0.1777,
+      "step": 8530
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.71010255456424e-05,
+      "loss": 0.1416,
+      "step": 8531
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.7100368667118394e-05,
+      "loss": 0.0,
+      "step": 8532
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.7099711722144435e-05,
+      "loss": 0.1138,
+      "step": 8533
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.709905471072413e-05,
+      "loss": 0.1582,
+      "step": 8534
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.709839763286109e-05,
+      "loss": 0.5273,
+      "step": 8535
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.7097740488558916e-05,
+      "loss": 0.0767,
+      "step": 8536
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.709708327782122e-05,
+      "loss": 0.5938,
+      "step": 8537
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.709642600065162e-05,
+      "loss": 0.0024,
+      "step": 8538
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.7095768657053715e-05,
+      "loss": 0.4473,
+      "step": 8539
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.709511124703112e-05,
+      "loss": 0.3535,
+      "step": 8540
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.709445377058744e-05,
+      "loss": 0.1021,
+      "step": 8541
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.709379622772629e-05,
+      "loss": 0.2275,
+      "step": 8542
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.7093138618451278e-05,
+      "loss": 0.6602,
+      "step": 8543
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.709248094276602e-05,
+      "loss": 0.0,
+      "step": 8544
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.7091823200674126e-05,
+      "loss": 0.1992,
+      "step": 8545
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.7091165392179206e-05,
+      "loss": 0.1523,
+      "step": 8546
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.7090507517284874e-05,
+      "loss": 0.0106,
+      "step": 8547
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.7089849575994737e-05,
+      "loss": 0.0908,
+      "step": 8548
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.7089191568312416e-05,
+      "loss": 2.2188,
+      "step": 8549
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.7088533494241526e-05,
+      "loss": 0.0005,
+      "step": 8550
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.7087875353785672e-05,
+      "loss": 0.9648,
+      "step": 8551
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.708721714694847e-05,
+      "loss": 0.0618,
+      "step": 8552
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.7086558873733544e-05,
+      "loss": 0.0029,
+      "step": 8553
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.70859005341445e-05,
+      "loss": 0.0004,
+      "step": 8554
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.7085242128184953e-05,
+      "loss": 0.0012,
+      "step": 8555
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.7084583655858524e-05,
+      "loss": 0.0021,
+      "step": 8556
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.7083925117168826e-05,
+      "loss": 0.0005,
+      "step": 8557
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.7083266512119477e-05,
+      "loss": 1.0312,
+      "step": 8558
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.7082607840714088e-05,
+      "loss": 0.0023,
+      "step": 8559
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.7081949102956282e-05,
+      "loss": 0.0003,
+      "step": 8560
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.7081290298849674e-05,
+      "loss": 0.0001,
+      "step": 8561
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.708063142839789e-05,
+      "loss": 0.6211,
+      "step": 8562
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.7079972491604534e-05,
+      "loss": 0.0003,
+      "step": 8563
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.7079313488473237e-05,
+      "loss": 0.0972,
+      "step": 8564
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.707865441900761e-05,
+      "loss": 0.0178,
+      "step": 8565
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.7077995283211273e-05,
+      "loss": 0.4688,
+      "step": 8566
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.7077336081087847e-05,
+      "loss": 0.5352,
+      "step": 8567
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.7076676812640962e-05,
+      "loss": 0.1914,
+      "step": 8568
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.707601747787422e-05,
+      "loss": 0.0,
+      "step": 8569
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.7075358076791255e-05,
+      "loss": 0.0001,
+      "step": 8570
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.7074698609395683e-05,
+      "loss": 0.0001,
+      "step": 8571
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.7074039075691127e-05,
+      "loss": 0.0,
+      "step": 8572
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.7073379475681208e-05,
+      "loss": 0.0,
+      "step": 8573
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.707271980936955e-05,
+      "loss": 0.1074,
+      "step": 8574
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.7072060076759774e-05,
+      "loss": 0.0001,
+      "step": 8575
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.7071400277855504e-05,
+      "loss": 0.7734,
+      "step": 8576
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.7070740412660358e-05,
+      "loss": 0.0,
+      "step": 8577
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.7070080481177973e-05,
+      "loss": 0.0,
+      "step": 8578
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.7069420483411956e-05,
+      "loss": 0.0,
+      "step": 8579
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.7068760419365946e-05,
+      "loss": 0.0,
+      "step": 8580
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.706810028904356e-05,
+      "loss": 0.0293,
+      "step": 8581
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.7067440092448422e-05,
+      "loss": 0.126,
+      "step": 8582
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.706677982958416e-05,
+      "loss": 0.0,
+      "step": 8583
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.7066119500454404e-05,
+      "loss": 0.6211,
+      "step": 8584
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.7065459105062773e-05,
+      "loss": 0.2207,
+      "step": 8585
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.70647986434129e-05,
+      "loss": 0.0034,
+      "step": 8586
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.7064138115508407e-05,
+      "loss": 0.6406,
+      "step": 8587
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.7063477521352922e-05,
+      "loss": 0.9102,
+      "step": 8588
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.706281686095008e-05,
+      "loss": 1.5625,
+      "step": 8589
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.7062156134303497e-05,
+      "loss": 0.0106,
+      "step": 8590
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.7061495341416807e-05,
+      "loss": 0.5078,
+      "step": 8591
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.7060834482293645e-05,
+      "loss": 1.8281,
+      "step": 8592
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.7060173556937632e-05,
+      "loss": 0.3848,
+      "step": 8593
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.7059512565352402e-05,
+      "loss": 0.3926,
+      "step": 8594
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.7058851507541575e-05,
+      "loss": 0.6719,
+      "step": 8595
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.7058190383508797e-05,
+      "loss": 0.0496,
+      "step": 8596
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.7057529193257688e-05,
+      "loss": 0.5898,
+      "step": 8597
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.7056867936791878e-05,
+      "loss": 0.064,
+      "step": 8598
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.7056206614115004e-05,
+      "loss": 0.3672,
+      "step": 8599
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.7055545225230697e-05,
+      "loss": 0.0002,
+      "step": 8600
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.705488377014259e-05,
+      "loss": 0.8164,
+      "step": 8601
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.7054222248854312e-05,
+      "loss": 0.0083,
+      "step": 8602
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.7053560661369494e-05,
+      "loss": 0.0026,
+      "step": 8603
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.7052899007691774e-05,
+      "loss": 0.0,
+      "step": 8604
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.705223728782478e-05,
+      "loss": 0.4102,
+      "step": 8605
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.7051575501772156e-05,
+      "loss": 0.0001,
+      "step": 8606
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.7050913649537527e-05,
+      "loss": 0.0057,
+      "step": 8607
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.705025173112453e-05,
+      "loss": 0.0001,
+      "step": 8608
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.7049589746536797e-05,
+      "loss": 0.0001,
+      "step": 8609
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.7048927695777974e-05,
+      "loss": 0.3066,
+      "step": 8610
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.704826557885168e-05,
+      "loss": 0.0483,
+      "step": 8611
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.704760339576157e-05,
+      "loss": 0.0,
+      "step": 8612
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.7046941146511264e-05,
+      "loss": 0.0718,
+      "step": 8613
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.704627883110441e-05,
+      "loss": 0.0,
+      "step": 8614
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.7045616449544635e-05,
+      "loss": 0.0011,
+      "step": 8615
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.704495400183559e-05,
+      "loss": 0.0001,
+      "step": 8616
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.70442914879809e-05,
+      "loss": 0.0,
+      "step": 8617
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.704362890798421e-05,
+      "loss": 0.0,
+      "step": 8618
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.7042966261849155e-05,
+      "loss": 0.0,
+      "step": 8619
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.704230354957938e-05,
+      "loss": 0.0001,
+      "step": 8620
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.7041640771178512e-05,
+      "loss": 0.0001,
+      "step": 8621
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.7040977926650207e-05,
+      "loss": 0.0001,
+      "step": 8622
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.7040315015998095e-05,
+      "loss": 0.2578,
+      "step": 8623
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.703965203922581e-05,
+      "loss": 0.0001,
+      "step": 8624
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.7038988996337013e-05,
+      "loss": 0.0,
+      "step": 8625
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.703832588733533e-05,
+      "loss": 0.0,
+      "step": 8626
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.7037662712224404e-05,
+      "loss": 0.0001,
+      "step": 8627
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.703699947100788e-05,
+      "loss": 0.0,
+      "step": 8628
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.70363361636894e-05,
+      "loss": 0.2695,
+      "step": 8629
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.70356727902726e-05,
+      "loss": 0.0001,
+      "step": 8630
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.7035009350761137e-05,
+      "loss": 0.1553,
+      "step": 8631
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.703434584515864e-05,
+      "loss": 0.1426,
+      "step": 8632
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.703368227346876e-05,
+      "loss": 0.1963,
+      "step": 8633
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.703301863569514e-05,
+      "loss": 1.1875,
+      "step": 8634
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.7032354931841426e-05,
+      "loss": 0.3711,
+      "step": 8635
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.7031691161911255e-05,
+      "loss": 1.2656,
+      "step": 8636
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.703102732590828e-05,
+      "loss": 0.9375,
+      "step": 8637
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.7030363423836147e-05,
+      "loss": 0.3711,
+      "step": 8638
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.7029699455698495e-05,
+      "loss": 0.3027,
+      "step": 8639
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.7029035421498974e-05,
+      "loss": 0.2256,
+      "step": 8640
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.7028371321241237e-05,
+      "loss": 0.2695,
+      "step": 8641
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.702770715492892e-05,
+      "loss": 1.0078,
+      "step": 8642
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.702704292256568e-05,
+      "loss": 1.2109,
+      "step": 8643
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.7026378624155156e-05,
+      "loss": 0.9062,
+      "step": 8644
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.7025714259701e-05,
+      "loss": 1.0156,
+      "step": 8645
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.702504982920686e-05,
+      "loss": 0.0308,
+      "step": 8646
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.702438533267639e-05,
+      "loss": 0.3691,
+      "step": 8647
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.702372077011323e-05,
+      "loss": 0.0344,
+      "step": 8648
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.7023056141521033e-05,
+      "loss": 1.4766,
+      "step": 8649
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.7022391446903452e-05,
+      "loss": 0.0012,
+      "step": 8650
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.7021726686264136e-05,
+      "loss": 0.0444,
+      "step": 8651
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.7021061859606732e-05,
+      "loss": 0.332,
+      "step": 8652
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.702039696693489e-05,
+      "loss": 0.0913,
+      "step": 8653
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.701973200825227e-05,
+      "loss": 0.1719,
+      "step": 8654
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.701906698356252e-05,
+      "loss": 0.0297,
+      "step": 8655
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.7018401892869283e-05,
+      "loss": 0.0001,
+      "step": 8656
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.7017736736176222e-05,
+      "loss": 1.1484,
+      "step": 8657
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.701707151348699e-05,
+      "loss": 0.3594,
+      "step": 8658
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.7016406224805232e-05,
+      "loss": 0.0001,
+      "step": 8659
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.701574087013461e-05,
+      "loss": 0.0002,
+      "step": 8660
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.7015075449478772e-05,
+      "loss": 0.2617,
+      "step": 8661
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.701440996284137e-05,
+      "loss": 0.0005,
+      "step": 8662
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.701374441022607e-05,
+      "loss": 0.0549,
+      "step": 8663
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.7013078791636515e-05,
+      "loss": 0.054,
+      "step": 8664
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.7012413107076364e-05,
+      "loss": 0.0001,
+      "step": 8665
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.7011747356549276e-05,
+      "loss": 0.2012,
+      "step": 8666
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.7011081540058908e-05,
+      "loss": 0.0003,
+      "step": 8667
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.7010415657608905e-05,
+      "loss": 0.0001,
+      "step": 8668
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.7009749709202937e-05,
+      "loss": 0.0017,
+      "step": 8669
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.7009083694844656e-05,
+      "loss": 0.0001,
+      "step": 8670
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.7008417614537715e-05,
+      "loss": 0.0001,
+      "step": 8671
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.700775146828578e-05,
+      "loss": 0.0,
+      "step": 8672
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.7007085256092504e-05,
+      "loss": 0.0004,
+      "step": 8673
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.7006418977961543e-05,
+      "loss": 0.0002,
+      "step": 8674
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.7005752633896564e-05,
+      "loss": 0.0089,
+      "step": 8675
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.700508622390122e-05,
+      "loss": 0.1079,
+      "step": 8676
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.700441974797918e-05,
+      "loss": 0.0001,
+      "step": 8677
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.700375320613409e-05,
+      "loss": 0.0571,
+      "step": 8678
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.7003086598369614e-05,
+      "loss": 0.1445,
+      "step": 8679
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.700241992468942e-05,
+      "loss": 0.2158,
+      "step": 8680
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.700175318509716e-05,
+      "loss": 0.0005,
+      "step": 8681
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.7001086379596507e-05,
+      "loss": 0.1738,
+      "step": 8682
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.700041950819111e-05,
+      "loss": 0.127,
+      "step": 8683
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.6999752570884642e-05,
+      "loss": 0.8008,
+      "step": 8684
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.699908556768076e-05,
+      "loss": 0.0369,
+      "step": 8685
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.6998418498583127e-05,
+      "loss": 0.0139,
+      "step": 8686
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.69977513635954e-05,
+      "loss": 1.4531,
+      "step": 8687
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.699708416272126e-05,
+      "loss": 0.3457,
+      "step": 8688
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.6996416895964355e-05,
+      "loss": 0.0106,
+      "step": 8689
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.6995749563328352e-05,
+      "loss": 0.3652,
+      "step": 8690
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.6995082164816924e-05,
+      "loss": 0.3945,
+      "step": 8691
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.6994414700433728e-05,
+      "loss": 0.7734,
+      "step": 8692
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.699374717018243e-05,
+      "loss": 0.2021,
+      "step": 8693
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.69930795740667e-05,
+      "loss": 0.0371,
+      "step": 8694
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.6992411912090203e-05,
+      "loss": 1.125,
+      "step": 8695
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.6991744184256603e-05,
+      "loss": 0.0012,
+      "step": 8696
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.699107639056957e-05,
+      "loss": 0.0297,
+      "step": 8697
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.6990408531032768e-05,
+      "loss": 0.1138,
+      "step": 8698
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.6989740605649865e-05,
+      "loss": 0.6484,
+      "step": 8699
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.6989072614424533e-05,
+      "loss": 0.0016,
+      "step": 8700
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.6988404557360434e-05,
+      "loss": 0.0177,
+      "step": 8701
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.698773643446124e-05,
+      "loss": 0.0003,
+      "step": 8702
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.698706824573062e-05,
+      "loss": 0.0349,
+      "step": 8703
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.698639999117225e-05,
+      "loss": 0.4316,
+      "step": 8704
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.698573167078979e-05,
+      "loss": 0.0003,
+      "step": 8705
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.6985063284586908e-05,
+      "loss": 0.0261,
+      "step": 8706
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.6984394832567284e-05,
+      "loss": 2.0469,
+      "step": 8707
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.6983726314734582e-05,
+      "loss": 0.0,
+      "step": 8708
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.698305773109248e-05,
+      "loss": 0.0025,
+      "step": 8709
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.6982389081644643e-05,
+      "loss": 0.8516,
+      "step": 8710
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.6981720366394744e-05,
+      "loss": 0.4121,
+      "step": 8711
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.6981051585346463e-05,
+      "loss": 0.0,
+      "step": 8712
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.6980382738503457e-05,
+      "loss": 0.0001,
+      "step": 8713
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.6979713825869413e-05,
+      "loss": 0.0121,
+      "step": 8714
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.6979044847448e-05,
+      "loss": 0.0212,
+      "step": 8715
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.6978375803242893e-05,
+      "loss": 0.0001,
+      "step": 8716
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.6977706693257762e-05,
+      "loss": 0.3477,
+      "step": 8717
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.6977037517496284e-05,
+      "loss": 0.0001,
+      "step": 8718
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.6976368275962135e-05,
+      "loss": 0.0012,
+      "step": 8719
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.6975698968658992e-05,
+      "loss": 0.5547,
+      "step": 8720
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.6975029595590523e-05,
+      "loss": 0.0002,
+      "step": 8721
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.697436015676041e-05,
+      "loss": 0.0002,
+      "step": 8722
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.6973690652172328e-05,
+      "loss": 0.0104,
+      "step": 8723
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.6973021081829954e-05,
+      "loss": 0.0,
+      "step": 8724
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.697235144573696e-05,
+      "loss": 0.0,
+      "step": 8725
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.6971681743897032e-05,
+      "loss": 0.7148,
+      "step": 8726
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.697101197631384e-05,
+      "loss": 0.1719,
+      "step": 8727
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.697034214299107e-05,
+      "loss": 0.0001,
+      "step": 8728
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.6969672243932394e-05,
+      "loss": 0.1089,
+      "step": 8729
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.6969002279141494e-05,
+      "loss": 0.1191,
+      "step": 8730
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.6968332248622044e-05,
+      "loss": 0.1196,
+      "step": 8731
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.696766215237773e-05,
+      "loss": 0.3184,
+      "step": 8732
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.6966991990412233e-05,
+      "loss": 0.1885,
+      "step": 8733
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.6966321762729222e-05,
+      "loss": 1.2656,
+      "step": 8734
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.696565146933239e-05,
+      "loss": 1.0,
+      "step": 8735
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.6964981110225412e-05,
+      "loss": 0.9609,
+      "step": 8736
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.696431068541197e-05,
+      "loss": 0.3613,
+      "step": 8737
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.6963640194895745e-05,
+      "loss": 0.0854,
+      "step": 8738
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.6962969638680424e-05,
+      "loss": 0.252,
+      "step": 8739
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.6962299016769685e-05,
+      "loss": 0.7227,
+      "step": 8740
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.6961628329167204e-05,
+      "loss": 0.9336,
+      "step": 8741
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.696095757587668e-05,
+      "loss": 1.2422,
+      "step": 8742
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.6960286756901786e-05,
+      "loss": 0.1079,
+      "step": 8743
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.6959615872246204e-05,
+      "loss": 0.7891,
+      "step": 8744
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.6958944921913625e-05,
+      "loss": 0.1709,
+      "step": 8745
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.695827390590773e-05,
+      "loss": 0.1182,
+      "step": 8746
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.6957602824232202e-05,
+      "loss": 0.209,
+      "step": 8747
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.6956931676890734e-05,
+      "loss": 0.6797,
+      "step": 8748
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.6956260463887e-05,
+      "loss": 0.0005,
+      "step": 8749
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.6955589185224696e-05,
+      "loss": 0.1118,
+      "step": 8750
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.6954917840907505e-05,
+      "loss": 0.0032,
+      "step": 8751
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.695424643093911e-05,
+      "loss": 0.1855,
+      "step": 8752
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.6953574955323204e-05,
+      "loss": 0.291,
+      "step": 8753
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.6952903414063473e-05,
+      "loss": 0.0077,
+      "step": 8754
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.6952231807163605e-05,
+      "loss": 1.7812,
+      "step": 8755
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.6951560134627283e-05,
+      "loss": 0.017,
+      "step": 8756
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.6950888396458203e-05,
+      "loss": 1.0938,
+      "step": 8757
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.695021659266005e-05,
+      "loss": 0.9492,
+      "step": 8758
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.694954472323651e-05,
+      "loss": 0.0,
+      "step": 8759
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.694887278819128e-05,
+      "loss": 0.4785,
+      "step": 8760
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.6948200787528046e-05,
+      "loss": 0.8789,
+      "step": 8761
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.69475287212505e-05,
+      "loss": 0.0114,
+      "step": 8762
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.6946856589362324e-05,
+      "loss": 0.0001,
+      "step": 8763
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.6946184391867226e-05,
+      "loss": 0.0635,
+      "step": 8764
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.6945512128768884e-05,
+      "loss": 0.0947,
+      "step": 8765
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.694483980007099e-05,
+      "loss": 0.0004,
+      "step": 8766
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.6944167405777244e-05,
+      "loss": 0.1143,
+      "step": 8767
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.6943494945891336e-05,
+      "loss": 0.0002,
+      "step": 8768
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.6942822420416952e-05,
+      "loss": 0.0,
+      "step": 8769
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.694214982935779e-05,
+      "loss": 0.0001,
+      "step": 8770
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.6941477172717546e-05,
+      "loss": 0.0016,
+      "step": 8771
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.6940804450499913e-05,
+      "loss": 0.0002,
+      "step": 8772
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.6940131662708584e-05,
+      "loss": 0.0,
+      "step": 8773
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.6939458809347252e-05,
+      "loss": 0.0,
+      "step": 8774
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.6938785890419614e-05,
+      "loss": 0.0189,
+      "step": 8775
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.6938112905929366e-05,
+      "loss": 0.0,
+      "step": 8776
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.69374398558802e-05,
+      "loss": 0.0923,
+      "step": 8777
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.6936766740275814e-05,
+      "loss": 0.1357,
+      "step": 8778
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.693609355911991e-05,
+      "loss": 0.0,
+      "step": 8779
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.6935420312416184e-05,
+      "loss": 0.0928,
+      "step": 8780
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.693474700016832e-05,
+      "loss": 0.0,
+      "step": 8781
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.693407362238003e-05,
+      "loss": 0.1128,
+      "step": 8782
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.6933400179055005e-05,
+      "loss": 0.1631,
+      "step": 8783
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.693272667019695e-05,
+      "loss": 0.4863,
+      "step": 8784
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.6932053095809554e-05,
+      "loss": 0.0864,
+      "step": 8785
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.693137945589652e-05,
+      "loss": 0.2832,
+      "step": 8786
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.6930705750461552e-05,
+      "loss": 1.0391,
+      "step": 8787
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.693003197950834e-05,
+      "loss": 0.5234,
+      "step": 8788
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.6929358143040597e-05,
+      "loss": 0.1328,
+      "step": 8789
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.692868424106201e-05,
+      "loss": 0.0192,
+      "step": 8790
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.6928010273576288e-05,
+      "loss": 0.7188,
+      "step": 8791
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.692733624058713e-05,
+      "loss": 0.1797,
+      "step": 8792
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.6926662142098237e-05,
+      "loss": 0.3477,
+      "step": 8793
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.6925987978113312e-05,
+      "loss": 1.5156,
+      "step": 8794
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.6925313748636057e-05,
+      "loss": 0.3027,
+      "step": 8795
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.6924639453670175e-05,
+      "loss": 0.0332,
+      "step": 8796
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.6923965093219367e-05,
+      "loss": 0.9141,
+      "step": 8797
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.6923290667287338e-05,
+      "loss": 0.1455,
+      "step": 8798
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.6922616175877795e-05,
+      "loss": 0.2773,
+      "step": 8799
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.692194161899443e-05,
+      "loss": 0.0003,
+      "step": 8800
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.6921266996640964e-05,
+      "loss": 0.1338,
+      "step": 8801
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.6920592308821086e-05,
+      "loss": 0.5195,
+      "step": 8802
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.6919917555538508e-05,
+      "loss": 0.0,
+      "step": 8803
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.691924273679694e-05,
+      "loss": 0.0003,
+      "step": 8804
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.6918567852600088e-05,
+      "loss": 0.3496,
+      "step": 8805
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.6917892902951647e-05,
+      "loss": 0.1167,
+      "step": 8806
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.691721788785533e-05,
+      "loss": 0.0004,
+      "step": 8807
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.691654280731485e-05,
+      "loss": 0.1338,
+      "step": 8808
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.6915867661333903e-05,
+      "loss": 0.0,
+      "step": 8809
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.6915192449916204e-05,
+      "loss": 0.0001,
+      "step": 8810
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.6914517173065458e-05,
+      "loss": 0.0,
+      "step": 8811
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.6913841830785377e-05,
+      "loss": 0.0845,
+      "step": 8812
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.691316642307966e-05,
+      "loss": 0.5859,
+      "step": 8813
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.691249094995203e-05,
+      "loss": 0.1143,
+      "step": 8814
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.691181541140619e-05,
+      "loss": 0.0,
+      "step": 8815
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.6911139807445843e-05,
+      "loss": 0.0,
+      "step": 8816
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.691046413807471e-05,
+      "loss": 0.0,
+      "step": 8817
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.6909788403296494e-05,
+      "loss": 0.0,
+      "step": 8818
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.6909112603114915e-05,
+      "loss": 0.0001,
+      "step": 8819
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.6908436737533674e-05,
+      "loss": 0.2656,
+      "step": 8820
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.6907760806556482e-05,
+      "loss": 0.0,
+      "step": 8821
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.6907084810187063e-05,
+      "loss": 0.0025,
+      "step": 8822
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.6906408748429116e-05,
+      "loss": 0.0391,
+      "step": 8823
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.690573262128636e-05,
+      "loss": 0.0,
+      "step": 8824
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.6905056428762512e-05,
+      "loss": 0.0,
+      "step": 8825
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.6904380170861275e-05,
+      "loss": 0.0001,
+      "step": 8826
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.690370384758637e-05,
+      "loss": 0.0001,
+      "step": 8827
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.690302745894151e-05,
+      "loss": 0.1221,
+      "step": 8828
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.690235100493041e-05,
+      "loss": 0.2432,
+      "step": 8829
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.690167448555678e-05,
+      "loss": 0.1748,
+      "step": 8830
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.690099790082434e-05,
+      "loss": 0.0889,
+      "step": 8831
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.6900321250736807e-05,
+      "loss": 0.1465,
+      "step": 8832
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.6899644535297897e-05,
+      "loss": 0.1875,
+      "step": 8833
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.6898967754511316e-05,
+      "loss": 1.4297,
+      "step": 8834
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.689829090838079e-05,
+      "loss": 0.2617,
+      "step": 8835
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.6897613996910037e-05,
+      "loss": 0.4121,
+      "step": 8836
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.689693702010277e-05,
+      "loss": 0.8711,
+      "step": 8837
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.6896259977962702e-05,
+      "loss": 0.0197,
+      "step": 8838
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.6895582870493566e-05,
+      "loss": 0.7188,
+      "step": 8839
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.689490569769907e-05,
+      "loss": 0.0001,
+      "step": 8840
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.6894228459582927e-05,
+      "loss": 0.6016,
+      "step": 8841
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.689355115614887e-05,
+      "loss": 0.8008,
+      "step": 8842
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.6892873787400605e-05,
+      "loss": 1.0703,
+      "step": 8843
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.6892196353341863e-05,
+      "loss": 1.0781,
+      "step": 8844
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.6891518853976362e-05,
+      "loss": 0.3887,
+      "step": 8845
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.6890841289307815e-05,
+      "loss": 0.5156,
+      "step": 8846
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.689016365933995e-05,
+      "loss": 0.0082,
+      "step": 8847
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.6889485964076488e-05,
+      "loss": 0.0243,
+      "step": 8848
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.6888808203521144e-05,
+      "loss": 0.0287,
+      "step": 8849
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.6888130377677652e-05,
+      "loss": 0.0039,
+      "step": 8850
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.688745248654972e-05,
+      "loss": 0.332,
+      "step": 8851
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.688677453014108e-05,
+      "loss": 0.0439,
+      "step": 8852
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.6886096508455456e-05,
+      "loss": 0.793,
+      "step": 8853
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.6885418421496562e-05,
+      "loss": 0.0,
+      "step": 8854
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.6884740269268135e-05,
+      "loss": 0.3223,
+      "step": 8855
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.688406205177389e-05,
+      "loss": 0.6484,
+      "step": 8856
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.6883383769017548e-05,
+      "loss": 0.0,
+      "step": 8857
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.688270542100285e-05,
+      "loss": 0.0,
+      "step": 8858
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.68820270077335e-05,
+      "loss": 0.3633,
+      "step": 8859
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.688134852921324e-05,
+      "loss": 0.0273,
+      "step": 8860
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.6880669985445788e-05,
+      "loss": 0.082,
+      "step": 8861
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.6879991376434875e-05,
+      "loss": 0.0,
+      "step": 8862
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.687931270218422e-05,
+      "loss": 0.0845,
+      "step": 8863
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.6878633962697564e-05,
+      "loss": 0.0001,
+      "step": 8864
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.6877955157978617e-05,
+      "loss": 0.0001,
+      "step": 8865
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.687727628803112e-05,
+      "loss": 0.0233,
+      "step": 8866
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.687659735285879e-05,
+      "loss": 0.0005,
+      "step": 8867
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.6875918352465365e-05,
+      "loss": 0.0001,
+      "step": 8868
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.687523928685457e-05,
+      "loss": 0.0,
+      "step": 8869
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.6874560156030136e-05,
+      "loss": 0.0001,
+      "step": 8870
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.687388095999579e-05,
+      "loss": 0.1021,
+      "step": 8871
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.6873201698755264e-05,
+      "loss": 0.0002,
+      "step": 8872
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.6872522372312286e-05,
+      "loss": 0.6719,
+      "step": 8873
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.6871842980670583e-05,
+      "loss": 0.0,
+      "step": 8874
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.6871163523833892e-05,
+      "loss": 0.083,
+      "step": 8875
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.6870484001805945e-05,
+      "loss": 0.0,
+      "step": 8876
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.686980441459047e-05,
+      "loss": 0.0001,
+      "step": 8877
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.6869124762191206e-05,
+      "loss": 0.1025,
+      "step": 8878
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.686844504461187e-05,
+      "loss": 0.0,
+      "step": 8879
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.6867765261856212e-05,
+      "loss": 0.0869,
+      "step": 8880
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.6867085413927957e-05,
+      "loss": 0.1572,
+      "step": 8881
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.6866405500830837e-05,
+      "loss": 0.1475,
+      "step": 8882
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.686572552256859e-05,
+      "loss": 0.1582,
+      "step": 8883
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.6865045479144943e-05,
+      "loss": 0.6797,
+      "step": 8884
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.6864365370563637e-05,
+      "loss": 0.5,
+      "step": 8885
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.6863685196828404e-05,
+      "loss": 0.6875,
+      "step": 8886
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.6863004957942985e-05,
+      "loss": 0.033,
+      "step": 8887
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.6862324653911107e-05,
+      "loss": 0.25,
+      "step": 8888
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.6861644284736513e-05,
+      "loss": 0.0615,
+      "step": 8889
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.6860963850422934e-05,
+      "loss": 0.4766,
+      "step": 8890
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.686028335097411e-05,
+      "loss": 0.4961,
+      "step": 8891
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.6859602786393772e-05,
+      "loss": 0.2676,
+      "step": 8892
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.685892215668567e-05,
+      "loss": 0.2129,
+      "step": 8893
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.6858241461853532e-05,
+      "loss": 0.1445,
+      "step": 8894
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.6857560701901095e-05,
+      "loss": 1.25,
+      "step": 8895
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.68568798768321e-05,
+      "loss": 0.0045,
+      "step": 8896
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.6856198986650293e-05,
+      "loss": 0.2793,
+      "step": 8897
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.6855518031359396e-05,
+      "loss": 0.084,
+      "step": 8898
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.6854837010963168e-05,
+      "loss": 0.1611,
+      "step": 8899
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.6854155925465333e-05,
+      "loss": 0.0,
+      "step": 8900
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.6853474774869642e-05,
+      "loss": 0.2246,
+      "step": 8901
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.685279355917983e-05,
+      "loss": 0.0576,
+      "step": 8902
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.685211227839964e-05,
+      "loss": 0.0232,
+      "step": 8903
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.6851430932532816e-05,
+      "loss": 1.9219,
+      "step": 8904
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.6850749521583093e-05,
+      "loss": 0.4805,
+      "step": 8905
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.685006804555422e-05,
+      "loss": 0.0,
+      "step": 8906
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.6849386504449932e-05,
+      "loss": 0.0177,
+      "step": 8907
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.6848704898273975e-05,
+      "loss": 0.0001,
+      "step": 8908
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.6848023227030092e-05,
+      "loss": 0.0084,
+      "step": 8909
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.6847341490722035e-05,
+      "loss": 0.0364,
+      "step": 8910
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.6846659689353532e-05,
+      "loss": 0.0,
+      "step": 8911
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.6845977822928338e-05,
+      "loss": 0.0,
+      "step": 8912
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.6845295891450198e-05,
+      "loss": 0.25,
+      "step": 8913
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.6844613894922844e-05,
+      "loss": 0.0001,
+      "step": 8914
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.684393183335004e-05,
+      "loss": 0.024,
+      "step": 8915
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.6843249706735517e-05,
+      "loss": 0.0,
+      "step": 8916
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.6842567515083028e-05,
+      "loss": 0.0005,
+      "step": 8917
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.684188525839632e-05,
+      "loss": 0.9648,
+      "step": 8918
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.684120293667913e-05,
+      "loss": 0.0001,
+      "step": 8919
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.684052054993522e-05,
+      "loss": 0.0001,
+      "step": 8920
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.6839838098168324e-05,
+      "loss": 0.0,
+      "step": 8921
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.68391555813822e-05,
+      "loss": 0.0,
+      "step": 8922
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.6838472999580587e-05,
+      "loss": 0.0283,
+      "step": 8923
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.6837790352767242e-05,
+      "loss": 0.7539,
+      "step": 8924
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.6837107640945904e-05,
+      "loss": 0.0104,
+      "step": 8925
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.683642486412033e-05,
+      "loss": 0.1138,
+      "step": 8926
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.6835742022294273e-05,
+      "loss": 0.0,
+      "step": 8927
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.683505911547147e-05,
+      "loss": 0.1299,
+      "step": 8928
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.683437614365568e-05,
+      "loss": 0.1079,
+      "step": 8929
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.6833693106850656e-05,
+      "loss": 0.1396,
+      "step": 8930
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.683301000506014e-05,
+      "loss": 0.1543,
+      "step": 8931
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.683232683828789e-05,
+      "loss": 0.1011,
+      "step": 8932
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.6831643606537654e-05,
+      "loss": 0.1553,
+      "step": 8933
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.6830960309813188e-05,
+      "loss": 0.918,
+      "step": 8934
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.6830276948118243e-05,
+      "loss": 0.5312,
+      "step": 8935
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.6829593521456568e-05,
+      "loss": 0.8789,
+      "step": 8936
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.682891002983192e-05,
+      "loss": 0.2676,
+      "step": 8937
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.6828226473248054e-05,
+      "loss": 0.0001,
+      "step": 8938
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.6827542851708723e-05,
+      "loss": 0.3691,
+      "step": 8939
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.6826859165217673e-05,
+      "loss": 0.0003,
+      "step": 8940
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.682617541377867e-05,
+      "loss": 1.1094,
+      "step": 8941
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.6825491597395463e-05,
+      "loss": 0.1816,
+      "step": 8942
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.6824807716071805e-05,
+      "loss": 0.0008,
+      "step": 8943
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.6824123769811456e-05,
+      "loss": 0.4238,
+      "step": 8944
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.6823439758618172e-05,
+      "loss": 1.2891,
+      "step": 8945
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.6822755682495707e-05,
+      "loss": 0.8672,
+      "step": 8946
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.6822071541447816e-05,
+      "loss": 1.0156,
+      "step": 8947
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.6821387335478263e-05,
+      "loss": 0.0002,
+      "step": 8948
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.6820703064590796e-05,
+      "loss": 0.0001,
+      "step": 8949
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.6820018728789184e-05,
+      "loss": 0.418,
+      "step": 8950
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.6819334328077174e-05,
+      "loss": 0.0,
+      "step": 8951
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.6818649862458527e-05,
+      "loss": 0.0,
+      "step": 8952
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.681796533193701e-05,
+      "loss": 0.0037,
+      "step": 8953
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.681728073651637e-05,
+      "loss": 0.4082,
+      "step": 8954
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.6816596076200377e-05,
+      "loss": 1.1719,
+      "step": 8955
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.6815911350992783e-05,
+      "loss": 0.4902,
+      "step": 8956
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.6815226560897355e-05,
+      "loss": 0.0,
+      "step": 8957
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.681454170591785e-05,
+      "loss": 0.1953,
+      "step": 8958
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.6813856786058024e-05,
+      "loss": 0.0022,
+      "step": 8959
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.681317180132165e-05,
+      "loss": 0.0,
+      "step": 8960
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.6812486751712474e-05,
+      "loss": 0.0,
+      "step": 8961
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.6811801637234276e-05,
+      "loss": 0.0,
+      "step": 8962
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.6811116457890804e-05,
+      "loss": 0.0,
+      "step": 8963
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.6810431213685828e-05,
+      "loss": 0.0004,
+      "step": 8964
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.680974590462311e-05,
+      "loss": 0.3184,
+      "step": 8965
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.6809060530706414e-05,
+      "loss": 0.0,
+      "step": 8966
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.6808375091939494e-05,
+      "loss": 0.0,
+      "step": 8967
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.680768958832613e-05,
+      "loss": 0.0,
+      "step": 8968
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.6807004019870075e-05,
+      "loss": 0.0,
+      "step": 8969
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.6806318386575104e-05,
+      "loss": 0.0,
+      "step": 8970
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.680563268844497e-05,
+      "loss": 0.0079,
+      "step": 8971
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.6804946925483445e-05,
+      "loss": 0.0276,
+      "step": 8972
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.6804261097694297e-05,
+      "loss": 0.0,
+      "step": 8973
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.680357520508129e-05,
+      "loss": 0.1914,
+      "step": 8974
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.6802889247648185e-05,
+      "loss": 0.167,
+      "step": 8975
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.6802203225398758e-05,
+      "loss": 0.0,
+      "step": 8976
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.6801517138336774e-05,
+      "loss": 0.0,
+      "step": 8977
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.6800830986466e-05,
+      "loss": 0.0001,
+      "step": 8978
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.68001447697902e-05,
+      "loss": 0.1094,
+      "step": 8979
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.6799458488313144e-05,
+      "loss": 0.1245,
+      "step": 8980
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.6798772142038608e-05,
+      "loss": 0.0879,
+      "step": 8981
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.6798085730970354e-05,
+      "loss": 0.0,
+      "step": 8982
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.679739925511215e-05,
+      "loss": 0.0,
+      "step": 8983
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.679671271446777e-05,
+      "loss": 0.6445,
+      "step": 8984
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.6796026109040987e-05,
+      "loss": 0.8477,
+      "step": 8985
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.6795339438835563e-05,
+      "loss": 0.8086,
+      "step": 8986
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.6794652703855277e-05,
+      "loss": 1.1328,
+      "step": 8987
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.6793965904103895e-05,
+      "loss": 0.3691,
+      "step": 8988
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.679327903958519e-05,
+      "loss": 0.9062,
+      "step": 8989
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.6792592110302937e-05,
+      "loss": 0.2559,
+      "step": 8990
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.6791905116260904e-05,
+      "loss": 0.1377,
+      "step": 8991
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.6791218057462862e-05,
+      "loss": 0.0099,
+      "step": 8992
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.679053093391259e-05,
+      "loss": 0.0003,
+      "step": 8993
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.6789843745613857e-05,
+      "loss": 0.1758,
+      "step": 8994
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.6789156492570442e-05,
+      "loss": 0.9297,
+      "step": 8995
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.6788469174786117e-05,
+      "loss": 0.0269,
+      "step": 8996
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.678778179226465e-05,
+      "loss": 0.0036,
+      "step": 8997
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.6787094345009825e-05,
+      "loss": 0.5742,
+      "step": 8998
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.678640683302541e-05,
+      "loss": 0.7422,
+      "step": 8999
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.678571925631518e-05,
+      "loss": 0.0001,
+      "step": 9000
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.678503161488292e-05,
+      "loss": 0.0156,
+      "step": 9001
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.67843439087324e-05,
+      "loss": 0.0011,
+      "step": 9002
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.67836561378674e-05,
+      "loss": 0.0,
+      "step": 9003
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.678296830229169e-05,
+      "loss": 0.0011,
+      "step": 9004
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.6782280402009055e-05,
+      "loss": 0.0067,
+      "step": 9005
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.678159243702326e-05,
+      "loss": 0.0001,
+      "step": 9006
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.6780904407338104e-05,
+      "loss": 0.0004,
+      "step": 9007
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.6780216312957344e-05,
+      "loss": 0.0007,
+      "step": 9008
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.677952815388477e-05,
+      "loss": 0.668,
+      "step": 9009
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.6778839930124164e-05,
+      "loss": 0.0004,
+      "step": 9010
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.6778151641679296e-05,
+      "loss": 0.0527,
+      "step": 9011
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.6777463288553955e-05,
+      "loss": 0.064,
+      "step": 9012
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.6776774870751913e-05,
+      "loss": 0.0036,
+      "step": 9013
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.6776086388276953e-05,
+      "loss": 0.0,
+      "step": 9014
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.677539784113286e-05,
+      "loss": 0.0,
+      "step": 9015
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.6774709229323414e-05,
+      "loss": 0.0,
+      "step": 9016
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.677402055285239e-05,
+      "loss": 0.2617,
+      "step": 9017
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.6773331811723574e-05,
+      "loss": 0.0869,
+      "step": 9018
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.6772643005940756e-05,
+      "loss": 0.0,
+      "step": 9019
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.6771954135507705e-05,
+      "loss": 0.0,
+      "step": 9020
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.6771265200428212e-05,
+      "loss": 0.0,
+      "step": 9021
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.6770576200706058e-05,
+      "loss": 0.0052,
+      "step": 9022
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.6769887136345027e-05,
+      "loss": 0.0,
+      "step": 9023
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.6769198007348904e-05,
+      "loss": 0.0,
+      "step": 9024
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.6768508813721474e-05,
+      "loss": 0.0,
+      "step": 9025
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.676781955546652e-05,
+      "loss": 0.0,
+      "step": 9026
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.6767130232587826e-05,
+      "loss": 0.0,
+      "step": 9027
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.6766440845089184e-05,
+      "loss": 0.1211,
+      "step": 9028
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.6765751392974373e-05,
+      "loss": 0.1099,
+      "step": 9029
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.676506187624718e-05,
+      "loss": 0.0,
+      "step": 9030
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.676437229491139e-05,
+      "loss": 0.1562,
+      "step": 9031
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.67636826489708e-05,
+      "loss": 0.1426,
+      "step": 9032
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.6762992938429184e-05,
+      "loss": 0.1416,
+      "step": 9033
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.6762303163290335e-05,
+      "loss": 0.8672,
+      "step": 9034
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.6761613323558045e-05,
+      "loss": 2.5469,
+      "step": 9035
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.6760923419236097e-05,
+      "loss": 0.1138,
+      "step": 9036
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.676023345032828e-05,
+      "loss": 1.8125,
+      "step": 9037
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.6759543416838386e-05,
+      "loss": 0.6484,
+      "step": 9038
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.67588533187702e-05,
+      "loss": 1.2266,
+      "step": 9039
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.675816315612752e-05,
+      "loss": 0.3652,
+      "step": 9040
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.6757472928914127e-05,
+      "loss": 0.0292,
+      "step": 9041
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.6756782637133814e-05,
+      "loss": 0.4121,
+      "step": 9042
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.6756092280790373e-05,
+      "loss": 0.6797,
+      "step": 9043
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.6755401859887598e-05,
+      "loss": 0.1416,
+      "step": 9044
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.6754711374429274e-05,
+      "loss": 0.1797,
+      "step": 9045
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.6754020824419197e-05,
+      "loss": 0.7891,
+      "step": 9046
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.675333020986116e-05,
+      "loss": 0.0001,
+      "step": 9047
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.675263953075895e-05,
+      "loss": 0.1963,
+      "step": 9048
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.675194878711637e-05,
+      "loss": 0.0003,
+      "step": 9049
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.6751257978937202e-05,
+      "loss": 0.0109,
+      "step": 9050
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.6750567106225244e-05,
+      "loss": 0.0,
+      "step": 9051
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.6749876168984293e-05,
+      "loss": 0.0001,
+      "step": 9052
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.6749185167218145e-05,
+      "loss": 0.0075,
+      "step": 9053
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.6748494100930583e-05,
+      "loss": 0.0,
+      "step": 9054
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.6747802970125412e-05,
+      "loss": 0.0001,
+      "step": 9055
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.674711177480643e-05,
+      "loss": 0.0,
+      "step": 9056
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.674642051497742e-05,
+      "loss": 0.0232,
+      "step": 9057
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.674572919064219e-05,
+      "loss": 0.6289,
+      "step": 9058
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.6745037801804532e-05,
+      "loss": 0.0,
+      "step": 9059
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.6744346348468246e-05,
+      "loss": 0.0,
+      "step": 9060
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.6743654830637127e-05,
+      "loss": 0.0,
+      "step": 9061
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.6742963248314966e-05,
+      "loss": 0.0157,
+      "step": 9062
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.6742271601505574e-05,
+      "loss": 0.0254,
+      "step": 9063
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.6741579890212736e-05,
+      "loss": 0.0,
+      "step": 9064
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.6740888114440257e-05,
+      "loss": 0.0,
+      "step": 9065
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.6740196274191935e-05,
+      "loss": 0.0105,
+      "step": 9066
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.6739504369471575e-05,
+      "loss": 0.0007,
+      "step": 9067
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.6738812400282965e-05,
+      "loss": 0.0,
+      "step": 9068
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.6738120366629916e-05,
+      "loss": 0.0,
+      "step": 9069
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.6737428268516223e-05,
+      "loss": 0.0,
+      "step": 9070
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.6736736105945685e-05,
+      "loss": 0.0003,
+      "step": 9071
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.673604387892211e-05,
+      "loss": 0.0,
+      "step": 9072
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.673535158744929e-05,
+      "loss": 0.0,
+      "step": 9073
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.673465923153103e-05,
+      "loss": 0.0,
+      "step": 9074
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.6733966811171143e-05,
+      "loss": 0.0947,
+      "step": 9075
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.6733274326373415e-05,
+      "loss": 0.0001,
+      "step": 9076
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.6732581777141657e-05,
+      "loss": 0.1094,
+      "step": 9077
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.6731889163479673e-05,
+      "loss": 0.0952,
+      "step": 9078
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.6731196485391266e-05,
+      "loss": 0.0,
+      "step": 9079
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.6730503742880234e-05,
+      "loss": 0.0,
+      "step": 9080
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.6729810935950387e-05,
+      "loss": 0.1445,
+      "step": 9081
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.6729118064605534e-05,
+      "loss": 0.126,
+      "step": 9082
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.6728425128849468e-05,
+      "loss": 0.2773,
+      "step": 9083
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.6727732128686005e-05,
+      "loss": 0.3164,
+      "step": 9084
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.6727039064118944e-05,
+      "loss": 0.7266,
+      "step": 9085
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.6726345935152094e-05,
+      "loss": 1.1875,
+      "step": 9086
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.6725652741789262e-05,
+      "loss": 0.0092,
+      "step": 9087
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.6724959484034254e-05,
+      "loss": 0.5352,
+      "step": 9088
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.672426616189087e-05,
+      "loss": 0.3418,
+      "step": 9089
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.672357277536293e-05,
+      "loss": 0.0986,
+      "step": 9090
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.6722879324454236e-05,
+      "loss": 0.7617,
+      "step": 9091
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.6722185809168598e-05,
+      "loss": 0.457,
+      "step": 9092
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.6721492229509814e-05,
+      "loss": 0.0007,
+      "step": 9093
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.672079858548171e-05,
+      "loss": 0.6914,
+      "step": 9094
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.672010487708808e-05,
+      "loss": 1.0859,
+      "step": 9095
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.6719411104332745e-05,
+      "loss": 0.0004,
+      "step": 9096
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.6718717267219507e-05,
+      "loss": 0.2617,
+      "step": 9097
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.671802336575218e-05,
+      "loss": 1.1328,
+      "step": 9098
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.671732939993457e-05,
+      "loss": 0.3652,
+      "step": 9099
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.67166353697705e-05,
+      "loss": 0.6562,
+      "step": 9100
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.6715941275263768e-05,
+      "loss": 0.0009,
+      "step": 9101
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.671524711641819e-05,
+      "loss": 0.3633,
+      "step": 9102
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.6714552893237584e-05,
+      "loss": 0.0001,
+      "step": 9103
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.671385860572575e-05,
+      "loss": 0.1582,
+      "step": 9104
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.671316425388651e-05,
+      "loss": 0.0135,
+      "step": 9105
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.6712469837723676e-05,
+      "loss": 0.0012,
+      "step": 9106
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.6711775357241057e-05,
+      "loss": 0.0042,
+      "step": 9107
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.6711080812442474e-05,
+      "loss": 0.0284,
+      "step": 9108
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.671038620333174e-05,
+      "loss": 2.2812,
+      "step": 9109
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.670969152991266e-05,
+      "loss": 1.0625,
+      "step": 9110
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.670899679218906e-05,
+      "loss": 0.0,
+      "step": 9111
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.6708301990164747e-05,
+      "loss": 0.0,
+      "step": 9112
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.6707607123843543e-05,
+      "loss": 0.126,
+      "step": 9113
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.670691219322926e-05,
+      "loss": 0.0,
+      "step": 9114
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.6706217198325716e-05,
+      "loss": 0.0035,
+      "step": 9115
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.6705522139136726e-05,
+      "loss": 0.0,
+      "step": 9116
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.6704827015666108e-05,
+      "loss": 0.0,
+      "step": 9117
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.670413182791768e-05,
+      "loss": 0.0,
+      "step": 9118
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.670343657589526e-05,
+      "loss": 0.0006,
+      "step": 9119
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.6702741259602662e-05,
+      "loss": 0.3086,
+      "step": 9120
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.670204587904371e-05,
+      "loss": 0.1807,
+      "step": 9121
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.6701350434222223e-05,
+      "loss": 0.0,
+      "step": 9122
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.6700654925142013e-05,
+      "loss": 0.0549,
+      "step": 9123
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.6699959351806908e-05,
+      "loss": 0.0,
+      "step": 9124
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.6699263714220715e-05,
+      "loss": 0.0,
+      "step": 9125
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.669856801238727e-05,
+      "loss": 0.0,
+      "step": 9126
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.6697872246310385e-05,
+      "loss": 0.1387,
+      "step": 9127
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.669717641599388e-05,
+      "loss": 0.0,
+      "step": 9128
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.6696480521441576e-05,
+      "loss": 0.1445,
+      "step": 9129
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.66957845626573e-05,
+      "loss": 0.0,
+      "step": 9130
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.669508853964487e-05,
+      "loss": 0.1621,
+      "step": 9131
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.6694392452408108e-05,
+      "loss": 0.209,
+      "step": 9132
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.669369630095084e-05,
+      "loss": 0.1226,
+      "step": 9133
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.6693000085276885e-05,
+      "loss": 1.4453,
+      "step": 9134
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.6692303805390063e-05,
+      "loss": 0.4277,
+      "step": 9135
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.6691607461294207e-05,
+      "loss": 0.375,
+      "step": 9136
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.6690911052993138e-05,
+      "loss": 0.3047,
+      "step": 9137
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.6690214580490675e-05,
+      "loss": 0.0039,
+      "step": 9138
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.668951804379065e-05,
+      "loss": 0.0245,
+      "step": 9139
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.6688821442896884e-05,
+      "loss": 0.4004,
+      "step": 9140
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.6688124777813196e-05,
+      "loss": 2.0469,
+      "step": 9141
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.668742804854343e-05,
+      "loss": 1.4375,
+      "step": 9142
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.6686731255091393e-05,
+      "loss": 2.0156,
+      "step": 9143
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.6686034397460918e-05,
+      "loss": 0.498,
+      "step": 9144
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.668533747565584e-05,
+      "loss": 0.0117,
+      "step": 9145
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.6684640489679977e-05,
+      "loss": 0.9062,
+      "step": 9146
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.6683943439537155e-05,
+      "loss": 1.0,
+      "step": 9147
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.668324632523121e-05,
+      "loss": 0.0021,
+      "step": 9148
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.6682549146765966e-05,
+      "loss": 0.3164,
+      "step": 9149
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.6681851904145247e-05,
+      "loss": 0.0,
+      "step": 9150
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.6681154597372892e-05,
+      "loss": 0.9375,
+      "step": 9151
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.6680457226452726e-05,
+      "loss": 0.0001,
+      "step": 9152
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.667975979138857e-05,
+      "loss": 0.5625,
+      "step": 9153
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.6679062292184272e-05,
+      "loss": 0.0002,
+      "step": 9154
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.6678364728843645e-05,
+      "loss": 0.0,
+      "step": 9155
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.6677667101370525e-05,
+      "loss": 0.334,
+      "step": 9156
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.667696940976875e-05,
+      "loss": 0.3125,
+      "step": 9157
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.6676271654042146e-05,
+      "loss": 0.0001,
+      "step": 9158
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.6675573834194542e-05,
+      "loss": 0.0,
+      "step": 9159
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.6674875950229772e-05,
+      "loss": 0.0,
+      "step": 9160
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.6674178002151673e-05,
+      "loss": 0.0,
+      "step": 9161
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.6673479989964072e-05,
+      "loss": 0.0,
+      "step": 9162
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.6672781913670806e-05,
+      "loss": 0.0001,
+      "step": 9163
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.6672083773275705e-05,
+      "loss": 0.0654,
+      "step": 9164
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.667138556878261e-05,
+      "loss": 0.0,
+      "step": 9165
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.6670687300195344e-05,
+      "loss": 0.0,
+      "step": 9166
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.6669988967517754e-05,
+      "loss": 0.0,
+      "step": 9167
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.666929057075366e-05,
+      "loss": 0.0522,
+      "step": 9168
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.666859210990691e-05,
+      "loss": 0.0,
+      "step": 9169
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.6667893584981336e-05,
+      "loss": 0.0,
+      "step": 9170
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.666719499598078e-05,
+      "loss": 0.0,
+      "step": 9171
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.6666496342909066e-05,
+      "loss": 0.0,
+      "step": 9172
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.6665797625770035e-05,
+      "loss": 0.0,
+      "step": 9173
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.666509884456753e-05,
+      "loss": 0.0,
+      "step": 9174
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.666439999930538e-05,
+      "loss": 0.0,
+      "step": 9175
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.666370108998743e-05,
+      "loss": 0.0474,
+      "step": 9176
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.6663002116617515e-05,
+      "loss": 0.0,
+      "step": 9177
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.6662303079199473e-05,
+      "loss": 0.0,
+      "step": 9178
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.6661603977737138e-05,
+      "loss": 1.2656,
+      "step": 9179
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.6660904812234362e-05,
+      "loss": 0.0,
+      "step": 9180
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.6660205582694974e-05,
+      "loss": 0.0933,
+      "step": 9181
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.6659506289122815e-05,
+      "loss": 0.1069,
+      "step": 9182
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.6658806931521732e-05,
+      "loss": 0.1367,
+      "step": 9183
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.6658107509895556e-05,
+      "loss": 0.3066,
+      "step": 9184
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.6657408024248136e-05,
+      "loss": 0.0092,
+      "step": 9185
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.6656708474583304e-05,
+      "loss": 0.4199,
+      "step": 9186
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.6656008860904913e-05,
+      "loss": 0.0466,
+      "step": 9187
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.6655309183216798e-05,
+      "loss": 1.0078,
+      "step": 9188
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.66546094415228e-05,
+      "loss": 0.0508,
+      "step": 9189
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.6653909635826766e-05,
+      "loss": 1.0469,
+      "step": 9190
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.665320976613254e-05,
+      "loss": 0.2812,
+      "step": 9191
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.6652509832443957e-05,
+      "loss": 0.207,
+      "step": 9192
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.6651809834764874e-05,
+      "loss": 0.0005,
+      "step": 9193
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.6651109773099125e-05,
+      "loss": 0.5977,
+      "step": 9194
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.6650409647450554e-05,
+      "loss": 1.7812,
+      "step": 9195
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.6649709457823014e-05,
+      "loss": 0.3633,
+      "step": 9196
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.6649009204220335e-05,
+      "loss": 0.1162,
+      "step": 9197
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.6648308886646383e-05,
+      "loss": 0.0791,
+      "step": 9198
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.664760850510499e-05,
+      "loss": 0.0199,
+      "step": 9199
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.6646908059600002e-05,
+      "loss": 1.2734,
+      "step": 9200
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.6646207550135277e-05,
+      "loss": 0.0,
+      "step": 9201
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.6645506976714648e-05,
+      "loss": 0.0366,
+      "step": 9202
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.6644806339341967e-05,
+      "loss": 0.4277,
+      "step": 9203
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.664410563802108e-05,
+      "loss": 0.1309,
+      "step": 9204
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.6643404872755844e-05,
+      "loss": 0.0193,
+      "step": 9205
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.66427040435501e-05,
+      "loss": 0.0002,
+      "step": 9206
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.66420031504077e-05,
+      "loss": 0.416,
+      "step": 9207
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.664130219333249e-05,
+      "loss": 1.6719,
+      "step": 9208
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.6640601172328318e-05,
+      "loss": 0.0,
+      "step": 9209
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.663990008739903e-05,
+      "loss": 0.0977,
+      "step": 9210
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.6639198938548486e-05,
+      "loss": 0.0019,
+      "step": 9211
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.6638497725780534e-05,
+      "loss": 0.208,
+      "step": 9212
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.6637796449099024e-05,
+      "loss": 0.0001,
+      "step": 9213
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.66370951085078e-05,
+      "loss": 0.0001,
+      "step": 9214
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.6636393704010725e-05,
+      "loss": 0.7383,
+      "step": 9215
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.6635692235611645e-05,
+      "loss": 0.0,
+      "step": 9216
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.663499070331441e-05,
+      "loss": 0.2891,
+      "step": 9217
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.6634289107122878e-05,
+      "loss": 0.0029,
+      "step": 9218
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.6633587447040896e-05,
+      "loss": 0.0251,
+      "step": 9219
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.6632885723072318e-05,
+      "loss": 0.0016,
+      "step": 9220
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.6632183935221006e-05,
+      "loss": 0.1465,
+      "step": 9221
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.6631482083490802e-05,
+      "loss": 0.0208,
+      "step": 9222
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.6630780167885567e-05,
+      "loss": 0.0071,
+      "step": 9223
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.6630078188409158e-05,
+      "loss": 0.0,
+      "step": 9224
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.662937614506542e-05,
+      "loss": 0.0002,
+      "step": 9225
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.662867403785822e-05,
+      "loss": 0.0,
+      "step": 9226
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.6627971866791408e-05,
+      "loss": 0.1152,
+      "step": 9227
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.662726963186884e-05,
+      "loss": 0.0,
+      "step": 9228
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.6626567333094368e-05,
+      "loss": 0.0,
+      "step": 9229
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.662586497047186e-05,
+      "loss": 0.0781,
+      "step": 9230
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.6625162544005162e-05,
+      "loss": 0.1172,
+      "step": 9231
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.6624460053698138e-05,
+      "loss": 0.1396,
+      "step": 9232
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.662375749955464e-05,
+      "loss": 0.207,
+      "step": 9233
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.6623054881578536e-05,
+      "loss": 0.1729,
+      "step": 9234
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.6622352199773672e-05,
+      "loss": 0.0889,
+      "step": 9235
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.6621649454143918e-05,
+      "loss": 0.4805,
+      "step": 9236
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.6620946644693124e-05,
+      "loss": 0.0742,
+      "step": 9237
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.6620243771425154e-05,
+      "loss": 0.8633,
+      "step": 9238
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.661954083434387e-05,
+      "loss": 0.1504,
+      "step": 9239
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.661883783345313e-05,
+      "loss": 0.2754,
+      "step": 9240
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.6618134768756794e-05,
+      "loss": 0.0625,
+      "step": 9241
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.6617431640258722e-05,
+      "loss": 0.0006,
+      "step": 9242
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.6616728447962782e-05,
+      "loss": 0.0001,
+      "step": 9243
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.6616025191872826e-05,
+      "loss": 0.1445,
+      "step": 9244
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.6615321871992718e-05,
+      "loss": 0.042,
+      "step": 9245
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.661461848832633e-05,
+      "loss": 0.4043,
+      "step": 9246
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.661391504087751e-05,
+      "loss": 0.0347,
+      "step": 9247
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.661321152965013e-05,
+      "loss": 1.8047,
+      "step": 9248
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.6612507954648052e-05,
+      "loss": 0.0009,
+      "step": 9249
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.661180431587514e-05,
+      "loss": 1.3438,
+      "step": 9250
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.6611100613335256e-05,
+      "loss": 0.4434,
+      "step": 9251
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.661039684703227e-05,
+      "loss": 0.0001,
+      "step": 9252
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.660969301697004e-05,
+      "loss": 0.0001,
+      "step": 9253
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.6608989123152433e-05,
+      "loss": 0.0442,
+      "step": 9254
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.6608285165583316e-05,
+      "loss": 0.0,
+      "step": 9255
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.6607581144266554e-05,
+      "loss": 0.0,
+      "step": 9256
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.660687705920601e-05,
+      "loss": 0.126,
+      "step": 9257
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.6606172910405563e-05,
+      "loss": 0.0,
+      "step": 9258
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.6605468697869063e-05,
+      "loss": 0.0398,
+      "step": 9259
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.6604764421600388e-05,
+      "loss": 0.0,
+      "step": 9260
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.66040600816034e-05,
+      "loss": 0.0,
+      "step": 9261
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.660335567788197e-05,
+      "loss": 0.0,
+      "step": 9262
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.6602651210439967e-05,
+      "loss": 0.1768,
+      "step": 9263
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.6601946679281256e-05,
+      "loss": 0.0002,
+      "step": 9264
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.6601242084409713e-05,
+      "loss": 0.0,
+      "step": 9265
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.6600537425829197e-05,
+      "loss": 0.0,
+      "step": 9266
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.6599832703543587e-05,
+      "loss": 2.3125,
+      "step": 9267
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.659912791755675e-05,
+      "loss": 0.1807,
+      "step": 9268
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.659842306787255e-05,
+      "loss": 0.6602,
+      "step": 9269
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.659771815449487e-05,
+      "loss": 0.4922,
+      "step": 9270
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.6597013177427573e-05,
+      "loss": 0.0,
+      "step": 9271
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.6596308136674533e-05,
+      "loss": 0.0,
+      "step": 9272
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.6595603032239617e-05,
+      "loss": 0.2871,
+      "step": 9273
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.6594897864126703e-05,
+      "loss": 0.6016,
+      "step": 9274
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.6594192632339662e-05,
+      "loss": 0.0,
+      "step": 9275
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.6593487336882364e-05,
+      "loss": 0.0005,
+      "step": 9276
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.6592781977758685e-05,
+      "loss": 0.0001,
+      "step": 9277
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.6592076554972497e-05,
+      "loss": 0.0,
+      "step": 9278
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.6591371068527678e-05,
+      "loss": 0.5039,
+      "step": 9279
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.6590665518428098e-05,
+      "loss": 0.0,
+      "step": 9280
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.6589959904677635e-05,
+      "loss": 0.1001,
+      "step": 9281
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.6589254227280153e-05,
+      "loss": 0.1357,
+      "step": 9282
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.6588548486239544e-05,
+      "loss": 0.1367,
+      "step": 9283
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.6587842681559672e-05,
+      "loss": 1.0156,
+      "step": 9284
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.6587136813244417e-05,
+      "loss": 0.7578,
+      "step": 9285
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.658643088129765e-05,
+      "loss": 0.5781,
+      "step": 9286
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.6585724885723262e-05,
+      "loss": 0.3145,
+      "step": 9287
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.6585018826525114e-05,
+      "loss": 0.1069,
+      "step": 9288
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.658431270370709e-05,
+      "loss": 0.9375,
+      "step": 9289
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.658360651727307e-05,
+      "loss": 0.5,
+      "step": 9290
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.6582900267226924e-05,
+      "loss": 0.0229,
+      "step": 9291
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.658219395357254e-05,
+      "loss": 0.7109,
+      "step": 9292
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.6581487576313795e-05,
+      "loss": 0.6758,
+      "step": 9293
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.6580781135454564e-05,
+      "loss": 0.4453,
+      "step": 9294
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.6580074630998724e-05,
+      "loss": 0.0403,
+      "step": 9295
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.657936806295016e-05,
+      "loss": 0.5859,
+      "step": 9296
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.6578661431312758e-05,
+      "loss": 0.0286,
+      "step": 9297
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.6577954736090384e-05,
+      "loss": 0.0022,
+      "step": 9298
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.657724797728693e-05,
+      "loss": 0.0874,
+      "step": 9299
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.6576541154906277e-05,
+      "loss": 1.0859,
+      "step": 9300
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.65758342689523e-05,
+      "loss": 0.6797,
+      "step": 9301
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.6575127319428884e-05,
+      "loss": 0.0005,
+      "step": 9302
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.657442030633991e-05,
+      "loss": 0.0129,
+      "step": 9303
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.657371322968927e-05,
+      "loss": 0.0002,
+      "step": 9304
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.657300608948083e-05,
+      "loss": 0.0002,
+      "step": 9305
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.6572298885718488e-05,
+      "loss": 0.0014,
+      "step": 9306
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.6571591618406117e-05,
+      "loss": 0.0006,
+      "step": 9307
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.657088428754761e-05,
+      "loss": 0.0,
+      "step": 9308
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.6570176893146847e-05,
+      "loss": 0.0005,
+      "step": 9309
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.6569469435207712e-05,
+      "loss": 0.0016,
+      "step": 9310
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.656876191373409e-05,
+      "loss": 0.2793,
+      "step": 9311
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.6568054328729866e-05,
+      "loss": 0.0001,
+      "step": 9312
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.6567346680198934e-05,
+      "loss": 0.0001,
+      "step": 9313
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.6566638968145172e-05,
+      "loss": 0.0001,
+      "step": 9314
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.6565931192572464e-05,
+      "loss": 0.0001,
+      "step": 9315
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.6565223353484705e-05,
+      "loss": 0.0156,
+      "step": 9316
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.6564515450885772e-05,
+      "loss": 0.0352,
+      "step": 9317
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.6563807484779565e-05,
+      "loss": 0.0,
+      "step": 9318
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.656309945516996e-05,
+      "loss": 0.4961,
+      "step": 9319
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.6562391362060852e-05,
+      "loss": 0.0,
+      "step": 9320
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.656168320545613e-05,
+      "loss": 0.0,
+      "step": 9321
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.656097498535968e-05,
+      "loss": 0.0001,
+      "step": 9322
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.656026670177539e-05,
+      "loss": 0.8633,
+      "step": 9323
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.6559558354707153e-05,
+      "loss": 0.0001,
+      "step": 9324
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.6558849944158856e-05,
+      "loss": 0.0464,
+      "step": 9325
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.6558141470134393e-05,
+      "loss": 0.1055,
+      "step": 9326
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.6557432932637653e-05,
+      "loss": 0.0004,
+      "step": 9327
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.655672433167252e-05,
+      "loss": 0.0001,
+      "step": 9328
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.6556015667242898e-05,
+      "loss": 0.0001,
+      "step": 9329
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.655530693935267e-05,
+      "loss": 0.0005,
+      "step": 9330
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.6554598148005732e-05,
+      "loss": 0.1934,
+      "step": 9331
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.6553889293205974e-05,
+      "loss": 0.2139,
+      "step": 9332
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.655318037495729e-05,
+      "loss": 0.0618,
+      "step": 9333
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.6552471393263575e-05,
+      "loss": 0.1514,
+      "step": 9334
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.655176234812872e-05,
+      "loss": 0.0303,
+      "step": 9335
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.655105323955661e-05,
+      "loss": 0.3965,
+      "step": 9336
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.6550344067551158e-05,
+      "loss": 0.3613,
+      "step": 9337
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.6549634832116243e-05,
+      "loss": 0.004,
+      "step": 9338
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.6548925533255766e-05,
+      "loss": 0.1523,
+      "step": 9339
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.6548216170973622e-05,
+      "loss": 0.8906,
+      "step": 9340
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.654750674527371e-05,
+      "loss": 1.7734,
+      "step": 9341
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.6546797256159917e-05,
+      "loss": 1.1406,
+      "step": 9342
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.6546087703636142e-05,
+      "loss": 0.3828,
+      "step": 9343
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.654537808770629e-05,
+      "loss": 1.5859,
+      "step": 9344
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.6544668408374244e-05,
+      "loss": 0.4023,
+      "step": 9345
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.6543958665643915e-05,
+      "loss": 0.416,
+      "step": 9346
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.654324885951919e-05,
+      "loss": 0.2217,
+      "step": 9347
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.654253899000397e-05,
+      "loss": 0.0063,
+      "step": 9348
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.6541829057102157e-05,
+      "loss": 0.001,
+      "step": 9349
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.654111906081765e-05,
+      "loss": 0.1904,
+      "step": 9350
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.654040900115434e-05,
+      "loss": 0.6641,
+      "step": 9351
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.6539698878116133e-05,
+      "loss": 1.1797,
+      "step": 9352
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.653898869170693e-05,
+      "loss": 0.7969,
+      "step": 9353
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.6538278441930627e-05,
+      "loss": 0.0012,
+      "step": 9354
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.6537568128791123e-05,
+      "loss": 0.0067,
+      "step": 9355
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.653685775229232e-05,
+      "loss": 0.8516,
+      "step": 9356
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.653614731243813e-05,
+      "loss": 0.8242,
+      "step": 9357
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.6535436809232434e-05,
+      "loss": 0.0211,
+      "step": 9358
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.653472624267915e-05,
+      "loss": 0.0376,
+      "step": 9359
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.6534015612782172e-05,
+      "loss": 0.0466,
+      "step": 9360
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.6533304919545408e-05,
+      "loss": 0.2441,
+      "step": 9361
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.6532594162972755e-05,
+      "loss": 0.0,
+      "step": 9362
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.6531883343068123e-05,
+      "loss": 0.0378,
+      "step": 9363
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.6531172459835408e-05,
+      "loss": 0.2695,
+      "step": 9364
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.6530461513278516e-05,
+      "loss": 0.0,
+      "step": 9365
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.6529750503401355e-05,
+      "loss": 0.0003,
+      "step": 9366
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.652903943020783e-05,
+      "loss": 0.0972,
+      "step": 9367
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.652832829370184e-05,
+      "loss": 0.0022,
+      "step": 9368
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.6527617093887293e-05,
+      "loss": 0.1001,
+      "step": 9369
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.6526905830768097e-05,
+      "loss": 0.0001,
+      "step": 9370
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.6526194504348154e-05,
+      "loss": 0.0001,
+      "step": 9371
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.6525483114631377e-05,
+      "loss": 0.1055,
+      "step": 9372
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.652477166162166e-05,
+      "loss": 0.0002,
+      "step": 9373
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.6524060145322925e-05,
+      "loss": 0.1504,
+      "step": 9374
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.652334856573907e-05,
+      "loss": 0.0001,
+      "step": 9375
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.6522636922874e-05,
+      "loss": 0.0,
+      "step": 9376
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.6521925216731634e-05,
+      "loss": 0.0,
+      "step": 9377
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.6521213447315872e-05,
+      "loss": 0.0002,
+      "step": 9378
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.652050161463063e-05,
+      "loss": 0.0001,
+      "step": 9379
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.6519789718679805e-05,
+      "loss": 0.1006,
+      "step": 9380
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.6519077759467317e-05,
+      "loss": 0.1143,
+      "step": 9381
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.6518365736997073e-05,
+      "loss": 0.1572,
+      "step": 9382
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.651765365127298e-05,
+      "loss": 0.1289,
+      "step": 9383
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.6516941502298953e-05,
+      "loss": 1.2344,
+      "step": 9384
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.65162292900789e-05,
+      "loss": 1.3125,
+      "step": 9385
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.651551701461673e-05,
+      "loss": 0.3184,
+      "step": 9386
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.6514804675916357e-05,
+      "loss": 1.1953,
+      "step": 9387
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.65140922739817e-05,
+      "loss": 1.9062,
+      "step": 9388
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.6513379808816662e-05,
+      "loss": 0.4141,
+      "step": 9389
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.6512667280425152e-05,
+      "loss": 0.1768,
+      "step": 9390
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.6511954688811096e-05,
+      "loss": 1.2422,
+      "step": 9391
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.6511242033978397e-05,
+      "loss": 0.2432,
+      "step": 9392
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.6510529315930972e-05,
+      "loss": 2.0781,
+      "step": 9393
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.6509816534672735e-05,
+      "loss": 1.3828,
+      "step": 9394
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.65091036902076e-05,
+      "loss": 0.7188,
+      "step": 9395
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.650839078253948e-05,
+      "loss": 1.2734,
+      "step": 9396
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6507677811672294e-05,
+      "loss": 1.25,
+      "step": 9397
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6506964777609954e-05,
+      "loss": 0.166,
+      "step": 9398
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.650625168035638e-05,
+      "loss": 0.0001,
+      "step": 9399
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.650553851991548e-05,
+      "loss": 0.4492,
+      "step": 9400
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6504825296291175e-05,
+      "loss": 0.0021,
+      "step": 9401
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6504112009487384e-05,
+      "loss": 0.252,
+      "step": 9402
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6503398659508017e-05,
+      "loss": 0.0,
+      "step": 9403
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6502685246357004e-05,
+      "loss": 0.0034,
+      "step": 9404
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6501971770038247e-05,
+      "loss": 0.0002,
+      "step": 9405
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.650125823055568e-05,
+      "loss": 0.0219,
+      "step": 9406
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6500544627913207e-05,
+      "loss": 1.6562,
+      "step": 9407
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6499830962114752e-05,
+      "loss": 0.0003,
+      "step": 9408
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6499117233164238e-05,
+      "loss": 0.1562,
+      "step": 9409
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6498403441065582e-05,
+      "loss": 0.0003,
+      "step": 9410
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6497689585822703e-05,
+      "loss": 0.0005,
+      "step": 9411
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.649697566743953e-05,
+      "loss": 0.0003,
+      "step": 9412
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6496261685919965e-05,
+      "loss": 0.0001,
+      "step": 9413
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6495547641267942e-05,
+      "loss": 0.0001,
+      "step": 9414
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.649483353348738e-05,
+      "loss": 0.0001,
+      "step": 9415
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.64941193625822e-05,
+      "loss": 0.0396,
+      "step": 9416
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.649340512855632e-05,
+      "loss": 0.0859,
+      "step": 9417
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.649269083141367e-05,
+      "loss": 0.3613,
+      "step": 9418
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.649197647115817e-05,
+      "loss": 0.0698,
+      "step": 9419
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6491262047793737e-05,
+      "loss": 0.0001,
+      "step": 9420
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6490547561324308e-05,
+      "loss": 0.0,
+      "step": 9421
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.648983301175379e-05,
+      "loss": 0.0001,
+      "step": 9422
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.648911839908612e-05,
+      "loss": 0.0,
+      "step": 9423
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6488403723325212e-05,
+      "loss": 0.0001,
+      "step": 9424
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6487688984475e-05,
+      "loss": 0.0001,
+      "step": 9425
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6486974182539403e-05,
+      "loss": 0.0002,
+      "step": 9426
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6486259317522347e-05,
+      "loss": 0.0,
+      "step": 9427
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6485544389427757e-05,
+      "loss": 0.0859,
+      "step": 9428
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6484829398259568e-05,
+      "loss": 0.0923,
+      "step": 9429
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6484114344021694e-05,
+      "loss": 0.0884,
+      "step": 9430
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.648339922671807e-05,
+      "loss": 0.1436,
+      "step": 9431
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6482684046352617e-05,
+      "loss": 0.1924,
+      "step": 9432
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6481968802929267e-05,
+      "loss": 0.1162,
+      "step": 9433
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6481253496451948e-05,
+      "loss": 0.2041,
+      "step": 9434
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6480538126924586e-05,
+      "loss": 1.9922,
+      "step": 9435
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.647982269435111e-05,
+      "loss": 1.0391,
+      "step": 9436
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6479107198735452e-05,
+      "loss": 0.1216,
+      "step": 9437
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6478391640081533e-05,
+      "loss": 1.4141,
+      "step": 9438
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.647767601839329e-05,
+      "loss": 0.3984,
+      "step": 9439
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.647696033367465e-05,
+      "loss": 0.2656,
+      "step": 9440
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6476244585929545e-05,
+      "loss": 0.1025,
+      "step": 9441
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.64755287751619e-05,
+      "loss": 0.2246,
+      "step": 9442
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6474812901375656e-05,
+      "loss": 0.7266,
+      "step": 9443
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6474096964574737e-05,
+      "loss": 0.2891,
+      "step": 9444
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6473380964763078e-05,
+      "loss": 0.0247,
+      "step": 9445
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.647266490194461e-05,
+      "loss": 1.0547,
+      "step": 9446
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6471948776123263e-05,
+      "loss": 0.0001,
+      "step": 9447
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6471232587302968e-05,
+      "loss": 0.0007,
+      "step": 9448
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6470516335487664e-05,
+      "loss": 2.25,
+      "step": 9449
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6469800020681277e-05,
+      "loss": 0.0004,
+      "step": 9450
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.646908364288775e-05,
+      "loss": 0.0311,
+      "step": 9451
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6468367202111015e-05,
+      "loss": 0.0004,
+      "step": 9452
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6467650698355e-05,
+      "loss": 1.4922,
+      "step": 9453
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6466934131623643e-05,
+      "loss": 0.3691,
+      "step": 9454
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6466217501920876e-05,
+      "loss": 0.418,
+      "step": 9455
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6465500809250642e-05,
+      "loss": 0.2812,
+      "step": 9456
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6464784053616875e-05,
+      "loss": 0.0004,
+      "step": 9457
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6464067235023507e-05,
+      "loss": 0.1719,
+      "step": 9458
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6463350353474474e-05,
+      "loss": 0.0016,
+      "step": 9459
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6462633408973714e-05,
+      "loss": 0.0006,
+      "step": 9460
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.646191640152517e-05,
+      "loss": 0.0,
+      "step": 9461
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.646119933113277e-05,
+      "loss": 0.0143,
+      "step": 9462
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.646048219780046e-05,
+      "loss": 0.875,
+      "step": 9463
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6459765001532172e-05,
+      "loss": 0.3887,
+      "step": 9464
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6459047742331846e-05,
+      "loss": 0.0003,
+      "step": 9465
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6458330420203422e-05,
+      "loss": 0.0091,
+      "step": 9466
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6457613035150842e-05,
+      "loss": 0.0001,
+      "step": 9467
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6456895587178042e-05,
+      "loss": 0.0001,
+      "step": 9468
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6456178076288963e-05,
+      "loss": 0.0001,
+      "step": 9469
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6455460502487546e-05,
+      "loss": 0.0,
+      "step": 9470
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.645474286577773e-05,
+      "loss": 0.0,
+      "step": 9471
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6454025166163456e-05,
+      "loss": 0.0007,
+      "step": 9472
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6453307403648664e-05,
+      "loss": 0.0,
+      "step": 9473
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6452589578237295e-05,
+      "loss": 0.0002,
+      "step": 9474
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.64518716899333e-05,
+      "loss": 0.0,
+      "step": 9475
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6451153738740615e-05,
+      "loss": 0.0001,
+      "step": 9476
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6450435724663177e-05,
+      "loss": 0.0001,
+      "step": 9477
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6449717647704938e-05,
+      "loss": 0.0001,
+      "step": 9478
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6448999507869836e-05,
+      "loss": 0.0991,
+      "step": 9479
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6448281305161817e-05,
+      "loss": 0.1221,
+      "step": 9480
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6447563039584827e-05,
+      "loss": 0.0041,
+      "step": 9481
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6446844711142807e-05,
+      "loss": 0.2041,
+      "step": 9482
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.64461263198397e-05,
+      "loss": 0.1055,
+      "step": 9483
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6445407865679458e-05,
+      "loss": 1.4375,
+      "step": 9484
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.644468934866602e-05,
+      "loss": 0.9414,
+      "step": 9485
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6443970768803336e-05,
+      "loss": 1.0703,
+      "step": 9486
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.644325212609535e-05,
+      "loss": 0.5898,
+      "step": 9487
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6442533420546004e-05,
+      "loss": 0.5547,
+      "step": 9488
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.644181465215925e-05,
+      "loss": 0.8008,
+      "step": 9489
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6441095820939036e-05,
+      "loss": 0.1138,
+      "step": 9490
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6440376926889312e-05,
+      "loss": 0.4785,
+      "step": 9491
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.643965797001402e-05,
+      "loss": 0.3105,
+      "step": 9492
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.643893895031711e-05,
+      "loss": 0.21,
+      "step": 9493
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.643821986780253e-05,
+      "loss": 1.2109,
+      "step": 9494
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.643750072247423e-05,
+      "loss": 1.0234,
+      "step": 9495
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6436781514336158e-05,
+      "loss": 0.5586,
+      "step": 9496
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6436062243392266e-05,
+      "loss": 0.1719,
+      "step": 9497
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6435342909646503e-05,
+      "loss": 0.0001,
+      "step": 9498
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6434623513102817e-05,
+      "loss": 0.6094,
+      "step": 9499
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6433904053765158e-05,
+      "loss": 0.0684,
+      "step": 9500
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6433184531637484e-05,
+      "loss": 1.4531,
+      "step": 9501
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6432464946723734e-05,
+      "loss": 0.0021,
+      "step": 9502
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6431745299027875e-05,
+      "loss": 0.0125,
+      "step": 9503
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.643102558855385e-05,
+      "loss": 0.0923,
+      "step": 9504
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6430305815305614e-05,
+      "loss": 0.0072,
+      "step": 9505
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6429585979287114e-05,
+      "loss": 0.0957,
+      "step": 9506
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6428866080502307e-05,
+      "loss": 0.0437,
+      "step": 9507
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.642814611895515e-05,
+      "loss": 0.0806,
+      "step": 9508
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6427426094649587e-05,
+      "loss": 0.0466,
+      "step": 9509
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6426706007589587e-05,
+      "loss": 0.0,
+      "step": 9510
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.642598585777909e-05,
+      "loss": 0.0126,
+      "step": 9511
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.642526564522206e-05,
+      "loss": 1.2266,
+      "step": 9512
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6424545369922443e-05,
+      "loss": 0.0007,
+      "step": 9513
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6423825031884205e-05,
+      "loss": 0.0001,
+      "step": 9514
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6423104631111295e-05,
+      "loss": 0.0,
+      "step": 9515
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.642238416760767e-05,
+      "loss": 0.0002,
+      "step": 9516
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.642166364137729e-05,
+      "loss": 0.1436,
+      "step": 9517
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6420943052424106e-05,
+      "loss": 0.0211,
+      "step": 9518
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6420222400752082e-05,
+      "loss": 0.0006,
+      "step": 9519
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.641950168636517e-05,
+      "loss": 0.0142,
+      "step": 9520
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6418780909267328e-05,
+      "loss": 0.0001,
+      "step": 9521
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.641806006946252e-05,
+      "loss": 0.0,
+      "step": 9522
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.64173391669547e-05,
+      "loss": 0.0001,
+      "step": 9523
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.641661820174782e-05,
+      "loss": 0.1533,
+      "step": 9524
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6415897173845856e-05,
+      "loss": 0.0947,
+      "step": 9525
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6415176083252756e-05,
+      "loss": 0.0015,
+      "step": 9526
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6414454929972476e-05,
+      "loss": 0.0008,
+      "step": 9527
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.641373371400899e-05,
+      "loss": 0.0003,
+      "step": 9528
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6413012435366247e-05,
+      "loss": 0.001,
+      "step": 9529
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6412291094048213e-05,
+      "loss": 0.1357,
+      "step": 9530
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6411569690058848e-05,
+      "loss": 0.083,
+      "step": 9531
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6410848223402113e-05,
+      "loss": 0.0001,
+      "step": 9532
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.641012669408197e-05,
+      "loss": 0.1416,
+      "step": 9533
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6409405102102383e-05,
+      "loss": 0.8711,
+      "step": 9534
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6408683447467316e-05,
+      "loss": 0.4434,
+      "step": 9535
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.640796173018073e-05,
+      "loss": 0.0513,
+      "step": 9536
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6407239950246582e-05,
+      "loss": 0.0008,
+      "step": 9537
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6406518107668846e-05,
+      "loss": 0.0991,
+      "step": 9538
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6405796202451482e-05,
+      "loss": 0.5703,
+      "step": 9539
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6405074234598454e-05,
+      "loss": 0.4805,
+      "step": 9540
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6404352204113727e-05,
+      "loss": 1.0234,
+      "step": 9541
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6403630111001267e-05,
+      "loss": 0.5156,
+      "step": 9542
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6402907955265035e-05,
+      "loss": 0.0003,
+      "step": 9543
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6402185736909003e-05,
+      "loss": 0.3477,
+      "step": 9544
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6401463455937137e-05,
+      "loss": 0.5703,
+      "step": 9545
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6400741112353396e-05,
+      "loss": 0.1738,
+      "step": 9546
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6400018706161757e-05,
+      "loss": 0.0386,
+      "step": 9547
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6399296237366183e-05,
+      "loss": 0.0259,
+      "step": 9548
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6398573705970636e-05,
+      "loss": 0.02,
+      "step": 9549
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6397851111979087e-05,
+      "loss": 0.02,
+      "step": 9550
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6397128455395506e-05,
+      "loss": 0.4883,
+      "step": 9551
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.639640573622386e-05,
+      "loss": 0.0013,
+      "step": 9552
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6395682954468115e-05,
+      "loss": 0.0327,
+      "step": 9553
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.639496011013225e-05,
+      "loss": 0.0006,
+      "step": 9554
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.639423720322023e-05,
+      "loss": 0.0001,
+      "step": 9555
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6393514233736017e-05,
+      "loss": 0.0457,
+      "step": 9556
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6392791201683594e-05,
+      "loss": 0.0023,
+      "step": 9557
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.639206810706692e-05,
+      "loss": 0.2695,
+      "step": 9558
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.639134494988997e-05,
+      "loss": 0.0129,
+      "step": 9559
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6390621730156724e-05,
+      "loss": 0.1011,
+      "step": 9560
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6389898447871138e-05,
+      "loss": 0.3281,
+      "step": 9561
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6389175103037194e-05,
+      "loss": 0.0001,
+      "step": 9562
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6388451695658865e-05,
+      "loss": 0.0002,
+      "step": 9563
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.638772822574012e-05,
+      "loss": 0.3008,
+      "step": 9564
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6387004693284928e-05,
+      "loss": 0.0001,
+      "step": 9565
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6386281098297272e-05,
+      "loss": 0.0002,
+      "step": 9566
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6385557440781122e-05,
+      "loss": 0.2471,
+      "step": 9567
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6384833720740445e-05,
+      "loss": 0.0,
+      "step": 9568
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6384109938179227e-05,
+      "loss": 0.0106,
+      "step": 9569
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6383386093101434e-05,
+      "loss": 0.0003,
+      "step": 9570
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6382662185511044e-05,
+      "loss": 0.0,
+      "step": 9571
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6381938215412028e-05,
+      "loss": 0.0002,
+      "step": 9572
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.638121418280837e-05,
+      "loss": 0.0,
+      "step": 9573
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6380490087704046e-05,
+      "loss": 0.0,
+      "step": 9574
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6379765930103022e-05,
+      "loss": 0.0,
+      "step": 9575
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6379041710009283e-05,
+      "loss": 0.0001,
+      "step": 9576
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6378317427426806e-05,
+      "loss": 0.0002,
+      "step": 9577
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6377593082359562e-05,
+      "loss": 0.0,
+      "step": 9578
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.637686867481154e-05,
+      "loss": 0.0,
+      "step": 9579
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.637614420478671e-05,
+      "loss": 0.1484,
+      "step": 9580
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6375419672289048e-05,
+      "loss": 0.0913,
+      "step": 9581
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.637469507732254e-05,
+      "loss": 0.2012,
+      "step": 9582
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6373970419891166e-05,
+      "loss": 0.166,
+      "step": 9583
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6373245699998893e-05,
+      "loss": 0.7109,
+      "step": 9584
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6372520917649715e-05,
+      "loss": 0.3105,
+      "step": 9585
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6371796072847608e-05,
+      "loss": 0.4551,
+      "step": 9586
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.637107116559655e-05,
+      "loss": 0.6641,
+      "step": 9587
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.637034619590052e-05,
+      "loss": 1.6797,
+      "step": 9588
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6369621163763507e-05,
+      "loss": 0.2139,
+      "step": 9589
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6368896069189483e-05,
+      "loss": 0.5586,
+      "step": 9590
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6368170912182433e-05,
+      "loss": 0.0728,
+      "step": 9591
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6367445692746347e-05,
+      "loss": 0.0034,
+      "step": 9592
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.63667204108852e-05,
+      "loss": 0.5391,
+      "step": 9593
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6365995066602978e-05,
+      "loss": 0.1245,
+      "step": 9594
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6365269659903662e-05,
+      "loss": 0.0869,
+      "step": 9595
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6364544190791232e-05,
+      "loss": 0.0356,
+      "step": 9596
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.636381865926968e-05,
+      "loss": 0.7109,
+      "step": 9597
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6363093065342987e-05,
+      "loss": 0.1289,
+      "step": 9598
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6362367409015134e-05,
+      "loss": 0.0403,
+      "step": 9599
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.636164169029011e-05,
+      "loss": 0.0292,
+      "step": 9600
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6360915909171906e-05,
+      "loss": 0.0001,
+      "step": 9601
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6360190065664496e-05,
+      "loss": 0.0226,
+      "step": 9602
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.635946415977187e-05,
+      "loss": 0.001,
+      "step": 9603
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6358738191498018e-05,
+      "loss": 0.8984,
+      "step": 9604
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6358012160846925e-05,
+      "loss": 0.0001,
+      "step": 9605
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6357286067822576e-05,
+      "loss": 0.0352,
+      "step": 9606
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.635655991242896e-05,
+      "loss": 0.0003,
+      "step": 9607
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6355833694670066e-05,
+      "loss": 0.083,
+      "step": 9608
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6355107414549878e-05,
+      "loss": 0.0001,
+      "step": 9609
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.635438107207239e-05,
+      "loss": 0.0342,
+      "step": 9610
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6353654667241587e-05,
+      "loss": 0.021,
+      "step": 9611
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6352928200061453e-05,
+      "loss": 0.3867,
+      "step": 9612
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6352201670535995e-05,
+      "loss": 0.1011,
+      "step": 9613
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.635147507866918e-05,
+      "loss": 0.3086,
+      "step": 9614
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6350748424465015e-05,
+      "loss": 0.019,
+      "step": 9615
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6350021707927485e-05,
+      "loss": 0.0001,
+      "step": 9616
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6349294929060576e-05,
+      "loss": 0.0001,
+      "step": 9617
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.634856808786829e-05,
+      "loss": 0.127,
+      "step": 9618
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6347841184354608e-05,
+      "loss": 0.0043,
+      "step": 9619
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6347114218523525e-05,
+      "loss": 0.0001,
+      "step": 9620
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6346387190379034e-05,
+      "loss": 0.0,
+      "step": 9621
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6345660099925132e-05,
+      "loss": 0.0,
+      "step": 9622
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6344932947165804e-05,
+      "loss": 0.0001,
+      "step": 9623
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.634420573210505e-05,
+      "loss": 0.0,
+      "step": 9624
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.634347845474686e-05,
+      "loss": 0.0,
+      "step": 9625
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6342751115095224e-05,
+      "loss": 0.106,
+      "step": 9626
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6342023713154144e-05,
+      "loss": 0.1143,
+      "step": 9627
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6341296248927605e-05,
+      "loss": 0.0001,
+      "step": 9628
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6340568722419617e-05,
+      "loss": 0.1641,
+      "step": 9629
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6339841133634162e-05,
+      "loss": 0.0981,
+      "step": 9630
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.633911348257524e-05,
+      "loss": 0.1016,
+      "step": 9631
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6338385769246843e-05,
+      "loss": 0.1328,
+      "step": 9632
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6337657993652978e-05,
+      "loss": 0.2754,
+      "step": 9633
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6336930155797634e-05,
+      "loss": 0.2598,
+      "step": 9634
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6336202255684806e-05,
+      "loss": 0.0273,
+      "step": 9635
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.633547429331849e-05,
+      "loss": 1.0547,
+      "step": 9636
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6334746268702693e-05,
+      "loss": 0.6875,
+      "step": 9637
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6334018181841404e-05,
+      "loss": 0.1514,
+      "step": 9638
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6333290032738626e-05,
+      "loss": 0.8008,
+      "step": 9639
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.633256182139836e-05,
+      "loss": 0.6719,
+      "step": 9640
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6331833547824594e-05,
+      "loss": 0.5156,
+      "step": 9641
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6331105212021342e-05,
+      "loss": 0.3105,
+      "step": 9642
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6330376813992596e-05,
+      "loss": 0.0552,
+      "step": 9643
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6329648353742356e-05,
+      "loss": 0.5938,
+      "step": 9644
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.632891983127462e-05,
+      "loss": 0.1108,
+      "step": 9645
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6328191246593397e-05,
+      "loss": 0.2119,
+      "step": 9646
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.632746259970268e-05,
+      "loss": 0.1279,
+      "step": 9647
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.632673389060647e-05,
+      "loss": 0.8906,
+      "step": 9648
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.632600511930878e-05,
+      "loss": 0.0194,
+      "step": 9649
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.63252762858136e-05,
+      "loss": 0.0187,
+      "step": 9650
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6324547390124934e-05,
+      "loss": 0.0012,
+      "step": 9651
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6323818432246794e-05,
+      "loss": 0.9062,
+      "step": 9652
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.632308941218317e-05,
+      "loss": 0.6055,
+      "step": 9653
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6322360329938074e-05,
+      "loss": 0.0004,
+      "step": 9654
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6321631185515506e-05,
+      "loss": 0.0003,
+      "step": 9655
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6320901978919474e-05,
+      "loss": 0.0003,
+      "step": 9656
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6320172710153977e-05,
+      "loss": 0.0001,
+      "step": 9657
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.631944337922303e-05,
+      "loss": 0.2812,
+      "step": 9658
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.631871398613063e-05,
+      "loss": 0.0012,
+      "step": 9659
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.631798453088078e-05,
+      "loss": 0.0237,
+      "step": 9660
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6317255013477487e-05,
+      "loss": 0.0256,
+      "step": 9661
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.631652543392477e-05,
+      "loss": 0.0009,
+      "step": 9662
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6315795792226615e-05,
+      "loss": 0.0391,
+      "step": 9663
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6315066088387043e-05,
+      "loss": 0.0026,
+      "step": 9664
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6314336322410062e-05,
+      "loss": 0.0,
+      "step": 9665
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6313606494299668e-05,
+      "loss": 0.0,
+      "step": 9666
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.631287660405988e-05,
+      "loss": 0.0,
+      "step": 9667
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6312146651694706e-05,
+      "loss": 0.1113,
+      "step": 9668
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6311416637208145e-05,
+      "loss": 0.0,
+      "step": 9669
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6310686560604213e-05,
+      "loss": 0.0001,
+      "step": 9670
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6309956421886918e-05,
+      "loss": 0.0,
+      "step": 9671
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.630922622106027e-05,
+      "loss": 0.7344,
+      "step": 9672
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.630849595812828e-05,
+      "loss": 0.0,
+      "step": 9673
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6307765633094956e-05,
+      "loss": 0.0001,
+      "step": 9674
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6307035245964307e-05,
+      "loss": 0.0,
+      "step": 9675
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.630630479674035e-05,
+      "loss": 0.0,
+      "step": 9676
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6305574285427088e-05,
+      "loss": 0.1108,
+      "step": 9677
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6304843712028544e-05,
+      "loss": 0.1426,
+      "step": 9678
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.630411307654872e-05,
+      "loss": 0.124,
+      "step": 9679
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.630338237899163e-05,
+      "loss": 0.127,
+      "step": 9680
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.630265161936129e-05,
+      "loss": 0.1621,
+      "step": 9681
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6301920797661712e-05,
+      "loss": 0.1797,
+      "step": 9682
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6301189913896903e-05,
+      "loss": 0.1562,
+      "step": 9683
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.630045896807089e-05,
+      "loss": 0.6484,
+      "step": 9684
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6299727960187678e-05,
+      "loss": 1.25,
+      "step": 9685
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6298996890251285e-05,
+      "loss": 0.0378,
+      "step": 9686
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6298265758265717e-05,
+      "loss": 0.0767,
+      "step": 9687
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6297534564235e-05,
+      "loss": 0.2891,
+      "step": 9688
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6296803308163143e-05,
+      "loss": 0.0111,
+      "step": 9689
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6296071990054167e-05,
+      "loss": 1.7344,
+      "step": 9690
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.629534060991208e-05,
+      "loss": 0.2432,
+      "step": 9691
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6294609167740904e-05,
+      "loss": 0.0015,
+      "step": 9692
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6293877663544658e-05,
+      "loss": 0.3281,
+      "step": 9693
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.629314609732736e-05,
+      "loss": 0.9688,
+      "step": 9694
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6292414469093017e-05,
+      "loss": 0.4395,
+      "step": 9695
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.629168277884565e-05,
+      "loss": 0.0466,
+      "step": 9696
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6290951026589288e-05,
+      "loss": 0.0049,
+      "step": 9697
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.629021921232794e-05,
+      "loss": 0.0003,
+      "step": 9698
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6289487336065625e-05,
+      "loss": 0.0645,
+      "step": 9699
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.628875539780637e-05,
+      "loss": 0.3477,
+      "step": 9700
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.628802339755418e-05,
+      "loss": 0.5508,
+      "step": 9701
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.628729133531309e-05,
+      "loss": 0.0408,
+      "step": 9702
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6286559211087107e-05,
+      "loss": 0.293,
+      "step": 9703
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6285827024880263e-05,
+      "loss": 0.0002,
+      "step": 9704
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6285094776696572e-05,
+      "loss": 0.0083,
+      "step": 9705
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6284362466540057e-05,
+      "loss": 0.0046,
+      "step": 9706
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6283630094414736e-05,
+      "loss": 0.5508,
+      "step": 9707
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.628289766032464e-05,
+      "loss": 0.0005,
+      "step": 9708
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.628216516427378e-05,
+      "loss": 0.0,
+      "step": 9709
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.628143260626619e-05,
+      "loss": 0.0535,
+      "step": 9710
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6280699986305883e-05,
+      "loss": 0.0007,
+      "step": 9711
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.627996730439689e-05,
+      "loss": 0.0,
+      "step": 9712
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6279234560543227e-05,
+      "loss": 0.1367,
+      "step": 9713
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.627850175474892e-05,
+      "loss": 0.1533,
+      "step": 9714
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6277768887017998e-05,
+      "loss": 0.0129,
+      "step": 9715
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6277035957354484e-05,
+      "loss": 0.0,
+      "step": 9716
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6276302965762402e-05,
+      "loss": 0.0693,
+      "step": 9717
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6275569912245775e-05,
+      "loss": 0.0,
+      "step": 9718
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.627483679680863e-05,
+      "loss": 0.006,
+      "step": 9719
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6274103619454992e-05,
+      "loss": 0.1758,
+      "step": 9720
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.627337038018889e-05,
+      "loss": 0.0001,
+      "step": 9721
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6272637079014348e-05,
+      "loss": 0.0,
+      "step": 9722
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6271903715935395e-05,
+      "loss": 0.0,
+      "step": 9723
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6271170290956056e-05,
+      "loss": 0.0283,
+      "step": 9724
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6270436804080364e-05,
+      "loss": 0.0,
+      "step": 9725
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.626970325531234e-05,
+      "loss": 0.5664,
+      "step": 9726
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6268969644656018e-05,
+      "loss": 0.0001,
+      "step": 9727
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.626823597211542e-05,
+      "loss": 0.4805,
+      "step": 9728
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6267502237694583e-05,
+      "loss": 0.0004,
+      "step": 9729
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.626676844139753e-05,
+      "loss": 0.0776,
+      "step": 9730
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6266034583228292e-05,
+      "loss": 0.2021,
+      "step": 9731
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6265300663190905e-05,
+      "loss": 0.167,
+      "step": 9732
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6264566681289392e-05,
+      "loss": 0.167,
+      "step": 9733
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6263832637527786e-05,
+      "loss": 1.1953,
+      "step": 9734
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6263098531910118e-05,
+      "loss": 1.0078,
+      "step": 9735
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.626236436444042e-05,
+      "loss": 0.0718,
+      "step": 9736
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6261630135122726e-05,
+      "loss": 0.4102,
+      "step": 9737
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.626089584396106e-05,
+      "loss": 0.832,
+      "step": 9738
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6260161490959466e-05,
+      "loss": 0.1699,
+      "step": 9739
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6259427076121965e-05,
+      "loss": 1.8906,
+      "step": 9740
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6258692599452604e-05,
+      "loss": 0.8164,
+      "step": 9741
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.62579580609554e-05,
+      "loss": 0.1953,
+      "step": 9742
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6257223460634395e-05,
+      "loss": 0.0009,
+      "step": 9743
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6256488798493628e-05,
+      "loss": 0.4902,
+      "step": 9744
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6255754074537125e-05,
+      "loss": 0.0991,
+      "step": 9745
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6255019288768928e-05,
+      "loss": 1.6328,
+      "step": 9746
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6254284441193065e-05,
+      "loss": 0.7578,
+      "step": 9747
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6253549531813576e-05,
+      "loss": 0.7148,
+      "step": 9748
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6252814560634498e-05,
+      "loss": 0.4902,
+      "step": 9749
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.625207952765986e-05,
+      "loss": 0.0645,
+      "step": 9750
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6251344432893705e-05,
+      "loss": 0.0005,
+      "step": 9751
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.625060927634007e-05,
+      "loss": 0.6797,
+      "step": 9752
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.624987405800299e-05,
+      "loss": 0.0481,
+      "step": 9753
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6249138777886503e-05,
+      "loss": 0.2773,
+      "step": 9754
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6248403435994647e-05,
+      "loss": 0.5234,
+      "step": 9755
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6247668032331453e-05,
+      "loss": 0.0002,
+      "step": 9756
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6246932566900973e-05,
+      "loss": 0.0086,
+      "step": 9757
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.624619703970724e-05,
+      "loss": 1.5078,
+      "step": 9758
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.624546145075429e-05,
+      "loss": 1.0234,
+      "step": 9759
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6244725800046164e-05,
+      "loss": 0.0,
+      "step": 9760
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6243990087586904e-05,
+      "loss": 0.0,
+      "step": 9761
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6243254313380548e-05,
+      "loss": 0.25,
+      "step": 9762
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.624251847743114e-05,
+      "loss": 0.4492,
+      "step": 9763
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6241782579742716e-05,
+      "loss": 0.001,
+      "step": 9764
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.624104662031932e-05,
+      "loss": 0.0,
+      "step": 9765
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6240310599164995e-05,
+      "loss": 0.8359,
+      "step": 9766
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6239574516283777e-05,
+      "loss": 0.0,
+      "step": 9767
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6238838371679715e-05,
+      "loss": 1.2734,
+      "step": 9768
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.623810216535685e-05,
+      "loss": 0.2441,
+      "step": 9769
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.623736589731922e-05,
+      "loss": 0.0393,
+      "step": 9770
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6236629567570877e-05,
+      "loss": 0.0,
+      "step": 9771
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6235893176115858e-05,
+      "loss": 0.0,
+      "step": 9772
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6235156722958208e-05,
+      "loss": 0.0,
+      "step": 9773
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6234420208101974e-05,
+      "loss": 0.0,
+      "step": 9774
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6233683631551197e-05,
+      "loss": 0.0,
+      "step": 9775
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6232946993309927e-05,
+      "loss": 0.0,
+      "step": 9776
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6232210293382197e-05,
+      "loss": 0.0952,
+      "step": 9777
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.623147353177207e-05,
+      "loss": 0.0001,
+      "step": 9778
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.623073670848358e-05,
+      "loss": 0.0,
+      "step": 9779
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6229999823520775e-05,
+      "loss": 0.4375,
+      "step": 9780
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6229262876887707e-05,
+      "loss": 0.2178,
+      "step": 9781
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6228525868588418e-05,
+      "loss": 0.0957,
+      "step": 9782
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6227788798626956e-05,
+      "loss": 0.125,
+      "step": 9783
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6227051667007368e-05,
+      "loss": 0.4922,
+      "step": 9784
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.622631447373371e-05,
+      "loss": 0.0505,
+      "step": 9785
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6225577218810015e-05,
+      "loss": 0.4961,
+      "step": 9786
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6224839902240346e-05,
+      "loss": 0.918,
+      "step": 9787
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6224102524028745e-05,
+      "loss": 1.5859,
+      "step": 9788
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.622336508417926e-05,
+      "loss": 0.3164,
+      "step": 9789
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.622262758269595e-05,
+      "loss": 1.9844,
+      "step": 9790
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6221890019582856e-05,
+      "loss": 0.0007,
+      "step": 9791
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6221152394844026e-05,
+      "loss": 1.1016,
+      "step": 9792
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.622041470848352e-05,
+      "loss": 0.2949,
+      "step": 9793
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6219676960505387e-05,
+      "loss": 0.25,
+      "step": 9794
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6218939150913672e-05,
+      "loss": 0.1953,
+      "step": 9795
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6218201279712433e-05,
+      "loss": 0.0131,
+      "step": 9796
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6217463346905718e-05,
+      "loss": 0.4727,
+      "step": 9797
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.621672535249758e-05,
+      "loss": 1.0078,
+      "step": 9798
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6215987296492078e-05,
+      "loss": 1.7734,
+      "step": 9799
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6215249178893257e-05,
+      "loss": 0.0542,
+      "step": 9800
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6214510999705176e-05,
+      "loss": 0.1973,
+      "step": 9801
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.621377275893188e-05,
+      "loss": 0.3145,
+      "step": 9802
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.6213034456577438e-05,
+      "loss": 0.3438,
+      "step": 9803
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.6212296092645888e-05,
+      "loss": 0.0002,
+      "step": 9804
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.6211557667141297e-05,
+      "loss": 0.0004,
+      "step": 9805
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.621081918006771e-05,
+      "loss": 0.0001,
+      "step": 9806
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.6210080631429198e-05,
+      "loss": 0.0009,
+      "step": 9807
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.6209342021229798e-05,
+      "loss": 0.0947,
+      "step": 9808
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.620860334947358e-05,
+      "loss": 0.0003,
+      "step": 9809
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.6207864616164595e-05,
+      "loss": 0.0007,
+      "step": 9810
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.62071258213069e-05,
+      "loss": 0.543,
+      "step": 9811
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.6206386964904553e-05,
+      "loss": 0.1318,
+      "step": 9812
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.620564804696161e-05,
+      "loss": 0.2637,
+      "step": 9813
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.6204909067482127e-05,
+      "loss": 0.0,
+      "step": 9814
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.620417002647017e-05,
+      "loss": 0.3008,
+      "step": 9815
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.620343092392979e-05,
+      "loss": 0.0206,
+      "step": 9816
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.6202691759865045e-05,
+      "loss": 0.0001,
+      "step": 9817
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.6201952534280004e-05,
+      "loss": 0.0001,
+      "step": 9818
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.6201213247178713e-05,
+      "loss": 0.0005,
+      "step": 9819
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.6200473898565243e-05,
+      "loss": 0.0928,
+      "step": 9820
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.6199734488443652e-05,
+      "loss": 0.0001,
+      "step": 9821
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.6198995016817992e-05,
+      "loss": 0.1289,
+      "step": 9822
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.6198255483692333e-05,
+      "loss": 0.0001,
+      "step": 9823
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.6197515889070737e-05,
+      "loss": 0.0001,
+      "step": 9824
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.619677623295726e-05,
+      "loss": 0.0106,
+      "step": 9825
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.6196036515355966e-05,
+      "loss": 0.0001,
+      "step": 9826
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.6195296736270916e-05,
+      "loss": 0.0,
+      "step": 9827
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.6194556895706176e-05,
+      "loss": 0.0,
+      "step": 9828
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.6193816993665803e-05,
+      "loss": 0.0001,
+      "step": 9829
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.6193077030153867e-05,
+      "loss": 0.0,
+      "step": 9830
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.6192337005174427e-05,
+      "loss": 0.1328,
+      "step": 9831
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.6191596918731546e-05,
+      "loss": 0.084,
+      "step": 9832
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.6190856770829293e-05,
+      "loss": 0.126,
+      "step": 9833
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.619011656147173e-05,
+      "loss": 0.3496,
+      "step": 9834
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.618937629066292e-05,
+      "loss": 0.0669,
+      "step": 9835
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.618863595840693e-05,
+      "loss": 1.0859,
+      "step": 9836
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.6187895564707826e-05,
+      "loss": 1.6016,
+      "step": 9837
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.6187155109569677e-05,
+      "loss": 1.5234,
+      "step": 9838
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.618641459299654e-05,
+      "loss": 0.373,
+      "step": 9839
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.6185674014992498e-05,
+      "loss": 0.8516,
+      "step": 9840
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.6184933375561603e-05,
+      "loss": 0.3262,
+      "step": 9841
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.618419267470792e-05,
+      "loss": 0.0064,
+      "step": 9842
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.618345191243553e-05,
+      "loss": 1.6484,
+      "step": 9843
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.6182711088748493e-05,
+      "loss": 0.9453,
+      "step": 9844
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.6181970203650873e-05,
+      "loss": 0.8242,
+      "step": 9845
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.6181229257146747e-05,
+      "loss": 0.0022,
+      "step": 9846
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.6180488249240185e-05,
+      "loss": 0.252,
+      "step": 9847
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.617974717993525e-05,
+      "loss": 0.8281,
+      "step": 9848
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.6179006049236014e-05,
+      "loss": 1.0156,
+      "step": 9849
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.617826485714655e-05,
+      "loss": 1.3438,
+      "step": 9850
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.6177523603670923e-05,
+      "loss": 0.0042,
+      "step": 9851
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.6176782288813204e-05,
+      "loss": 0.1128,
+      "step": 9852
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.617604091257747e-05,
+      "loss": 0.2021,
+      "step": 9853
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.6175299474967783e-05,
+      "loss": 0.0031,
+      "step": 9854
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.6174557975988222e-05,
+      "loss": 0.0067,
+      "step": 9855
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.617381641564286e-05,
+      "loss": 0.0006,
+      "step": 9856
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.617307479393576e-05,
+      "loss": 0.0013,
+      "step": 9857
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.6172333110871006e-05,
+      "loss": 0.1011,
+      "step": 9858
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.6171591366452664e-05,
+      "loss": 0.0004,
+      "step": 9859
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.617084956068481e-05,
+      "loss": 0.0001,
+      "step": 9860
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.6170107693571517e-05,
+      "loss": 0.0,
+      "step": 9861
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.6169365765116857e-05,
+      "loss": 0.0034,
+      "step": 9862
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.6168623775324908e-05,
+      "loss": 0.3203,
+      "step": 9863
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.616788172419974e-05,
+      "loss": 0.0,
+      "step": 9864
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.616713961174544e-05,
+      "loss": 0.0001,
+      "step": 9865
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.6166397437966065e-05,
+      "loss": 0.0001,
+      "step": 9866
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.61656552028657e-05,
+      "loss": 0.0102,
+      "step": 9867
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.6164912906448426e-05,
+      "loss": 0.0033,
+      "step": 9868
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.616417054871831e-05,
+      "loss": 0.0001,
+      "step": 9869
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.6163428129679437e-05,
+      "loss": 0.0063,
+      "step": 9870
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.616268564933588e-05,
+      "loss": 0.2334,
+      "step": 9871
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.6161943107691713e-05,
+      "loss": 0.0303,
+      "step": 9872
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.616120050475102e-05,
+      "loss": 0.0001,
+      "step": 9873
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.6160457840517875e-05,
+      "loss": 0.0,
+      "step": 9874
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.615971511499636e-05,
+      "loss": 0.0,
+      "step": 9875
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.615897232819055e-05,
+      "loss": 0.1572,
+      "step": 9876
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.6158229480104525e-05,
+      "loss": 0.0,
+      "step": 9877
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.6157486570742363e-05,
+      "loss": 0.0,
+      "step": 9878
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.615674360010815e-05,
+      "loss": 0.0,
+      "step": 9879
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.615600056820596e-05,
+      "loss": 0.167,
+      "step": 9880
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.6155257475039874e-05,
+      "loss": 0.1191,
+      "step": 9881
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.6154514320613975e-05,
+      "loss": 0.2676,
+      "step": 9882
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.6153771104932336e-05,
+      "loss": 0.1494,
+      "step": 9883
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.6153027827999055e-05,
+      "loss": 0.4199,
+      "step": 9884
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.61522844898182e-05,
+      "loss": 0.3223,
+      "step": 9885
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.6151541090393856e-05,
+      "loss": 0.4941,
+      "step": 9886
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.615079762973011e-05,
+      "loss": 0.9727,
+      "step": 9887
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.615005410783104e-05,
+      "loss": 0.7383,
+      "step": 9888
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.614931052470073e-05,
+      "loss": 0.6133,
+      "step": 9889
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.6148566880343266e-05,
+      "loss": 0.7188,
+      "step": 9890
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.6147823174762725e-05,
+      "loss": 1.0312,
+      "step": 9891
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.61470794079632e-05,
+      "loss": 0.0008,
+      "step": 9892
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.6146335579948768e-05,
+      "loss": 0.0762,
+      "step": 9893
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.6145591690723518e-05,
+      "loss": 0.7188,
+      "step": 9894
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.614484774029154e-05,
+      "loss": 0.1865,
+      "step": 9895
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.6144103728656902e-05,
+      "loss": 0.0063,
+      "step": 9896
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.614335965582371e-05,
+      "loss": 0.2891,
+      "step": 9897
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.614261552179604e-05,
+      "loss": 0.7656,
+      "step": 9898
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.614187132657798e-05,
+      "loss": 0.293,
+      "step": 9899
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.6141127070173617e-05,
+      "loss": 1.3359,
+      "step": 9900
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.6140382752587034e-05,
+      "loss": 0.1138,
+      "step": 9901
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.6139638373822325e-05,
+      "loss": 1.5469,
+      "step": 9902
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.6138893933883578e-05,
+      "loss": 0.4277,
+      "step": 9903
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.6138149432774873e-05,
+      "loss": 0.0,
+      "step": 9904
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.6137404870500306e-05,
+      "loss": 0.0002,
+      "step": 9905
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.6136660247063963e-05,
+      "loss": 0.0021,
+      "step": 9906
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.613591556246993e-05,
+      "loss": 0.7109,
+      "step": 9907
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.6135170816722308e-05,
+      "loss": 0.0,
+      "step": 9908
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.6134426009825175e-05,
+      "loss": 0.0,
+      "step": 9909
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.6133681141782627e-05,
+      "loss": 0.0024,
+      "step": 9910
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.613293621259875e-05,
+      "loss": 0.3242,
+      "step": 9911
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.613219122227764e-05,
+      "loss": 0.0,
+      "step": 9912
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.6131446170823382e-05,
+      "loss": 0.0026,
+      "step": 9913
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.6130701058240074e-05,
+      "loss": 0.0,
+      "step": 9914
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.6129955884531806e-05,
+      "loss": 0.0001,
+      "step": 9915
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.6129210649702668e-05,
+      "loss": 0.1963,
+      "step": 9916
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.6128465353756754e-05,
+      "loss": 0.105,
+      "step": 9917
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.6127719996698156e-05,
+      "loss": 0.0,
+      "step": 9918
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.6126974578530965e-05,
+      "loss": 0.2695,
+      "step": 9919
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.612622909925928e-05,
+      "loss": 0.0002,
+      "step": 9920
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.6125483558887195e-05,
+      "loss": 0.0,
+      "step": 9921
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.6124737957418795e-05,
+      "loss": 0.0,
+      "step": 9922
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.6123992294858184e-05,
+      "loss": 0.0,
+      "step": 9923
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.612324657120945e-05,
+      "loss": 0.0,
+      "step": 9924
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.6122500786476694e-05,
+      "loss": 0.0,
+      "step": 9925
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.6121754940664015e-05,
+      "loss": 0.0,
+      "step": 9926
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.6121009033775497e-05,
+      "loss": 0.0,
+      "step": 9927
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.6120263065815243e-05,
+      "loss": 0.0001,
+      "step": 9928
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.611951703678735e-05,
+      "loss": 0.0001,
+      "step": 9929
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.611877094669591e-05,
+      "loss": 0.0001,
+      "step": 9930
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.6118024795545032e-05,
+      "loss": 0.1338,
+      "step": 9931
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.61172785833388e-05,
+      "loss": 0.1138,
+      "step": 9932
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.6116532310081317e-05,
+      "loss": 0.1484,
+      "step": 9933
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.6115785975776684e-05,
+      "loss": 1.0781,
+      "step": 9934
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.6115039580428995e-05,
+      "loss": 0.4375,
+      "step": 9935
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.611429312404235e-05,
+      "loss": 0.9023,
+      "step": 9936
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.6113546606620852e-05,
+      "loss": 0.3262,
+      "step": 9937
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.6112800028168594e-05,
+      "loss": 0.7617,
+      "step": 9938
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.6112053388689686e-05,
+      "loss": 1.0391,
+      "step": 9939
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.6111306688188213e-05,
+      "loss": 0.0076,
+      "step": 9940
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.6110559926668293e-05,
+      "loss": 0.6133,
+      "step": 9941
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.6109813104134015e-05,
+      "loss": 0.5977,
+      "step": 9942
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.610906622058948e-05,
+      "loss": 0.7812,
+      "step": 9943
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.61083192760388e-05,
+      "loss": 0.0003,
+      "step": 9944
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.6107572270486064e-05,
+      "loss": 0.3047,
+      "step": 9945
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.6106825203935383e-05,
+      "loss": 0.1846,
+      "step": 9946
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.610607807639086e-05,
+      "loss": 0.0486,
+      "step": 9947
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.6105330887856592e-05,
+      "loss": 0.4707,
+      "step": 9948
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.610458363833668e-05,
+      "loss": 0.0986,
+      "step": 9949
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.6103836327835236e-05,
+      "loss": 0.0537,
+      "step": 9950
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.610308895635636e-05,
+      "loss": 0.2793,
+      "step": 9951
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.6102341523904157e-05,
+      "loss": 0.0009,
+      "step": 9952
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.6101594030482735e-05,
+      "loss": 0.543,
+      "step": 9953
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.6100846476096188e-05,
+      "loss": 0.8945,
+      "step": 9954
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.6100098860748636e-05,
+      "loss": 0.0038,
+      "step": 9955
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.6099351184444167e-05,
+      "loss": 0.0004,
+      "step": 9956
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.6098603447186907e-05,
+      "loss": 0.0006,
+      "step": 9957
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.6097855648980947e-05,
+      "loss": 0.0005,
+      "step": 9958
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.60971077898304e-05,
+      "loss": 0.1797,
+      "step": 9959
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.609635986973937e-05,
+      "loss": 0.0618,
+      "step": 9960
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.6095611888711967e-05,
+      "loss": 0.0549,
+      "step": 9961
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.6094863846752293e-05,
+      "loss": 0.3848,
+      "step": 9962
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.6094115743864472e-05,
+      "loss": 0.0032,
+      "step": 9963
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.609336758005259e-05,
+      "loss": 0.085,
+      "step": 9964
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.6092619355320767e-05,
+      "loss": 0.1504,
+      "step": 9965
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.6091871069673116e-05,
+      "loss": 0.0091,
+      "step": 9966
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.6091122723113737e-05,
+      "loss": 0.0071,
+      "step": 9967
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.6090374315646742e-05,
+      "loss": 0.0079,
+      "step": 9968
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.608962584727625e-05,
+      "loss": 0.0,
+      "step": 9969
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.608887731800636e-05,
+      "loss": 0.0028,
+      "step": 9970
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.6088128727841187e-05,
+      "loss": 0.0001,
+      "step": 9971
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.6087380076784843e-05,
+      "loss": 0.5156,
+      "step": 9972
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.6086631364841438e-05,
+      "loss": 0.0,
+      "step": 9973
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.6085882592015082e-05,
+      "loss": 0.0045,
+      "step": 9974
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.6085133758309887e-05,
+      "loss": 0.0,
+      "step": 9975
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.6084384863729972e-05,
+      "loss": 0.0,
+      "step": 9976
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.608363590827944e-05,
+      "loss": 0.0,
+      "step": 9977
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.6082886891962406e-05,
+      "loss": 0.0,
+      "step": 9978
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.608213781478299e-05,
+      "loss": 0.0,
+      "step": 9979
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.6081388676745302e-05,
+      "loss": 0.1162,
+      "step": 9980
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.608063947785345e-05,
+      "loss": 0.1455,
+      "step": 9981
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.607989021811156e-05,
+      "loss": 0.1426,
+      "step": 9982
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.6079140897523736e-05,
+      "loss": 0.1758,
+      "step": 9983
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.6078391516094095e-05,
+      "loss": 0.332,
+      "step": 9984
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.6077642073826755e-05,
+      "loss": 0.3965,
+      "step": 9985
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.6076892570725832e-05,
+      "loss": 0.2773,
+      "step": 9986
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.6076143006795437e-05,
+      "loss": 0.0913,
+      "step": 9987
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.6075393382039698e-05,
+      "loss": 0.0952,
+      "step": 9988
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.6074643696462717e-05,
+      "loss": 0.2158,
+      "step": 9989
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.6073893950068622e-05,
+      "loss": 0.0171,
+      "step": 9990
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.6073144142861525e-05,
+      "loss": 0.3027,
+      "step": 9991
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.6072394274845547e-05,
+      "loss": 0.8008,
+      "step": 9992
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.6071644346024797e-05,
+      "loss": 0.6992,
+      "step": 9993
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.6070894356403402e-05,
+      "loss": 0.0942,
+      "step": 9994
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.607014430598548e-05,
+      "loss": 0.3125,
+      "step": 9995
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.606939419477515e-05,
+      "loss": 0.1895,
+      "step": 9996
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.6068644022776523e-05,
+      "loss": 1.8672,
+      "step": 9997
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.606789378999373e-05,
+      "loss": 0.3125,
+      "step": 9998
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.6067143496430887e-05,
+      "loss": 0.4102,
+      "step": 9999
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.6066393142092107e-05,
+      "loss": 1.375,
+      "step": 10000
+    },
+    {
+      "epoch": 1.47,
+      "eval_loss": 0.28706327080726624,
+      "eval_runtime": 268.1173,
+      "eval_samples_per_second": 2.667,
+      "eval_steps_per_second": 2.667,
+      "step": 10000
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.6065642726981526e-05,
+      "loss": 0.0016,
+      "step": 10001
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.606489225110325e-05,
+      "loss": 0.5781,
+      "step": 10002
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.6064141714461407e-05,
+      "loss": 0.054,
+      "step": 10003
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.606339111706012e-05,
+      "loss": 0.0447,
+      "step": 10004
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.6062640458903507e-05,
+      "loss": 0.4004,
+      "step": 10005
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.606188973999569e-05,
+      "loss": 0.0005,
+      "step": 10006
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.6061138960340796e-05,
+      "loss": 0.0,
+      "step": 10007
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.6060388119942948e-05,
+      "loss": 0.0918,
+      "step": 10008
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.6059637218806264e-05,
+      "loss": 0.2061,
+      "step": 10009
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.6058886256934872e-05,
+      "loss": 0.0072,
+      "step": 10010
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.6058135234332894e-05,
+      "loss": 0.0029,
+      "step": 10011
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.6057384151004455e-05,
+      "loss": 0.0205,
+      "step": 10012
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.605663300695368e-05,
+      "loss": 0.8281,
+      "step": 10013
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.6055881802184694e-05,
+      "loss": 0.0432,
+      "step": 10014
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.6055130536701616e-05,
+      "loss": 0.0015,
+      "step": 10015
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.6054379210508584e-05,
+      "loss": 0.0053,
+      "step": 10016
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.6053627823609716e-05,
+      "loss": 0.0012,
+      "step": 10017
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.6052876376009137e-05,
+      "loss": 0.0,
+      "step": 10018
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.605212486771098e-05,
+      "loss": 0.0552,
+      "step": 10019
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.6051373298719364e-05,
+      "loss": 0.0001,
+      "step": 10020
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.6050621669038428e-05,
+      "loss": 0.2246,
+      "step": 10021
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.6049869978672284e-05,
+      "loss": 0.0039,
+      "step": 10022
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.604911822762507e-05,
+      "loss": 0.0,
+      "step": 10023
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.604836641590092e-05,
+      "loss": 0.0,
+      "step": 10024
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.604761454350394e-05,
+      "loss": 0.0,
+      "step": 10025
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.6046862610438287e-05,
+      "loss": 0.0,
+      "step": 10026
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.604611061670807e-05,
+      "loss": 0.1709,
+      "step": 10027
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.6045358562317426e-05,
+      "loss": 0.0,
+      "step": 10028
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.604460644727049e-05,
+      "loss": 0.0001,
+      "step": 10029
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.6043854271571384e-05,
+      "loss": 0.1221,
+      "step": 10030
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.604310203522424e-05,
+      "loss": 0.0001,
+      "step": 10031
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.6042349738233195e-05,
+      "loss": 0.3926,
+      "step": 10032
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.6041597380602373e-05,
+      "loss": 0.1309,
+      "step": 10033
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.6040844962335907e-05,
+      "loss": 0.0593,
+      "step": 10034
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.6040092483437932e-05,
+      "loss": 1.2031,
+      "step": 10035
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.603933994391258e-05,
+      "loss": 1.0703,
+      "step": 10036
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.603858734376398e-05,
+      "loss": 0.6992,
+      "step": 10037
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.6037834682996272e-05,
+      "loss": 0.7969,
+      "step": 10038
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.603708196161358e-05,
+      "loss": 0.4258,
+      "step": 10039
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.6036329179620048e-05,
+      "loss": 2.375,
+      "step": 10040
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.6035576337019793e-05,
+      "loss": 0.0825,
+      "step": 10041
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.603482343381697e-05,
+      "loss": 0.0991,
+      "step": 10042
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.60340704700157e-05,
+      "loss": 0.0,
+      "step": 10043
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.6033317445620124e-05,
+      "loss": 0.0229,
+      "step": 10044
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.6032564360634378e-05,
+      "loss": 0.332,
+      "step": 10045
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.603181121506259e-05,
+      "loss": 0.6758,
+      "step": 10046
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.60310580089089e-05,
+      "loss": 0.3262,
+      "step": 10047
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.603030474217745e-05,
+      "loss": 0.4082,
+      "step": 10048
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.6029551414872372e-05,
+      "loss": 1.0703,
+      "step": 10049
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.6028798026997798e-05,
+      "loss": 0.7656,
+      "step": 10050
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.6028044578557875e-05,
+      "loss": 0.0002,
+      "step": 10051
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.6027291069556732e-05,
+      "loss": 0.625,
+      "step": 10052
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.602653749999851e-05,
+      "loss": 0.9062,
+      "step": 10053
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.602578386988735e-05,
+      "loss": 0.2256,
+      "step": 10054
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.602503017922739e-05,
+      "loss": 0.0038,
+      "step": 10055
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.6024276428022765e-05,
+      "loss": 0.0002,
+      "step": 10056
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.602352261627762e-05,
+      "loss": 0.0001,
+      "step": 10057
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.6022768743996085e-05,
+      "loss": 0.3945,
+      "step": 10058
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.602201481118231e-05,
+      "loss": 0.0776,
+      "step": 10059
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.602126081784043e-05,
+      "loss": 0.0162,
+      "step": 10060
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.602050676397459e-05,
+      "loss": 0.0,
+      "step": 10061
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.6019752649588927e-05,
+      "loss": 0.0001,
+      "step": 10062
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.6018998474687585e-05,
+      "loss": 0.0,
+      "step": 10063
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.6018244239274696e-05,
+      "loss": 0.0128,
+      "step": 10064
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.601748994335442e-05,
+      "loss": 0.0,
+      "step": 10065
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.601673558693088e-05,
+      "loss": 0.0102,
+      "step": 10066
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.6015981170008235e-05,
+      "loss": 0.0051,
+      "step": 10067
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.601522669259062e-05,
+      "loss": 0.2236,
+      "step": 10068
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.601447215468217e-05,
+      "loss": 0.0001,
+      "step": 10069
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.601371755628705e-05,
+      "loss": 0.0101,
+      "step": 10070
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.6012962897409384e-05,
+      "loss": 0.0,
+      "step": 10071
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.6012208178053322e-05,
+      "loss": 0.0,
+      "step": 10072
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.6011453398223017e-05,
+      "loss": 0.0,
+      "step": 10073
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.60106985579226e-05,
+      "loss": 0.1079,
+      "step": 10074
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.600994365715623e-05,
+      "loss": 0.0002,
+      "step": 10075
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.6009188695928045e-05,
+      "loss": 0.0001,
+      "step": 10076
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.6008433674242184e-05,
+      "loss": 0.1055,
+      "step": 10077
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.600767859210281e-05,
+      "loss": 0.1377,
+      "step": 10078
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.6006923449514056e-05,
+      "loss": 0.1699,
+      "step": 10079
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.600616824648008e-05,
+      "loss": 0.1699,
+      "step": 10080
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.6005412983005015e-05,
+      "loss": 0.1235,
+      "step": 10081
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.6004657659093017e-05,
+      "loss": 0.0001,
+      "step": 10082
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.600390227474824e-05,
+      "loss": 0.1875,
+      "step": 10083
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.600314682997482e-05,
+      "loss": 0.3691,
+      "step": 10084
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.6002391324776914e-05,
+      "loss": 0.2451,
+      "step": 10085
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.600163575915867e-05,
+      "loss": 0.0006,
+      "step": 10086
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.600088013312423e-05,
+      "loss": 0.0403,
+      "step": 10087
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.600012444667775e-05,
+      "loss": 0.7188,
+      "step": 10088
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5999368699823384e-05,
+      "loss": 0.0003,
+      "step": 10089
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5998612892565276e-05,
+      "loss": 0.7461,
+      "step": 10090
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.599785702490757e-05,
+      "loss": 0.9141,
+      "step": 10091
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.599710109685443e-05,
+      "loss": 1.0547,
+      "step": 10092
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5996345108410006e-05,
+      "loss": 0.5078,
+      "step": 10093
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5995589059578436e-05,
+      "loss": 0.2988,
+      "step": 10094
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5994832950363887e-05,
+      "loss": 0.0513,
+      "step": 10095
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5994076780770505e-05,
+      "loss": 0.3281,
+      "step": 10096
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5993320550802444e-05,
+      "loss": 0.5117,
+      "step": 10097
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.599256426046385e-05,
+      "loss": 0.0315,
+      "step": 10098
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5991807909758888e-05,
+      "loss": 0.7852,
+      "step": 10099
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5991051498691703e-05,
+      "loss": 0.0004,
+      "step": 10100
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5990295027266452e-05,
+      "loss": 0.0028,
+      "step": 10101
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.598953849548729e-05,
+      "loss": 0.375,
+      "step": 10102
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.598878190335837e-05,
+      "loss": 0.0403,
+      "step": 10103
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.598802525088385e-05,
+      "loss": 0.0002,
+      "step": 10104
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5987268538067875e-05,
+      "loss": 0.0009,
+      "step": 10105
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5986511764914614e-05,
+      "loss": 0.0,
+      "step": 10106
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5985754931428215e-05,
+      "loss": 0.0,
+      "step": 10107
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5984998037612834e-05,
+      "loss": 0.0004,
+      "step": 10108
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5984241083472627e-05,
+      "loss": 0.0041,
+      "step": 10109
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5983484069011758e-05,
+      "loss": 0.0,
+      "step": 10110
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5982726994234384e-05,
+      "loss": 0.0003,
+      "step": 10111
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5981969859144653e-05,
+      "loss": 0.1602,
+      "step": 10112
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5981212663746724e-05,
+      "loss": 0.0,
+      "step": 10113
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5980455408044766e-05,
+      "loss": 0.1816,
+      "step": 10114
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5979698092042926e-05,
+      "loss": 0.0,
+      "step": 10115
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5978940715745367e-05,
+      "loss": 0.3027,
+      "step": 10116
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5978183279156254e-05,
+      "loss": 0.0,
+      "step": 10117
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5977425782279736e-05,
+      "loss": 0.0,
+      "step": 10118
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.597666822511998e-05,
+      "loss": 0.0022,
+      "step": 10119
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5975910607681142e-05,
+      "loss": 0.8008,
+      "step": 10120
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5975152929967387e-05,
+      "loss": 0.0,
+      "step": 10121
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5974395191982877e-05,
+      "loss": 0.0,
+      "step": 10122
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5973637393731762e-05,
+      "loss": 0.0,
+      "step": 10123
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5972879535218214e-05,
+      "loss": 0.0,
+      "step": 10124
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5972121616446397e-05,
+      "loss": 0.0002,
+      "step": 10125
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5971363637420464e-05,
+      "loss": 0.0,
+      "step": 10126
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5970605598144583e-05,
+      "loss": 0.0,
+      "step": 10127
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.596984749862291e-05,
+      "loss": 0.0289,
+      "step": 10128
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5969089338859622e-05,
+      "loss": 0.1201,
+      "step": 10129
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.596833111885887e-05,
+      "loss": 0.1187,
+      "step": 10130
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.596757283862482e-05,
+      "loss": 0.1221,
+      "step": 10131
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.596681449816164e-05,
+      "loss": 0.1172,
+      "step": 10132
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.596605609747349e-05,
+      "loss": 0.1719,
+      "step": 10133
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5965297636564547e-05,
+      "loss": 1.0547,
+      "step": 10134
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5964539115438958e-05,
+      "loss": 0.1245,
+      "step": 10135
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.59637805341009e-05,
+      "loss": 0.0095,
+      "step": 10136
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5963021892554533e-05,
+      "loss": 0.6758,
+      "step": 10137
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5962263190804027e-05,
+      "loss": 0.5391,
+      "step": 10138
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5961504428853547e-05,
+      "loss": 0.2812,
+      "step": 10139
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.596074560670726e-05,
+      "loss": 0.0166,
+      "step": 10140
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5959986724369336e-05,
+      "loss": 0.0757,
+      "step": 10141
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5959227781843937e-05,
+      "loss": 0.0099,
+      "step": 10142
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5958468779135235e-05,
+      "loss": 0.9336,
+      "step": 10143
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5957709716247393e-05,
+      "loss": 0.6562,
+      "step": 10144
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.595695059318459e-05,
+      "loss": 0.1934,
+      "step": 10145
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5956191409950983e-05,
+      "loss": 0.0417,
+      "step": 10146
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5955432166550746e-05,
+      "loss": 0.0019,
+      "step": 10147
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5954672862988047e-05,
+      "loss": 0.3613,
+      "step": 10148
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.595391349926706e-05,
+      "loss": 0.4785,
+      "step": 10149
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.595315407539195e-05,
+      "loss": 0.0042,
+      "step": 10150
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5952394591366893e-05,
+      "loss": 0.9609,
+      "step": 10151
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5951635047196056e-05,
+      "loss": 0.0078,
+      "step": 10152
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5950875442883605e-05,
+      "loss": 0.3652,
+      "step": 10153
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5950115778433723e-05,
+      "loss": 0.0021,
+      "step": 10154
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5949356053850573e-05,
+      "loss": 0.0014,
+      "step": 10155
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5948596269138332e-05,
+      "loss": 1.2109,
+      "step": 10156
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.594783642430117e-05,
+      "loss": 0.0,
+      "step": 10157
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5947076519343254e-05,
+      "loss": 0.0061,
+      "step": 10158
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.594631655426877e-05,
+      "loss": 0.3008,
+      "step": 10159
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.594555652908188e-05,
+      "loss": 0.2061,
+      "step": 10160
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5944796443786763e-05,
+      "loss": 0.0347,
+      "step": 10161
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5944036298387596e-05,
+      "loss": 0.4395,
+      "step": 10162
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5943276092888547e-05,
+      "loss": 0.0,
+      "step": 10163
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.594251582729379e-05,
+      "loss": 0.0001,
+      "step": 10164
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5941755501607507e-05,
+      "loss": 0.103,
+      "step": 10165
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5940995115833867e-05,
+      "loss": 0.0,
+      "step": 10166
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.594023466997705e-05,
+      "loss": 0.0,
+      "step": 10167
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5939474164041232e-05,
+      "loss": 0.0,
+      "step": 10168
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5938713598030585e-05,
+      "loss": 0.0,
+      "step": 10169
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5937952971949287e-05,
+      "loss": 0.1963,
+      "step": 10170
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.593719228580152e-05,
+      "loss": 0.0,
+      "step": 10171
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5936431539591456e-05,
+      "loss": 0.0,
+      "step": 10172
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5935670733323275e-05,
+      "loss": 0.0,
+      "step": 10173
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.593490986700116e-05,
+      "loss": 0.0,
+      "step": 10174
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5934148940629272e-05,
+      "loss": 0.0,
+      "step": 10175
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.593338795421181e-05,
+      "loss": 0.0,
+      "step": 10176
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5932626907752945e-05,
+      "loss": 0.0,
+      "step": 10177
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.593186580125685e-05,
+      "loss": 0.1172,
+      "step": 10178
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5931104634727718e-05,
+      "loss": 0.1094,
+      "step": 10179
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5930343408169716e-05,
+      "loss": 0.0859,
+      "step": 10180
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.592958212158703e-05,
+      "loss": 0.0001,
+      "step": 10181
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.592882077498384e-05,
+      "loss": 0.8633,
+      "step": 10182
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5928059368364333e-05,
+      "loss": 0.0635,
+      "step": 10183
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5927297901732685e-05,
+      "loss": 0.8164,
+      "step": 10184
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5926536375093072e-05,
+      "loss": 0.5469,
+      "step": 10185
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5925774788449683e-05,
+      "loss": 0.5938,
+      "step": 10186
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.59250131418067e-05,
+      "loss": 0.0344,
+      "step": 10187
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5924251435168304e-05,
+      "loss": 0.707,
+      "step": 10188
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5923489668538677e-05,
+      "loss": 0.2324,
+      "step": 10189
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5922727841922007e-05,
+      "loss": 0.875,
+      "step": 10190
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5921965955322472e-05,
+      "loss": 0.0006,
+      "step": 10191
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5921204008744256e-05,
+      "loss": 0.1123,
+      "step": 10192
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.592044200219155e-05,
+      "loss": 0.7148,
+      "step": 10193
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5919679935668533e-05,
+      "loss": 0.0001,
+      "step": 10194
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5918917809179387e-05,
+      "loss": 0.1309,
+      "step": 10195
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5918155622728305e-05,
+      "loss": 0.0996,
+      "step": 10196
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.591739337631947e-05,
+      "loss": 0.6328,
+      "step": 10197
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.591663106995706e-05,
+      "loss": 0.1523,
+      "step": 10198
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5915868703645275e-05,
+      "loss": 0.3477,
+      "step": 10199
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5915106277388293e-05,
+      "loss": 0.3086,
+      "step": 10200
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.59143437911903e-05,
+      "loss": 0.0221,
+      "step": 10201
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.591358124505549e-05,
+      "loss": 0.0026,
+      "step": 10202
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5912818638988045e-05,
+      "loss": 0.793,
+      "step": 10203
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5912055972992153e-05,
+      "loss": 0.0014,
+      "step": 10204
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5911293247072007e-05,
+      "loss": 0.0,
+      "step": 10205
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.591053046123179e-05,
+      "loss": 0.0001,
+      "step": 10206
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5909767615475688e-05,
+      "loss": 0.0001,
+      "step": 10207
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.59090047098079e-05,
+      "loss": 2.0156,
+      "step": 10208
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.5908241744232613e-05,
+      "loss": 0.0,
+      "step": 10209
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.5907478718754014e-05,
+      "loss": 0.0,
+      "step": 10210
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.590671563337629e-05,
+      "loss": 0.0,
+      "step": 10211
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.5905952488103635e-05,
+      "loss": 0.0,
+      "step": 10212
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.5905189282940243e-05,
+      "loss": 0.0073,
+      "step": 10213
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.5904426017890305e-05,
+      "loss": 0.1055,
+      "step": 10214
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.5903662692958e-05,
+      "loss": 0.0002,
+      "step": 10215
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.590289930814754e-05,
+      "loss": 0.0,
+      "step": 10216
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.5902135863463102e-05,
+      "loss": 0.0001,
+      "step": 10217
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.5901372358908886e-05,
+      "loss": 0.0,
+      "step": 10218
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.590060879448908e-05,
+      "loss": 0.0,
+      "step": 10219
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.589984517020788e-05,
+      "loss": 0.0,
+      "step": 10220
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.5899081486069483e-05,
+      "loss": 0.0,
+      "step": 10221
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.5898317742078076e-05,
+      "loss": 0.0,
+      "step": 10222
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.5897553938237852e-05,
+      "loss": 0.0,
+      "step": 10223
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.5896790074553013e-05,
+      "loss": 0.0,
+      "step": 10224
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.589602615102775e-05,
+      "loss": 0.0,
+      "step": 10225
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.5895262167666257e-05,
+      "loss": 0.7695,
+      "step": 10226
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.589449812447273e-05,
+      "loss": 0.0,
+      "step": 10227
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.589373402145137e-05,
+      "loss": 0.0,
+      "step": 10228
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.5892969858606364e-05,
+      "loss": 0.1289,
+      "step": 10229
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.5892205635941914e-05,
+      "loss": 0.0,
+      "step": 10230
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.5891441353462217e-05,
+      "loss": 0.1128,
+      "step": 10231
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.589067701117147e-05,
+      "loss": 0.1216,
+      "step": 10232
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.5889912609073868e-05,
+      "loss": 0.1709,
+      "step": 10233
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.588914814717361e-05,
+      "loss": 0.4961,
+      "step": 10234
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.5888383625474896e-05,
+      "loss": 0.7812,
+      "step": 10235
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.5887619043981923e-05,
+      "loss": 1.0234,
+      "step": 10236
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.588685440269889e-05,
+      "loss": 0.0703,
+      "step": 10237
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.588608970162999e-05,
+      "loss": 0.543,
+      "step": 10238
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.588532494077943e-05,
+      "loss": 2.0312,
+      "step": 10239
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.5884560120151414e-05,
+      "loss": 0.8984,
+      "step": 10240
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.5883795239750132e-05,
+      "loss": 1.1328,
+      "step": 10241
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.5883030299579787e-05,
+      "loss": 0.0003,
+      "step": 10242
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.5882265299644582e-05,
+      "loss": 0.7383,
+      "step": 10243
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.5881500239948716e-05,
+      "loss": 0.0011,
+      "step": 10244
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.588073512049639e-05,
+      "loss": 0.5859,
+      "step": 10245
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.587996994129181e-05,
+      "loss": 0.5742,
+      "step": 10246
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.5879204702339173e-05,
+      "loss": 0.4512,
+      "step": 10247
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.587843940364268e-05,
+      "loss": 0.0072,
+      "step": 10248
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.5877674045206545e-05,
+      "loss": 0.0299,
+      "step": 10249
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.5876908627034962e-05,
+      "loss": 0.8945,
+      "step": 10250
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.587614314913213e-05,
+      "loss": 0.25,
+      "step": 10251
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.5875377611502263e-05,
+      "loss": 0.0015,
+      "step": 10252
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.5874612014149554e-05,
+      "loss": 1.3594,
+      "step": 10253
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.5873846357078217e-05,
+      "loss": 0.0488,
+      "step": 10254
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.5873080640292452e-05,
+      "loss": 0.0003,
+      "step": 10255
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.5872314863796473e-05,
+      "loss": 0.0378,
+      "step": 10256
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.5871549027594465e-05,
+      "loss": 0.012,
+      "step": 10257
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.587078313169065e-05,
+      "loss": 0.0002,
+      "step": 10258
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.5870017176089237e-05,
+      "loss": 0.3359,
+      "step": 10259
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.586925116079442e-05,
+      "loss": 0.0,
+      "step": 10260
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.5868485085810408e-05,
+      "loss": 0.0001,
+      "step": 10261
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.5867718951141417e-05,
+      "loss": 0.8203,
+      "step": 10262
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.5866952756791643e-05,
+      "loss": 0.0,
+      "step": 10263
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.58661865027653e-05,
+      "loss": 0.043,
+      "step": 10264
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.5865420189066596e-05,
+      "loss": 0.0001,
+      "step": 10265
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.5864653815699736e-05,
+      "loss": 0.4629,
+      "step": 10266
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.5863887382668932e-05,
+      "loss": 0.0,
+      "step": 10267
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.586312088997839e-05,
+      "loss": 0.1992,
+      "step": 10268
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.5862354337632326e-05,
+      "loss": 0.377,
+      "step": 10269
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.5861587725634936e-05,
+      "loss": 0.0378,
+      "step": 10270
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.5860821053990448e-05,
+      "loss": 1.0625,
+      "step": 10271
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.5860054322703054e-05,
+      "loss": 0.0,
+      "step": 10272
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.5859287531776977e-05,
+      "loss": 0.0,
+      "step": 10273
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.585852068121642e-05,
+      "loss": 0.0,
+      "step": 10274
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.58577537710256e-05,
+      "loss": 0.0,
+      "step": 10275
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.585698680120873e-05,
+      "loss": 0.0,
+      "step": 10276
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.5856219771770014e-05,
+      "loss": 0.0,
+      "step": 10277
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.585545268271367e-05,
+      "loss": 0.103,
+      "step": 10278
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.585468553404391e-05,
+      "loss": 0.0001,
+      "step": 10279
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.5853918325764944e-05,
+      "loss": 0.0,
+      "step": 10280
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.5853151057880988e-05,
+      "loss": 0.1309,
+      "step": 10281
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.5852383730396256e-05,
+      "loss": 0.1187,
+      "step": 10282
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.5851616343314958e-05,
+      "loss": 0.1484,
+      "step": 10283
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.5850848896641313e-05,
+      "loss": 0.5117,
+      "step": 10284
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.5850081390379535e-05,
+      "loss": 0.5391,
+      "step": 10285
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.584931382453383e-05,
+      "loss": 0.0018,
+      "step": 10286
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.5848546199108427e-05,
+      "loss": 0.9961,
+      "step": 10287
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.584777851410753e-05,
+      "loss": 0.4531,
+      "step": 10288
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.584701076953536e-05,
+      "loss": 1.1094,
+      "step": 10289
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.5846242965396133e-05,
+      "loss": 0.3242,
+      "step": 10290
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.5845475101694068e-05,
+      "loss": 0.1611,
+      "step": 10291
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.5844707178433375e-05,
+      "loss": 0.0111,
+      "step": 10292
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.5843939195618276e-05,
+      "loss": 0.0001,
+      "step": 10293
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.5843171153252988e-05,
+      "loss": 0.001,
+      "step": 10294
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.5842403051341726e-05,
+      "loss": 1.0234,
+      "step": 10295
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.584163488988871e-05,
+      "loss": 0.5469,
+      "step": 10296
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.584086666889816e-05,
+      "loss": 0.291,
+      "step": 10297
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.5840098388374294e-05,
+      "loss": 1.375,
+      "step": 10298
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.5839330048321326e-05,
+      "loss": 0.0713,
+      "step": 10299
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.5838561648743487e-05,
+      "loss": 0.5547,
+      "step": 10300
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.5837793189644984e-05,
+      "loss": 0.5469,
+      "step": 10301
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.583702467103004e-05,
+      "loss": 0.1826,
+      "step": 10302
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.5836256092902884e-05,
+      "loss": 0.012,
+      "step": 10303
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.5835487455267725e-05,
+      "loss": 0.2891,
+      "step": 10304
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.5834718758128794e-05,
+      "loss": 0.005,
+      "step": 10305
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.5833950001490304e-05,
+      "loss": 0.3594,
+      "step": 10306
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.5833181185356484e-05,
+      "loss": 0.0002,
+      "step": 10307
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.583241230973155e-05,
+      "loss": 0.0,
+      "step": 10308
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.5831643374619728e-05,
+      "loss": 0.0001,
+      "step": 10309
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.5830874380025237e-05,
+      "loss": 0.0,
+      "step": 10310
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.5830105325952307e-05,
+      "loss": 0.0253,
+      "step": 10311
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.5829336212405152e-05,
+      "loss": 0.0001,
+      "step": 10312
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.5828567039388006e-05,
+      "loss": 0.5703,
+      "step": 10313
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.582779780690508e-05,
+      "loss": 0.0728,
+      "step": 10314
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.582702851496061e-05,
+      "loss": 0.0001,
+      "step": 10315
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.5826259163558818e-05,
+      "loss": 0.3652,
+      "step": 10316
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.5825489752703925e-05,
+      "loss": 0.0145,
+      "step": 10317
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.582472028240016e-05,
+      "loss": 0.0254,
+      "step": 10318
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.582395075265174e-05,
+      "loss": 0.0,
+      "step": 10319
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.5823181163462907e-05,
+      "loss": 0.0,
+      "step": 10320
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.5822411514837873e-05,
+      "loss": 0.0,
+      "step": 10321
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.5821641806780873e-05,
+      "loss": 0.0669,
+      "step": 10322
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.5820872039296127e-05,
+      "loss": 0.0,
+      "step": 10323
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.582010221238787e-05,
+      "loss": 0.0,
+      "step": 10324
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.5819332326060325e-05,
+      "loss": 0.0,
+      "step": 10325
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.581856238031772e-05,
+      "loss": 0.0762,
+      "step": 10326
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.5817792375164286e-05,
+      "loss": 0.0003,
+      "step": 10327
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.581702231060424e-05,
+      "loss": 0.0674,
+      "step": 10328
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.5816252186641827e-05,
+      "loss": 0.0,
+      "step": 10329
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.581548200328127e-05,
+      "loss": 0.0991,
+      "step": 10330
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.5814711760526796e-05,
+      "loss": 0.1543,
+      "step": 10331
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.5813941458382636e-05,
+      "loss": 0.165,
+      "step": 10332
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.581317109685302e-05,
+      "loss": 0.1553,
+      "step": 10333
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.581240067594218e-05,
+      "loss": 0.625,
+      "step": 10334
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.581163019565435e-05,
+      "loss": 1.1172,
+      "step": 10335
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.581085965599375e-05,
+      "loss": 0.0001,
+      "step": 10336
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.5810089056964625e-05,
+      "loss": 0.543,
+      "step": 10337
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.5809318398571197e-05,
+      "loss": 0.5156,
+      "step": 10338
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.58085476808177e-05,
+      "loss": 0.1016,
+      "step": 10339
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.580777690370837e-05,
+      "loss": 1.1641,
+      "step": 10340
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.5807006067247442e-05,
+      "loss": 0.3027,
+      "step": 10341
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.580623517143914e-05,
+      "loss": 0.2373,
+      "step": 10342
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.58054642162877e-05,
+      "loss": 0.0432,
+      "step": 10343
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.580469320179736e-05,
+      "loss": 0.0061,
+      "step": 10344
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.5803922127972356e-05,
+      "loss": 0.0001,
+      "step": 10345
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.5803150994816914e-05,
+      "loss": 0.2314,
+      "step": 10346
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.580237980233527e-05,
+      "loss": 0.0154,
+      "step": 10347
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.580160855053167e-05,
+      "loss": 0.4551,
+      "step": 10348
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.580083723941034e-05,
+      "loss": 0.0003,
+      "step": 10349
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.5800065868975517e-05,
+      "loss": 0.0085,
+      "step": 10350
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.5799294439231434e-05,
+      "loss": 0.0003,
+      "step": 10351
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.5798522950182337e-05,
+      "loss": 0.0002,
+      "step": 10352
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.579775140183245e-05,
+      "loss": 0.7266,
+      "step": 10353
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.579697979418602e-05,
+      "loss": 0.1484,
+      "step": 10354
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.5796208127247277e-05,
+      "loss": 0.5195,
+      "step": 10355
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.5795436401020463e-05,
+      "loss": 0.3477,
+      "step": 10356
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.579466461550982e-05,
+      "loss": 0.1904,
+      "step": 10357
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.5793892770719584e-05,
+      "loss": 0.0001,
+      "step": 10358
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.5793120866653984e-05,
+      "loss": 0.6094,
+      "step": 10359
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.579234890331727e-05,
+      "loss": 0.0002,
+      "step": 10360
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.5791576880713677e-05,
+      "loss": 0.0461,
+      "step": 10361
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.5790804798847443e-05,
+      "loss": 0.1445,
+      "step": 10362
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.579003265772281e-05,
+      "loss": 0.0,
+      "step": 10363
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.578926045734402e-05,
+      "loss": 0.0001,
+      "step": 10364
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.578848819771531e-05,
+      "loss": 0.0,
+      "step": 10365
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.578771587884093e-05,
+      "loss": 0.1602,
+      "step": 10366
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.5786943500725105e-05,
+      "loss": 0.0001,
+      "step": 10367
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.5786171063372093e-05,
+      "loss": 0.0001,
+      "step": 10368
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.5785398566786127e-05,
+      "loss": 0.0,
+      "step": 10369
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.5784626010971447e-05,
+      "loss": 0.0035,
+      "step": 10370
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.57838533959323e-05,
+      "loss": 0.0,
+      "step": 10371
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.578308072167293e-05,
+      "loss": 0.0635,
+      "step": 10372
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.5782307988197586e-05,
+      "loss": 0.0,
+      "step": 10373
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.5781535195510488e-05,
+      "loss": 0.0001,
+      "step": 10374
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.5780762343615908e-05,
+      "loss": 0.0001,
+      "step": 10375
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.5779989432518076e-05,
+      "loss": 0.0,
+      "step": 10376
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.5779216462221233e-05,
+      "loss": 0.0001,
+      "step": 10377
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.5778443432729635e-05,
+      "loss": 0.0,
+      "step": 10378
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.5777670344047517e-05,
+      "loss": 0.0,
+      "step": 10379
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.577689719617913e-05,
+      "loss": 0.1216,
+      "step": 10380
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.577612398912872e-05,
+      "loss": 0.1187,
+      "step": 10381
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.5775350722900534e-05,
+      "loss": 0.1357,
+      "step": 10382
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.5774577397498812e-05,
+      "loss": 0.1807,
+      "step": 10383
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.5773804012927808e-05,
+      "loss": 0.3203,
+      "step": 10384
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.5773030569191766e-05,
+      "loss": 1.0703,
+      "step": 10385
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.577225706629493e-05,
+      "loss": 1.0859,
+      "step": 10386
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.5771483504241552e-05,
+      "loss": 0.1162,
+      "step": 10387
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.577070988303588e-05,
+      "loss": 0.6562,
+      "step": 10388
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.5769936202682164e-05,
+      "loss": 0.249,
+      "step": 10389
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.5769162463184646e-05,
+      "loss": 0.8398,
+      "step": 10390
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.5768388664547586e-05,
+      "loss": 0.016,
+      "step": 10391
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.5767614806775225e-05,
+      "loss": 0.3496,
+      "step": 10392
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.576684088987181e-05,
+      "loss": 0.0001,
+      "step": 10393
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.5766066913841594e-05,
+      "loss": 0.1318,
+      "step": 10394
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.5765292878688835e-05,
+      "loss": 0.0015,
+      "step": 10395
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.5764518784417774e-05,
+      "loss": 0.4277,
+      "step": 10396
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.576374463103267e-05,
+      "loss": 0.3828,
+      "step": 10397
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.5762970418537765e-05,
+      "loss": 0.0493,
+      "step": 10398
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.5762196146937317e-05,
+      "loss": 0.1084,
+      "step": 10399
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.5761421816235576e-05,
+      "loss": 0.0135,
+      "step": 10400
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.5760647426436796e-05,
+      "loss": 0.4004,
+      "step": 10401
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.5759872977545228e-05,
+      "loss": 0.4414,
+      "step": 10402
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.575909846956512e-05,
+      "loss": 0.252,
+      "step": 10403
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.5758323902500737e-05,
+      "loss": 0.103,
+      "step": 10404
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.5757549276356326e-05,
+      "loss": 0.0002,
+      "step": 10405
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.5756774591136138e-05,
+      "loss": 0.0381,
+      "step": 10406
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.5755999846844434e-05,
+      "loss": 0.0967,
+      "step": 10407
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.575522504348546e-05,
+      "loss": 0.0003,
+      "step": 10408
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.5754450181063476e-05,
+      "loss": 0.2578,
+      "step": 10409
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.575367525958274e-05,
+      "loss": 0.0001,
+      "step": 10410
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.5752900279047502e-05,
+      "loss": 0.0,
+      "step": 10411
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.5752125239462024e-05,
+      "loss": 0.0001,
+      "step": 10412
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.5751350140830553e-05,
+      "loss": 0.0232,
+      "step": 10413
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.5750574983157355e-05,
+      "loss": 0.0001,
+      "step": 10414
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.574979976644668e-05,
+      "loss": 0.051,
+      "step": 10415
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.5749024490702787e-05,
+      "loss": 0.0432,
+      "step": 10416
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.574824915592994e-05,
+      "loss": 0.0003,
+      "step": 10417
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.574747376213239e-05,
+      "loss": 0.0991,
+      "step": 10418
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.574669830931439e-05,
+      "loss": 0.0001,
+      "step": 10419
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.5745922797480212e-05,
+      "loss": 0.0003,
+      "step": 10420
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.57451472266341e-05,
+      "loss": 0.0,
+      "step": 10421
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.574437159678033e-05,
+      "loss": 0.0182,
+      "step": 10422
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.5743595907923146e-05,
+      "loss": 0.0001,
+      "step": 10423
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.5742820160066816e-05,
+      "loss": 0.0001,
+      "step": 10424
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.5742044353215597e-05,
+      "loss": 0.0002,
+      "step": 10425
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.574126848737375e-05,
+      "loss": 0.0002,
+      "step": 10426
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.5740492562545534e-05,
+      "loss": 0.0001,
+      "step": 10427
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.5739716578735214e-05,
+      "loss": 0.168,
+      "step": 10428
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.5738940535947048e-05,
+      "loss": 0.1021,
+      "step": 10429
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.5738164434185306e-05,
+      "loss": 0.0947,
+      "step": 10430
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.5737388273454232e-05,
+      "loss": 0.1338,
+      "step": 10431
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.573661205375811e-05,
+      "loss": 0.1582,
+      "step": 10432
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.5735835775101186e-05,
+      "loss": 0.0001,
+      "step": 10433
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.5735059437487728e-05,
+      "loss": 1.2422,
+      "step": 10434
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.5734283040922004e-05,
+      "loss": 0.4902,
+      "step": 10435
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.573350658540827e-05,
+      "loss": 0.2305,
+      "step": 10436
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.5732730070950797e-05,
+      "loss": 0.4922,
+      "step": 10437
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.573195349755385e-05,
+      "loss": 0.5859,
+      "step": 10438
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.5731176865221685e-05,
+      "loss": 0.4824,
+      "step": 10439
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.5730400173958575e-05,
+      "loss": 0.4199,
+      "step": 10440
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.572962342376878e-05,
+      "loss": 1.0078,
+      "step": 10441
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.5728846614656566e-05,
+      "loss": 0.9141,
+      "step": 10442
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.57280697466262e-05,
+      "loss": 0.4277,
+      "step": 10443
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.5727292819681953e-05,
+      "loss": 0.0094,
+      "step": 10444
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.5726515833828083e-05,
+      "loss": 0.0688,
+      "step": 10445
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.5725738789068864e-05,
+      "loss": 0.001,
+      "step": 10446
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.572496168540856e-05,
+      "loss": 0.0378,
+      "step": 10447
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.5724184522851435e-05,
+      "loss": 0.0537,
+      "step": 10448
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.5723407301401764e-05,
+      "loss": 1.6328,
+      "step": 10449
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.572263002106381e-05,
+      "loss": 0.625,
+      "step": 10450
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.5721852681841845e-05,
+      "loss": 0.0111,
+      "step": 10451
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.5721075283740142e-05,
+      "loss": 0.3926,
+      "step": 10452
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.5720297826762958e-05,
+      "loss": 1.2422,
+      "step": 10453
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.5719520310914565e-05,
+      "loss": 0.4141,
+      "step": 10454
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.5718742736199245e-05,
+      "loss": 0.0005,
+      "step": 10455
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.571796510262125e-05,
+      "loss": 0.1787,
+      "step": 10456
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.571718741018487e-05,
+      "loss": 0.043,
+      "step": 10457
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.571640965889436e-05,
+      "loss": 0.7734,
+      "step": 10458
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.5715631848754e-05,
+      "loss": 0.0,
+      "step": 10459
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.5714853979768053e-05,
+      "loss": 1.0859,
+      "step": 10460
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.5714076051940798e-05,
+      "loss": 0.0148,
+      "step": 10461
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.5713298065276507e-05,
+      "loss": 0.0001,
+      "step": 10462
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.5712520019779452e-05,
+      "loss": 0.0001,
+      "step": 10463
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.57117419154539e-05,
+      "loss": 0.1069,
+      "step": 10464
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.5710963752304127e-05,
+      "loss": 0.0001,
+      "step": 10465
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.571018553033441e-05,
+      "loss": 0.0845,
+      "step": 10466
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.570940724954902e-05,
+      "loss": 0.0001,
+      "step": 10467
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.570862890995223e-05,
+      "loss": 0.0116,
+      "step": 10468
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.5707850511548318e-05,
+      "loss": 0.0,
+      "step": 10469
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.5707072054341555e-05,
+      "loss": 0.0001,
+      "step": 10470
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.5706293538336217e-05,
+      "loss": 0.0001,
+      "step": 10471
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.5705514963536574e-05,
+      "loss": 0.0001,
+      "step": 10472
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.5704736329946916e-05,
+      "loss": 0.0001,
+      "step": 10473
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.5703957637571504e-05,
+      "loss": 0.0,
+      "step": 10474
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.5703178886414617e-05,
+      "loss": 0.0001,
+      "step": 10475
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.570240007648054e-05,
+      "loss": 0.0001,
+      "step": 10476
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.5701621207773542e-05,
+      "loss": 0.0001,
+      "step": 10477
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.57008422802979e-05,
+      "loss": 0.0001,
+      "step": 10478
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.5700063294057903e-05,
+      "loss": 0.0001,
+      "step": 10479
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5699284249057813e-05,
+      "loss": 0.1377,
+      "step": 10480
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5698505145301917e-05,
+      "loss": 0.0918,
+      "step": 10481
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5697725982794488e-05,
+      "loss": 0.1816,
+      "step": 10482
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5696946761539815e-05,
+      "loss": 0.1592,
+      "step": 10483
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5696167481542164e-05,
+      "loss": 0.2168,
+      "step": 10484
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5695388142805823e-05,
+      "loss": 0.8789,
+      "step": 10485
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5694608745335073e-05,
+      "loss": 0.4531,
+      "step": 10486
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.569382928913419e-05,
+      "loss": 0.2373,
+      "step": 10487
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.569304977420745e-05,
+      "loss": 0.0879,
+      "step": 10488
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.569227020055915e-05,
+      "loss": 0.0972,
+      "step": 10489
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.569149056819355e-05,
+      "loss": 0.0996,
+      "step": 10490
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5690710877114945e-05,
+      "loss": 0.0046,
+      "step": 10491
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5689931127327613e-05,
+      "loss": 0.0013,
+      "step": 10492
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5689151318835836e-05,
+      "loss": 0.3418,
+      "step": 10493
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.56883714516439e-05,
+      "loss": 0.1973,
+      "step": 10494
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.568759152575608e-05,
+      "loss": 1.5938,
+      "step": 10495
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5686811541176668e-05,
+      "loss": 0.127,
+      "step": 10496
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.568603149790994e-05,
+      "loss": 0.0272,
+      "step": 10497
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5685251395960187e-05,
+      "loss": 0.0347,
+      "step": 10498
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5684471235331684e-05,
+      "loss": 0.5977,
+      "step": 10499
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5683691016028718e-05,
+      "loss": 0.0004,
+      "step": 10500
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5682910738055577e-05,
+      "loss": 0.5,
+      "step": 10501
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5682130401416545e-05,
+      "loss": 1.4688,
+      "step": 10502
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5681350006115908e-05,
+      "loss": 0.0376,
+      "step": 10503
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5680569552157947e-05,
+      "loss": 0.0011,
+      "step": 10504
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5679789039546953e-05,
+      "loss": 0.2119,
+      "step": 10505
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5679008468287208e-05,
+      "loss": 0.0078,
+      "step": 10506
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5678227838383004e-05,
+      "loss": 0.1836,
+      "step": 10507
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5677447149838625e-05,
+      "loss": 0.0058,
+      "step": 10508
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.567666640265836e-05,
+      "loss": 0.0046,
+      "step": 10509
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5675885596846487e-05,
+      "loss": 0.0537,
+      "step": 10510
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5675104732407304e-05,
+      "loss": 0.1338,
+      "step": 10511
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5674323809345103e-05,
+      "loss": 0.0,
+      "step": 10512
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5673542827664156e-05,
+      "loss": 0.0001,
+      "step": 10513
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.567276178736877e-05,
+      "loss": 0.0004,
+      "step": 10514
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5671980688463222e-05,
+      "loss": 0.0001,
+      "step": 10515
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5671199530951805e-05,
+      "loss": 0.0332,
+      "step": 10516
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5670418314838813e-05,
+      "loss": 0.0002,
+      "step": 10517
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5669637040128528e-05,
+      "loss": 0.0001,
+      "step": 10518
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5668855706825245e-05,
+      "loss": 0.0003,
+      "step": 10519
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5668074314933258e-05,
+      "loss": 0.0001,
+      "step": 10520
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5667292864456854e-05,
+      "loss": 0.0002,
+      "step": 10521
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5666511355400323e-05,
+      "loss": 0.0,
+      "step": 10522
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.566572978776796e-05,
+      "loss": 0.0007,
+      "step": 10523
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5664948161564054e-05,
+      "loss": 0.0001,
+      "step": 10524
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5664166476792895e-05,
+      "loss": 0.0001,
+      "step": 10525
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5663384733458787e-05,
+      "loss": 0.0011,
+      "step": 10526
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.566260293156601e-05,
+      "loss": 0.0003,
+      "step": 10527
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5661821071118868e-05,
+      "loss": 0.125,
+      "step": 10528
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5661039152121646e-05,
+      "loss": 0.0,
+      "step": 10529
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5660257174578645e-05,
+      "loss": 0.1147,
+      "step": 10530
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5659475138494155e-05,
+      "loss": 0.1226,
+      "step": 10531
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5658693043872468e-05,
+      "loss": 0.0001,
+      "step": 10532
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5657910890717887e-05,
+      "loss": 0.1182,
+      "step": 10533
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5657128679034698e-05,
+      "loss": 1.6797,
+      "step": 10534
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5656346408827204e-05,
+      "loss": 0.0099,
+      "step": 10535
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5655564080099695e-05,
+      "loss": 1.0,
+      "step": 10536
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5654781692856473e-05,
+      "loss": 0.7461,
+      "step": 10537
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5653999247101828e-05,
+      "loss": 1.9844,
+      "step": 10538
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.565321674284007e-05,
+      "loss": 0.7461,
+      "step": 10539
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.565243418007548e-05,
+      "loss": 0.8438,
+      "step": 10540
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.565165155881236e-05,
+      "loss": 0.6328,
+      "step": 10541
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5650868879055012e-05,
+      "loss": 0.0029,
+      "step": 10542
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5650086140807734e-05,
+      "loss": 0.7695,
+      "step": 10543
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.564930334407482e-05,
+      "loss": 0.3555,
+      "step": 10544
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5648520488860574e-05,
+      "loss": 0.0006,
+      "step": 10545
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5647737575169293e-05,
+      "loss": 0.0118,
+      "step": 10546
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.5646954603005277e-05,
+      "loss": 0.4707,
+      "step": 10547
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.5646171572372823e-05,
+      "loss": 0.0006,
+      "step": 10548
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.564538848327623e-05,
+      "loss": 0.3457,
+      "step": 10549
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.5644605335719803e-05,
+      "loss": 0.0006,
+      "step": 10550
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.5643822129707846e-05,
+      "loss": 0.004,
+      "step": 10551
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.564303886524465e-05,
+      "loss": 0.5078,
+      "step": 10552
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.5642255542334522e-05,
+      "loss": 0.0013,
+      "step": 10553
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.564147216098176e-05,
+      "loss": 0.002,
+      "step": 10554
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.5640688721190672e-05,
+      "loss": 0.04,
+      "step": 10555
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.5639905222965562e-05,
+      "loss": 0.0001,
+      "step": 10556
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.563912166631072e-05,
+      "loss": 0.0001,
+      "step": 10557
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.563833805123046e-05,
+      "loss": 1.125,
+      "step": 10558
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.5637554377729082e-05,
+      "loss": 0.0009,
+      "step": 10559
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.563677064581089e-05,
+      "loss": 0.0001,
+      "step": 10560
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.5635986855480185e-05,
+      "loss": 0.4473,
+      "step": 10561
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.5635203006741277e-05,
+      "loss": 0.0001,
+      "step": 10562
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.5634419099598466e-05,
+      "loss": 0.0004,
+      "step": 10563
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.563363513405606e-05,
+      "loss": 0.5156,
+      "step": 10564
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.563285111011836e-05,
+      "loss": 0.2256,
+      "step": 10565
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.5632067027789675e-05,
+      "loss": 0.0591,
+      "step": 10566
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.563128288707431e-05,
+      "loss": 1.8281,
+      "step": 10567
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.563049868797657e-05,
+      "loss": 0.0002,
+      "step": 10568
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.5629714430500766e-05,
+      "loss": 0.0002,
+      "step": 10569
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.5628930114651198e-05,
+      "loss": 0.0007,
+      "step": 10570
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.562814574043218e-05,
+      "loss": 0.0001,
+      "step": 10571
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.562736130784801e-05,
+      "loss": 0.0002,
+      "step": 10572
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.5626576816903003e-05,
+      "loss": 0.0001,
+      "step": 10573
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.5625792267601468e-05,
+      "loss": 0.0635,
+      "step": 10574
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.562500765994771e-05,
+      "loss": 0.0,
+      "step": 10575
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.562422299394604e-05,
+      "loss": 0.0001,
+      "step": 10576
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.5623438269600768e-05,
+      "loss": 0.4492,
+      "step": 10577
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.5622653486916198e-05,
+      "loss": 0.0001,
+      "step": 10578
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.562186864589664e-05,
+      "loss": 0.1001,
+      "step": 10579
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.562108374654641e-05,
+      "loss": 0.0554,
+      "step": 10580
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.5620298788869814e-05,
+      "loss": 0.1826,
+      "step": 10581
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.5619513772871167e-05,
+      "loss": 0.167,
+      "step": 10582
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.5618728698554776e-05,
+      "loss": 0.2266,
+      "step": 10583
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.561794356592495e-05,
+      "loss": 0.2412,
+      "step": 10584
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.5617158374986007e-05,
+      "loss": 0.5977,
+      "step": 10585
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.5616373125742255e-05,
+      "loss": 0.3281,
+      "step": 10586
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.5615587818198008e-05,
+      "loss": 0.4004,
+      "step": 10587
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.5614802452357572e-05,
+      "loss": 0.7305,
+      "step": 10588
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.561401702822527e-05,
+      "loss": 0.4492,
+      "step": 10589
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.5613231545805414e-05,
+      "loss": 0.3008,
+      "step": 10590
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.5612446005102308e-05,
+      "loss": 0.6523,
+      "step": 10591
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.5611660406120275e-05,
+      "loss": 0.0093,
+      "step": 10592
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.5610874748863622e-05,
+      "loss": 1.9375,
+      "step": 10593
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.5610089033336673e-05,
+      "loss": 0.2676,
+      "step": 10594
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.5609303259543733e-05,
+      "loss": 0.625,
+      "step": 10595
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.5608517427489123e-05,
+      "loss": 0.459,
+      "step": 10596
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.560773153717716e-05,
+      "loss": 0.0092,
+      "step": 10597
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.560694558861215e-05,
+      "loss": 0.8594,
+      "step": 10598
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.560615958179842e-05,
+      "loss": 0.0001,
+      "step": 10599
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.5605373516740283e-05,
+      "loss": 0.4785,
+      "step": 10600
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.5604587393442054e-05,
+      "loss": 0.0408,
+      "step": 10601
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.560380121190805e-05,
+      "loss": 1.1875,
+      "step": 10602
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.5603014972142594e-05,
+      "loss": 0.0713,
+      "step": 10603
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.5602228674149996e-05,
+      "loss": 0.0001,
+      "step": 10604
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.5601442317934576e-05,
+      "loss": 0.8438,
+      "step": 10605
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.560065590350065e-05,
+      "loss": 0.008,
+      "step": 10606
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.5599869430852545e-05,
+      "loss": 0.0003,
+      "step": 10607
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.5599082899994574e-05,
+      "loss": 0.0003,
+      "step": 10608
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.5598296310931057e-05,
+      "loss": 0.1147,
+      "step": 10609
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.5597509663666318e-05,
+      "loss": 0.0038,
+      "step": 10610
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.5596722958204668e-05,
+      "loss": 0.4258,
+      "step": 10611
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.5595936194550436e-05,
+      "loss": 0.5625,
+      "step": 10612
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.5595149372707933e-05,
+      "loss": 0.1147,
+      "step": 10613
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.5594362492681487e-05,
+      "loss": 0.4141,
+      "step": 10614
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.5593575554475418e-05,
+      "loss": 0.1504,
+      "step": 10615
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.5592788558094048e-05,
+      "loss": 0.2236,
+      "step": 10616
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.55920015035417e-05,
+      "loss": 0.0269,
+      "step": 10617
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.5591214390822694e-05,
+      "loss": 0.0,
+      "step": 10618
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.559042721994135e-05,
+      "loss": 0.0,
+      "step": 10619
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.5589639990901995e-05,
+      "loss": 0.0001,
+      "step": 10620
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.558885270370895e-05,
+      "loss": 0.2871,
+      "step": 10621
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.558806535836654e-05,
+      "loss": 0.0001,
+      "step": 10622
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.5587277954879088e-05,
+      "loss": 0.0001,
+      "step": 10623
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.5586490493250916e-05,
+      "loss": 0.0001,
+      "step": 10624
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.5585702973486354e-05,
+      "loss": 0.0162,
+      "step": 10625
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.5584915395589717e-05,
+      "loss": 0.0006,
+      "step": 10626
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.5584127759565342e-05,
+      "loss": 0.0,
+      "step": 10627
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.5583340065417546e-05,
+      "loss": 0.1406,
+      "step": 10628
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.5582552313150656e-05,
+      "loss": 0.1797,
+      "step": 10629
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.5581764502769e-05,
+      "loss": 0.166,
+      "step": 10630
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.5580976634276904e-05,
+      "loss": 0.1152,
+      "step": 10631
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.5580188707678694e-05,
+      "loss": 0.1631,
+      "step": 10632
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.5579400722978695e-05,
+      "loss": 0.1396,
+      "step": 10633
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.5578612680181242e-05,
+      "loss": 1.2109,
+      "step": 10634
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.557782457929065e-05,
+      "loss": 0.6953,
+      "step": 10635
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.5577036420311255e-05,
+      "loss": 0.7852,
+      "step": 10636
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.557624820324739e-05,
+      "loss": 0.7227,
+      "step": 10637
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.5575459928103372e-05,
+      "loss": 0.5781,
+      "step": 10638
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.5574671594883533e-05,
+      "loss": 0.9648,
+      "step": 10639
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.557388320359221e-05,
+      "loss": 1.1562,
+      "step": 10640
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.5573094754233723e-05,
+      "loss": 0.248,
+      "step": 10641
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.5572306246812406e-05,
+      "loss": 1.3047,
+      "step": 10642
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.557151768133259e-05,
+      "loss": 0.1045,
+      "step": 10643
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.5570729057798606e-05,
+      "loss": 0.6602,
+      "step": 10644
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.5569940376214783e-05,
+      "loss": 0.9141,
+      "step": 10645
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.556915163658545e-05,
+      "loss": 0.5547,
+      "step": 10646
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.5568362838914943e-05,
+      "loss": 0.3145,
+      "step": 10647
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.5567573983207593e-05,
+      "loss": 0.0186,
+      "step": 10648
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.5566785069467726e-05,
+      "loss": 1.1641,
+      "step": 10649
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.556599609769968e-05,
+      "loss": 0.0261,
+      "step": 10650
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.556520706790779e-05,
+      "loss": 1.3359,
+      "step": 10651
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.5564417980096384e-05,
+      "loss": 0.6953,
+      "step": 10652
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.5563628834269797e-05,
+      "loss": 0.0001,
+      "step": 10653
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.556283963043236e-05,
+      "loss": 0.0022,
+      "step": 10654
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.5562050368588413e-05,
+      "loss": 0.0165,
+      "step": 10655
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.5561261048742286e-05,
+      "loss": 0.0002,
+      "step": 10656
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.5560471670898315e-05,
+      "loss": 0.0002,
+      "step": 10657
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.555968223506083e-05,
+      "loss": 0.0002,
+      "step": 10658
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.5558892741234173e-05,
+      "loss": 0.0002,
+      "step": 10659
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.555810318942268e-05,
+      "loss": 0.0004,
+      "step": 10660
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.555731357963068e-05,
+      "loss": 0.0002,
+      "step": 10661
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.5556523911862516e-05,
+      "loss": 0.001,
+      "step": 10662
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.555573418612252e-05,
+      "loss": 0.0001,
+      "step": 10663
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.5554944402415035e-05,
+      "loss": 0.0859,
+      "step": 10664
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.5554154560744387e-05,
+      "loss": 0.0,
+      "step": 10665
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.555336466111492e-05,
+      "loss": 1.5234,
+      "step": 10666
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.555257470353098e-05,
+      "loss": 0.2393,
+      "step": 10667
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.555178468799689e-05,
+      "loss": 0.1177,
+      "step": 10668
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.5550994614517e-05,
+      "loss": 0.0024,
+      "step": 10669
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.5550204483095642e-05,
+      "loss": 0.0001,
+      "step": 10670
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.5549414293737153e-05,
+      "loss": 0.0,
+      "step": 10671
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.5548624046445883e-05,
+      "loss": 0.0352,
+      "step": 10672
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.5547833741226162e-05,
+      "loss": 0.0258,
+      "step": 10673
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.5547043378082333e-05,
+      "loss": 0.0002,
+      "step": 10674
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.554625295701874e-05,
+      "loss": 0.0002,
+      "step": 10675
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.5545462478039717e-05,
+      "loss": 0.0947,
+      "step": 10676
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.554467194114961e-05,
+      "loss": 0.0,
+      "step": 10677
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.5543881346352757e-05,
+      "loss": 0.1318,
+      "step": 10678
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.5543090693653507e-05,
+      "loss": 0.127,
+      "step": 10679
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.5542299983056188e-05,
+      "loss": 0.106,
+      "step": 10680
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.5541509214565152e-05,
+      "loss": 0.1006,
+      "step": 10681
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.5540718388184744e-05,
+      "loss": 0.1543,
+      "step": 10682
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.55399275039193e-05,
+      "loss": 0.1133,
+      "step": 10683
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.5539136561773168e-05,
+      "loss": 0.543,
+      "step": 10684
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.5538345561750688e-05,
+      "loss": 0.1787,
+      "step": 10685
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.5537554503856205e-05,
+      "loss": 0.0427,
+      "step": 10686
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.5536763388094065e-05,
+      "loss": 0.8086,
+      "step": 10687
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.553597221446861e-05,
+      "loss": 1.0391,
+      "step": 10688
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.5535180982984183e-05,
+      "loss": 0.2002,
+      "step": 10689
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.5534389693645134e-05,
+      "loss": 0.0294,
+      "step": 10690
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.5533598346455805e-05,
+      "loss": 0.0544,
+      "step": 10691
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.5532806941420543e-05,
+      "loss": 0.0393,
+      "step": 10692
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.5532015478543698e-05,
+      "loss": 0.4746,
+      "step": 10693
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.5531223957829607e-05,
+      "loss": 0.4297,
+      "step": 10694
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.5530432379282624e-05,
+      "loss": 0.8906,
+      "step": 10695
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.5529640742907095e-05,
+      "loss": 0.0309,
+      "step": 10696
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.552884904870737e-05,
+      "loss": 0.0052,
+      "step": 10697
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.5528057296687784e-05,
+      "loss": 0.0002,
+      "step": 10698
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.55272654868527e-05,
+      "loss": 0.0002,
+      "step": 10699
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.552647361920646e-05,
+      "loss": 0.0005,
+      "step": 10700
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.5525681693753414e-05,
+      "loss": 0.0048,
+      "step": 10701
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.5524889710497904e-05,
+      "loss": 0.0,
+      "step": 10702
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.552409766944429e-05,
+      "loss": 0.0028,
+      "step": 10703
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.552330557059691e-05,
+      "loss": 0.3105,
+      "step": 10704
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.552251341396013e-05,
+      "loss": 0.0001,
+      "step": 10705
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.5521721199538286e-05,
+      "loss": 0.5469,
+      "step": 10706
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.5520928927335735e-05,
+      "loss": 0.0302,
+      "step": 10707
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.552013659735682e-05,
+      "loss": 0.0,
+      "step": 10708
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.5519344209605908e-05,
+      "loss": 0.0005,
+      "step": 10709
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.551855176408733e-05,
+      "loss": 0.0,
+      "step": 10710
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.5517759260805455e-05,
+      "loss": 0.3555,
+      "step": 10711
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.5516966699764625e-05,
+      "loss": 0.0001,
+      "step": 10712
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.5516174080969194e-05,
+      "loss": 0.0001,
+      "step": 10713
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.551538140442352e-05,
+      "loss": 0.0001,
+      "step": 10714
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.551458867013195e-05,
+      "loss": 0.21,
+      "step": 10715
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.5513795878098842e-05,
+      "loss": 0.0,
+      "step": 10716
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.5513003028328545e-05,
+      "loss": 0.0,
+      "step": 10717
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.551221012082542e-05,
+      "loss": 0.0,
+      "step": 10718
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.5511417155593814e-05,
+      "loss": 0.0,
+      "step": 10719
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.5510624132638084e-05,
+      "loss": 0.0,
+      "step": 10720
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.5509831051962587e-05,
+      "loss": 0.0,
+      "step": 10721
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.550903791357167e-05,
+      "loss": 0.0001,
+      "step": 10722
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.5508244717469705e-05,
+      "loss": 0.3008,
+      "step": 10723
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.550745146366103e-05,
+      "loss": 0.0,
+      "step": 10724
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.5506658152150014e-05,
+      "loss": 0.0001,
+      "step": 10725
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.5505864782941007e-05,
+      "loss": 0.0,
+      "step": 10726
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.5505071356038368e-05,
+      "loss": 0.0001,
+      "step": 10727
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.5504277871446456e-05,
+      "loss": 0.0001,
+      "step": 10728
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.550348432916962e-05,
+      "loss": 0.1963,
+      "step": 10729
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.550269072921223e-05,
+      "loss": 0.1162,
+      "step": 10730
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.5501897071578633e-05,
+      "loss": 0.0003,
+      "step": 10731
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.5501103356273198e-05,
+      "loss": 0.1367,
+      "step": 10732
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.5500309583300273e-05,
+      "loss": 0.1729,
+      "step": 10733
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.5499515752664227e-05,
+      "loss": 0.003,
+      "step": 10734
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.549872186436941e-05,
+      "loss": 0.4824,
+      "step": 10735
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.5497927918420192e-05,
+      "loss": 0.1016,
+      "step": 10736
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.5497133914820923e-05,
+      "loss": 0.2354,
+      "step": 10737
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.5496339853575967e-05,
+      "loss": 1.6016,
+      "step": 10738
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.549554573468969e-05,
+      "loss": 0.2695,
+      "step": 10739
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.5494751558166446e-05,
+      "loss": 0.4375,
+      "step": 10740
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.5493957324010596e-05,
+      "loss": 0.5586,
+      "step": 10741
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.549316303222651e-05,
+      "loss": 0.3086,
+      "step": 10742
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.5492368682818537e-05,
+      "loss": 0.8008,
+      "step": 10743
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.549157427579105e-05,
+      "loss": 0.0884,
+      "step": 10744
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.5490779811148406e-05,
+      "loss": 0.0483,
+      "step": 10745
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.548998528889497e-05,
+      "loss": 0.5391,
+      "step": 10746
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.5489190709035108e-05,
+      "loss": 0.5273,
+      "step": 10747
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.5488396071573177e-05,
+      "loss": 0.0037,
+      "step": 10748
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.5487601376513544e-05,
+      "loss": 1.4609,
+      "step": 10749
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.5486806623860577e-05,
+      "loss": 0.0447,
+      "step": 10750
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.5486011813618636e-05,
+      "loss": 0.002,
+      "step": 10751
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.548521694579208e-05,
+      "loss": 0.3477,
+      "step": 10752
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.548442202038529e-05,
+      "loss": 0.0003,
+      "step": 10753
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.548362703740262e-05,
+      "loss": 1.3359,
+      "step": 10754
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.548283199684843e-05,
+      "loss": 0.0001,
+      "step": 10755
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.54820368987271e-05,
+      "loss": 0.0,
+      "step": 10756
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.5481241743042988e-05,
+      "loss": 0.0181,
+      "step": 10757
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.548044652980046e-05,
+      "loss": 0.1875,
+      "step": 10758
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.547965125900389e-05,
+      "loss": 0.418,
+      "step": 10759
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.547885593065764e-05,
+      "loss": 0.0001,
+      "step": 10760
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.547806054476608e-05,
+      "loss": 0.0001,
+      "step": 10761
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.5477265101333572e-05,
+      "loss": 0.0182,
+      "step": 10762
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.5476469600364492e-05,
+      "loss": 0.0693,
+      "step": 10763
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.5475674041863204e-05,
+      "loss": 0.0001,
+      "step": 10764
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.547487842583408e-05,
+      "loss": 0.0,
+      "step": 10765
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.547408275228148e-05,
+      "loss": 0.0015,
+      "step": 10766
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.547328702120979e-05,
+      "loss": 0.0073,
+      "step": 10767
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.5472491232623365e-05,
+      "loss": 0.1816,
+      "step": 10768
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.547169538652658e-05,
+      "loss": 0.0067,
+      "step": 10769
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.5470899482923808e-05,
+      "loss": 0.0,
+      "step": 10770
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.5470103521819418e-05,
+      "loss": 0.0,
+      "step": 10771
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.546930750321778e-05,
+      "loss": 0.0,
+      "step": 10772
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.5468511427123264e-05,
+      "loss": 0.0,
+      "step": 10773
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.5467715293540245e-05,
+      "loss": 0.625,
+      "step": 10774
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.5466919102473098e-05,
+      "loss": 0.1699,
+      "step": 10775
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.546612285392619e-05,
+      "loss": 0.0811,
+      "step": 10776
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.546532654790389e-05,
+      "loss": 0.0,
+      "step": 10777
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.546453018441058e-05,
+      "loss": 0.0918,
+      "step": 10778
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.546373376345063e-05,
+      "loss": 0.0,
+      "step": 10779
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.546293728502841e-05,
+      "loss": 0.0003,
+      "step": 10780
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.5462140749148296e-05,
+      "loss": 0.0043,
+      "step": 10781
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.546134415581467e-05,
+      "loss": 0.1611,
+      "step": 10782
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.5460547505031892e-05,
+      "loss": 0.1226,
+      "step": 10783
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.5459750796804347e-05,
+      "loss": 1.1094,
+      "step": 10784
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.5458954031136408e-05,
+      "loss": 0.3613,
+      "step": 10785
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.5458157208032453e-05,
+      "loss": 0.5625,
+      "step": 10786
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.5457360327496848e-05,
+      "loss": 0.7734,
+      "step": 10787
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.545656338953398e-05,
+      "loss": 0.1924,
+      "step": 10788
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.545576639414822e-05,
+      "loss": 0.5742,
+      "step": 10789
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.545496934134395e-05,
+      "loss": 0.0,
+      "step": 10790
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.5454172231125538e-05,
+      "loss": 0.1104,
+      "step": 10791
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.545337506349737e-05,
+      "loss": 0.2852,
+      "step": 10792
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.5452577838463814e-05,
+      "loss": 0.0002,
+      "step": 10793
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.545178055602926e-05,
+      "loss": 0.9648,
+      "step": 10794
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.545098321619808e-05,
+      "loss": 0.0001,
+      "step": 10795
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.545018581897465e-05,
+      "loss": 0.9453,
+      "step": 10796
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.5449388364363356e-05,
+      "loss": 0.0002,
+      "step": 10797
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.5448590852368575e-05,
+      "loss": 0.0001,
+      "step": 10798
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.5447793282994675e-05,
+      "loss": 0.2871,
+      "step": 10799
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.5446995656246055e-05,
+      "loss": 0.0002,
+      "step": 10800
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.5446197972127088e-05,
+      "loss": 0.0001,
+      "step": 10801
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.5445400230642147e-05,
+      "loss": 0.0547,
+      "step": 10802
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.544460243179562e-05,
+      "loss": 0.0811,
+      "step": 10803
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.5443804575591888e-05,
+      "loss": 0.0,
+      "step": 10804
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.5443006662035334e-05,
+      "loss": 0.0008,
+      "step": 10805
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.544220869113033e-05,
+      "loss": 0.3105,
+      "step": 10806
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.544141066288127e-05,
+      "loss": 0.0001,
+      "step": 10807
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.544061257729253e-05,
+      "loss": 0.0001,
+      "step": 10808
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.5439814434368495e-05,
+      "loss": 0.064,
+      "step": 10809
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.5439016234113546e-05,
+      "loss": 0.0103,
+      "step": 10810
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.5438217976532067e-05,
+      "loss": 0.0001,
+      "step": 10811
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.5437419661628443e-05,
+      "loss": 0.0588,
+      "step": 10812
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.543662128940706e-05,
+      "loss": 0.064,
+      "step": 10813
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.5435822859872298e-05,
+      "loss": 0.0,
+      "step": 10814
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.543502437302854e-05,
+      "loss": 0.1147,
+      "step": 10815
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.543422582888018e-05,
+      "loss": 0.0771,
+      "step": 10816
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.5433427227431596e-05,
+      "loss": 0.0991,
+      "step": 10817
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.5432628568687175e-05,
+      "loss": 0.0059,
+      "step": 10818
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.5431829852651302e-05,
+      "loss": 0.0,
+      "step": 10819
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.543103107932836e-05,
+      "loss": 0.0,
+      "step": 10820
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.5430232248722745e-05,
+      "loss": 0.0001,
+      "step": 10821
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.542943336083884e-05,
+      "loss": 0.0,
+      "step": 10822
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.5428634415681028e-05,
+      "loss": 0.0,
+      "step": 10823
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.54278354132537e-05,
+      "loss": 0.0,
+      "step": 10824
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.542703635356124e-05,
+      "loss": 0.0,
+      "step": 10825
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.5426237236608044e-05,
+      "loss": 0.0,
+      "step": 10826
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.542543806239849e-05,
+      "loss": 0.0001,
+      "step": 10827
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.5424638830936974e-05,
+      "loss": 0.0,
+      "step": 10828
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.5423839542227882e-05,
+      "loss": 0.0001,
+      "step": 10829
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.5423040196275606e-05,
+      "loss": 0.0,
+      "step": 10830
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.5422240793084536e-05,
+      "loss": 0.0962,
+      "step": 10831
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.5421441332659057e-05,
+      "loss": 0.1689,
+      "step": 10832
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.5420641815003563e-05,
+      "loss": 0.1035,
+      "step": 10833
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.5419842240122445e-05,
+      "loss": 0.8281,
+      "step": 10834
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.541904260802009e-05,
+      "loss": 0.1079,
+      "step": 10835
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.54182429187009e-05,
+      "loss": 0.0771,
+      "step": 10836
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.5417443172169248e-05,
+      "loss": 1.3125,
+      "step": 10837
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.5416643368429544e-05,
+      "loss": 0.6523,
+      "step": 10838
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.5415843507486173e-05,
+      "loss": 0.4336,
+      "step": 10839
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.541504358934352e-05,
+      "loss": 0.1108,
+      "step": 10840
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.5414243614005993e-05,
+      "loss": 0.0193,
+      "step": 10841
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.5413443581477975e-05,
+      "loss": 0.3613,
+      "step": 10842
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.5412643491763862e-05,
+      "loss": 0.3984,
+      "step": 10843
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.5411843344868042e-05,
+      "loss": 0.0002,
+      "step": 10844
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.541104314079492e-05,
+      "loss": 0.0349,
+      "step": 10845
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.541024287954888e-05,
+      "loss": 0.416,
+      "step": 10846
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.5409442561134326e-05,
+      "loss": 0.015,
+      "step": 10847
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.5408642185555646e-05,
+      "loss": 0.6719,
+      "step": 10848
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.5407841752817242e-05,
+      "loss": 0.5195,
+      "step": 10849
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.5407041262923497e-05,
+      "loss": 0.0002,
+      "step": 10850
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.5406240715878823e-05,
+      "loss": 0.2715,
+      "step": 10851
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.5405440111687605e-05,
+      "loss": 0.0003,
+      "step": 10852
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.5404639450354242e-05,
+      "loss": 0.0015,
+      "step": 10853
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.5403838731883134e-05,
+      "loss": 1.2188,
+      "step": 10854
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.5403037956278678e-05,
+      "loss": 0.004,
+      "step": 10855
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.5402237123545268e-05,
+      "loss": 0.0002,
+      "step": 10856
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.5401436233687302e-05,
+      "loss": 0.0003,
+      "step": 10857
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.5400635286709178e-05,
+      "loss": 0.6914,
+      "step": 10858
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.53998342826153e-05,
+      "loss": 0.015,
+      "step": 10859
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.5399033221410064e-05,
+      "loss": 0.1187,
+      "step": 10860
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.5398232103097863e-05,
+      "loss": 0.0115,
+      "step": 10861
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.5397430927683105e-05,
+      "loss": 0.0001,
+      "step": 10862
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.5396629695170185e-05,
+      "loss": 0.0889,
+      "step": 10863
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.5395828405563504e-05,
+      "loss": 0.7266,
+      "step": 10864
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.5395027058867462e-05,
+      "loss": 0.3086,
+      "step": 10865
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.5394225655086462e-05,
+      "loss": 0.0126,
+      "step": 10866
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.53934241942249e-05,
+      "loss": 0.0,
+      "step": 10867
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.5392622676287183e-05,
+      "loss": 0.0,
+      "step": 10868
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.5391821101277714e-05,
+      "loss": 0.0,
+      "step": 10869
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.5391019469200884e-05,
+      "loss": 0.0002,
+      "step": 10870
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.5390217780061104e-05,
+      "loss": 0.0977,
+      "step": 10871
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.5389416033862773e-05,
+      "loss": 0.0,
+      "step": 10872
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.5388614230610298e-05,
+      "loss": 0.0,
+      "step": 10873
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.538781237030808e-05,
+      "loss": 0.0,
+      "step": 10874
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.5387010452960518e-05,
+      "loss": 0.0,
+      "step": 10875
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.5386208478572022e-05,
+      "loss": 0.0001,
+      "step": 10876
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.5385406447146994e-05,
+      "loss": 0.0,
+      "step": 10877
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.5384604358689835e-05,
+      "loss": 0.0001,
+      "step": 10878
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.538380221320496e-05,
+      "loss": 0.0977,
+      "step": 10879
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.5383000010696763e-05,
+      "loss": 0.1245,
+      "step": 10880
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.5382197751169652e-05,
+      "loss": 0.1475,
+      "step": 10881
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.5381395434628035e-05,
+      "loss": 0.1367,
+      "step": 10882
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.5380593061076317e-05,
+      "loss": 0.1914,
+      "step": 10883
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.5379790630518908e-05,
+      "loss": 1.0234,
+      "step": 10884
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.5378988142960208e-05,
+      "loss": 0.1187,
+      "step": 10885
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.537818559840463e-05,
+      "loss": 0.4512,
+      "step": 10886
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.5377382996856568e-05,
+      "loss": 0.1709,
+      "step": 10887
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.537658033832045e-05,
+      "loss": 1.375,
+      "step": 10888
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.537577762280067e-05,
+      "loss": 0.0708,
+      "step": 10889
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.537497485030164e-05,
+      "loss": 0.75,
+      "step": 10890
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.5374172020827766e-05,
+      "loss": 0.4805,
+      "step": 10891
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.537336913438346e-05,
+      "loss": 0.1484,
+      "step": 10892
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.537256619097313e-05,
+      "loss": 0.5781,
+      "step": 10893
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.5371763190601185e-05,
+      "loss": 0.0698,
+      "step": 10894
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.5370960133272034e-05,
+      "loss": 0.3105,
+      "step": 10895
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.5370157018990095e-05,
+      "loss": 0.3711,
+      "step": 10896
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.5369353847759765e-05,
+      "loss": 0.0001,
+      "step": 10897
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.536855061958546e-05,
+      "loss": 0.0004,
+      "step": 10898
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.53677473344716e-05,
+      "loss": 0.5078,
+      "step": 10899
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.536694399242258e-05,
+      "loss": 0.0393,
+      "step": 10900
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.536614059344282e-05,
+      "loss": 0.1963,
+      "step": 10901
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.5365337137536738e-05,
+      "loss": 0.0038,
+      "step": 10902
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.5364533624708734e-05,
+      "loss": 0.0015,
+      "step": 10903
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.536373005496323e-05,
+      "loss": 0.6602,
+      "step": 10904
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.5362926428304634e-05,
+      "loss": 0.0178,
+      "step": 10905
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.5362122744737367e-05,
+      "loss": 0.0065,
+      "step": 10906
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.536131900426583e-05,
+      "loss": 0.0079,
+      "step": 10907
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.5360515206894445e-05,
+      "loss": 0.0184,
+      "step": 10908
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.5359711352627623e-05,
+      "loss": 1.5078,
+      "step": 10909
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.535890744146978e-05,
+      "loss": 0.6758,
+      "step": 10910
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.535810347342533e-05,
+      "loss": 0.0,
+      "step": 10911
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.535729944849869e-05,
+      "loss": 0.0,
+      "step": 10912
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.535649536669427e-05,
+      "loss": 0.0361,
+      "step": 10913
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.5355691228016493e-05,
+      "loss": 0.0,
+      "step": 10914
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.5354887032469773e-05,
+      "loss": 0.0,
+      "step": 10915
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.5354082780058522e-05,
+      "loss": 0.0,
+      "step": 10916
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.5353278470787156e-05,
+      "loss": 0.0801,
+      "step": 10917
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.5352474104660096e-05,
+      "loss": 0.0,
+      "step": 10918
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.535166968168176e-05,
+      "loss": 0.0,
+      "step": 10919
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.5350865201856566e-05,
+      "loss": 0.9297,
+      "step": 10920
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.5350060665188933e-05,
+      "loss": 0.0001,
+      "step": 10921
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.5349256071683264e-05,
+      "loss": 0.0,
+      "step": 10922
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.5348451421344e-05,
+      "loss": 0.0,
+      "step": 10923
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.5347646714175545e-05,
+      "loss": 0.0,
+      "step": 10924
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.5346841950182324e-05,
+      "loss": 0.0,
+      "step": 10925
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.5346037129368754e-05,
+      "loss": 0.0,
+      "step": 10926
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.5345232251739256e-05,
+      "loss": 0.0001,
+      "step": 10927
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.5344427317298245e-05,
+      "loss": 0.1328,
+      "step": 10928
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.534362232605015e-05,
+      "loss": 0.1738,
+      "step": 10929
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.534281727799939e-05,
+      "loss": 0.127,
+      "step": 10930
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.5342012173150375e-05,
+      "loss": 0.1504,
+      "step": 10931
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.5341207011507542e-05,
+      "loss": 0.1797,
+      "step": 10932
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.5340401793075303e-05,
+      "loss": 0.1196,
+      "step": 10933
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.5339596517858085e-05,
+      "loss": 0.2383,
+      "step": 10934
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.53387911858603e-05,
+      "loss": 0.9961,
+      "step": 10935
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.5337985797086385e-05,
+      "loss": 0.7188,
+      "step": 10936
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.5337180351540758e-05,
+      "loss": 0.6992,
+      "step": 10937
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.5336374849227832e-05,
+      "loss": 0.0493,
+      "step": 10938
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.533556929015204e-05,
+      "loss": 0.4688,
+      "step": 10939
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.5334763674317812e-05,
+      "loss": 2.125,
+      "step": 10940
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.5333958001729553e-05,
+      "loss": 1.25,
+      "step": 10941
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.5333152272391707e-05,
+      "loss": 0.0342,
+      "step": 10942
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.5332346486308687e-05,
+      "loss": 0.2178,
+      "step": 10943
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.5331540643484922e-05,
+      "loss": 1.2188,
+      "step": 10944
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.5330734743924837e-05,
+      "loss": 0.0,
+      "step": 10945
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.532992878763286e-05,
+      "loss": 1.1797,
+      "step": 10946
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.532912277461341e-05,
+      "loss": 0.8945,
+      "step": 10947
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.5328316704870918e-05,
+      "loss": 0.0,
+      "step": 10948
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.5327510578409813e-05,
+      "loss": 0.1025,
+      "step": 10949
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.532670439523452e-05,
+      "loss": 0.0121,
+      "step": 10950
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.5325898155349463e-05,
+      "loss": 0.0005,
+      "step": 10951
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.5325091858759076e-05,
+      "loss": 0.123,
+      "step": 10952
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.532428550546778e-05,
+      "loss": 0.0036,
+      "step": 10953
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.5323479095479998e-05,
+      "loss": 0.4727,
+      "step": 10954
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.5322672628800174e-05,
+      "loss": 1.2578,
+      "step": 10955
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.5321866105432728e-05,
+      "loss": 0.0,
+      "step": 10956
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.5321059525382094e-05,
+      "loss": 0.0075,
+      "step": 10957
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.5320252888652695e-05,
+      "loss": 0.0,
+      "step": 10958
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.5319446195248958e-05,
+      "loss": 0.0011,
+      "step": 10959
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.5318639445175328e-05,
+      "loss": 0.1504,
+      "step": 10960
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.5317832638436214e-05,
+      "loss": 0.2539,
+      "step": 10961
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.531702577503607e-05,
+      "loss": 0.3105,
+      "step": 10962
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.5316218854979305e-05,
+      "loss": 0.0104,
+      "step": 10963
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.5315411878270365e-05,
+      "loss": 0.0,
+      "step": 10964
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.531460484491368e-05,
+      "loss": 0.0,
+      "step": 10965
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.5313797754913672e-05,
+      "loss": 0.0004,
+      "step": 10966
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.5312990608274785e-05,
+      "loss": 0.0027,
+      "step": 10967
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.5312183405001446e-05,
+      "loss": 0.0,
+      "step": 10968
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.5311376145098086e-05,
+      "loss": 0.013,
+      "step": 10969
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.5310568828569144e-05,
+      "loss": 0.1797,
+      "step": 10970
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.530976145541905e-05,
+      "loss": 0.0374,
+      "step": 10971
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.5308954025652232e-05,
+      "loss": 0.0,
+      "step": 10972
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.5308146539273133e-05,
+      "loss": 0.0029,
+      "step": 10973
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.5307338996286188e-05,
+      "loss": 0.3242,
+      "step": 10974
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.530653139669582e-05,
+      "loss": 0.0,
+      "step": 10975
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.5305723740506476e-05,
+      "loss": 0.0001,
+      "step": 10976
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.5304916027722588e-05,
+      "loss": 0.0,
+      "step": 10977
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.5304108258348592e-05,
+      "loss": 0.0,
+      "step": 10978
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.5303300432388916e-05,
+      "loss": 0.0,
+      "step": 10979
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.5302492549848007e-05,
+      "loss": 0.0,
+      "step": 10980
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.53016846107303e-05,
+      "loss": 0.0,
+      "step": 10981
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.5300876615040223e-05,
+      "loss": 0.0957,
+      "step": 10982
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.5300068562782223e-05,
+      "loss": 0.1211,
+      "step": 10983
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.5299260453960732e-05,
+      "loss": 0.7461,
+      "step": 10984
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.529845228858019e-05,
+      "loss": 0.5391,
+      "step": 10985
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.5297644066645037e-05,
+      "loss": 0.6562,
+      "step": 10986
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.5296835788159706e-05,
+      "loss": 0.5117,
+      "step": 10987
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.5296027453128637e-05,
+      "loss": 0.7461,
+      "step": 10988
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.5295219061556273e-05,
+      "loss": 0.0239,
+      "step": 10989
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.5294410613447058e-05,
+      "loss": 0.2393,
+      "step": 10990
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.5293602108805414e-05,
+      "loss": 1.0625,
+      "step": 10991
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.5292793547635796e-05,
+      "loss": 0.4551,
+      "step": 10992
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.529198492994264e-05,
+      "loss": 0.5703,
+      "step": 10993
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.5291176255730387e-05,
+      "loss": 0.0332,
+      "step": 10994
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.5290367525003476e-05,
+      "loss": 0.3242,
+      "step": 10995
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.5289558737766355e-05,
+      "loss": 1.1172,
+      "step": 10996
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.5288749894023453e-05,
+      "loss": 0.2188,
+      "step": 10997
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.5287940993779225e-05,
+      "loss": 0.1963,
+      "step": 10998
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.5287132037038103e-05,
+      "loss": 0.0003,
+      "step": 10999
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.528632302380453e-05,
+      "loss": 0.0002,
+      "step": 11000
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.528551395408296e-05,
+      "loss": 0.0001,
+      "step": 11001
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.5284704827877822e-05,
+      "loss": 1.0156,
+      "step": 11002
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.528389564519357e-05,
+      "loss": 0.0072,
+      "step": 11003
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.5283086406034642e-05,
+      "loss": 0.007,
+      "step": 11004
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.5282277110405484e-05,
+      "loss": 0.0,
+      "step": 11005
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.528146775831054e-05,
+      "loss": 0.0356,
+      "step": 11006
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.5280658349754255e-05,
+      "loss": 0.0,
+      "step": 11007
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.527984888474107e-05,
+      "loss": 0.0,
+      "step": 11008
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.5279039363275436e-05,
+      "loss": 1.5234,
+      "step": 11009
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.5278229785361794e-05,
+      "loss": 0.0003,
+      "step": 11010
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.5277420151004596e-05,
+      "loss": 0.0,
+      "step": 11011
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.527661046020828e-05,
+      "loss": 0.0,
+      "step": 11012
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.5275800712977298e-05,
+      "loss": 0.0041,
+      "step": 11013
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.5274990909316093e-05,
+      "loss": 0.1318,
+      "step": 11014
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.527418104922912e-05,
+      "loss": 0.0,
+      "step": 11015
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.527337113272081e-05,
+      "loss": 0.0,
+      "step": 11016
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.527256115979563e-05,
+      "loss": 0.1504,
+      "step": 11017
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.5271751130458016e-05,
+      "loss": 0.4258,
+      "step": 11018
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.5270941044712424e-05,
+      "loss": 0.0001,
+      "step": 11019
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.5270130902563295e-05,
+      "loss": 0.0,
+      "step": 11020
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.526932070401508e-05,
+      "loss": 0.0,
+      "step": 11021
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.5268510449072235e-05,
+      "loss": 0.0,
+      "step": 11022
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.52677001377392e-05,
+      "loss": 0.0001,
+      "step": 11023
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.5266889770020427e-05,
+      "loss": 0.3496,
+      "step": 11024
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.5266079345920373e-05,
+      "loss": 0.0001,
+      "step": 11025
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.5265268865443482e-05,
+      "loss": 0.0,
+      "step": 11026
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.5264458328594208e-05,
+      "loss": 0.0001,
+      "step": 11027
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.5263647735376998e-05,
+      "loss": 0.0,
+      "step": 11028
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.5262837085796305e-05,
+      "loss": 0.0002,
+      "step": 11029
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.526202637985658e-05,
+      "loss": 0.0938,
+      "step": 11030
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.5261215617562288e-05,
+      "loss": 0.1426,
+      "step": 11031
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.526040479891786e-05,
+      "loss": 0.1128,
+      "step": 11032
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.5259593923927762e-05,
+      "loss": 0.1187,
+      "step": 11033
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.5258782992596444e-05,
+      "loss": 0.2676,
+      "step": 11034
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.5257972004928357e-05,
+      "loss": 0.0574,
+      "step": 11035
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.525716096092796e-05,
+      "loss": 0.1387,
+      "step": 11036
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.5256349860599703e-05,
+      "loss": 1.7109,
+      "step": 11037
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.525553870394804e-05,
+      "loss": 0.9414,
+      "step": 11038
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.525472749097742e-05,
+      "loss": 0.3691,
+      "step": 11039
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.5253916221692316e-05,
+      "loss": 0.5469,
+      "step": 11040
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.5253104896097166e-05,
+      "loss": 1.2656,
+      "step": 11041
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.5252293514196432e-05,
+      "loss": 0.2637,
+      "step": 11042
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.5251482075994564e-05,
+      "loss": 0.7148,
+      "step": 11043
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.5250670581496026e-05,
+      "loss": 0.582,
+      "step": 11044
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.524985903070527e-05,
+      "loss": 0.2637,
+      "step": 11045
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.5249047423626758e-05,
+      "loss": 1.0312,
+      "step": 11046
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.5248235760264934e-05,
+      "loss": 0.7227,
+      "step": 11047
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.5247424040624273e-05,
+      "loss": 0.6797,
+      "step": 11048
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.5246612264709217e-05,
+      "loss": 0.0126,
+      "step": 11049
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.5245800432524233e-05,
+      "loss": 0.0085,
+      "step": 11050
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.524498854407378e-05,
+      "loss": 0.0,
+      "step": 11051
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.5244176599362305e-05,
+      "loss": 0.0048,
+      "step": 11052
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.524336459839428e-05,
+      "loss": 0.0,
+      "step": 11053
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.524255254117416e-05,
+      "loss": 0.0096,
+      "step": 11054
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.5241740427706402e-05,
+      "loss": 0.01,
+      "step": 11055
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.524092825799547e-05,
+      "loss": 2.1094,
+      "step": 11056
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.5240116032045817e-05,
+      "loss": 0.0,
+      "step": 11057
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.5239303749861913e-05,
+      "loss": 0.0537,
+      "step": 11058
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.5238491411448214e-05,
+      "loss": 0.0027,
+      "step": 11059
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.5237679016809172e-05,
+      "loss": 0.1309,
+      "step": 11060
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.5236866565949264e-05,
+      "loss": 0.0,
+      "step": 11061
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.5236054058872945e-05,
+      "loss": 0.0,
+      "step": 11062
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.5235241495584677e-05,
+      "loss": 0.8516,
+      "step": 11063
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.5234428876088915e-05,
+      "loss": 0.0002,
+      "step": 11064
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.5233616200390135e-05,
+      "loss": 0.0344,
+      "step": 11065
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.523280346849279e-05,
+      "loss": 0.0,
+      "step": 11066
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.523199068040135e-05,
+      "loss": 0.0574,
+      "step": 11067
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.5231177836120273e-05,
+      "loss": 0.0001,
+      "step": 11068
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.5230364935654023e-05,
+      "loss": 0.0001,
+      "step": 11069
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.522955197900707e-05,
+      "loss": 0.0007,
+      "step": 11070
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.5228738966183868e-05,
+      "loss": 0.0001,
+      "step": 11071
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.5227925897188893e-05,
+      "loss": 0.0,
+      "step": 11072
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.52271127720266e-05,
+      "loss": 0.0,
+      "step": 11073
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.522629959070146e-05,
+      "loss": 0.0,
+      "step": 11074
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.5225486353217943e-05,
+      "loss": 0.0,
+      "step": 11075
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.5224673059580503e-05,
+      "loss": 0.0001,
+      "step": 11076
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.522385970979362e-05,
+      "loss": 0.0002,
+      "step": 11077
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.5223046303861748e-05,
+      "loss": 0.103,
+      "step": 11078
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.5222232841789364e-05,
+      "loss": 0.0001,
+      "step": 11079
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.5221419323580925e-05,
+      "loss": 0.0002,
+      "step": 11080
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.5220605749240907e-05,
+      "loss": 0.1318,
+      "step": 11081
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.521979211877378e-05,
+      "loss": 0.105,
+      "step": 11082
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.5218978432184003e-05,
+      "loss": 0.2002,
+      "step": 11083
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.5218164689476044e-05,
+      "loss": 0.5391,
+      "step": 11084
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.5217350890654382e-05,
+      "loss": 0.3906,
+      "step": 11085
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.521653703572347e-05,
+      "loss": 0.8125,
+      "step": 11086
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.52157231246878e-05,
+      "loss": 0.332,
+      "step": 11087
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.5214909157551824e-05,
+      "loss": 0.543,
+      "step": 11088
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.5214095134320018e-05,
+      "loss": 0.0417,
+      "step": 11089
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.521328105499685e-05,
+      "loss": 0.0168,
+      "step": 11090
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.5212466919586798e-05,
+      "loss": 0.0057,
+      "step": 11091
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.5211652728094315e-05,
+      "loss": 0.2598,
+      "step": 11092
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.5210838480523895e-05,
+      "loss": 0.6289,
+      "step": 11093
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.5210024176879995e-05,
+      "loss": 0.377,
+      "step": 11094
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.5209209817167088e-05,
+      "loss": 0.7539,
+      "step": 11095
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.5208395401389653e-05,
+      "loss": 0.0001,
+      "step": 11096
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.520758092955215e-05,
+      "loss": 0.4082,
+      "step": 11097
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.5206766401659064e-05,
+      "loss": 1.9219,
+      "step": 11098
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.520595181771487e-05,
+      "loss": 0.6992,
+      "step": 11099
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.5205137177724024e-05,
+      "loss": 0.8438,
+      "step": 11100
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.5204322481691017e-05,
+      "loss": 0.0001,
+      "step": 11101
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.5203507729620316e-05,
+      "loss": 0.0,
+      "step": 11102
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.5202692921516396e-05,
+      "loss": 0.0,
+      "step": 11103
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.520187805738373e-05,
+      "loss": 0.0164,
+      "step": 11104
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.5201063137226792e-05,
+      "loss": 0.0045,
+      "step": 11105
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.5200248161050065e-05,
+      "loss": 0.0,
+      "step": 11106
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.5199433128858016e-05,
+      "loss": 1.4219,
+      "step": 11107
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.5198618040655124e-05,
+      "loss": 0.0001,
+      "step": 11108
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.519780289644586e-05,
+      "loss": 0.0013,
+      "step": 11109
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.519698769623471e-05,
+      "loss": 0.4277,
+      "step": 11110
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.519617244002615e-05,
+      "loss": 0.0,
+      "step": 11111
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.5195357127824643e-05,
+      "loss": 0.0061,
+      "step": 11112
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.5194541759634685e-05,
+      "loss": 0.0,
+      "step": 11113
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.519372633546074e-05,
+      "loss": 0.0,
+      "step": 11114
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.5192910855307295e-05,
+      "loss": 0.0012,
+      "step": 11115
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.519209531917882e-05,
+      "loss": 0.0001,
+      "step": 11116
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.5191279727079804e-05,
+      "loss": 0.001,
+      "step": 11117
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.5190464079014713e-05,
+      "loss": 0.0007,
+      "step": 11118
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.5189648374988037e-05,
+      "loss": 0.0001,
+      "step": 11119
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.5188832615004246e-05,
+      "loss": 0.0002,
+      "step": 11120
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.518801679906783e-05,
+      "loss": 0.0001,
+      "step": 11121
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.5187200927183262e-05,
+      "loss": 0.0016,
+      "step": 11122
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.5186384999355027e-05,
+      "loss": 0.0,
+      "step": 11123
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.5185569015587602e-05,
+      "loss": 0.0,
+      "step": 11124
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.5184752975885468e-05,
+      "loss": 0.0,
+      "step": 11125
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.518393688025311e-05,
+      "loss": 0.0,
+      "step": 11126
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.5183120728695005e-05,
+      "loss": 0.0,
+      "step": 11127
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.518230452121564e-05,
+      "loss": 0.0,
+      "step": 11128
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.5181488257819495e-05,
+      "loss": 0.0,
+      "step": 11129
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.518067193851105e-05,
+      "loss": 0.1216,
+      "step": 11130
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.517985556329479e-05,
+      "loss": 0.0869,
+      "step": 11131
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.51790391321752e-05,
+      "loss": 0.0825,
+      "step": 11132
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.517822264515676e-05,
+      "loss": 0.1855,
+      "step": 11133
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.5177406102243958e-05,
+      "loss": 1.9688,
+      "step": 11134
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.5176589503441275e-05,
+      "loss": 0.5977,
+      "step": 11135
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.517577284875319e-05,
+      "loss": 1.0625,
+      "step": 11136
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.51749561381842e-05,
+      "loss": 0.5117,
+      "step": 11137
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.5174139371738783e-05,
+      "loss": 0.126,
+      "step": 11138
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.517332254942142e-05,
+      "loss": 0.9258,
+      "step": 11139
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.5172505671236606e-05,
+      "loss": 0.1143,
+      "step": 11140
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.5171688737188822e-05,
+      "loss": 0.3125,
+      "step": 11141
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.5170871747282553e-05,
+      "loss": 0.0129,
+      "step": 11142
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.517005470152229e-05,
+      "loss": 0.002,
+      "step": 11143
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.516923759991252e-05,
+      "loss": 0.0383,
+      "step": 11144
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.5168420442457722e-05,
+      "loss": 0.0123,
+      "step": 11145
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.516760322916239e-05,
+      "loss": 0.0004,
+      "step": 11146
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.516678596003101e-05,
+      "loss": 0.0311,
+      "step": 11147
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.5165968635068073e-05,
+      "loss": 0.4453,
+      "step": 11148
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.5165151254278064e-05,
+      "loss": 0.0003,
+      "step": 11149
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.5164333817665477e-05,
+      "loss": 0.0132,
+      "step": 11150
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.516351632523479e-05,
+      "loss": 0.0001,
+      "step": 11151
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.5162698776990505e-05,
+      "loss": 0.293,
+      "step": 11152
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.5161881172937103e-05,
+      "loss": 0.0261,
+      "step": 11153
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.5161063513079075e-05,
+      "loss": 0.3555,
+      "step": 11154
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.5160245797420913e-05,
+      "loss": 0.0,
+      "step": 11155
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.515942802596711e-05,
+      "loss": 0.0001,
+      "step": 11156
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.5158610198722156e-05,
+      "loss": 1.0547,
+      "step": 11157
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.5157792315690538e-05,
+      "loss": 0.0001,
+      "step": 11158
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.5156974376876748e-05,
+      "loss": 0.0,
+      "step": 11159
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.5156156382285283e-05,
+      "loss": 0.0004,
+      "step": 11160
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.5155338331920628e-05,
+      "loss": 0.0001,
+      "step": 11161
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.5154520225787284e-05,
+      "loss": 0.0001,
+      "step": 11162
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.5153702063889735e-05,
+      "loss": 0.0046,
+      "step": 11163
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.515288384623248e-05,
+      "loss": 0.0009,
+      "step": 11164
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.515206557282001e-05,
+      "loss": 0.0,
+      "step": 11165
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.5151247243656816e-05,
+      "loss": 0.1001,
+      "step": 11166
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.5150428858747395e-05,
+      "loss": 0.0,
+      "step": 11167
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.5149610418096243e-05,
+      "loss": 0.0001,
+      "step": 11168
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.514879192170785e-05,
+      "loss": 0.0,
+      "step": 11169
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.5147973369586716e-05,
+      "loss": 0.0,
+      "step": 11170
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.5147154761737335e-05,
+      "loss": 0.0,
+      "step": 11171
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.5146336098164195e-05,
+      "loss": 0.0,
+      "step": 11172
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.51455173788718e-05,
+      "loss": 0.0023,
+      "step": 11173
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.5144698603864646e-05,
+      "loss": 0.0,
+      "step": 11174
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.5143879773147223e-05,
+      "loss": 0.0,
+      "step": 11175
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.514306088672403e-05,
+      "loss": 0.0001,
+      "step": 11176
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.5142241944599572e-05,
+      "loss": 0.0,
+      "step": 11177
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.5141422946778337e-05,
+      "loss": 0.084,
+      "step": 11178
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.5140603893264825e-05,
+      "loss": 0.1064,
+      "step": 11179
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.5139784784063534e-05,
+      "loss": 0.1196,
+      "step": 11180
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.5138965619178962e-05,
+      "loss": 0.1455,
+      "step": 11181
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.513814639861561e-05,
+      "loss": 0.0513,
+      "step": 11182
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.5137327122377973e-05,
+      "loss": 0.0996,
+      "step": 11183
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.513650779047055e-05,
+      "loss": 1.6875,
+      "step": 11184
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.5135688402897846e-05,
+      "loss": 0.0903,
+      "step": 11185
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.5134868959664353e-05,
+      "loss": 0.1328,
+      "step": 11186
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.513404946077458e-05,
+      "loss": 0.4902,
+      "step": 11187
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.5133229906233023e-05,
+      "loss": 0.4238,
+      "step": 11188
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.5132410296044176e-05,
+      "loss": 1.3672,
+      "step": 11189
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.5131590630212548e-05,
+      "loss": 0.0112,
+      "step": 11190
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.513077090874264e-05,
+      "loss": 0.3867,
+      "step": 11191
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.5129951131638953e-05,
+      "loss": 0.062,
+      "step": 11192
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.5129131298905982e-05,
+      "loss": 0.5391,
+      "step": 11193
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.5128311410548238e-05,
+      "loss": 1.0703,
+      "step": 11194
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.512749146657022e-05,
+      "loss": 0.625,
+      "step": 11195
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.512667146697643e-05,
+      "loss": 0.4219,
+      "step": 11196
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.5125851411771378e-05,
+      "loss": 0.3105,
+      "step": 11197
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.5125031300959557e-05,
+      "loss": 0.0306,
+      "step": 11198
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.512421113454547e-05,
+      "loss": 0.3105,
+      "step": 11199
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.5123390912533632e-05,
+      "loss": 0.0991,
+      "step": 11200
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.5122570634928542e-05,
+      "loss": 0.127,
+      "step": 11201
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.5121750301734702e-05,
+      "loss": 0.5898,
+      "step": 11202
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.5120929912956618e-05,
+      "loss": 0.0281,
+      "step": 11203
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.5120109468598798e-05,
+      "loss": 1.4141,
+      "step": 11204
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.5119288968665746e-05,
+      "loss": 0.0022,
+      "step": 11205
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.5118468413161966e-05,
+      "loss": 0.1328,
+      "step": 11206
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.5117647802091968e-05,
+      "loss": 0.0006,
+      "step": 11207
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.5116827135460254e-05,
+      "loss": 0.1289,
+      "step": 11208
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.5116006413271333e-05,
+      "loss": 0.0,
+      "step": 11209
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.5115185635529713e-05,
+      "loss": 0.0002,
+      "step": 11210
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.5114364802239903e-05,
+      "loss": 0.0001,
+      "step": 11211
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.5113543913406402e-05,
+      "loss": 0.0012,
+      "step": 11212
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.5112722969033726e-05,
+      "loss": 0.0,
+      "step": 11213
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.5111901969126382e-05,
+      "loss": 0.0,
+      "step": 11214
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.5111080913688876e-05,
+      "loss": 0.0001,
+      "step": 11215
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.5110259802725718e-05,
+      "loss": 0.3594,
+      "step": 11216
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.5109438636241424e-05,
+      "loss": 0.0,
+      "step": 11217
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.510861741424049e-05,
+      "loss": 0.0001,
+      "step": 11218
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.5107796136727438e-05,
+      "loss": 0.0062,
+      "step": 11219
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.510697480370677e-05,
+      "loss": 0.0003,
+      "step": 11220
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.5106153415182996e-05,
+      "loss": 0.0001,
+      "step": 11221
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.510533197116064e-05,
+      "loss": 0.0014,
+      "step": 11222
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.5104510471644195e-05,
+      "loss": 0.8516,
+      "step": 11223
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.5103688916638183e-05,
+      "loss": 0.0,
+      "step": 11224
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.5102867306147116e-05,
+      "loss": 0.0249,
+      "step": 11225
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.5102045640175502e-05,
+      "loss": 0.1709,
+      "step": 11226
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.5101223918727855e-05,
+      "loss": 0.082,
+      "step": 11227
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.5100402141808687e-05,
+      "loss": 0.0,
+      "step": 11228
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.509958030942251e-05,
+      "loss": 0.1309,
+      "step": 11229
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.5098758421573833e-05,
+      "loss": 0.1357,
+      "step": 11230
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.509793647826718e-05,
+      "loss": 0.0,
+      "step": 11231
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.5097114479507055e-05,
+      "loss": 0.127,
+      "step": 11232
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.509629242529798e-05,
+      "loss": 0.1455,
+      "step": 11233
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.5095470315644465e-05,
+      "loss": 0.1748,
+      "step": 11234
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.5094648150551026e-05,
+      "loss": 0.8242,
+      "step": 11235
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.5093825930022175e-05,
+      "loss": 1.3984,
+      "step": 11236
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.509300365406243e-05,
+      "loss": 0.5898,
+      "step": 11237
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.5092181322676308e-05,
+      "loss": 0.2246,
+      "step": 11238
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.509135893586832e-05,
+      "loss": 0.5703,
+      "step": 11239
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.5090536493642987e-05,
+      "loss": 2.2969,
+      "step": 11240
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.5089713996004823e-05,
+      "loss": 0.0527,
+      "step": 11241
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.5088891442958345e-05,
+      "loss": 0.064,
+      "step": 11242
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.5088068834508068e-05,
+      "loss": 0.0011,
+      "step": 11243
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.5087246170658516e-05,
+      "loss": 0.1826,
+      "step": 11244
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.5086423451414202e-05,
+      "loss": 0.0135,
+      "step": 11245
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.5085600676779646e-05,
+      "loss": 0.375,
+      "step": 11246
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.508477784675936e-05,
+      "loss": 0.0004,
+      "step": 11247
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.5083954961357872e-05,
+      "loss": 0.0001,
+      "step": 11248
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.508313202057969e-05,
+      "loss": 0.0532,
+      "step": 11249
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.508230902442935e-05,
+      "loss": 0.1006,
+      "step": 11250
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.5081485972911356e-05,
+      "loss": 0.0041,
+      "step": 11251
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.5080662866030234e-05,
+      "loss": 0.0532,
+      "step": 11252
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.50798397037905e-05,
+      "loss": 0.0002,
+      "step": 11253
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.507901648619668e-05,
+      "loss": 0.3242,
+      "step": 11254
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.5078193213253295e-05,
+      "loss": 0.1138,
+      "step": 11255
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.507736988496486e-05,
+      "loss": 0.0,
+      "step": 11256
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.5076546501335903e-05,
+      "loss": 0.0,
+      "step": 11257
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.5075723062370936e-05,
+      "loss": 0.0002,
+      "step": 11258
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.5074899568074494e-05,
+      "loss": 0.1338,
+      "step": 11259
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.507407601845109e-05,
+      "loss": 0.373,
+      "step": 11260
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.507325241350525e-05,
+      "loss": 0.0,
+      "step": 11261
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.5072428753241494e-05,
+      "loss": 0.0001,
+      "step": 11262
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.5071605037664353e-05,
+      "loss": 0.0,
+      "step": 11263
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.507078126677834e-05,
+      "loss": 0.0,
+      "step": 11264
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.5069957440587987e-05,
+      "loss": 0.0,
+      "step": 11265
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.5069133559097813e-05,
+      "loss": 0.0091,
+      "step": 11266
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.5068309622312345e-05,
+      "loss": 0.0,
+      "step": 11267
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.5067485630236104e-05,
+      "loss": 0.4473,
+      "step": 11268
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.5066661582873622e-05,
+      "loss": 0.0001,
+      "step": 11269
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.5065837480229417e-05,
+      "loss": 0.0,
+      "step": 11270
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.506501332230802e-05,
+      "loss": 0.0,
+      "step": 11271
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.5064189109113954e-05,
+      "loss": 0.0,
+      "step": 11272
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.506336484065175e-05,
+      "loss": 0.0,
+      "step": 11273
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.5062540516925924e-05,
+      "loss": 0.0232,
+      "step": 11274
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.506171613794101e-05,
+      "loss": 0.0001,
+      "step": 11275
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.5060891703701537e-05,
+      "loss": 0.0,
+      "step": 11276
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.5060067214212027e-05,
+      "loss": 0.0,
+      "step": 11277
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.5059242669477014e-05,
+      "loss": 0.0001,
+      "step": 11278
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.5058418069501025e-05,
+      "loss": 0.0001,
+      "step": 11279
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.505759341428858e-05,
+      "loss": 0.0,
+      "step": 11280
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.5056768703844212e-05,
+      "loss": 0.125,
+      "step": 11281
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.505594393817246e-05,
+      "loss": 0.167,
+      "step": 11282
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.505511911727784e-05,
+      "loss": 0.248,
+      "step": 11283
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.5054294241164885e-05,
+      "loss": 1.4531,
+      "step": 11284
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.5053469309838132e-05,
+      "loss": 0.7891,
+      "step": 11285
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.50526443233021e-05,
+      "loss": 0.1855,
+      "step": 11286
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.5051819281561326e-05,
+      "loss": 0.5742,
+      "step": 11287
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.505099418462034e-05,
+      "loss": 0.4844,
+      "step": 11288
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.505016903248367e-05,
+      "loss": 0.2432,
+      "step": 11289
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.5049343825155853e-05,
+      "loss": 0.3164,
+      "step": 11290
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.504851856264142e-05,
+      "loss": 0.3086,
+      "step": 11291
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.5047693244944894e-05,
+      "loss": 0.0004,
+      "step": 11292
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.5046867872070823e-05,
+      "loss": 0.3242,
+      "step": 11293
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.5046042444023726e-05,
+      "loss": 0.3828,
+      "step": 11294
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.504521696080814e-05,
+      "loss": 0.6836,
+      "step": 11295
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.50443914224286e-05,
+      "loss": 1.2031,
+      "step": 11296
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.5043565828889637e-05,
+      "loss": 0.0001,
+      "step": 11297
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.504274018019579e-05,
+      "loss": 0.5742,
+      "step": 11298
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.5041914476351583e-05,
+      "loss": 0.4766,
+      "step": 11299
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.5041088717361558e-05,
+      "loss": 0.2246,
+      "step": 11300
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.5040262903230254e-05,
+      "loss": 1.2031,
+      "step": 11301
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.5039437033962193e-05,
+      "loss": 0.0002,
+      "step": 11302
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.503861110956192e-05,
+      "loss": 0.0125,
+      "step": 11303
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.503778513003397e-05,
+      "loss": 0.0001,
+      "step": 11304
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.5036959095382875e-05,
+      "loss": 0.0001,
+      "step": 11305
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.503613300561317e-05,
+      "loss": 0.1855,
+      "step": 11306
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.5035306860729403e-05,
+      "loss": 0.0,
+      "step": 11307
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.5034480660736098e-05,
+      "loss": 0.0001,
+      "step": 11308
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.5033654405637795e-05,
+      "loss": 0.5117,
+      "step": 11309
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.503282809543904e-05,
+      "loss": 0.7109,
+      "step": 11310
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.5032001730144355e-05,
+      "loss": 0.0187,
+      "step": 11311
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.5031175309758292e-05,
+      "loss": 0.7383,
+      "step": 11312
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.5030348834285387e-05,
+      "loss": 0.5391,
+      "step": 11313
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.5029522303730174e-05,
+      "loss": 0.0,
+      "step": 11314
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.502869571809719e-05,
+      "loss": 0.123,
+      "step": 11315
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.502786907739098e-05,
+      "loss": 0.2812,
+      "step": 11316
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.5027042381616086e-05,
+      "loss": 0.3926,
+      "step": 11317
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.5026215630777042e-05,
+      "loss": 0.5,
+      "step": 11318
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.5025388824878387e-05,
+      "loss": 0.0579,
+      "step": 11319
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.5024561963924664e-05,
+      "loss": 0.0,
+      "step": 11320
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.5023735047920416e-05,
+      "loss": 0.0,
+      "step": 11321
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.502290807687018e-05,
+      "loss": 0.0002,
+      "step": 11322
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.5022081050778504e-05,
+      "loss": 0.0,
+      "step": 11323
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.502125396964992e-05,
+      "loss": 0.8398,
+      "step": 11324
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.5020426833488975e-05,
+      "loss": 0.0,
+      "step": 11325
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.5019599642300216e-05,
+      "loss": 0.0001,
+      "step": 11326
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.501877239608818e-05,
+      "loss": 0.0001,
+      "step": 11327
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.5017945094857407e-05,
+      "loss": 0.0007,
+      "step": 11328
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.5017117738612447e-05,
+      "loss": 0.1147,
+      "step": 11329
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.5016290327357837e-05,
+      "loss": 0.104,
+      "step": 11330
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.5015462861098127e-05,
+      "loss": 0.1963,
+      "step": 11331
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.5014635339837857e-05,
+      "loss": 0.168,
+      "step": 11332
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.5013807763581574e-05,
+      "loss": 0.1621,
+      "step": 11333
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.5012980132333822e-05,
+      "loss": 0.4141,
+      "step": 11334
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.5012152446099144e-05,
+      "loss": 0.2656,
+      "step": 11335
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.5011324704882087e-05,
+      "loss": 0.4922,
+      "step": 11336
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.5010496908687198e-05,
+      "loss": 0.9961,
+      "step": 11337
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.5009669057519016e-05,
+      "loss": 0.1504,
+      "step": 11338
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.5008841151382097e-05,
+      "loss": 0.0491,
+      "step": 11339
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.5008013190280982e-05,
+      "loss": 0.4102,
+      "step": 11340
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.5007185174220216e-05,
+      "loss": 0.2715,
+      "step": 11341
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.500635710320435e-05,
+      "loss": 0.2715,
+      "step": 11342
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.5005528977237937e-05,
+      "loss": 0.0006,
+      "step": 11343
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.5004700796325508e-05,
+      "loss": 1.5312,
+      "step": 11344
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.5003872560471625e-05,
+      "loss": 0.1885,
+      "step": 11345
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.500304426968083e-05,
+      "loss": 0.2578,
+      "step": 11346
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.5002215923957678e-05,
+      "loss": 0.4746,
+      "step": 11347
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.5001387523306708e-05,
+      "loss": 0.9141,
+      "step": 11348
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.5000559067732482e-05,
+      "loss": 0.0233,
+      "step": 11349
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.4999730557239536e-05,
+      "loss": 0.0016,
+      "step": 11350
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.499890199183243e-05,
+      "loss": 0.4531,
+      "step": 11351
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.4998073371515708e-05,
+      "loss": 0.0006,
+      "step": 11352
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.4997244696293926e-05,
+      "loss": 0.0,
+      "step": 11353
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.4996415966171626e-05,
+      "loss": 0.0004,
+      "step": 11354
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.499558718115337e-05,
+      "loss": 0.0012,
+      "step": 11355
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.49947583412437e-05,
+      "loss": 0.0161,
+      "step": 11356
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.4993929446447174e-05,
+      "loss": 0.0,
+      "step": 11357
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.4993100496768343e-05,
+      "loss": 0.0,
+      "step": 11358
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.499227149221176e-05,
+      "loss": 0.0083,
+      "step": 11359
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.4991442432781965e-05,
+      "loss": 0.0005,
+      "step": 11360
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.4990613318483527e-05,
+      "loss": 0.0,
+      "step": 11361
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.4989784149320994e-05,
+      "loss": 0.0,
+      "step": 11362
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.4988954925298915e-05,
+      "loss": 0.0001,
+      "step": 11363
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.4988125646421854e-05,
+      "loss": 0.7109,
+      "step": 11364
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.498729631269435e-05,
+      "loss": 0.2285,
+      "step": 11365
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.498646692412097e-05,
+      "loss": 0.1001,
+      "step": 11366
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.4985637480706266e-05,
+      "loss": 0.0525,
+      "step": 11367
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.498480798245479e-05,
+      "loss": 0.0771,
+      "step": 11368
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.4983978429371097e-05,
+      "loss": 0.5898,
+      "step": 11369
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.498314882145974e-05,
+      "loss": 0.019,
+      "step": 11370
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.498231915872529e-05,
+      "loss": 0.0013,
+      "step": 11371
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.4981489441172284e-05,
+      "loss": 0.0001,
+      "step": 11372
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.4980659668805286e-05,
+      "loss": 0.0,
+      "step": 11373
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.4979829841628857e-05,
+      "loss": 0.0001,
+      "step": 11374
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.497899995964755e-05,
+      "loss": 0.0077,
+      "step": 11375
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.497817002286592e-05,
+      "loss": 0.0654,
+      "step": 11376
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.4977340031288535e-05,
+      "loss": 0.0001,
+      "step": 11377
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.4976509984919935e-05,
+      "loss": 0.0,
+      "step": 11378
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.497567988376469e-05,
+      "loss": 0.082,
+      "step": 11379
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.4974849727827362e-05,
+      "loss": 0.1494,
+      "step": 11380
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.4974019517112503e-05,
+      "loss": 0.1641,
+      "step": 11381
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.4973189251624673e-05,
+      "loss": 0.1855,
+      "step": 11382
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.4972358931368437e-05,
+      "loss": 0.1572,
+      "step": 11383
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.4971528556348343e-05,
+      "loss": 0.9219,
+      "step": 11384
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.497069812656896e-05,
+      "loss": 0.1465,
+      "step": 11385
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.4969867642034853e-05,
+      "loss": 0.75,
+      "step": 11386
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.496903710275057e-05,
+      "loss": 0.0786,
+      "step": 11387
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.496820650872068e-05,
+      "loss": 0.0015,
+      "step": 11388
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.496737585994974e-05,
+      "loss": 0.3203,
+      "step": 11389
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.4966545156442322e-05,
+      "loss": 1.0234,
+      "step": 11390
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.4965714398202974e-05,
+      "loss": 0.2217,
+      "step": 11391
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.4964883585236265e-05,
+      "loss": 0.062,
+      "step": 11392
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.4964052717546755e-05,
+      "loss": 0.0165,
+      "step": 11393
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.4963221795139012e-05,
+      "loss": 0.4297,
+      "step": 11394
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.4962390818017596e-05,
+      "loss": 0.0065,
+      "step": 11395
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.4961559786187067e-05,
+      "loss": 0.0084,
+      "step": 11396
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.496072869965199e-05,
+      "loss": 1.2969,
+      "step": 11397
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.4959897558416933e-05,
+      "loss": 1.8125,
+      "step": 11398
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.4959066362486455e-05,
+      "loss": 0.0048,
+      "step": 11399
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.4958235111865127e-05,
+      "loss": 0.0004,
+      "step": 11400
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.495740380655751e-05,
+      "loss": 0.2236,
+      "step": 11401
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.4956572446568168e-05,
+      "loss": 0.25,
+      "step": 11402
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.4955741031901667e-05,
+      "loss": 0.0003,
+      "step": 11403
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.4954909562562575e-05,
+      "loss": 0.0001,
+      "step": 11404
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.4954078038555454e-05,
+      "loss": 0.0005,
+      "step": 11405
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.4953246459884876e-05,
+      "loss": 0.0177,
+      "step": 11406
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.4952414826555404e-05,
+      "loss": 0.0089,
+      "step": 11407
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.495158313857161e-05,
+      "loss": 0.0002,
+      "step": 11408
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.4950751395938048e-05,
+      "loss": 1.0312,
+      "step": 11409
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.4949919598659297e-05,
+      "loss": 0.4941,
+      "step": 11410
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.4949087746739926e-05,
+      "loss": 0.1162,
+      "step": 11411
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.4948255840184496e-05,
+      "loss": 0.0042,
+      "step": 11412
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.494742387899758e-05,
+      "loss": 0.002,
+      "step": 11413
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.4946591863183746e-05,
+      "loss": 0.0002,
+      "step": 11414
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.4945759792747563e-05,
+      "loss": 0.0,
+      "step": 11415
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.49449276676936e-05,
+      "loss": 0.2891,
+      "step": 11416
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.494409548802643e-05,
+      "loss": 0.0003,
+      "step": 11417
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.4943263253750618e-05,
+      "loss": 0.0,
+      "step": 11418
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.494243096487073e-05,
+      "loss": 0.0001,
+      "step": 11419
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.494159862139135e-05,
+      "loss": 0.0,
+      "step": 11420
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.4940766223317035e-05,
+      "loss": 0.0021,
+      "step": 11421
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.4939933770652366e-05,
+      "loss": 0.1602,
+      "step": 11422
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.4939101263401916e-05,
+      "loss": 0.7109,
+      "step": 11423
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.4938268701570245e-05,
+      "loss": 0.0001,
+      "step": 11424
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.4937436085161937e-05,
+      "loss": 0.0654,
+      "step": 11425
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.4936603414181553e-05,
+      "loss": 0.0003,
+      "step": 11426
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.4935770688633677e-05,
+      "loss": 0.0,
+      "step": 11427
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.493493790852288e-05,
+      "loss": 0.0027,
+      "step": 11428
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.4934105073853724e-05,
+      "loss": 0.0,
+      "step": 11429
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.4933272184630798e-05,
+      "loss": 0.1885,
+      "step": 11430
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.4932439240858666e-05,
+      "loss": 0.0017,
+      "step": 11431
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.49316062425419e-05,
+      "loss": 0.124,
+      "step": 11432
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.4930773189685088e-05,
+      "loss": 0.1445,
+      "step": 11433
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.492994008229279e-05,
+      "loss": 0.75,
+      "step": 11434
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.492910692036959e-05,
+      "loss": 0.1855,
+      "step": 11435
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.492827370392006e-05,
+      "loss": 0.8203,
+      "step": 11436
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.4927440432948776e-05,
+      "loss": 1.3125,
+      "step": 11437
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.492660710746031e-05,
+      "loss": 0.2949,
+      "step": 11438
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.4925773727459248e-05,
+      "loss": 0.4531,
+      "step": 11439
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.492494029295016e-05,
+      "loss": 0.7734,
+      "step": 11440
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.4924106803937622e-05,
+      "loss": 0.7617,
+      "step": 11441
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.4923273260426214e-05,
+      "loss": 0.0155,
+      "step": 11442
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.492243966242051e-05,
+      "loss": 0.543,
+      "step": 11443
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.4921606009925097e-05,
+      "loss": 0.4863,
+      "step": 11444
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.492077230294454e-05,
+      "loss": 0.0001,
+      "step": 11445
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.4919938541483425e-05,
+      "loss": 0.9062,
+      "step": 11446
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.491910472554633e-05,
+      "loss": 0.0364,
+      "step": 11447
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.4918270855137832e-05,
+      "loss": 0.0002,
+      "step": 11448
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.491743693026251e-05,
+      "loss": 0.2373,
+      "step": 11449
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.491660295092495e-05,
+      "loss": 0.1118,
+      "step": 11450
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.4915768917129723e-05,
+      "loss": 0.002,
+      "step": 11451
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.491493482888141e-05,
+      "loss": 0.0039,
+      "step": 11452
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.4914100686184605e-05,
+      "loss": 0.0208,
+      "step": 11453
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.4913266489043865e-05,
+      "loss": 0.0001,
+      "step": 11454
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.4912432237463796e-05,
+      "loss": 0.0001,
+      "step": 11455
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.491159793144896e-05,
+      "loss": 0.1187,
+      "step": 11456
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.491076357100395e-05,
+      "loss": 0.457,
+      "step": 11457
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.4909929156133342e-05,
+      "loss": 0.8242,
+      "step": 11458
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.4909094686841723e-05,
+      "loss": 0.0265,
+      "step": 11459
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.4908260163133667e-05,
+      "loss": 0.0042,
+      "step": 11460
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.4907425585013767e-05,
+      "loss": 0.05,
+      "step": 11461
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.4906590952486603e-05,
+      "loss": 0.3672,
+      "step": 11462
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.4905756265556753e-05,
+      "loss": 0.2148,
+      "step": 11463
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.4904921524228813e-05,
+      "loss": 0.0,
+      "step": 11464
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.4904086728507355e-05,
+      "loss": 0.0,
+      "step": 11465
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.490325187839697e-05,
+      "loss": 0.0,
+      "step": 11466
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.490241697390223e-05,
+      "loss": 0.0001,
+      "step": 11467
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.4901582015027743e-05,
+      "loss": 0.0,
+      "step": 11468
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.4900747001778073e-05,
+      "loss": 0.0001,
+      "step": 11469
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.4899911934157816e-05,
+      "loss": 0.1143,
+      "step": 11470
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.4899076812171556e-05,
+      "loss": 0.1348,
+      "step": 11471
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.489824163582388e-05,
+      "loss": 0.0001,
+      "step": 11472
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.4897406405119368e-05,
+      "loss": 0.0001,
+      "step": 11473
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.489657112006262e-05,
+      "loss": 0.0003,
+      "step": 11474
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.489573578065821e-05,
+      "loss": 0.0,
+      "step": 11475
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.489490038691073e-05,
+      "loss": 0.0,
+      "step": 11476
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.4894064938824767e-05,
+      "loss": 0.0,
+      "step": 11477
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.489322943640491e-05,
+      "loss": 0.0002,
+      "step": 11478
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.4892393879655747e-05,
+      "loss": 0.0703,
+      "step": 11479
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.4891558268581875e-05,
+      "loss": 0.123,
+      "step": 11480
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.4890722603187865e-05,
+      "loss": 0.166,
+      "step": 11481
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.4889886883478314e-05,
+      "loss": 0.1147,
+      "step": 11482
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.488905110945782e-05,
+      "loss": 0.1523,
+      "step": 11483
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.488821528113096e-05,
+      "loss": 0.3926,
+      "step": 11484
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.488737939850233e-05,
+      "loss": 0.4297,
+      "step": 11485
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.4886543461576524e-05,
+      "loss": 0.0173,
+      "step": 11486
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.4885707470358128e-05,
+      "loss": 0.0659,
+      "step": 11487
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.488487142485173e-05,
+      "loss": 1.3672,
+      "step": 11488
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.4884035325061928e-05,
+      "loss": 1.1797,
+      "step": 11489
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.488319917099331e-05,
+      "loss": 0.3125,
+      "step": 11490
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.4882362962650468e-05,
+      "loss": 0.5469,
+      "step": 11491
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.488152670003799e-05,
+      "loss": 0.6133,
+      "step": 11492
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.4880690383160473e-05,
+      "loss": 0.0118,
+      "step": 11493
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.487985401202251e-05,
+      "loss": 0.668,
+      "step": 11494
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.487901758662869e-05,
+      "loss": 0.0503,
+      "step": 11495
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.4878181106983615e-05,
+      "loss": 0.7969,
+      "step": 11496
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.487734457309187e-05,
+      "loss": 0.0776,
+      "step": 11497
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.487650798495805e-05,
+      "loss": 0.0002,
+      "step": 11498
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.4875671342586755e-05,
+      "loss": 0.0036,
+      "step": 11499
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.4874834645982573e-05,
+      "loss": 0.0002,
+      "step": 11500
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.4873997895150102e-05,
+      "loss": 0.0002,
+      "step": 11501
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.4873161090093935e-05,
+      "loss": 0.0,
+      "step": 11502
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.487232423081867e-05,
+      "loss": 0.0718,
+      "step": 11503
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.48714873173289e-05,
+      "loss": 0.0001,
+      "step": 11504
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.4870650349629223e-05,
+      "loss": 0.1396,
+      "step": 11505
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.4869813327724236e-05,
+      "loss": 0.1187,
+      "step": 11506
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.4868976251618533e-05,
+      "loss": 0.126,
+      "step": 11507
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.4868139121316714e-05,
+      "loss": 0.0131,
+      "step": 11508
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.486730193682337e-05,
+      "loss": 0.0,
+      "step": 11509
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.4866464698143107e-05,
+      "loss": 0.5234,
+      "step": 11510
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.486562740528051e-05,
+      "loss": 0.0004,
+      "step": 11511
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.486479005824019e-05,
+      "loss": 0.3125,
+      "step": 11512
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.4863952657026743e-05,
+      "loss": 0.0,
+      "step": 11513
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.4863115201644765e-05,
+      "loss": 0.0001,
+      "step": 11514
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.4862277692098855e-05,
+      "loss": 0.0001,
+      "step": 11515
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.486144012839361e-05,
+      "loss": 0.1709,
+      "step": 11516
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.486060251053363e-05,
+      "loss": 0.0,
+      "step": 11517
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.4859764838523522e-05,
+      "loss": 0.0001,
+      "step": 11518
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.4858927112367883e-05,
+      "loss": 0.0001,
+      "step": 11519
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.4858089332071307e-05,
+      "loss": 0.083,
+      "step": 11520
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.4857251497638398e-05,
+      "loss": 0.0001,
+      "step": 11521
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.4856413609073766e-05,
+      "loss": 0.0,
+      "step": 11522
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.485557566638199e-05,
+      "loss": 1.1562,
+      "step": 11523
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.48547376695677e-05,
+      "loss": 0.0001,
+      "step": 11524
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.485389961863548e-05,
+      "loss": 0.1328,
+      "step": 11525
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.4853061513589934e-05,
+      "loss": 0.0,
+      "step": 11526
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.4852223354435667e-05,
+      "loss": 0.0,
+      "step": 11527
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.485138514117728e-05,
+      "loss": 0.8789,
+      "step": 11528
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.4850546873819383e-05,
+      "loss": 0.0786,
+      "step": 11529
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.484970855236657e-05,
+      "loss": 0.0,
+      "step": 11530
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.484887017682345e-05,
+      "loss": 0.064,
+      "step": 11531
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.4848031747194625e-05,
+      "loss": 0.0649,
+      "step": 11532
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.4847193263484698e-05,
+      "loss": 0.1152,
+      "step": 11533
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.484635472569828e-05,
+      "loss": 0.5117,
+      "step": 11534
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.484551613383997e-05,
+      "loss": 0.7734,
+      "step": 11535
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.4844677487914373e-05,
+      "loss": 0.1484,
+      "step": 11536
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.48438387879261e-05,
+      "loss": 0.0226,
+      "step": 11537
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.484300003387975e-05,
+      "loss": 0.3555,
+      "step": 11538
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.484216122577993e-05,
+      "loss": 0.1807,
+      "step": 11539
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.484132236363125e-05,
+      "loss": 0.0231,
+      "step": 11540
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.4840483447438318e-05,
+      "loss": 0.0002,
+      "step": 11541
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.4839644477205734e-05,
+      "loss": 0.0586,
+      "step": 11542
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.483880545293811e-05,
+      "loss": 0.2393,
+      "step": 11543
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.4837966374640056e-05,
+      "loss": 0.6914,
+      "step": 11544
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.4837127242316174e-05,
+      "loss": 0.0508,
+      "step": 11545
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.4836288055971072e-05,
+      "loss": 0.1104,
+      "step": 11546
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.4835448815609367e-05,
+      "loss": 0.0052,
+      "step": 11547
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.4834609521235662e-05,
+      "loss": 1.1016,
+      "step": 11548
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.4833770172854563e-05,
+      "loss": 0.1211,
+      "step": 11549
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.4832930770470682e-05,
+      "loss": 0.0938,
+      "step": 11550
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.4832091314088627e-05,
+      "loss": 0.3418,
+      "step": 11551
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.4831251803713016e-05,
+      "loss": 0.0107,
+      "step": 11552
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.483041223934845e-05,
+      "loss": 0.0026,
+      "step": 11553
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.4829572620999543e-05,
+      "loss": 2.3438,
+      "step": 11554
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.4828732948670903e-05,
+      "loss": 0.0076,
+      "step": 11555
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.482789322236715e-05,
+      "loss": 0.0811,
+      "step": 11556
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.482705344209288e-05,
+      "loss": 0.0383,
+      "step": 11557
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.482621360785272e-05,
+      "loss": 0.0,
+      "step": 11558
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.4825373719651278e-05,
+      "loss": 0.0,
+      "step": 11559
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.4824533777493158e-05,
+      "loss": 0.0104,
+      "step": 11560
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.482369378138298e-05,
+      "loss": 0.0,
+      "step": 11561
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.4822853731325357e-05,
+      "loss": 0.0447,
+      "step": 11562
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.48220136273249e-05,
+      "loss": 0.0,
+      "step": 11563
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.482117346938622e-05,
+      "loss": 0.0552,
+      "step": 11564
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.482033325751394e-05,
+      "loss": 0.0,
+      "step": 11565
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.4819492991712663e-05,
+      "loss": 0.0,
+      "step": 11566
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.4818652671987014e-05,
+      "loss": 0.0,
+      "step": 11567
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.4817812298341597e-05,
+      "loss": 0.0786,
+      "step": 11568
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.4816971870781034e-05,
+      "loss": 0.0001,
+      "step": 11569
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.481613138930994e-05,
+      "loss": 0.0001,
+      "step": 11570
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.4815290853932924e-05,
+      "loss": 0.2129,
+      "step": 11571
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.4814450264654612e-05,
+      "loss": 0.0557,
+      "step": 11572
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.4813609621479617e-05,
+      "loss": 0.0001,
+      "step": 11573
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.4812768924412548e-05,
+      "loss": 0.0,
+      "step": 11574
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.4811928173458025e-05,
+      "loss": 0.2637,
+      "step": 11575
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.481108736862067e-05,
+      "loss": 0.0001,
+      "step": 11576
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.4810246509905097e-05,
+      "loss": 0.0,
+      "step": 11577
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.4809405597315925e-05,
+      "loss": 0.0854,
+      "step": 11578
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.4808564630857772e-05,
+      "loss": 0.0,
+      "step": 11579
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.4807723610535256e-05,
+      "loss": 0.1377,
+      "step": 11580
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.480688253635299e-05,
+      "loss": 0.0,
+      "step": 11581
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.48060414083156e-05,
+      "loss": 0.1611,
+      "step": 11582
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.4805200226427703e-05,
+      "loss": 0.1826,
+      "step": 11583
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.4804358990693917e-05,
+      "loss": 0.4766,
+      "step": 11584
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.4803517701118868e-05,
+      "loss": 1.2344,
+      "step": 11585
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.4802676357707163e-05,
+      "loss": 0.4766,
+      "step": 11586
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.4801834960463435e-05,
+      "loss": 0.0889,
+      "step": 11587
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.4800993509392302e-05,
+      "loss": 0.1416,
+      "step": 11588
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.4800152004498374e-05,
+      "loss": 0.1035,
+      "step": 11589
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.479931044578629e-05,
+      "loss": 0.002,
+      "step": 11590
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.4798468833260658e-05,
+      "loss": 0.4902,
+      "step": 11591
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.47976271669261e-05,
+      "loss": 0.0586,
+      "step": 11592
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.479678544678725e-05,
+      "loss": 0.6758,
+      "step": 11593
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.4795943672848718e-05,
+      "loss": 0.5586,
+      "step": 11594
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.4795101845115128e-05,
+      "loss": 0.375,
+      "step": 11595
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.4794259963591107e-05,
+      "loss": 0.6992,
+      "step": 11596
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.4793418028281286e-05,
+      "loss": 0.6055,
+      "step": 11597
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.479257603919027e-05,
+      "loss": 0.0547,
+      "step": 11598
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.4791733996322696e-05,
+      "loss": 0.0,
+      "step": 11599
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.4790891899683183e-05,
+      "loss": 0.0003,
+      "step": 11600
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.479004974927636e-05,
+      "loss": 0.0001,
+      "step": 11601
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.478920754510685e-05,
+      "loss": 0.0002,
+      "step": 11602
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.4788365287179272e-05,
+      "loss": 0.0562,
+      "step": 11603
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.4787522975498263e-05,
+      "loss": 0.6484,
+      "step": 11604
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.4786680610068433e-05,
+      "loss": 0.042,
+      "step": 11605
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.4785838190894423e-05,
+      "loss": 0.0012,
+      "step": 11606
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.478499571798085e-05,
+      "loss": 0.0013,
+      "step": 11607
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.4784153191332346e-05,
+      "loss": 0.2715,
+      "step": 11608
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.4783310610953533e-05,
+      "loss": 0.0001,
+      "step": 11609
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.4782467976849042e-05,
+      "loss": 0.5898,
+      "step": 11610
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.4781625289023496e-05,
+      "loss": 0.0002,
+      "step": 11611
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.478078254748153e-05,
+      "loss": 0.1416,
+      "step": 11612
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.4779939752227765e-05,
+      "loss": 0.0001,
+      "step": 11613
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.4779096903266835e-05,
+      "loss": 0.0001,
+      "step": 11614
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.4778254000603356e-05,
+      "loss": 0.0,
+      "step": 11615
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.4777411044241972e-05,
+      "loss": 0.0525,
+      "step": 11616
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.4776568034187305e-05,
+      "loss": 0.0001,
+      "step": 11617
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.4775724970443987e-05,
+      "loss": 0.0,
+      "step": 11618
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.4774881853016646e-05,
+      "loss": 0.0001,
+      "step": 11619
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.4774038681909913e-05,
+      "loss": 0.0,
+      "step": 11620
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.4773195457128413e-05,
+      "loss": 0.001,
+      "step": 11621
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.4772352178676785e-05,
+      "loss": 0.0,
+      "step": 11622
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.477150884655966e-05,
+      "loss": 0.0001,
+      "step": 11623
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.4770665460781655e-05,
+      "loss": 0.0001,
+      "step": 11624
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.476982202134742e-05,
+      "loss": 0.0,
+      "step": 11625
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.476897852826158e-05,
+      "loss": 0.0,
+      "step": 11626
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.4768134981528763e-05,
+      "loss": 0.1084,
+      "step": 11627
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.4767291381153603e-05,
+      "loss": 0.0012,
+      "step": 11628
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.4766447727140735e-05,
+      "loss": 0.1338,
+      "step": 11629
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.4765604019494792e-05,
+      "loss": 0.1709,
+      "step": 11630
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.4764760258220407e-05,
+      "loss": 0.1758,
+      "step": 11631
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.4763916443322207e-05,
+      "loss": 0.1455,
+      "step": 11632
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.4763072574804835e-05,
+      "loss": 0.1768,
+      "step": 11633
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.4762228652672928e-05,
+      "loss": 0.3574,
+      "step": 11634
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.4761384676931106e-05,
+      "loss": 0.2695,
+      "step": 11635
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.4760540647584014e-05,
+      "loss": 0.8555,
+      "step": 11636
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.4759696564636284e-05,
+      "loss": 1.0156,
+      "step": 11637
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.475885242809256e-05,
+      "loss": 0.0962,
+      "step": 11638
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.475800823795746e-05,
+      "loss": 0.7422,
+      "step": 11639
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.4757163994235638e-05,
+      "loss": 0.0479,
+      "step": 11640
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.4756319696931717e-05,
+      "loss": 0.4551,
+      "step": 11641
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.475547534605034e-05,
+      "loss": 0.4199,
+      "step": 11642
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.475463094159614e-05,
+      "loss": 0.3301,
+      "step": 11643
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.475378648357376e-05,
+      "loss": 1.5938,
+      "step": 11644
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.4752941971987834e-05,
+      "loss": 1.0234,
+      "step": 11645
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.4752097406842998e-05,
+      "loss": 0.0152,
+      "step": 11646
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.475125278814389e-05,
+      "loss": 0.249,
+      "step": 11647
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.4750408115895152e-05,
+      "loss": 0.4492,
+      "step": 11648
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.474956339010142e-05,
+      "loss": 0.9922,
+      "step": 11649
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.4748718610767335e-05,
+      "loss": 0.5391,
+      "step": 11650
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.4747873777897532e-05,
+      "loss": 0.0,
+      "step": 11651
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.4747028891496656e-05,
+      "loss": 1.7656,
+      "step": 11652
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.474618395156934e-05,
+      "loss": 1.7109,
+      "step": 11653
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.4745338958120233e-05,
+      "loss": 0.0003,
+      "step": 11654
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.4744493911153962e-05,
+      "loss": 0.0002,
+      "step": 11655
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.4743648810675182e-05,
+      "loss": 0.0942,
+      "step": 11656
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.4742803656688528e-05,
+      "loss": 0.208,
+      "step": 11657
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.474195844919864e-05,
+      "loss": 0.0004,
+      "step": 11658
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.4741113188210157e-05,
+      "loss": 0.0106,
+      "step": 11659
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.4740267873727727e-05,
+      "loss": 0.6953,
+      "step": 11660
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.473942250575599e-05,
+      "loss": 0.0002,
+      "step": 11661
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.4738577084299583e-05,
+      "loss": 0.0012,
+      "step": 11662
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.473773160936316e-05,
+      "loss": 0.2305,
+      "step": 11663
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.4736886080951354e-05,
+      "loss": 0.2031,
+      "step": 11664
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.473604049906881e-05,
+      "loss": 0.0,
+      "step": 11665
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.4735194863720174e-05,
+      "loss": 0.0002,
+      "step": 11666
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.4734349174910093e-05,
+      "loss": 0.2217,
+      "step": 11667
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.47335034326432e-05,
+      "loss": 0.0889,
+      "step": 11668
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.473265763692415e-05,
+      "loss": 0.1289,
+      "step": 11669
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.473181178775759e-05,
+      "loss": 0.0088,
+      "step": 11670
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.473096588514815e-05,
+      "loss": 0.0001,
+      "step": 11671
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.4730119929100493e-05,
+      "loss": 0.0,
+      "step": 11672
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.472927391961925e-05,
+      "loss": 0.0,
+      "step": 11673
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.4728427856709078e-05,
+      "loss": 0.0,
+      "step": 11674
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.4727581740374614e-05,
+      "loss": 0.1084,
+      "step": 11675
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.4726735570620513e-05,
+      "loss": 0.0,
+      "step": 11676
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.472588934745141e-05,
+      "loss": 0.1387,
+      "step": 11677
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.472504307087197e-05,
+      "loss": 0.0864,
+      "step": 11678
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.472419674088683e-05,
+      "loss": 0.0,
+      "step": 11679
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.4723350357500634e-05,
+      "loss": 0.0,
+      "step": 11680
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.4722503920718028e-05,
+      "loss": 0.1719,
+      "step": 11681
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.4721657430543672e-05,
+      "loss": 0.1309,
+      "step": 11682
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.472081088698221e-05,
+      "loss": 0.1553,
+      "step": 11683
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.471996429003829e-05,
+      "loss": 0.2871,
+      "step": 11684
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.4719117639716556e-05,
+      "loss": 1.9062,
+      "step": 11685
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.4718270936021668e-05,
+      "loss": 1.0234,
+      "step": 11686
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.471742417895827e-05,
+      "loss": 0.3398,
+      "step": 11687
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.4716577368531008e-05,
+      "loss": 0.0645,
+      "step": 11688
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.4715730504744538e-05,
+      "loss": 0.6406,
+      "step": 11689
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.4714883587603506e-05,
+      "loss": 0.9766,
+      "step": 11690
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.4714036617112565e-05,
+      "loss": 0.0135,
+      "step": 11691
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.4713189593276373e-05,
+      "loss": 0.6602,
+      "step": 11692
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.471234251609957e-05,
+      "loss": 0.0669,
+      "step": 11693
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.4711495385586814e-05,
+      "loss": 0.104,
+      "step": 11694
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.4710648201742758e-05,
+      "loss": 0.1514,
+      "step": 11695
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.4709800964572052e-05,
+      "loss": 1.7031,
+      "step": 11696
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.470895367407935e-05,
+      "loss": 0.625,
+      "step": 11697
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.4708106330269304e-05,
+      "loss": 1.3984,
+      "step": 11698
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.4707258933146567e-05,
+      "loss": 0.4258,
+      "step": 11699
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.470641148271579e-05,
+      "loss": 0.0035,
+      "step": 11700
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.4705563978981634e-05,
+      "loss": 0.6602,
+      "step": 11701
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.470471642194875e-05,
+      "loss": 0.1963,
+      "step": 11702
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.4703868811621788e-05,
+      "loss": 0.0006,
+      "step": 11703
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.4703021148005405e-05,
+      "loss": 0.5547,
+      "step": 11704
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.470217343110426e-05,
+      "loss": 0.0,
+      "step": 11705
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.4701325660923006e-05,
+      "loss": 0.0043,
+      "step": 11706
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.470047783746629e-05,
+      "loss": 0.0,
+      "step": 11707
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.4699629960738785e-05,
+      "loss": 0.0,
+      "step": 11708
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.4698782030745137e-05,
+      "loss": 0.0061,
+      "step": 11709
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.4697934047490002e-05,
+      "loss": 0.0,
+      "step": 11710
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.4697086010978038e-05,
+      "loss": 0.2891,
+      "step": 11711
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.46962379212139e-05,
+      "loss": 0.0002,
+      "step": 11712
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.4695389778202254e-05,
+      "loss": 0.0103,
+      "step": 11713
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.4694541581947743e-05,
+      "loss": 0.0001,
+      "step": 11714
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.4693693332455036e-05,
+      "loss": 0.0002,
+      "step": 11715
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.469284502972879e-05,
+      "loss": 0.4512,
+      "step": 11716
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.4691996673773665e-05,
+      "loss": 0.0,
+      "step": 11717
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.469114826459431e-05,
+      "loss": 0.0028,
+      "step": 11718
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.4690299802195394e-05,
+      "loss": 0.4043,
+      "step": 11719
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.4689451286581574e-05,
+      "loss": 0.123,
+      "step": 11720
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.468860271775751e-05,
+      "loss": 0.0,
+      "step": 11721
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.4687754095727854e-05,
+      "loss": 0.0996,
+      "step": 11722
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.4686905420497275e-05,
+      "loss": 0.0,
+      "step": 11723
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.4686056692070438e-05,
+      "loss": 0.0,
+      "step": 11724
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.4685207910451993e-05,
+      "loss": 0.0,
+      "step": 11725
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.4684359075646603e-05,
+      "loss": 0.2598,
+      "step": 11726
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.468351018765894e-05,
+      "loss": 0.0,
+      "step": 11727
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.468266124649365e-05,
+      "loss": 0.9883,
+      "step": 11728
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.4681812252155407e-05,
+      "loss": 0.0005,
+      "step": 11729
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.4680963204648868e-05,
+      "loss": 0.1484,
+      "step": 11730
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.468011410397869e-05,
+      "loss": 0.4395,
+      "step": 11731
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.4679264950149553e-05,
+      "loss": 0.0003,
+      "step": 11732
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.4678415743166106e-05,
+      "loss": 0.1943,
+      "step": 11733
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.4677566483033016e-05,
+      "loss": 1.2344,
+      "step": 11734
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.4676717169754947e-05,
+      "loss": 0.4414,
+      "step": 11735
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.4675867803336566e-05,
+      "loss": 0.0874,
+      "step": 11736
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.467501838378253e-05,
+      "loss": 0.3535,
+      "step": 11737
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.467416891109751e-05,
+      "loss": 0.6367,
+      "step": 11738
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.4673319385286165e-05,
+      "loss": 0.4648,
+      "step": 11739
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.467246980635317e-05,
+      "loss": 0.248,
+      "step": 11740
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.467162017430318e-05,
+      "loss": 1.3516,
+      "step": 11741
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.4670770489140872e-05,
+      "loss": 0.3926,
+      "step": 11742
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.4669920750870896e-05,
+      "loss": 0.3691,
+      "step": 11743
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.4669070959497932e-05,
+      "loss": 0.1865,
+      "step": 11744
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.466822111502665e-05,
+      "loss": 0.3789,
+      "step": 11745
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.46673712174617e-05,
+      "loss": 0.2676,
+      "step": 11746
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.4666521266807764e-05,
+      "loss": 0.8242,
+      "step": 11747
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.4665671263069505e-05,
+      "loss": 0.0664,
+      "step": 11748
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.466482120625159e-05,
+      "loss": 0.1226,
+      "step": 11749
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.4663971096358682e-05,
+      "loss": 0.2754,
+      "step": 11750
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.466312093339546e-05,
+      "loss": 0.1895,
+      "step": 11751
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.4662270717366586e-05,
+      "loss": 0.625,
+      "step": 11752
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.466142044827673e-05,
+      "loss": 0.0593,
+      "step": 11753
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.4660570126130565e-05,
+      "loss": 0.0001,
+      "step": 11754
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.4659719750932755e-05,
+      "loss": 2.5156,
+      "step": 11755
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.4658869322687973e-05,
+      "loss": 0.7461,
+      "step": 11756
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.465801884140089e-05,
+      "loss": 0.0012,
+      "step": 11757
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.4657168307076173e-05,
+      "loss": 0.0001,
+      "step": 11758
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.4656317719718498e-05,
+      "loss": 0.1445,
+      "step": 11759
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.4655467079332528e-05,
+      "loss": 0.0022,
+      "step": 11760
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.4654616385922946e-05,
+      "loss": 0.0,
+      "step": 11761
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.4653765639494407e-05,
+      "loss": 0.1357,
+      "step": 11762
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.4652914840051604e-05,
+      "loss": 0.0021,
+      "step": 11763
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.465206398759919e-05,
+      "loss": 0.0015,
+      "step": 11764
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.465121308214185e-05,
+      "loss": 0.0013,
+      "step": 11765
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.465036212368425e-05,
+      "loss": 0.0457,
+      "step": 11766
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.4649511112231068e-05,
+      "loss": 0.0002,
+      "step": 11767
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.4648660047786974e-05,
+      "loss": 0.1016,
+      "step": 11768
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.464780893035664e-05,
+      "loss": 0.5977,
+      "step": 11769
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.4646957759944745e-05,
+      "loss": 0.0003,
+      "step": 11770
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.4646106536555962e-05,
+      "loss": 0.0002,
+      "step": 11771
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.4645255260194958e-05,
+      "loss": 0.0049,
+      "step": 11772
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.464440393086642e-05,
+      "loss": 0.0001,
+      "step": 11773
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.4643552548575018e-05,
+      "loss": 0.0479,
+      "step": 11774
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.4642701113325422e-05,
+      "loss": 0.0192,
+      "step": 11775
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.4641849625122315e-05,
+      "loss": 0.0002,
+      "step": 11776
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.4640998083970372e-05,
+      "loss": 0.0001,
+      "step": 11777
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.4640146489874265e-05,
+      "loss": 0.0001,
+      "step": 11778
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.4639294842838674e-05,
+      "loss": 0.0001,
+      "step": 11779
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.463844314286828e-05,
+      "loss": 0.0996,
+      "step": 11780
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.463759138996775e-05,
+      "loss": 0.1436,
+      "step": 11781
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.4636739584141765e-05,
+      "loss": 0.0002,
+      "step": 11782
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.4635887725395005e-05,
+      "loss": 0.1245,
+      "step": 11783
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.463503581373215e-05,
+      "loss": 0.8555,
+      "step": 11784
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.4634183849157877e-05,
+      "loss": 0.4902,
+      "step": 11785
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.463333183167686e-05,
+      "loss": 0.3652,
+      "step": 11786
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.4632479761293783e-05,
+      "loss": 0.6719,
+      "step": 11787
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.4631627638013328e-05,
+      "loss": 0.5117,
+      "step": 11788
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.4630775461840167e-05,
+      "loss": 0.0216,
+      "step": 11789
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.4629923232778978e-05,
+      "loss": 0.2148,
+      "step": 11790
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.4629070950834452e-05,
+      "loss": 0.75,
+      "step": 11791
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.462821861601126e-05,
+      "loss": 0.8945,
+      "step": 11792
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.4627366228314086e-05,
+      "loss": 0.5117,
+      "step": 11793
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.4626513787747615e-05,
+      "loss": 0.9492,
+      "step": 11794
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.4625661294316523e-05,
+      "loss": 1.7188,
+      "step": 11795
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.462480874802549e-05,
+      "loss": 0.4941,
+      "step": 11796
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.46239561488792e-05,
+      "loss": 0.0302,
+      "step": 11797
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.4623103496882337e-05,
+      "loss": 0.3457,
+      "step": 11798
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.4622250792039583e-05,
+      "loss": 0.2266,
+      "step": 11799
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.4621398034355618e-05,
+      "loss": 1.9141,
+      "step": 11800
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.4620545223835126e-05,
+      "loss": 0.625,
+      "step": 11801
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.4619692360482795e-05,
+      "loss": 0.9141,
+      "step": 11802
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.46188394443033e-05,
+      "loss": 0.3066,
+      "step": 11803
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.4617986475301333e-05,
+      "loss": 0.0007,
+      "step": 11804
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.4617133453481574e-05,
+      "loss": 0.0109,
+      "step": 11805
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.4616280378848705e-05,
+      "loss": 1.4844,
+      "step": 11806
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.4615427251407414e-05,
+      "loss": 0.0006,
+      "step": 11807
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.4614574071162387e-05,
+      "loss": 0.0176,
+      "step": 11808
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.461372083811831e-05,
+      "loss": 0.064,
+      "step": 11809
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.4612867552279863e-05,
+      "loss": 0.0103,
+      "step": 11810
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.4612014213651737e-05,
+      "loss": 0.0002,
+      "step": 11811
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.4611160822238615e-05,
+      "loss": 0.0002,
+      "step": 11812
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.4610307378045183e-05,
+      "loss": 0.0001,
+      "step": 11813
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.460945388107613e-05,
+      "loss": 0.9414,
+      "step": 11814
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.460860033133615e-05,
+      "loss": 0.0001,
+      "step": 11815
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.4607746728829917e-05,
+      "loss": 0.2373,
+      "step": 11816
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.4606893073562125e-05,
+      "loss": 0.0004,
+      "step": 11817
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.4606039365537466e-05,
+      "loss": 0.0002,
+      "step": 11818
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.4605185604760615e-05,
+      "loss": 1.1328,
+      "step": 11819
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.4604331791236277e-05,
+      "loss": 0.0,
+      "step": 11820
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.460347792496913e-05,
+      "loss": 0.0001,
+      "step": 11821
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.4602624005963867e-05,
+      "loss": 0.0003,
+      "step": 11822
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.4601770034225173e-05,
+      "loss": 0.2773,
+      "step": 11823
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.460091600975775e-05,
+      "loss": 0.0006,
+      "step": 11824
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.460006193256627e-05,
+      "loss": 0.0005,
+      "step": 11825
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.4599207802655433e-05,
+      "loss": 0.0002,
+      "step": 11826
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.4598353620029933e-05,
+      "loss": 0.0,
+      "step": 11827
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.4597499384694453e-05,
+      "loss": 0.1982,
+      "step": 11828
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.4596645096653687e-05,
+      "loss": 0.127,
+      "step": 11829
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.4595790755912333e-05,
+      "loss": 0.1475,
+      "step": 11830
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.459493636247507e-05,
+      "loss": 0.1299,
+      "step": 11831
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.45940819163466e-05,
+      "loss": 0.1436,
+      "step": 11832
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.459322741753161e-05,
+      "loss": 0.2461,
+      "step": 11833
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.45923728660348e-05,
+      "loss": 0.0459,
+      "step": 11834
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.4591518261860848e-05,
+      "loss": 0.0845,
+      "step": 11835
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.4590663605014462e-05,
+      "loss": 0.5977,
+      "step": 11836
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.458980889550033e-05,
+      "loss": 0.1631,
+      "step": 11837
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.458895413332314e-05,
+      "loss": 0.0036,
+      "step": 11838
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.4588099318487597e-05,
+      "loss": 0.1206,
+      "step": 11839
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.4587244450998387e-05,
+      "loss": 0.3809,
+      "step": 11840
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.4586389530860212e-05,
+      "loss": 0.0077,
+      "step": 11841
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.4585534558077757e-05,
+      "loss": 0.6133,
+      "step": 11842
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.458467953265572e-05,
+      "loss": 0.707,
+      "step": 11843
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.4583824454598804e-05,
+      "loss": 0.1875,
+      "step": 11844
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.45829693239117e-05,
+      "loss": 0.1235,
+      "step": 11845
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.4582114140599102e-05,
+      "loss": 0.0099,
+      "step": 11846
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.4581258904665706e-05,
+      "loss": 0.5898,
+      "step": 11847
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.4580403616116212e-05,
+      "loss": 0.0011,
+      "step": 11848
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.4579548274955316e-05,
+      "loss": 0.0427,
+      "step": 11849
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.4578692881187708e-05,
+      "loss": 0.0544,
+      "step": 11850
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.4577837434818097e-05,
+      "loss": 0.0134,
+      "step": 11851
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.4576981935851175e-05,
+      "loss": 0.0006,
+      "step": 11852
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.457612638429164e-05,
+      "loss": 0.0,
+      "step": 11853
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.457527078014419e-05,
+      "loss": 0.0008,
+      "step": 11854
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.4574415123413524e-05,
+      "loss": 0.0393,
+      "step": 11855
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.4573559414104344e-05,
+      "loss": 0.1514,
+      "step": 11856
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.4572703652221347e-05,
+      "loss": 0.0075,
+      "step": 11857
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.4571847837769232e-05,
+      "loss": 0.0869,
+      "step": 11858
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.4570991970752695e-05,
+      "loss": 0.0001,
+      "step": 11859
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.4570136051176442e-05,
+      "loss": 0.0,
+      "step": 11860
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.456928007904517e-05,
+      "loss": 0.5625,
+      "step": 11861
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.4568424054363583e-05,
+      "loss": 0.0023,
+      "step": 11862
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.4567567977136376e-05,
+      "loss": 0.0001,
+      "step": 11863
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.4566711847368257e-05,
+      "loss": 0.0001,
+      "step": 11864
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.4565855665063922e-05,
+      "loss": 0.0381,
+      "step": 11865
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.4564999430228078e-05,
+      "loss": 0.0004,
+      "step": 11866
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.456414314286542e-05,
+      "loss": 0.0001,
+      "step": 11867
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.456328680298066e-05,
+      "loss": 2.25,
+      "step": 11868
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.4562430410578492e-05,
+      "loss": 0.3945,
+      "step": 11869
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.4561573965663623e-05,
+      "loss": 0.0003,
+      "step": 11870
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.4560717468240755e-05,
+      "loss": 1.0391,
+      "step": 11871
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.455986091831459e-05,
+      "loss": 0.7188,
+      "step": 11872
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.4559004315889836e-05,
+      "loss": 0.2129,
+      "step": 11873
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.4558147660971193e-05,
+      "loss": 0.0001,
+      "step": 11874
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.455729095356337e-05,
+      "loss": 0.0037,
+      "step": 11875
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.4556434193671065e-05,
+      "loss": 0.0,
+      "step": 11876
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.4555577381298992e-05,
+      "loss": 0.0884,
+      "step": 11877
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.455472051645185e-05,
+      "loss": 0.0001,
+      "step": 11878
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.455386359913434e-05,
+      "loss": 0.0,
+      "step": 11879
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.4553006629351175e-05,
+      "loss": 0.0001,
+      "step": 11880
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.4552149607107062e-05,
+      "loss": 0.1357,
+      "step": 11881
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.4551292532406707e-05,
+      "loss": 0.1738,
+      "step": 11882
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.4550435405254813e-05,
+      "loss": 0.0,
+      "step": 11883
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.4549578225656085e-05,
+      "loss": 1.6484,
+      "step": 11884
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.4548720993615237e-05,
+      "loss": 0.8125,
+      "step": 11885
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.4547863709136974e-05,
+      "loss": 0.7539,
+      "step": 11886
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.4547006372226e-05,
+      "loss": 0.8398,
+      "step": 11887
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.4546148982887023e-05,
+      "loss": 1.3516,
+      "step": 11888
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.454529154112476e-05,
+      "loss": 1.4766,
+      "step": 11889
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.4544434046943916e-05,
+      "loss": 1.0625,
+      "step": 11890
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.4543576500349193e-05,
+      "loss": 0.0262,
+      "step": 11891
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.454271890134531e-05,
+      "loss": 0.3945,
+      "step": 11892
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.454186124993697e-05,
+      "loss": 0.0203,
+      "step": 11893
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.4541003546128886e-05,
+      "loss": 0.3086,
+      "step": 11894
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.4540145789925768e-05,
+      "loss": 0.1177,
+      "step": 11895
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.4539287981332323e-05,
+      "loss": 0.0403,
+      "step": 11896
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.4538430120353262e-05,
+      "loss": 0.0058,
+      "step": 11897
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.4537572206993304e-05,
+      "loss": 0.9336,
+      "step": 11898
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.4536714241257155e-05,
+      "loss": 0.582,
+      "step": 11899
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.4535856223149525e-05,
+      "loss": 0.6016,
+      "step": 11900
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.4534998152675123e-05,
+      "loss": 0.4531,
+      "step": 11901
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.453414002983867e-05,
+      "loss": 0.0006,
+      "step": 11902
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.4533281854644866e-05,
+      "loss": 0.0002,
+      "step": 11903
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.4532423627098436e-05,
+      "loss": 0.0033,
+      "step": 11904
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.453156534720409e-05,
+      "loss": 0.0001,
+      "step": 11905
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.4530707014966538e-05,
+      "loss": 0.0001,
+      "step": 11906
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.4529848630390496e-05,
+      "loss": 0.6445,
+      "step": 11907
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.4528990193480676e-05,
+      "loss": 0.0001,
+      "step": 11908
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.4528131704241798e-05,
+      "loss": 0.0649,
+      "step": 11909
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.4527273162678564e-05,
+      "loss": 0.0001,
+      "step": 11910
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.45264145687957e-05,
+      "loss": 0.0042,
+      "step": 11911
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.452555592259792e-05,
+      "loss": 0.0491,
+      "step": 11912
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.4524697224089933e-05,
+      "loss": 0.063,
+      "step": 11913
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.452383847327646e-05,
+      "loss": 0.0001,
+      "step": 11914
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.452297967016221e-05,
+      "loss": 0.0001,
+      "step": 11915
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.4522120814751913e-05,
+      "loss": 0.5312,
+      "step": 11916
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.452126190705027e-05,
+      "loss": 0.0,
+      "step": 11917
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.452040294706201e-05,
+      "loss": 0.5195,
+      "step": 11918
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.4519543934791844e-05,
+      "loss": 1.4141,
+      "step": 11919
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.451868487024449e-05,
+      "loss": 0.1357,
+      "step": 11920
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.4517825753424662e-05,
+      "loss": 0.25,
+      "step": 11921
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.4516966584337083e-05,
+      "loss": 0.0,
+      "step": 11922
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.451610736298647e-05,
+      "loss": 0.0,
+      "step": 11923
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.451524808937754e-05,
+      "loss": 0.0,
+      "step": 11924
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.4514388763515015e-05,
+      "loss": 0.0,
+      "step": 11925
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.4513529385403614e-05,
+      "loss": 0.0001,
+      "step": 11926
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.451266995504805e-05,
+      "loss": 2.3125,
+      "step": 11927
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.451181047245305e-05,
+      "loss": 0.0977,
+      "step": 11928
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.4510950937623324e-05,
+      "loss": 0.0,
+      "step": 11929
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.4510091350563606e-05,
+      "loss": 0.0771,
+      "step": 11930
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.4509231711278608e-05,
+      "loss": 0.0,
+      "step": 11931
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.450837201977305e-05,
+      "loss": 0.1289,
+      "step": 11932
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.4507512276051658e-05,
+      "loss": 0.167,
+      "step": 11933
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.450665248011915e-05,
+      "loss": 0.9844,
+      "step": 11934
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.4505792631980247e-05,
+      "loss": 1.0781,
+      "step": 11935
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.450493273163967e-05,
+      "loss": 0.7812,
+      "step": 11936
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.450407277910215e-05,
+      "loss": 1.1172,
+      "step": 11937
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.4503212774372397e-05,
+      "loss": 0.3086,
+      "step": 11938
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.4502352717455144e-05,
+      "loss": 0.2988,
+      "step": 11939
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.4501492608355106e-05,
+      "loss": 0.2217,
+      "step": 11940
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.450063244707701e-05,
+      "loss": 0.4473,
+      "step": 11941
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.4499772233625584e-05,
+      "loss": 0.3867,
+      "step": 11942
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.4498911968005543e-05,
+      "loss": 0.0,
+      "step": 11943
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.4498051650221618e-05,
+      "loss": 0.6914,
+      "step": 11944
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.4497191280278532e-05,
+      "loss": 1.8828,
+      "step": 11945
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.4496330858181005e-05,
+      "loss": 0.0012,
+      "step": 11946
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.4495470383933768e-05,
+      "loss": 0.0002,
+      "step": 11947
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.4494609857541545e-05,
+      "loss": 0.1729,
+      "step": 11948
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.4493749279009058e-05,
+      "loss": 0.001,
+      "step": 11949
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.4492888648341042e-05,
+      "loss": 0.2344,
+      "step": 11950
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.4492027965542217e-05,
+      "loss": 0.0,
+      "step": 11951
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.4491167230617307e-05,
+      "loss": 0.0,
+      "step": 11952
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.449030644357104e-05,
+      "loss": 0.0513,
+      "step": 11953
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.4489445604408145e-05,
+      "loss": 0.0007,
+      "step": 11954
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.448858471313335e-05,
+      "loss": 0.0588,
+      "step": 11955
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.448772376975138e-05,
+      "loss": 0.0,
+      "step": 11956
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.448686277426697e-05,
+      "loss": 0.0005,
+      "step": 11957
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.4486001726684834e-05,
+      "loss": 0.1699,
+      "step": 11958
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.4485140627009715e-05,
+      "loss": 0.0212,
+      "step": 11959
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.4484279475246336e-05,
+      "loss": 0.0415,
+      "step": 11960
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.4483418271399425e-05,
+      "loss": 0.0002,
+      "step": 11961
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.448255701547371e-05,
+      "loss": 0.0031,
+      "step": 11962
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.4481695707473925e-05,
+      "loss": 0.028,
+      "step": 11963
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.4480834347404798e-05,
+      "loss": 0.0001,
+      "step": 11964
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.447997293527106e-05,
+      "loss": 0.0579,
+      "step": 11965
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.4479111471077442e-05,
+      "loss": 0.0,
+      "step": 11966
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.447824995482867e-05,
+      "loss": 0.0,
+      "step": 11967
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.447738838652948e-05,
+      "loss": 0.0001,
+      "step": 11968
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.44765267661846e-05,
+      "loss": 0.0,
+      "step": 11969
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.4475665093798766e-05,
+      "loss": 0.0,
+      "step": 11970
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.4474803369376706e-05,
+      "loss": 0.0,
+      "step": 11971
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.4473941592923156e-05,
+      "loss": 0.0,
+      "step": 11972
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.447307976444284e-05,
+      "loss": 0.1338,
+      "step": 11973
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.4472217883940503e-05,
+      "loss": 0.1133,
+      "step": 11974
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.4471355951420875e-05,
+      "loss": 0.0,
+      "step": 11975
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.4470493966888678e-05,
+      "loss": 0.1523,
+      "step": 11976
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.4469631930348658e-05,
+      "loss": 0.1328,
+      "step": 11977
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.4468769841805548e-05,
+      "loss": 0.0894,
+      "step": 11978
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.4467907701264075e-05,
+      "loss": 0.1338,
+      "step": 11979
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.4467045508728975e-05,
+      "loss": 0.0781,
+      "step": 11980
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.4466183264204992e-05,
+      "loss": 0.1641,
+      "step": 11981
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.4465320967696852e-05,
+      "loss": 0.1768,
+      "step": 11982
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.4464458619209292e-05,
+      "loss": 0.1396,
+      "step": 11983
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.446359621874705e-05,
+      "loss": 0.0044,
+      "step": 11984
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.446273376631486e-05,
+      "loss": 0.0452,
+      "step": 11985
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.4461871261917455e-05,
+      "loss": 0.0991,
+      "step": 11986
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.446100870555958e-05,
+      "loss": 0.3027,
+      "step": 11987
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.4460146097245964e-05,
+      "loss": 0.033,
+      "step": 11988
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.445928343698135e-05,
+      "loss": 0.4062,
+      "step": 11989
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.445842072477047e-05,
+      "loss": 0.7461,
+      "step": 11990
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.4457557960618067e-05,
+      "loss": 0.3203,
+      "step": 11991
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.445669514452887e-05,
+      "loss": 0.5312,
+      "step": 11992
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.445583227650763e-05,
+      "loss": 0.0204,
+      "step": 11993
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.4454969356559072e-05,
+      "loss": 0.498,
+      "step": 11994
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.4454106384687948e-05,
+      "loss": 0.3359,
+      "step": 11995
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.445324336089899e-05,
+      "loss": 0.2988,
+      "step": 11996
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.445238028519693e-05,
+      "loss": 0.668,
+      "step": 11997
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.4451517157586527e-05,
+      "loss": 0.7578,
+      "step": 11998
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.4450653978072503e-05,
+      "loss": 0.0051,
+      "step": 11999
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.4449790746659604e-05,
+      "loss": 0.0356,
+      "step": 12000
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.444892746335257e-05,
+      "loss": 0.0962,
+      "step": 12001
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.4448064128156145e-05,
+      "loss": 0.106,
+      "step": 12002
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.4447200741075067e-05,
+      "loss": 0.0053,
+      "step": 12003
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.444633730211408e-05,
+      "loss": 0.0015,
+      "step": 12004
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.4445473811277922e-05,
+      "loss": 0.0001,
+      "step": 12005
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.4444610268571336e-05,
+      "loss": 0.0001,
+      "step": 12006
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.444374667399907e-05,
+      "loss": 0.0006,
+      "step": 12007
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.4442883027565857e-05,
+      "loss": 1.25,
+      "step": 12008
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.4442019329276445e-05,
+      "loss": 0.0002,
+      "step": 12009
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.4441155579135582e-05,
+      "loss": 0.0,
+      "step": 12010
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.4440291777147995e-05,
+      "loss": 0.0,
+      "step": 12011
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.4439427923318448e-05,
+      "loss": 0.0007,
+      "step": 12012
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.443856401765167e-05,
+      "loss": 0.0,
+      "step": 12013
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.443770006015241e-05,
+      "loss": 0.0028,
+      "step": 12014
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.4436836050825414e-05,
+      "loss": 0.0135,
+      "step": 12015
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.443597198967543e-05,
+      "loss": 0.0001,
+      "step": 12016
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.443510787670719e-05,
+      "loss": 0.0,
+      "step": 12017
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.4434243711925454e-05,
+      "loss": 0.0,
+      "step": 12018
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.4433379495334962e-05,
+      "loss": 0.1924,
+      "step": 12019
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.443251522694046e-05,
+      "loss": 0.0,
+      "step": 12020
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.4431650906746688e-05,
+      "loss": 0.0001,
+      "step": 12021
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.44307865347584e-05,
+      "loss": 0.0001,
+      "step": 12022
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.442992211098035e-05,
+      "loss": 0.7148,
+      "step": 12023
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.4429057635417265e-05,
+      "loss": 0.0001,
+      "step": 12024
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.4428193108073905e-05,
+      "loss": 0.0,
+      "step": 12025
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.4427328528955022e-05,
+      "loss": 0.0,
+      "step": 12026
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.442646389806535e-05,
+      "loss": 0.0,
+      "step": 12027
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.442559921540965e-05,
+      "loss": 1.0234,
+      "step": 12028
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.4424734480992663e-05,
+      "loss": 0.0001,
+      "step": 12029
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.442386969481914e-05,
+      "loss": 0.1377,
+      "step": 12030
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.4423004856893835e-05,
+      "loss": 0.0864,
+      "step": 12031
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.442213996722149e-05,
+      "loss": 0.1494,
+      "step": 12032
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.4421275025806855e-05,
+      "loss": 0.1025,
+      "step": 12033
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.4420410032654687e-05,
+      "loss": 1.0703,
+      "step": 12034
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.441954498776972e-05,
+      "loss": 0.7852,
+      "step": 12035
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.4418679891156726e-05,
+      "loss": 0.5898,
+      "step": 12036
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.441781474282044e-05,
+      "loss": 0.9922,
+      "step": 12037
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.4416949542765625e-05,
+      "loss": 1.8984,
+      "step": 12038
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.4416084290997018e-05,
+      "loss": 0.1143,
+      "step": 12039
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.4415218987519384e-05,
+      "loss": 0.0,
+      "step": 12040
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.441435363233747e-05,
+      "loss": 0.5352,
+      "step": 12041
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.4413488225456023e-05,
+      "loss": 0.2461,
+      "step": 12042
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.4412622766879803e-05,
+      "loss": 0.625,
+      "step": 12043
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.4411757256613556e-05,
+      "loss": 0.0481,
+      "step": 12044
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.441089169466204e-05,
+      "loss": 0.7266,
+      "step": 12045
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.4410026081030007e-05,
+      "loss": 1.7578,
+      "step": 12046
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.440916041572221e-05,
+      "loss": 0.0009,
+      "step": 12047
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.4408294698743402e-05,
+      "loss": 0.3496,
+      "step": 12048
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.4407428930098337e-05,
+      "loss": 2.125,
+      "step": 12049
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.4406563109791774e-05,
+      "loss": 0.0815,
+      "step": 12050
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.4405697237828465e-05,
+      "loss": 0.5234,
+      "step": 12051
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.4404831314213162e-05,
+      "loss": 0.1367,
+      "step": 12052
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.440396533895063e-05,
+      "loss": 0.0041,
+      "step": 12053
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.4403099312045604e-05,
+      "loss": 1.8438,
+      "step": 12054
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.4402233233502862e-05,
+      "loss": 0.668,
+      "step": 12055
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.4401367103327147e-05,
+      "loss": 0.0075,
+      "step": 12056
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.4400500921523225e-05,
+      "loss": 0.3516,
+      "step": 12057
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.4399634688095843e-05,
+      "loss": 0.0742,
+      "step": 12058
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.4398768403049766e-05,
+      "loss": 0.0,
+      "step": 12059
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.4397902066389742e-05,
+      "loss": 0.0016,
+      "step": 12060
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.439703567812054e-05,
+      "loss": 0.2852,
+      "step": 12061
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.439616923824691e-05,
+      "loss": 0.0,
+      "step": 12062
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.439530274677361e-05,
+      "loss": 0.0004,
+      "step": 12063
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.4394436203705403e-05,
+      "loss": 0.1533,
+      "step": 12064
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.439356960904704e-05,
+      "loss": 0.0,
+      "step": 12065
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.4392702962803293e-05,
+      "loss": 0.0001,
+      "step": 12066
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.439183626497891e-05,
+      "loss": 0.0,
+      "step": 12067
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.439096951557865e-05,
+      "loss": 0.0005,
+      "step": 12068
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.4390102714607282e-05,
+      "loss": 0.0,
+      "step": 12069
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.438923586206956e-05,
+      "loss": 0.0096,
+      "step": 12070
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.4388368957970242e-05,
+      "loss": 0.0001,
+      "step": 12071
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.4387502002314096e-05,
+      "loss": 0.0,
+      "step": 12072
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.4386634995105878e-05,
+      "loss": 0.2695,
+      "step": 12073
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.4385767936350345e-05,
+      "loss": 0.0,
+      "step": 12074
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.4384900826052267e-05,
+      "loss": 0.0,
+      "step": 12075
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.4384033664216402e-05,
+      "loss": 0.0001,
+      "step": 12076
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.438316645084751e-05,
+      "loss": 0.1689,
+      "step": 12077
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.4382299185950362e-05,
+      "loss": 0.1582,
+      "step": 12078
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.438143186952971e-05,
+      "loss": 0.1128,
+      "step": 12079
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.4380564501590323e-05,
+      "loss": 0.1367,
+      "step": 12080
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.437969708213696e-05,
+      "loss": 0.1455,
+      "step": 12081
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.437882961117439e-05,
+      "loss": 0.1387,
+      "step": 12082
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.437796208870737e-05,
+      "loss": 0.0,
+      "step": 12083
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.437709451474067e-05,
+      "loss": 1.5312,
+      "step": 12084
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.437622688927905e-05,
+      "loss": 0.5312,
+      "step": 12085
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.4375359212327285e-05,
+      "loss": 2.2812,
+      "step": 12086
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.437449148389012e-05,
+      "loss": 0.4727,
+      "step": 12087
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.437362370397234e-05,
+      "loss": 0.0806,
+      "step": 12088
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.4372755872578704e-05,
+      "loss": 0.0156,
+      "step": 12089
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.4371887989713967e-05,
+      "loss": 0.6797,
+      "step": 12090
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.437102005538291e-05,
+      "loss": 0.0156,
+      "step": 12091
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.437015206959029e-05,
+      "loss": 0.1167,
+      "step": 12092
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.436928403234088e-05,
+      "loss": 0.3457,
+      "step": 12093
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.4368415943639443e-05,
+      "loss": 0.3398,
+      "step": 12094
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.4367547803490746e-05,
+      "loss": 1.0,
+      "step": 12095
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.436667961189956e-05,
+      "loss": 0.1338,
+      "step": 12096
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.4365811368870648e-05,
+      "loss": 0.0005,
+      "step": 12097
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.436494307440878e-05,
+      "loss": 0.7266,
+      "step": 12098
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.4364074728518728e-05,
+      "loss": 0.5234,
+      "step": 12099
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.4363206331205258e-05,
+      "loss": 0.0002,
+      "step": 12100
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.436233788247313e-05,
+      "loss": 0.0002,
+      "step": 12101
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.4361469382327127e-05,
+      "loss": 0.0,
+      "step": 12102
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.436060083077201e-05,
+      "loss": 0.0016,
+      "step": 12103
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.435973222781255e-05,
+      "loss": 0.3359,
+      "step": 12104
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.435886357345353e-05,
+      "loss": 0.0,
+      "step": 12105
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.4357994867699697e-05,
+      "loss": 0.0,
+      "step": 12106
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.4357126110555833e-05,
+      "loss": 1.1094,
+      "step": 12107
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.4356257302026716e-05,
+      "loss": 0.0422,
+      "step": 12108
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.43553884421171e-05,
+      "loss": 0.0029,
+      "step": 12109
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.435451953083177e-05,
+      "loss": 0.1465,
+      "step": 12110
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.43536505681755e-05,
+      "loss": 0.0086,
+      "step": 12111
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.435278155415305e-05,
+      "loss": 0.0635,
+      "step": 12112
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.43519124887692e-05,
+      "loss": 0.0,
+      "step": 12113
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.4351043372028724e-05,
+      "loss": 0.6094,
+      "step": 12114
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.4350174203936387e-05,
+      "loss": 0.0183,
+      "step": 12115
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.434930498449697e-05,
+      "loss": 0.0,
+      "step": 12116
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.4348435713715245e-05,
+      "loss": 0.7422,
+      "step": 12117
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.4347566391595977e-05,
+      "loss": 0.0,
+      "step": 12118
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.4346697018143954e-05,
+      "loss": 0.0002,
+      "step": 12119
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.4345827593363937e-05,
+      "loss": 0.0,
+      "step": 12120
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.4344958117260707e-05,
+      "loss": 0.0,
+      "step": 12121
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.4344088589839043e-05,
+      "loss": 0.0017,
+      "step": 12122
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.434321901110371e-05,
+      "loss": 0.0,
+      "step": 12123
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.4342349381059492e-05,
+      "loss": 0.1348,
+      "step": 12124
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.4341479699711156e-05,
+      "loss": 0.0,
+      "step": 12125
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.434060996706349e-05,
+      "loss": 0.2314,
+      "step": 12126
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.433974018312126e-05,
+      "loss": 0.0845,
+      "step": 12127
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.433887034788925e-05,
+      "loss": 0.0,
+      "step": 12128
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.4338000461372226e-05,
+      "loss": 0.1348,
+      "step": 12129
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.433713052357498e-05,
+      "loss": 0.0459,
+      "step": 12130
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.433626053450227e-05,
+      "loss": 0.0,
+      "step": 12131
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.4335390494158893e-05,
+      "loss": 0.0938,
+      "step": 12132
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.4334520402549613e-05,
+      "loss": 0.0001,
+      "step": 12133
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.4333650259679218e-05,
+      "loss": 0.6914,
+      "step": 12134
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.433278006555248e-05,
+      "loss": 0.025,
+      "step": 12135
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.433190982017418e-05,
+      "loss": 0.2021,
+      "step": 12136
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.4331039523549098e-05,
+      "loss": 0.0309,
+      "step": 12137
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.4330169175682007e-05,
+      "loss": 0.0236,
+      "step": 12138
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.43292987765777e-05,
+      "loss": 0.2539,
+      "step": 12139
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.4328428326240946e-05,
+      "loss": 0.6289,
+      "step": 12140
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.4327557824676522e-05,
+      "loss": 0.9375,
+      "step": 12141
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.4326687271889217e-05,
+      "loss": 0.0603,
+      "step": 12142
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.4325816667883812e-05,
+      "loss": 0.0569,
+      "step": 12143
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.4324946012665085e-05,
+      "loss": 0.0045,
+      "step": 12144
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.432407530623781e-05,
+      "loss": 0.1328,
+      "step": 12145
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.4323204548606786e-05,
+      "loss": 0.0001,
+      "step": 12146
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.4322333739776773e-05,
+      "loss": 0.0005,
+      "step": 12147
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.4321462879752573e-05,
+      "loss": 0.1211,
+      "step": 12148
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.4320591968538954e-05,
+      "loss": 0.0006,
+      "step": 12149
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.431972100614071e-05,
+      "loss": 0.0001,
+      "step": 12150
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.4318849992562615e-05,
+      "loss": 0.0312,
+      "step": 12151
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.4317978927809455e-05,
+      "loss": 0.2734,
+      "step": 12152
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.4317107811886014e-05,
+      "loss": 0.0001,
+      "step": 12153
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.431623664479708e-05,
+      "loss": 0.0003,
+      "step": 12154
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.431536542654743e-05,
+      "loss": 0.0176,
+      "step": 12155
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.4314494157141853e-05,
+      "loss": 0.0003,
+      "step": 12156
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.431362283658513e-05,
+      "loss": 0.0006,
+      "step": 12157
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.4312751464882047e-05,
+      "loss": 0.0001,
+      "step": 12158
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.431188004203739e-05,
+      "loss": 0.0001,
+      "step": 12159
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.4311008568055948e-05,
+      "loss": 0.0,
+      "step": 12160
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.43101370429425e-05,
+      "loss": 0.0,
+      "step": 12161
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.4309265466701837e-05,
+      "loss": 0.0,
+      "step": 12162
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.4308393839338747e-05,
+      "loss": 0.0023,
+      "step": 12163
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.4307522160858012e-05,
+      "loss": 0.0854,
+      "step": 12164
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.4306650431264418e-05,
+      "loss": 0.0,
+      "step": 12165
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.4305778650562754e-05,
+      "loss": 0.0,
+      "step": 12166
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.4304906818757806e-05,
+      "loss": 0.0928,
+      "step": 12167
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.4304034935854366e-05,
+      "loss": 0.0,
+      "step": 12168
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.430316300185722e-05,
+      "loss": 0.1348,
+      "step": 12169
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.4302291016771156e-05,
+      "loss": 0.0001,
+      "step": 12170
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.4301418980600963e-05,
+      "loss": 0.0,
+      "step": 12171
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.4300546893351427e-05,
+      "loss": 0.0,
+      "step": 12172
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.429967475502734e-05,
+      "loss": 0.0,
+      "step": 12173
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.4298802565633492e-05,
+      "loss": 0.0,
+      "step": 12174
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.429793032517467e-05,
+      "loss": 0.0,
+      "step": 12175
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.4297058033655668e-05,
+      "loss": 0.1104,
+      "step": 12176
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.429618569108127e-05,
+      "loss": 0.1157,
+      "step": 12177
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.429531329745627e-05,
+      "loss": 0.0002,
+      "step": 12178
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.429444085278546e-05,
+      "loss": 0.0,
+      "step": 12179
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.4293568357073634e-05,
+      "loss": 0.1172,
+      "step": 12180
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.4292695810325573e-05,
+      "loss": 0.1631,
+      "step": 12181
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.429182321254608e-05,
+      "loss": 0.1074,
+      "step": 12182
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.4290950563739937e-05,
+      "loss": 0.0,
+      "step": 12183
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.429007786391195e-05,
+      "loss": 1.2188,
+      "step": 12184
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.4289205113066896e-05,
+      "loss": 0.9727,
+      "step": 12185
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.4288332311209575e-05,
+      "loss": 0.668,
+      "step": 12186
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.4287459458344773e-05,
+      "loss": 0.2412,
+      "step": 12187
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.42865865544773e-05,
+      "loss": 0.1621,
+      "step": 12188
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.428571359961194e-05,
+      "loss": 0.5312,
+      "step": 12189
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.4284840593753482e-05,
+      "loss": 0.0962,
+      "step": 12190
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.4283967536906724e-05,
+      "loss": 0.0396,
+      "step": 12191
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.4283094429076464e-05,
+      "loss": 0.1445,
+      "step": 12192
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.4282221270267485e-05,
+      "loss": 0.2949,
+      "step": 12193
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.42813480604846e-05,
+      "loss": 0.2363,
+      "step": 12194
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.428047479973259e-05,
+      "loss": 0.032,
+      "step": 12195
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.4279601488016257e-05,
+      "loss": 1.3359,
+      "step": 12196
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.42787281253404e-05,
+      "loss": 0.0371,
+      "step": 12197
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.4277854711709806e-05,
+      "loss": 1.2578,
+      "step": 12198
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.4276981247129275e-05,
+      "loss": 0.0864,
+      "step": 12199
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.4276107731603602e-05,
+      "loss": 0.1094,
+      "step": 12200
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.427523416513759e-05,
+      "loss": 0.0066,
+      "step": 12201
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.4274360547736034e-05,
+      "loss": 0.0001,
+      "step": 12202
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.4273486879403733e-05,
+      "loss": 0.0002,
+      "step": 12203
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.4272613160145474e-05,
+      "loss": 0.3555,
+      "step": 12204
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.427173938996607e-05,
+      "loss": 0.2617,
+      "step": 12205
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.427086556887031e-05,
+      "loss": 0.0,
+      "step": 12206
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.4269991696862996e-05,
+      "loss": 0.0001,
+      "step": 12207
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.4269117773948926e-05,
+      "loss": 1.8438,
+      "step": 12208
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.4268243800132903e-05,
+      "loss": 0.0942,
+      "step": 12209
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.4267369775419716e-05,
+      "loss": 0.0698,
+      "step": 12210
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.4266495699814174e-05,
+      "loss": 0.0003,
+      "step": 12211
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.4265621573321075e-05,
+      "loss": 0.1602,
+      "step": 12212
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.426474739594522e-05,
+      "loss": 0.0,
+      "step": 12213
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.4263873167691408e-05,
+      "loss": 0.0,
+      "step": 12214
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.4262998888564442e-05,
+      "loss": 0.0022,
+      "step": 12215
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.426212455856912e-05,
+      "loss": 0.0762,
+      "step": 12216
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.426125017771024e-05,
+      "loss": 0.0,
+      "step": 12217
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.4260375745992614e-05,
+      "loss": 0.5703,
+      "step": 12218
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.425950126342104e-05,
+      "loss": 0.0,
+      "step": 12219
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.425862673000032e-05,
+      "loss": 0.0,
+      "step": 12220
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.4257752145735253e-05,
+      "loss": 0.0,
+      "step": 12221
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.425687751063064e-05,
+      "loss": 0.0,
+      "step": 12222
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.4256002824691296e-05,
+      "loss": 0.0001,
+      "step": 12223
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.4255128087922014e-05,
+      "loss": 0.0063,
+      "step": 12224
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.42542533003276e-05,
+      "loss": 1.3047,
+      "step": 12225
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.4253378461912857e-05,
+      "loss": 0.0,
+      "step": 12226
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.4252503572682594e-05,
+      "loss": 0.0,
+      "step": 12227
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.425162863264161e-05,
+      "loss": 0.0933,
+      "step": 12228
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.4250753641794716e-05,
+      "loss": 0.1108,
+      "step": 12229
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.4249878600146712e-05,
+      "loss": 0.1582,
+      "step": 12230
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.42490035077024e-05,
+      "loss": 0.1221,
+      "step": 12231
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.424812836446659e-05,
+      "loss": 0.0001,
+      "step": 12232
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.4247253170444095e-05,
+      "loss": 0.0776,
+      "step": 12233
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.424637792563971e-05,
+      "loss": 0.2754,
+      "step": 12234
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.4245502630058243e-05,
+      "loss": 0.0117,
+      "step": 12235
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.424462728370451e-05,
+      "loss": 0.4707,
+      "step": 12236
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.424375188658331e-05,
+      "loss": 0.7344,
+      "step": 12237
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.4242876438699447e-05,
+      "loss": 0.1348,
+      "step": 12238
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.4242000940057734e-05,
+      "loss": 0.4121,
+      "step": 12239
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.4241125390662982e-05,
+      "loss": 0.0674,
+      "step": 12240
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.424024979051999e-05,
+      "loss": 0.8984,
+      "step": 12241
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.4239374139633577e-05,
+      "loss": 1.4219,
+      "step": 12242
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.4238498438008546e-05,
+      "loss": 0.8633,
+      "step": 12243
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.4237622685649705e-05,
+      "loss": 0.5117,
+      "step": 12244
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.4236746882561867e-05,
+      "loss": 0.5234,
+      "step": 12245
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.4235871028749837e-05,
+      "loss": 0.9375,
+      "step": 12246
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.4234995124218426e-05,
+      "loss": 0.2256,
+      "step": 12247
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.4234119168972445e-05,
+      "loss": 0.6133,
+      "step": 12248
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.4233243163016706e-05,
+      "loss": 0.3867,
+      "step": 12249
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.4232367106356017e-05,
+      "loss": 0.6328,
+      "step": 12250
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.423149099899519e-05,
+      "loss": 0.0001,
+      "step": 12251
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.423061484093904e-05,
+      "loss": 0.3711,
+      "step": 12252
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.422973863219237e-05,
+      "loss": 0.0305,
+      "step": 12253
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.422886237276e-05,
+      "loss": 0.042,
+      "step": 12254
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.422798606264673e-05,
+      "loss": 0.3555,
+      "step": 12255
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.422710970185739e-05,
+      "loss": 0.0093,
+      "step": 12256
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.422623329039678e-05,
+      "loss": 0.5742,
+      "step": 12257
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.4225356828269717e-05,
+      "loss": 0.104,
+      "step": 12258
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.4224480315481013e-05,
+      "loss": 0.0,
+      "step": 12259
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.4223603752035483e-05,
+      "loss": 0.0,
+      "step": 12260
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.4222727137937938e-05,
+      "loss": 0.0,
+      "step": 12261
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.4221850473193193e-05,
+      "loss": 0.0,
+      "step": 12262
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.4220973757806066e-05,
+      "loss": 0.0,
+      "step": 12263
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.4220096991781365e-05,
+      "loss": 0.014,
+      "step": 12264
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.421922017512391e-05,
+      "loss": 0.2109,
+      "step": 12265
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.4218343307838515e-05,
+      "loss": 0.0,
+      "step": 12266
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.421746638992999e-05,
+      "loss": 0.1406,
+      "step": 12267
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.421658942140316e-05,
+      "loss": 0.0,
+      "step": 12268
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.4215712402262833e-05,
+      "loss": 0.0001,
+      "step": 12269
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.4214835332513832e-05,
+      "loss": 0.1416,
+      "step": 12270
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.4213958212160968e-05,
+      "loss": 0.0,
+      "step": 12271
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.421308104120906e-05,
+      "loss": 0.0,
+      "step": 12272
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.4212203819662924e-05,
+      "loss": 0.1602,
+      "step": 12273
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.4211326547527377e-05,
+      "loss": 0.0,
+      "step": 12274
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.421044922480724e-05,
+      "loss": 0.0,
+      "step": 12275
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.4209571851507328e-05,
+      "loss": 0.003,
+      "step": 12276
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.4208694427632455e-05,
+      "loss": 0.1006,
+      "step": 12277
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.4207816953187452e-05,
+      "loss": 0.0,
+      "step": 12278
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.420693942817712e-05,
+      "loss": 0.0,
+      "step": 12279
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.4206061852606297e-05,
+      "loss": 1.1797,
+      "step": 12280
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.420518422647979e-05,
+      "loss": 0.1572,
+      "step": 12281
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.4204306549802423e-05,
+      "loss": 0.0,
+      "step": 12282
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.420342882257901e-05,
+      "loss": 0.2402,
+      "step": 12283
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.420255104481438e-05,
+      "loss": 0.7109,
+      "step": 12284
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.4201673216513343e-05,
+      "loss": 0.1436,
+      "step": 12285
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.4200795337680726e-05,
+      "loss": 0.0603,
+      "step": 12286
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.4199917408321358e-05,
+      "loss": 0.3828,
+      "step": 12287
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.4199039428440046e-05,
+      "loss": 0.957,
+      "step": 12288
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.4198161398041614e-05,
+      "loss": 0.0996,
+      "step": 12289
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.419728331713089e-05,
+      "loss": 0.3516,
+      "step": 12290
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.4196405185712697e-05,
+      "loss": 0.0084,
+      "step": 12291
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.4195527003791848e-05,
+      "loss": 0.8008,
+      "step": 12292
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.419464877137317e-05,
+      "loss": 0.3574,
+      "step": 12293
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.4193770488461485e-05,
+      "loss": 0.8164,
+      "step": 12294
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.4192892155061627e-05,
+      "loss": 0.0141,
+      "step": 12295
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.4192013771178406e-05,
+      "loss": 0.0132,
+      "step": 12296
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.4191135336816646e-05,
+      "loss": 0.0104,
+      "step": 12297
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.419025685198118e-05,
+      "loss": 0.125,
+      "step": 12298
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.418937831667682e-05,
+      "loss": 0.0,
+      "step": 12299
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.41884997309084e-05,
+      "loss": 0.0001,
+      "step": 12300
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.4187621094680746e-05,
+      "loss": 0.0,
+      "step": 12301
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.418674240799868e-05,
+      "loss": 0.5859,
+      "step": 12302
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.4185863670867025e-05,
+      "loss": 0.0815,
+      "step": 12303
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.4184984883290612e-05,
+      "loss": 0.0249,
+      "step": 12304
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.418410604527426e-05,
+      "loss": 0.0,
+      "step": 12305
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.41832271568228e-05,
+      "loss": 0.085,
+      "step": 12306
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.4182348217941057e-05,
+      "loss": 0.0386,
+      "step": 12307
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.4181469228633858e-05,
+      "loss": 0.0,
+      "step": 12308
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.418059018890603e-05,
+      "loss": 0.0,
+      "step": 12309
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.4179711098762397e-05,
+      "loss": 0.0,
+      "step": 12310
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 2.4178831958207795e-05,
+      "loss": 0.0698,
+      "step": 12311
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.4177952767247045e-05,
+      "loss": 0.0,
+      "step": 12312
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.417707352588497e-05,
+      "loss": 0.0254,
+      "step": 12313
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.4176194234126416e-05,
+      "loss": 0.498,
+      "step": 12314
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.4175314891976197e-05,
+      "loss": 0.8281,
+      "step": 12315
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.4174435499439143e-05,
+      "loss": 0.0,
+      "step": 12316
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.4173556056520087e-05,
+      "loss": 0.2246,
+      "step": 12317
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.4172676563223863e-05,
+      "loss": 0.0,
+      "step": 12318
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.417179701955529e-05,
+      "loss": 0.1797,
+      "step": 12319
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.4170917425519204e-05,
+      "loss": 0.0,
+      "step": 12320
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.417003778112044e-05,
+      "loss": 0.0,
+      "step": 12321
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.4169158086363815e-05,
+      "loss": 0.0,
+      "step": 12322
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.4168278341254176e-05,
+      "loss": 0.0,
+      "step": 12323
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.416739854579634e-05,
+      "loss": 0.0182,
+      "step": 12324
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.4166518699995152e-05,
+      "loss": 0.0,
+      "step": 12325
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.416563880385543e-05,
+      "loss": 0.0,
+      "step": 12326
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.4164758857382017e-05,
+      "loss": 0.0,
+      "step": 12327
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.4163878860579742e-05,
+      "loss": 0.0,
+      "step": 12328
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.4162998813453437e-05,
+      "loss": 0.0,
+      "step": 12329
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.4162118716007927e-05,
+      "loss": 0.0811,
+      "step": 12330
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.416123856824806e-05,
+      "loss": 0.103,
+      "step": 12331
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.416035837017866e-05,
+      "loss": 0.1514,
+      "step": 12332
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.4159478121804562e-05,
+      "loss": 0.1992,
+      "step": 12333
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.4158597823130604e-05,
+      "loss": 0.668,
+      "step": 12334
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.415771747416161e-05,
+      "loss": 0.3984,
+      "step": 12335
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.4156837074902426e-05,
+      "loss": 0.4141,
+      "step": 12336
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.415595662535788e-05,
+      "loss": 0.582,
+      "step": 12337
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.4155076125532813e-05,
+      "loss": 0.0747,
+      "step": 12338
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.4154195575432056e-05,
+      "loss": 0.0003,
+      "step": 12339
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.415331497506044e-05,
+      "loss": 0.2188,
+      "step": 12340
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.415243432442281e-05,
+      "loss": 0.4062,
+      "step": 12341
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.4151553623523994e-05,
+      "loss": 0.2812,
+      "step": 12342
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.4150672872368837e-05,
+      "loss": 0.1631,
+      "step": 12343
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.414979207096217e-05,
+      "loss": 0.016,
+      "step": 12344
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.4148911219308826e-05,
+      "loss": 1.2266,
+      "step": 12345
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.4148030317413652e-05,
+      "loss": 0.0042,
+      "step": 12346
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.4147149365281483e-05,
+      "loss": 0.0011,
+      "step": 12347
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.4146268362917147e-05,
+      "loss": 0.0383,
+      "step": 12348
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.4145387310325498e-05,
+      "loss": 0.0309,
+      "step": 12349
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.414450620751136e-05,
+      "loss": 0.0001,
+      "step": 12350
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.4143625054479582e-05,
+      "loss": 0.9062,
+      "step": 12351
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.4142743851235e-05,
+      "loss": 0.0217,
+      "step": 12352
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.414186259778245e-05,
+      "loss": 0.0009,
+      "step": 12353
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.414098129412677e-05,
+      "loss": 0.0952,
+      "step": 12354
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.4140099940272806e-05,
+      "loss": 0.0776,
+      "step": 12355
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.4139218536225395e-05,
+      "loss": 0.0,
+      "step": 12356
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.4138337081989378e-05,
+      "loss": 0.0981,
+      "step": 12357
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.4137455577569596e-05,
+      "loss": 0.0009,
+      "step": 12358
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.4136574022970887e-05,
+      "loss": 1.4844,
+      "step": 12359
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.4135692418198094e-05,
+      "loss": 0.033,
+      "step": 12360
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.4134810763256058e-05,
+      "loss": 0.0,
+      "step": 12361
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.413392905814962e-05,
+      "loss": 0.0,
+      "step": 12362
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.4133047302883623e-05,
+      "loss": 0.0464,
+      "step": 12363
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.413216549746291e-05,
+      "loss": 0.1089,
+      "step": 12364
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.4131283641892325e-05,
+      "loss": 0.0,
+      "step": 12365
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.4130401736176705e-05,
+      "loss": 0.0108,
+      "step": 12366
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.41295197803209e-05,
+      "loss": 0.0001,
+      "step": 12367
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.4128637774329744e-05,
+      "loss": 0.0072,
+      "step": 12368
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.4127755718208092e-05,
+      "loss": 0.022,
+      "step": 12369
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.412687361196078e-05,
+      "loss": 0.0001,
+      "step": 12370
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.4125991455592654e-05,
+      "loss": 0.0,
+      "step": 12371
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.4125109249108556e-05,
+      "loss": 0.6797,
+      "step": 12372
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.412422699251334e-05,
+      "loss": 0.8242,
+      "step": 12373
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.4123344685811837e-05,
+      "loss": 0.0001,
+      "step": 12374
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.4122462329008906e-05,
+      "loss": 0.0001,
+      "step": 12375
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.4121579922109383e-05,
+      "loss": 0.0,
+      "step": 12376
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.4120697465118117e-05,
+      "loss": 0.0,
+      "step": 12377
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 2.4119814958039953e-05,
+      "loss": 0.0708,
+      "step": 12378
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.411893240087974e-05,
+      "loss": 0.1445,
+      "step": 12379
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.4118049793642325e-05,
+      "loss": 0.2012,
+      "step": 12380
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.4117167136332548e-05,
+      "loss": 0.1553,
+      "step": 12381
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.4116284428955263e-05,
+      "loss": 0.1279,
+      "step": 12382
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.411540167151532e-05,
+      "loss": 0.1377,
+      "step": 12383
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.4114518864017554e-05,
+      "loss": 0.6289,
+      "step": 12384
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.411363600646683e-05,
+      "loss": 0.5156,
+      "step": 12385
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.4112753098867985e-05,
+      "loss": 0.3164,
+      "step": 12386
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.411187014122586e-05,
+      "loss": 0.9531,
+      "step": 12387
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.4110987133545326e-05,
+      "loss": 0.6758,
+      "step": 12388
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.4110104075831217e-05,
+      "loss": 0.1074,
+      "step": 12389
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.4109220968088383e-05,
+      "loss": 0.4355,
+      "step": 12390
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.4108337810321677e-05,
+      "loss": 0.4336,
+      "step": 12391
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.4107454602535945e-05,
+      "loss": 0.9297,
+      "step": 12392
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.4106571344736044e-05,
+      "loss": 1.7344,
+      "step": 12393
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.4105688036926824e-05,
+      "loss": 0.7539,
+      "step": 12394
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.4104804679113125e-05,
+      "loss": 1.2891,
+      "step": 12395
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.410392127129981e-05,
+      "loss": 1.5547,
+      "step": 12396
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.410303781349172e-05,
+      "loss": 0.1729,
+      "step": 12397
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.4102154305693718e-05,
+      "loss": 0.2363,
+      "step": 12398
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.410127074791065e-05,
+      "loss": 0.106,
+      "step": 12399
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.4100387140147363e-05,
+      "loss": 0.0552,
+      "step": 12400
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.4099503482408716e-05,
+      "loss": 0.0,
+      "step": 12401
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.4098619774699568e-05,
+      "loss": 0.2344,
+      "step": 12402
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.4097736017024756e-05,
+      "loss": 0.2754,
+      "step": 12403
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.409685220938914e-05,
+      "loss": 0.1982,
+      "step": 12404
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.409596835179758e-05,
+      "loss": 0.0,
+      "step": 12405
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.409508444425492e-05,
+      "loss": 0.0001,
+      "step": 12406
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.409420048676602e-05,
+      "loss": 0.0001,
+      "step": 12407
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.4093316479335737e-05,
+      "loss": 0.0001,
+      "step": 12408
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.4092432421968918e-05,
+      "loss": 0.0737,
+      "step": 12409
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.4091548314670422e-05,
+      "loss": 0.0001,
+      "step": 12410
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.4090664157445103e-05,
+      "loss": 0.0542,
+      "step": 12411
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.4089779950297817e-05,
+      "loss": 0.0,
+      "step": 12412
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.4088895693233423e-05,
+      "loss": 0.0001,
+      "step": 12413
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.4088011386256762e-05,
+      "loss": 0.2637,
+      "step": 12414
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.4087127029372713e-05,
+      "loss": 1.8984,
+      "step": 12415
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.4086242622586118e-05,
+      "loss": 0.0001,
+      "step": 12416
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.4085358165901835e-05,
+      "loss": 0.0688,
+      "step": 12417
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.408447365932473e-05,
+      "loss": 0.103,
+      "step": 12418
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.4083589102859647e-05,
+      "loss": 0.0028,
+      "step": 12419
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.408270449651145e-05,
+      "loss": 0.0,
+      "step": 12420
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.4081819840285e-05,
+      "loss": 0.0,
+      "step": 12421
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.4080935134185148e-05,
+      "loss": 0.0001,
+      "step": 12422
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.408005037821676e-05,
+      "loss": 0.0258,
+      "step": 12423
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.407916557238469e-05,
+      "loss": 0.0002,
+      "step": 12424
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.4078280716693796e-05,
+      "loss": 0.0,
+      "step": 12425
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.4077395811148943e-05,
+      "loss": 0.0,
+      "step": 12426
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.4076510855754984e-05,
+      "loss": 0.0,
+      "step": 12427
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.4075625850516786e-05,
+      "loss": 0.124,
+      "step": 12428
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.40747407954392e-05,
+      "loss": 0.0009,
+      "step": 12429
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.407385569052709e-05,
+      "loss": 0.6523,
+      "step": 12430
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.407297053578532e-05,
+      "loss": 0.1079,
+      "step": 12431
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.407208533121875e-05,
+      "loss": 0.1855,
+      "step": 12432
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.4071200076832235e-05,
+      "loss": 0.1079,
+      "step": 12433
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.4070314772630648e-05,
+      "loss": 0.7305,
+      "step": 12434
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.406942941861884e-05,
+      "loss": 0.0011,
+      "step": 12435
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.4068544014801676e-05,
+      "loss": 0.8984,
+      "step": 12436
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.4067658561184022e-05,
+      "loss": 0.5234,
+      "step": 12437
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.4066773057770734e-05,
+      "loss": 0.0123,
+      "step": 12438
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.406588750456668e-05,
+      "loss": 0.1416,
+      "step": 12439
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.4065001901576723e-05,
+      "loss": 0.2295,
+      "step": 12440
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.4064116248805723e-05,
+      "loss": 0.2188,
+      "step": 12441
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.4063230546258547e-05,
+      "loss": 0.0659,
+      "step": 12442
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.4062344793940053e-05,
+      "loss": 0.543,
+      "step": 12443
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.4061458991855115e-05,
+      "loss": 0.0284,
+      "step": 12444
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.4060573140008587e-05,
+      "loss": 0.5195,
+      "step": 12445
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 2.405968723840534e-05,
+      "loss": 0.0057,
+      "step": 12446
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.405880128705024e-05,
+      "loss": 0.4316,
+      "step": 12447
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.4057915285948148e-05,
+      "loss": 0.0613,
+      "step": 12448
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.4057029235103935e-05,
+      "loss": 0.0287,
+      "step": 12449
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.405614313452246e-05,
+      "loss": 0.5352,
+      "step": 12450
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.405525698420859e-05,
+      "loss": 0.543,
+      "step": 12451
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.4054370784167198e-05,
+      "loss": 0.832,
+      "step": 12452
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.405348453440314e-05,
+      "loss": 0.0003,
+      "step": 12453
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.4052598234921297e-05,
+      "loss": 0.4375,
+      "step": 12454
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.4051711885726522e-05,
+      "loss": 0.0188,
+      "step": 12455
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.4050825486823694e-05,
+      "loss": 0.0002,
+      "step": 12456
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.4049939038217674e-05,
+      "loss": 0.0002,
+      "step": 12457
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.404905253991333e-05,
+      "loss": 0.0029,
+      "step": 12458
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.4048165991915536e-05,
+      "loss": 0.0408,
+      "step": 12459
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.404727939422915e-05,
+      "loss": 0.0272,
+      "step": 12460
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.4046392746859052e-05,
+      "loss": 0.0001,
+      "step": 12461
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.40455060498101e-05,
+      "loss": 0.2295,
+      "step": 12462
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.4044619303087178e-05,
+      "loss": 0.0001,
+      "step": 12463
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.404373250669514e-05,
+      "loss": 0.0,
+      "step": 12464
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.4042845660638867e-05,
+      "loss": 0.0,
+      "step": 12465
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.4041958764923223e-05,
+      "loss": 0.0206,
+      "step": 12466
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.404107181955308e-05,
+      "loss": 0.1514,
+      "step": 12467
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.4040184824533313e-05,
+      "loss": 0.2598,
+      "step": 12468
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.4039297779868788e-05,
+      "loss": 0.0002,
+      "step": 12469
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.4038410685564375e-05,
+      "loss": 0.0408,
+      "step": 12470
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.403752354162495e-05,
+      "loss": 0.0001,
+      "step": 12471
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.403663634805538e-05,
+      "loss": 0.0,
+      "step": 12472
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.4035749104860546e-05,
+      "loss": 0.0,
+      "step": 12473
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.4034861812045312e-05,
+      "loss": 0.0001,
+      "step": 12474
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.4033974469614554e-05,
+      "loss": 0.0,
+      "step": 12475
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.4033087077573144e-05,
+      "loss": 0.0,
+      "step": 12476
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.4032199635925947e-05,
+      "loss": 0.0103,
+      "step": 12477
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.4031312144677853e-05,
+      "loss": 0.2812,
+      "step": 12478
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.4030424603833723e-05,
+      "loss": 0.1426,
+      "step": 12479
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.402953701339844e-05,
+      "loss": 0.1177,
+      "step": 12480
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.4028649373376873e-05,
+      "loss": 0.1289,
+      "step": 12481
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.4027761683773896e-05,
+      "loss": 0.0869,
+      "step": 12482
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.4026873944594383e-05,
+      "loss": 0.0815,
+      "step": 12483
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.4025986155843212e-05,
+      "loss": 0.2695,
+      "step": 12484
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.402509831752526e-05,
+      "loss": 0.025,
+      "step": 12485
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.4024210429645394e-05,
+      "loss": 0.2832,
+      "step": 12486
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.40233224922085e-05,
+      "loss": 0.209,
+      "step": 12487
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.402243450521945e-05,
+      "loss": 0.8242,
+      "step": 12488
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.402154646868312e-05,
+      "loss": 0.5547,
+      "step": 12489
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.4020658382604386e-05,
+      "loss": 0.3906,
+      "step": 12490
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.4019770246988124e-05,
+      "loss": 0.4336,
+      "step": 12491
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.4018882061839215e-05,
+      "loss": 0.373,
+      "step": 12492
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.4017993827162537e-05,
+      "loss": 0.0004,
+      "step": 12493
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.4017105542962967e-05,
+      "loss": 0.0229,
+      "step": 12494
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.4016217209245377e-05,
+      "loss": 0.0,
+      "step": 12495
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.401532882601465e-05,
+      "loss": 0.3984,
+      "step": 12496
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.4014440393275667e-05,
+      "loss": 0.0002,
+      "step": 12497
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.401355191103331e-05,
+      "loss": 0.6016,
+      "step": 12498
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.4012663379292446e-05,
+      "loss": 0.0006,
+      "step": 12499
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.4011774798057958e-05,
+      "loss": 0.9141,
+      "step": 12500
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.4010886167334736e-05,
+      "loss": 0.0079,
+      "step": 12501
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.400999748712765e-05,
+      "loss": 0.126,
+      "step": 12502
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.4009108757441583e-05,
+      "loss": 0.0201,
+      "step": 12503
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.4008219978281418e-05,
+      "loss": 0.0001,
+      "step": 12504
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.4007331149652027e-05,
+      "loss": 0.1797,
+      "step": 12505
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.4006442271558305e-05,
+      "loss": 0.1514,
+      "step": 12506
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.4005553344005122e-05,
+      "loss": 0.0,
+      "step": 12507
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.4004664366997366e-05,
+      "loss": 0.0068,
+      "step": 12508
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.4003775340539914e-05,
+      "loss": 0.0,
+      "step": 12509
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.400288626463765e-05,
+      "loss": 0.2461,
+      "step": 12510
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.4001997139295456e-05,
+      "loss": 0.0084,
+      "step": 12511
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.4001107964518215e-05,
+      "loss": 0.0,
+      "step": 12512
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.4000218740310814e-05,
+      "loss": 0.0,
+      "step": 12513
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 2.3999329466678127e-05,
+      "loss": 0.0,
+      "step": 12514
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.3998440143625048e-05,
+      "loss": 0.1245,
+      "step": 12515
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.399755077115646e-05,
+      "loss": 0.0,
+      "step": 12516
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.3996661349277234e-05,
+      "loss": 0.0,
+      "step": 12517
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.3995771877992265e-05,
+      "loss": 0.0,
+      "step": 12518
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.3994882357306443e-05,
+      "loss": 0.0001,
+      "step": 12519
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.3993992787224638e-05,
+      "loss": 0.0,
+      "step": 12520
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.3993103167751745e-05,
+      "loss": 0.0083,
+      "step": 12521
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.3992213498892647e-05,
+      "loss": 0.0001,
+      "step": 12522
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.3991323780652234e-05,
+      "loss": 0.0001,
+      "step": 12523
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.3990434013035386e-05,
+      "loss": 0.0047,
+      "step": 12524
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.3989544196046987e-05,
+      "loss": 0.0001,
+      "step": 12525
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.3988654329691935e-05,
+      "loss": 0.0002,
+      "step": 12526
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.3987764413975104e-05,
+      "loss": 0.0,
+      "step": 12527
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.3986874448901388e-05,
+      "loss": 0.106,
+      "step": 12528
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.3985984434475674e-05,
+      "loss": 0.0002,
+      "step": 12529
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.398509437070285e-05,
+      "loss": 0.085,
+      "step": 12530
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.3984204257587795e-05,
+      "loss": 0.0806,
+      "step": 12531
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.398331409513541e-05,
+      "loss": 0.25,
+      "step": 12532
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.3982423883350578e-05,
+      "loss": 0.0,
+      "step": 12533
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.398153362223818e-05,
+      "loss": 0.2412,
+      "step": 12534
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.3980643311803117e-05,
+      "loss": 0.6211,
+      "step": 12535
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.397975295205028e-05,
+      "loss": 0.2969,
+      "step": 12536
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.397886254298454e-05,
+      "loss": 0.1611,
+      "step": 12537
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.397797208461081e-05,
+      "loss": 0.6328,
+      "step": 12538
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.397708157693396e-05,
+      "loss": 0.4551,
+      "step": 12539
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.3976191019958893e-05,
+      "loss": 0.3418,
+      "step": 12540
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.3975300413690496e-05,
+      "loss": 0.5625,
+      "step": 12541
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.3974409758133655e-05,
+      "loss": 0.0289,
+      "step": 12542
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.3973519053293272e-05,
+      "loss": 0.0002,
+      "step": 12543
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.3972628299174227e-05,
+      "loss": 0.8828,
+      "step": 12544
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.3971737495781416e-05,
+      "loss": 1.8203,
+      "step": 12545
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.397084664311973e-05,
+      "loss": 0.7422,
+      "step": 12546
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.3969955741194066e-05,
+      "loss": 1.7266,
+      "step": 12547
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.3969064790009312e-05,
+      "loss": 0.0002,
+      "step": 12548
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.396817378957036e-05,
+      "loss": 0.0,
+      "step": 12549
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.3967282739882108e-05,
+      "loss": 0.0003,
+      "step": 12550
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.3966391640949442e-05,
+      "loss": 0.8867,
+      "step": 12551
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.3965500492777266e-05,
+      "loss": 0.4785,
+      "step": 12552
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.396460929537046e-05,
+      "loss": 0.0001,
+      "step": 12553
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.3963718048733927e-05,
+      "loss": 0.0089,
+      "step": 12554
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.396282675287256e-05,
+      "loss": 0.0003,
+      "step": 12555
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.3961935407791254e-05,
+      "loss": 0.0229,
+      "step": 12556
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.3961044013494905e-05,
+      "loss": 0.2275,
+      "step": 12557
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.3960152569988403e-05,
+      "loss": 0.0243,
+      "step": 12558
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.3959261077276648e-05,
+      "loss": 0.0173,
+      "step": 12559
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.3958369535364534e-05,
+      "loss": 0.0001,
+      "step": 12560
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.3957477944256957e-05,
+      "loss": 0.0,
+      "step": 12561
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.3956586303958812e-05,
+      "loss": 0.1963,
+      "step": 12562
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.3955694614475e-05,
+      "loss": 0.0,
+      "step": 12563
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.3954802875810416e-05,
+      "loss": 0.0001,
+      "step": 12564
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.395391108796995e-05,
+      "loss": 0.0,
+      "step": 12565
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.395301925095851e-05,
+      "loss": 0.0001,
+      "step": 12566
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.3952127364780988e-05,
+      "loss": 0.0115,
+      "step": 12567
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.3951235429442283e-05,
+      "loss": 0.543,
+      "step": 12568
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.395034344494729e-05,
+      "loss": 0.0001,
+      "step": 12569
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.3949451411300914e-05,
+      "loss": 0.0001,
+      "step": 12570
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.3948559328508045e-05,
+      "loss": 0.0757,
+      "step": 12571
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.394766719657359e-05,
+      "loss": 0.0001,
+      "step": 12572
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.3946775015502442e-05,
+      "loss": 0.0001,
+      "step": 12573
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.394588278529951e-05,
+      "loss": 0.0001,
+      "step": 12574
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.394499050596968e-05,
+      "loss": 0.0,
+      "step": 12575
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.3944098177517863e-05,
+      "loss": 0.1226,
+      "step": 12576
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.3943205799948952e-05,
+      "loss": 0.0,
+      "step": 12577
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.3942313373267852e-05,
+      "loss": 0.0,
+      "step": 12578
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.3941420897479464e-05,
+      "loss": 0.1865,
+      "step": 12579
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.394052837258869e-05,
+      "loss": 0.1709,
+      "step": 12580
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.3939635798600423e-05,
+      "loss": 0.0835,
+      "step": 12581
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.3938743175519575e-05,
+      "loss": 0.1328,
+      "step": 12582
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.3937850503351047e-05,
+      "loss": 0.2285,
+      "step": 12583
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.3936957782099734e-05,
+      "loss": 0.3027,
+      "step": 12584
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.393606501177054e-05,
+      "loss": 0.5547,
+      "step": 12585
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.3935172192368373e-05,
+      "loss": 0.5195,
+      "step": 12586
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.393427932389813e-05,
+      "loss": 0.8359,
+      "step": 12587
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.3933386406364723e-05,
+      "loss": 0.2695,
+      "step": 12588
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.3932493439773044e-05,
+      "loss": 2.2344,
+      "step": 12589
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.3931600424128007e-05,
+      "loss": 0.0104,
+      "step": 12590
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.3930707359434508e-05,
+      "loss": 0.7539,
+      "step": 12591
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.3929814245697455e-05,
+      "loss": 0.4258,
+      "step": 12592
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.392892108292176e-05,
+      "loss": 0.0208,
+      "step": 12593
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.3928027871112308e-05,
+      "loss": 0.5391,
+      "step": 12594
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.3927134610274024e-05,
+      "loss": 1.3047,
+      "step": 12595
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.392624130041181e-05,
+      "loss": 0.0025,
+      "step": 12596
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.3925347941530556e-05,
+      "loss": 0.5742,
+      "step": 12597
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.3924454533635187e-05,
+      "loss": 0.0879,
+      "step": 12598
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.39235610767306e-05,
+      "loss": 0.0001,
+      "step": 12599
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.3922667570821707e-05,
+      "loss": 0.2734,
+      "step": 12600
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.3921774015913405e-05,
+      "loss": 0.0017,
+      "step": 12601
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.3920880412010614e-05,
+      "loss": 0.0011,
+      "step": 12602
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.3919986759118225e-05,
+      "loss": 0.5859,
+      "step": 12603
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.391909305724116e-05,
+      "loss": 0.0001,
+      "step": 12604
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.391819930638432e-05,
+      "loss": 0.0216,
+      "step": 12605
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.3917305506552617e-05,
+      "loss": 0.5742,
+      "step": 12606
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.3916411657750957e-05,
+      "loss": 0.0001,
+      "step": 12607
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.3915517759984247e-05,
+      "loss": 0.0106,
+      "step": 12608
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.39146238132574e-05,
+      "loss": 0.0091,
+      "step": 12609
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.3913729817575317e-05,
+      "loss": 0.0006,
+      "step": 12610
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.391283577294292e-05,
+      "loss": 0.7383,
+      "step": 12611
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.391194167936511e-05,
+      "loss": 0.0,
+      "step": 12612
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.3911047536846796e-05,
+      "loss": 0.1035,
+      "step": 12613
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.3910153345392895e-05,
+      "loss": 0.0,
+      "step": 12614
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.3909259105008314e-05,
+      "loss": 0.0033,
+      "step": 12615
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.3908364815697957e-05,
+      "loss": 0.0003,
+      "step": 12616
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.390747047746675e-05,
+      "loss": 0.0,
+      "step": 12617
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.3906576090319587e-05,
+      "loss": 0.0,
+      "step": 12618
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.3905681654261393e-05,
+      "loss": 0.2793,
+      "step": 12619
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.3904787169297075e-05,
+      "loss": 0.0,
+      "step": 12620
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.3903892635431545e-05,
+      "loss": 0.0082,
+      "step": 12621
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.3902998052669714e-05,
+      "loss": 0.1807,
+      "step": 12622
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.39021034210165e-05,
+      "loss": 0.0001,
+      "step": 12623
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.3901208740476808e-05,
+      "loss": 0.0,
+      "step": 12624
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.3900314011055556e-05,
+      "loss": 0.0101,
+      "step": 12625
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.3899419232757663e-05,
+      "loss": 0.0,
+      "step": 12626
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.3898524405588033e-05,
+      "loss": 0.0,
+      "step": 12627
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.389762952955158e-05,
+      "loss": 0.0,
+      "step": 12628
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.3896734604653227e-05,
+      "loss": 0.0,
+      "step": 12629
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.3895839630897884e-05,
+      "loss": 0.0008,
+      "step": 12630
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.389494460829046e-05,
+      "loss": 0.0001,
+      "step": 12631
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.389404953683588e-05,
+      "loss": 0.1455,
+      "step": 12632
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.389315441653905e-05,
+      "loss": 0.1934,
+      "step": 12633
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.3892259247404893e-05,
+      "loss": 1.1328,
+      "step": 12634
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.3891364029438323e-05,
+      "loss": 1.1562,
+      "step": 12635
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.3890468762644254e-05,
+      "loss": 0.2949,
+      "step": 12636
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.3889573447027606e-05,
+      "loss": 0.7422,
+      "step": 12637
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.3888678082593292e-05,
+      "loss": 1.5,
+      "step": 12638
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.3887782669346234e-05,
+      "loss": 0.0273,
+      "step": 12639
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.3886887207291342e-05,
+      "loss": 0.0347,
+      "step": 12640
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.3885991696433532e-05,
+      "loss": 0.4727,
+      "step": 12641
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.3885096136777734e-05,
+      "loss": 0.4883,
+      "step": 12642
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.3884200528328856e-05,
+      "loss": 0.4727,
+      "step": 12643
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.388330487109182e-05,
+      "loss": 0.0967,
+      "step": 12644
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.3882409165071548e-05,
+      "loss": 0.6641,
+      "step": 12645
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.3881513410272948e-05,
+      "loss": 0.2832,
+      "step": 12646
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.3880617606700948e-05,
+      "loss": 0.5625,
+      "step": 12647
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.3879721754360467e-05,
+      "loss": 0.0001,
+      "step": 12648
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.387882585325642e-05,
+      "loss": 0.001,
+      "step": 12649
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.387792990339373e-05,
+      "loss": 0.0006,
+      "step": 12650
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.3877033904777313e-05,
+      "loss": 0.0004,
+      "step": 12651
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.38761378574121e-05,
+      "loss": 0.0002,
+      "step": 12652
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.3875241761302998e-05,
+      "loss": 0.0064,
+      "step": 12653
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.3874345616454942e-05,
+      "loss": 0.0017,
+      "step": 12654
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.3873449422872842e-05,
+      "loss": 0.0041,
+      "step": 12655
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.3872553180561624e-05,
+      "loss": 0.4941,
+      "step": 12656
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.387165688952621e-05,
+      "loss": 0.0002,
+      "step": 12657
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.3870760549771524e-05,
+      "loss": 0.0,
+      "step": 12658
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.3869864161302478e-05,
+      "loss": 0.0,
+      "step": 12659
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.3868967724124012e-05,
+      "loss": 0.0,
+      "step": 12660
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.386807123824103e-05,
+      "loss": 0.5586,
+      "step": 12661
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.386717470365847e-05,
+      "loss": 0.0,
+      "step": 12662
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.386627812038125e-05,
+      "loss": 0.0054,
+      "step": 12663
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.386538148841429e-05,
+      "loss": 0.0525,
+      "step": 12664
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.386448480776252e-05,
+      "loss": 0.0,
+      "step": 12665
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.3863588078430857e-05,
+      "loss": 0.0,
+      "step": 12666
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.3862691300424233e-05,
+      "loss": 0.0,
+      "step": 12667
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.3861794473747566e-05,
+      "loss": 0.0352,
+      "step": 12668
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.3860897598405788e-05,
+      "loss": 0.0,
+      "step": 12669
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.3860000674403823e-05,
+      "loss": 0.0,
+      "step": 12670
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.3859103701746588e-05,
+      "loss": 0.1406,
+      "step": 12671
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.385820668043902e-05,
+      "loss": 0.0008,
+      "step": 12672
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.385730961048604e-05,
+      "loss": 0.0,
+      "step": 12673
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.385641249189257e-05,
+      "loss": 0.0,
+      "step": 12674
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.3855515324663548e-05,
+      "loss": 1.7109,
+      "step": 12675
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.3854618108803885e-05,
+      "loss": 0.0,
+      "step": 12676
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.385372084431852e-05,
+      "loss": 0.0767,
+      "step": 12677
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.3852823531212382e-05,
+      "loss": 0.0723,
+      "step": 12678
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.385192616949039e-05,
+      "loss": 0.0013,
+      "step": 12679
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.3851028759157476e-05,
+      "loss": 0.0,
+      "step": 12680
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.385013130021857e-05,
+      "loss": 0.0,
+      "step": 12681
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.3849233792678595e-05,
+      "loss": 0.0825,
+      "step": 12682
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.3848336236542487e-05,
+      "loss": 0.1309,
+      "step": 12683
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.384743863181517e-05,
+      "loss": 0.2051,
+      "step": 12684
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.3846540978501574e-05,
+      "loss": 0.5078,
+      "step": 12685
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.3845643276606627e-05,
+      "loss": 1.3047,
+      "step": 12686
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.3844745526135266e-05,
+      "loss": 0.1045,
+      "step": 12687
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.384384772709241e-05,
+      "loss": 0.4297,
+      "step": 12688
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.3842949879483e-05,
+      "loss": 0.25,
+      "step": 12689
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.3842051983311962e-05,
+      "loss": 0.1025,
+      "step": 12690
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.384115403858422e-05,
+      "loss": 0.3047,
+      "step": 12691
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.384025604530472e-05,
+      "loss": 1.4219,
+      "step": 12692
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.3839358003478377e-05,
+      "loss": 0.5117,
+      "step": 12693
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.3838459913110137e-05,
+      "loss": 1.1719,
+      "step": 12694
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.3837561774204923e-05,
+      "loss": 0.1543,
+      "step": 12695
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.3836663586767672e-05,
+      "loss": 0.3438,
+      "step": 12696
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.3835765350803314e-05,
+      "loss": 1.7266,
+      "step": 12697
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.383486706631678e-05,
+      "loss": 1.1719,
+      "step": 12698
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.3833968733313006e-05,
+      "loss": 0.0002,
+      "step": 12699
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.3833070351796925e-05,
+      "loss": 0.2324,
+      "step": 12700
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.3832171921773467e-05,
+      "loss": 0.1147,
+      "step": 12701
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.3831273443247574e-05,
+      "loss": 0.707,
+      "step": 12702
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.383037491622417e-05,
+      "loss": 0.0001,
+      "step": 12703
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.3829476340708195e-05,
+      "loss": 0.3223,
+      "step": 12704
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.3828577716704583e-05,
+      "loss": 0.1553,
+      "step": 12705
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.382767904421827e-05,
+      "loss": 0.0498,
+      "step": 12706
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.3826780323254185e-05,
+      "loss": 0.0,
+      "step": 12707
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.3825881553817273e-05,
+      "loss": 0.0129,
+      "step": 12708
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.382498273591246e-05,
+      "loss": 0.8789,
+      "step": 12709
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.382408386954469e-05,
+      "loss": 0.0,
+      "step": 12710
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.3823184954718894e-05,
+      "loss": 0.0024,
+      "step": 12711
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.3822285991440013e-05,
+      "loss": 0.2598,
+      "step": 12712
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.3821386979712975e-05,
+      "loss": 0.0,
+      "step": 12713
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.382048791954273e-05,
+      "loss": 0.165,
+      "step": 12714
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.3819588810934204e-05,
+      "loss": 0.007,
+      "step": 12715
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.3818689653892338e-05,
+      "loss": 1.1094,
+      "step": 12716
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.381779044842207e-05,
+      "loss": 0.0,
+      "step": 12717
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.3816891194528344e-05,
+      "loss": 0.0092,
+      "step": 12718
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.3815991892216086e-05,
+      "loss": 0.0,
+      "step": 12719
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.3815092541490247e-05,
+      "loss": 0.0,
+      "step": 12720
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.381419314235576e-05,
+      "loss": 0.3691,
+      "step": 12721
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.3813293694817558e-05,
+      "loss": 0.1973,
+      "step": 12722
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.3812394198880588e-05,
+      "loss": 0.0,
+      "step": 12723
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.381149465454979e-05,
+      "loss": 0.2832,
+      "step": 12724
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.3810595061830103e-05,
+      "loss": 0.0,
+      "step": 12725
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.380969542072647e-05,
+      "loss": 0.0,
+      "step": 12726
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.380879573124382e-05,
+      "loss": 0.0,
+      "step": 12727
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.3807895993387104e-05,
+      "loss": 0.1494,
+      "step": 12728
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.380699620716126e-05,
+      "loss": 0.0,
+      "step": 12729
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.3806096372571235e-05,
+      "loss": 0.1309,
+      "step": 12730
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.3805196489621957e-05,
+      "loss": 0.1099,
+      "step": 12731
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.380429655831838e-05,
+      "loss": 0.3359,
+      "step": 12732
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.3803396578665444e-05,
+      "loss": 0.2061,
+      "step": 12733
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.3802496550668086e-05,
+      "loss": 0.8438,
+      "step": 12734
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.3801596474331252e-05,
+      "loss": 0.0315,
+      "step": 12735
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.3800696349659886e-05,
+      "loss": 0.0026,
+      "step": 12736
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.379979617665893e-05,
+      "loss": 0.4238,
+      "step": 12737
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.3798895955333326e-05,
+      "loss": 0.1719,
+      "step": 12738
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.3797995685688022e-05,
+      "loss": 0.0002,
+      "step": 12739
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.3797095367727956e-05,
+      "loss": 0.4609,
+      "step": 12740
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.3796195001458073e-05,
+      "loss": 0.3027,
+      "step": 12741
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.3795294586883318e-05,
+      "loss": 1.1953,
+      "step": 12742
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.379439412400864e-05,
+      "loss": 0.6641,
+      "step": 12743
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.3793493612838982e-05,
+      "loss": 0.2197,
+      "step": 12744
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.3792593053379285e-05,
+      "loss": 0.0635,
+      "step": 12745
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.3791692445634502e-05,
+      "loss": 0.7539,
+      "step": 12746
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.3790791789609573e-05,
+      "loss": 0.0149,
+      "step": 12747
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.378989108530944e-05,
+      "loss": 0.0884,
+      "step": 12748
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.378899033273906e-05,
+      "loss": 0.0001,
+      "step": 12749
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.3788089531903372e-05,
+      "loss": 0.0131,
+      "step": 12750
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.3787188682807325e-05,
+      "loss": 0.0006,
+      "step": 12751
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.3786287785455865e-05,
+      "loss": 0.0001,
+      "step": 12752
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.3785386839853943e-05,
+      "loss": 0.4551,
+      "step": 12753
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.37844858460065e-05,
+      "loss": 0.0349,
+      "step": 12754
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.3783584803918493e-05,
+      "loss": 0.0,
+      "step": 12755
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.3782683713594862e-05,
+      "loss": 1.125,
+      "step": 12756
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.3781782575040555e-05,
+      "loss": 0.0012,
+      "step": 12757
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.378088138826053e-05,
+      "loss": 0.0981,
+      "step": 12758
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.3779980153259727e-05,
+      "loss": 0.0,
+      "step": 12759
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.3779078870043096e-05,
+      "loss": 0.8867,
+      "step": 12760
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.3778177538615592e-05,
+      "loss": 0.0728,
+      "step": 12761
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.3777276158982163e-05,
+      "loss": 0.1226,
+      "step": 12762
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.377637473114775e-05,
+      "loss": 0.0396,
+      "step": 12763
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.3775473255117314e-05,
+      "loss": 0.0,
+      "step": 12764
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.3774571730895808e-05,
+      "loss": 0.0,
+      "step": 12765
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.377367015848817e-05,
+      "loss": 0.0,
+      "step": 12766
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.377276853789936e-05,
+      "loss": 0.0547,
+      "step": 12767
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.3771866869134323e-05,
+      "loss": 0.3125,
+      "step": 12768
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.3770965152198018e-05,
+      "loss": 0.0,
+      "step": 12769
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.3770063387095392e-05,
+      "loss": 0.0334,
+      "step": 12770
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.3769161573831404e-05,
+      "loss": 0.0253,
+      "step": 12771
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.3768259712410998e-05,
+      "loss": 0.0002,
+      "step": 12772
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.3767357802839126e-05,
+      "loss": 0.0,
+      "step": 12773
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.376645584512075e-05,
+      "loss": 0.0,
+      "step": 12774
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.3765553839260813e-05,
+      "loss": 0.0,
+      "step": 12775
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.3764651785264275e-05,
+      "loss": 0.0,
+      "step": 12776
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.376374968313609e-05,
+      "loss": 0.0,
+      "step": 12777
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.3762847532881205e-05,
+      "loss": 0.0,
+      "step": 12778
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.376194533450458e-05,
+      "loss": 0.0,
+      "step": 12779
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.3761043088011175e-05,
+      "loss": 0.0024,
+      "step": 12780
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.3760140793405932e-05,
+      "loss": 0.1235,
+      "step": 12781
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.3759238450693812e-05,
+      "loss": 0.1729,
+      "step": 12782
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.3758336059879772e-05,
+      "loss": 0.1738,
+      "step": 12783
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.375743362096877e-05,
+      "loss": 0.3418,
+      "step": 12784
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.3756531133965752e-05,
+      "loss": 1.5391,
+      "step": 12785
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.3755628598875682e-05,
+      "loss": 0.9766,
+      "step": 12786
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.3754726015703512e-05,
+      "loss": 1.5938,
+      "step": 12787
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.37538233844542e-05,
+      "loss": 0.5195,
+      "step": 12788
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.375292070513271e-05,
+      "loss": 0.0933,
+      "step": 12789
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.375201797774399e-05,
+      "loss": 0.6094,
+      "step": 12790
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.3751115202292994e-05,
+      "loss": 0.3047,
+      "step": 12791
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.3750212378784695e-05,
+      "loss": 0.0038,
+      "step": 12792
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.374930950722403e-05,
+      "loss": 0.291,
+      "step": 12793
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.3748406587615982e-05,
+      "loss": 0.418,
+      "step": 12794
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.3747503619965492e-05,
+      "loss": 0.9844,
+      "step": 12795
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.3746600604277523e-05,
+      "loss": 0.6484,
+      "step": 12796
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.374569754055703e-05,
+      "loss": 0.2695,
+      "step": 12797
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.374479442880898e-05,
+      "loss": 0.4238,
+      "step": 12798
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.374389126903833e-05,
+      "loss": 1.1953,
+      "step": 12799
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.3742988061250037e-05,
+      "loss": 0.4844,
+      "step": 12800
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.374208480544906e-05,
+      "loss": 0.0023,
+      "step": 12801
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.3741181501640362e-05,
+      "loss": 0.0001,
+      "step": 12802
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.3740278149828907e-05,
+      "loss": 0.0315,
+      "step": 12803
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.3739374750019645e-05,
+      "loss": 0.0001,
+      "step": 12804
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.3738471302217555e-05,
+      "loss": 0.4492,
+      "step": 12805
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.373756780642758e-05,
+      "loss": 0.1426,
+      "step": 12806
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.3736664262654692e-05,
+      "loss": 0.0,
+      "step": 12807
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.373576067090385e-05,
+      "loss": 0.0,
+      "step": 12808
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.3734857031180016e-05,
+      "loss": 0.0001,
+      "step": 12809
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.3733953343488152e-05,
+      "loss": 0.0249,
+      "step": 12810
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.373304960783322e-05,
+      "loss": 0.0,
+      "step": 12811
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.3732145824220184e-05,
+      "loss": 0.0,
+      "step": 12812
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.3731241992654014e-05,
+      "loss": 0.4219,
+      "step": 12813
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.3730338113139657e-05,
+      "loss": 0.7891,
+      "step": 12814
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.3729434185682094e-05,
+      "loss": 0.2158,
+      "step": 12815
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.3728530210286277e-05,
+      "loss": 0.0,
+      "step": 12816
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.3727626186957177e-05,
+      "loss": 0.0378,
+      "step": 12817
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.3726722115699755e-05,
+      "loss": 0.1436,
+      "step": 12818
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.372581799651898e-05,
+      "loss": 0.0,
+      "step": 12819
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.3724913829419808e-05,
+      "loss": 0.0,
+      "step": 12820
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.3724009614407217e-05,
+      "loss": 0.0,
+      "step": 12821
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.372310535148616e-05,
+      "loss": 0.0001,
+      "step": 12822
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.3722201040661614e-05,
+      "loss": 0.0,
+      "step": 12823
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.3721296681938537e-05,
+      "loss": 0.0,
+      "step": 12824
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.37203922753219e-05,
+      "loss": 0.0,
+      "step": 12825
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.3719487820816664e-05,
+      "loss": 0.0001,
+      "step": 12826
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.3718583318427803e-05,
+      "loss": 0.123,
+      "step": 12827
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.371767876816028e-05,
+      "loss": 0.0986,
+      "step": 12828
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.3716774170019064e-05,
+      "loss": 0.9414,
+      "step": 12829
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.3715869524009123e-05,
+      "loss": 0.1182,
+      "step": 12830
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.371496483013542e-05,
+      "loss": 0.1553,
+      "step": 12831
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.3714060088402926e-05,
+      "loss": 0.0884,
+      "step": 12832
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.3713155298816614e-05,
+      "loss": 0.1221,
+      "step": 12833
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.3712250461381448e-05,
+      "loss": 0.0101,
+      "step": 12834
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.3711345576102396e-05,
+      "loss": 0.3574,
+      "step": 12835
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.3710440642984427e-05,
+      "loss": 0.2451,
+      "step": 12836
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.370953566203252e-05,
+      "loss": 0.2256,
+      "step": 12837
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.3708630633251637e-05,
+      "loss": 0.1641,
+      "step": 12838
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.3707725556646744e-05,
+      "loss": 0.0007,
+      "step": 12839
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.370682043222282e-05,
+      "loss": 0.1738,
+      "step": 12840
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.3705915259984827e-05,
+      "loss": 0.0791,
+      "step": 12841
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.3705010039937744e-05,
+      "loss": 1.1875,
+      "step": 12842
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.3704104772086533e-05,
+      "loss": 0.0052,
+      "step": 12843
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.3703199456436176e-05,
+      "loss": 0.5234,
+      "step": 12844
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.3702294092991633e-05,
+      "loss": 0.4727,
+      "step": 12845
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.370138868175789e-05,
+      "loss": 0.0,
+      "step": 12846
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.3700483222739912e-05,
+      "loss": 0.8828,
+      "step": 12847
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.369957771594266e-05,
+      "loss": 0.957,
+      "step": 12848
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.3698672161371124e-05,
+      "loss": 0.2344,
+      "step": 12849
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.3697766559030272e-05,
+      "loss": 0.0654,
+      "step": 12850
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.369686090892507e-05,
+      "loss": 0.0337,
+      "step": 12851
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.3695955211060497e-05,
+      "loss": 0.0005,
+      "step": 12852
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.369504946544153e-05,
+      "loss": 0.0003,
+      "step": 12853
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.369414367207314e-05,
+      "loss": 0.054,
+      "step": 12854
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.36932378309603e-05,
+      "loss": 0.0002,
+      "step": 12855
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.3692331942107982e-05,
+      "loss": 0.0009,
+      "step": 12856
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.369142600552117e-05,
+      "loss": 0.001,
+      "step": 12857
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.3690520021204827e-05,
+      "loss": 0.0,
+      "step": 12858
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.3689613989163934e-05,
+      "loss": 0.008,
+      "step": 12859
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.368870790940347e-05,
+      "loss": 0.0228,
+      "step": 12860
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.3687801781928406e-05,
+      "loss": 0.3926,
+      "step": 12861
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.3686895606743722e-05,
+      "loss": 0.0,
+      "step": 12862
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.3685989383854386e-05,
+      "loss": 0.0,
+      "step": 12863
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.3685083113265386e-05,
+      "loss": 0.0001,
+      "step": 12864
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.3684176794981692e-05,
+      "loss": 0.0082,
+      "step": 12865
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.368327042900828e-05,
+      "loss": 0.0532,
+      "step": 12866
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.3682364015350135e-05,
+      "loss": 1.0156,
+      "step": 12867
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.3681457554012222e-05,
+      "loss": 0.0001,
+      "step": 12868
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.368055104499953e-05,
+      "loss": 0.0001,
+      "step": 12869
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.3679644488317035e-05,
+      "loss": 0.0,
+      "step": 12870
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.367873788396971e-05,
+      "loss": 0.0815,
+      "step": 12871
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.3677831231962538e-05,
+      "loss": 0.0001,
+      "step": 12872
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.36769245323005e-05,
+      "loss": 0.0001,
+      "step": 12873
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.367601778498857e-05,
+      "loss": 0.0,
+      "step": 12874
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.3675110990031737e-05,
+      "loss": 0.2344,
+      "step": 12875
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.3674204147434967e-05,
+      "loss": 0.0,
+      "step": 12876
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.367329725720325e-05,
+      "loss": 0.0001,
+      "step": 12877
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.3672390319341558e-05,
+      "loss": 0.0811,
+      "step": 12878
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.3671483333854884e-05,
+      "loss": 0.0845,
+      "step": 12879
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.3670576300748194e-05,
+      "loss": 0.1357,
+      "step": 12880
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.366966922002648e-05,
+      "loss": 0.1367,
+      "step": 12881
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.3668762091694724e-05,
+      "loss": 0.1001,
+      "step": 12882
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.3667854915757895e-05,
+      "loss": 0.1318,
+      "step": 12883
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.3666947692220987e-05,
+      "loss": 1.5859,
+      "step": 12884
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.3666040421088982e-05,
+      "loss": 0.7578,
+      "step": 12885
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.3665133102366847e-05,
+      "loss": 0.4844,
+      "step": 12886
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.3664225736059588e-05,
+      "loss": 0.7812,
+      "step": 12887
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.366331832217217e-05,
+      "loss": 0.2637,
+      "step": 12888
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.3662410860709585e-05,
+      "loss": 0.3262,
+      "step": 12889
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.3661503351676813e-05,
+      "loss": 0.2119,
+      "step": 12890
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.366059579507883e-05,
+      "loss": 0.3457,
+      "step": 12891
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.3659688190920637e-05,
+      "loss": 0.4043,
+      "step": 12892
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.3658780539207208e-05,
+      "loss": 0.0522,
+      "step": 12893
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.3657872839943528e-05,
+      "loss": 0.0525,
+      "step": 12894
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.365696509313458e-05,
+      "loss": 0.6602,
+      "step": 12895
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.3656057298785354e-05,
+      "loss": 0.1357,
+      "step": 12896
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.3655149456900828e-05,
+      "loss": 0.0005,
+      "step": 12897
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.3654241567485996e-05,
+      "loss": 0.25,
+      "step": 12898
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.365333363054584e-05,
+      "loss": 0.2334,
+      "step": 12899
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.3652425646085344e-05,
+      "loss": 0.3008,
+      "step": 12900
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.3651517614109494e-05,
+      "loss": 0.2559,
+      "step": 12901
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.365060953462328e-05,
+      "loss": 0.0001,
+      "step": 12902
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.364970140763169e-05,
+      "loss": 0.1523,
+      "step": 12903
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.3648793233139707e-05,
+      "loss": 0.0006,
+      "step": 12904
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.3647885011152317e-05,
+      "loss": 0.0,
+      "step": 12905
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.364697674167451e-05,
+      "loss": 0.1699,
+      "step": 12906
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.3646068424711278e-05,
+      "loss": 0.0,
+      "step": 12907
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.36451600602676e-05,
+      "loss": 0.0219,
+      "step": 12908
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.3644251648348473e-05,
+      "loss": 0.0029,
+      "step": 12909
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.364334318895888e-05,
+      "loss": 0.0,
+      "step": 12910
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.364243468210381e-05,
+      "loss": 0.0001,
+      "step": 12911
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.3641526127788257e-05,
+      "loss": 0.0474,
+      "step": 12912
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.3640617526017206e-05,
+      "loss": 0.0,
+      "step": 12913
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.3639708876795647e-05,
+      "loss": 0.0001,
+      "step": 12914
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.363880018012857e-05,
+      "loss": 0.1875,
+      "step": 12915
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.363789143602097e-05,
+      "loss": 0.0001,
+      "step": 12916
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.3636982644477833e-05,
+      "loss": 0.0001,
+      "step": 12917
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.3636073805504147e-05,
+      "loss": 0.0,
+      "step": 12918
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.363516491910491e-05,
+      "loss": 0.0,
+      "step": 12919
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.3634255985285104e-05,
+      "loss": 0.0001,
+      "step": 12920
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.363334700404973e-05,
+      "loss": 0.0,
+      "step": 12921
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.3632437975403774e-05,
+      "loss": 0.4863,
+      "step": 12922
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.3631528899352227e-05,
+      "loss": 0.1309,
+      "step": 12923
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.3630619775900085e-05,
+      "loss": 0.1289,
+      "step": 12924
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.362971060505234e-05,
+      "loss": 0.0,
+      "step": 12925
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.3628801386813986e-05,
+      "loss": 0.1289,
+      "step": 12926
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.362789212119001e-05,
+      "loss": 0.0,
+      "step": 12927
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.362698280818541e-05,
+      "loss": 0.1475,
+      "step": 12928
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.362607344780518e-05,
+      "loss": 0.1216,
+      "step": 12929
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.3625164040054318e-05,
+      "loss": 0.1201,
+      "step": 12930
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.3624254584937803e-05,
+      "loss": 0.2031,
+      "step": 12931
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.3623345082460644e-05,
+      "loss": 0.2207,
+      "step": 12932
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.362243553262783e-05,
+      "loss": 0.25,
+      "step": 12933
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.3621525935444358e-05,
+      "loss": 1.5234,
+      "step": 12934
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.362061629091522e-05,
+      "loss": 0.3633,
+      "step": 12935
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.361970659904541e-05,
+      "loss": 0.6562,
+      "step": 12936
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.361879685983993e-05,
+      "loss": 0.0205,
+      "step": 12937
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.3617887073303777e-05,
+      "loss": 0.7227,
+      "step": 12938
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.3616977239441935e-05,
+      "loss": 0.0145,
+      "step": 12939
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.361606735825941e-05,
+      "loss": 0.0086,
+      "step": 12940
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.36151574297612e-05,
+      "loss": 0.2119,
+      "step": 12941
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.3614247453952296e-05,
+      "loss": 0.7539,
+      "step": 12942
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.3613337430837695e-05,
+      "loss": 0.0093,
+      "step": 12943
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.3612427360422404e-05,
+      "loss": 0.4062,
+      "step": 12944
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.3611517242711415e-05,
+      "loss": 0.416,
+      "step": 12945
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.3610607077709713e-05,
+      "loss": 0.2129,
+      "step": 12946
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.3609696865422318e-05,
+      "loss": 0.0005,
+      "step": 12947
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.360878660585422e-05,
+      "loss": 0.9219,
+      "step": 12948
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.3607876299010405e-05,
+      "loss": 0.4746,
+      "step": 12949
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.3606965944895895e-05,
+      "loss": 0.25,
+      "step": 12950
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.360605554351568e-05,
+      "loss": 0.0098,
+      "step": 12951
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.3605145094874746e-05,
+      "loss": 0.0001,
+      "step": 12952
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.360423459897811e-05,
+      "loss": 1.2734,
+      "step": 12953
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.3603324055830768e-05,
+      "loss": 1.75,
+      "step": 12954
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.3602413465437715e-05,
+      "loss": 0.0732,
+      "step": 12955
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.3601502827803954e-05,
+      "loss": 3.2656,
+      "step": 12956
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.360059214293449e-05,
+      "loss": 0.0,
+      "step": 12957
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.359968141083432e-05,
+      "loss": 0.0001,
+      "step": 12958
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.3598770631508443e-05,
+      "loss": 0.2969,
+      "step": 12959
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.359785980496187e-05,
+      "loss": 0.0001,
+      "step": 12960
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.359694893119959e-05,
+      "loss": 0.0,
+      "step": 12961
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.3596038010226615e-05,
+      "loss": 0.0,
+      "step": 12962
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.3595127042047946e-05,
+      "loss": 0.0,
+      "step": 12963
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.359421602666858e-05,
+      "loss": 0.0,
+      "step": 12964
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.3593304964093525e-05,
+      "loss": 0.7383,
+      "step": 12965
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.3592393854327784e-05,
+      "loss": 0.1055,
+      "step": 12966
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.3591482697376364e-05,
+      "loss": 0.4883,
+      "step": 12967
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.3590571493244254e-05,
+      "loss": 0.0001,
+      "step": 12968
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.3589660241936475e-05,
+      "loss": 0.0,
+      "step": 12969
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.3588748943458017e-05,
+      "loss": 0.0,
+      "step": 12970
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.35878375978139e-05,
+      "loss": 0.0027,
+      "step": 12971
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.3586926205009116e-05,
+      "loss": 0.0005,
+      "step": 12972
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.3586014765048675e-05,
+      "loss": 0.1069,
+      "step": 12973
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.358510327793758e-05,
+      "loss": 0.0,
+      "step": 12974
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.3584191743680842e-05,
+      "loss": 0.0,
+      "step": 12975
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.358328016228346e-05,
+      "loss": 0.0,
+      "step": 12976
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.3582368533750445e-05,
+      "loss": 0.0923,
+      "step": 12977
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.3581456858086796e-05,
+      "loss": 0.0,
+      "step": 12978
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.3580545135297527e-05,
+      "loss": 0.1504,
+      "step": 12979
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.3579633365387646e-05,
+      "loss": 0.1533,
+      "step": 12980
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.3578721548362146e-05,
+      "loss": 0.1357,
+      "step": 12981
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.3577809684226053e-05,
+      "loss": 0.1348,
+      "step": 12982
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.3576897772984368e-05,
+      "loss": 0.1504,
+      "step": 12983
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.3575985814642094e-05,
+      "loss": 0.0972,
+      "step": 12984
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.3575073809204243e-05,
+      "loss": 0.7656,
+      "step": 12985
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.3574161756675818e-05,
+      "loss": 0.2637,
+      "step": 12986
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.3573249657061835e-05,
+      "loss": 0.3125,
+      "step": 12987
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.3572337510367304e-05,
+      "loss": 1.875,
+      "step": 12988
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 2.3571425316597224e-05,
+      "loss": 0.1138,
+      "step": 12989
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.357051307575661e-05,
+      "loss": 0.3516,
+      "step": 12990
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.356960078785048e-05,
+      "loss": 0.291,
+      "step": 12991
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.356868845288383e-05,
+      "loss": 0.2734,
+      "step": 12992
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.3567776070861674e-05,
+      "loss": 0.5508,
+      "step": 12993
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.3566863641789026e-05,
+      "loss": 0.9805,
+      "step": 12994
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.35659511656709e-05,
+      "loss": 0.5977,
+      "step": 12995
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.3565038642512298e-05,
+      "loss": 1.8984,
+      "step": 12996
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.3564126072318234e-05,
+      "loss": 0.2852,
+      "step": 12997
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.3563213455093724e-05,
+      "loss": 0.0281,
+      "step": 12998
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.3562300790843773e-05,
+      "loss": 0.0693,
+      "step": 12999
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.3561388079573394e-05,
+      "loss": 0.0199,
+      "step": 13000
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.3560475321287607e-05,
+      "loss": 0.0001,
+      "step": 13001
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.3559562515991418e-05,
+      "loss": 0.75,
+      "step": 13002
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.3558649663689843e-05,
+      "loss": 0.0001,
+      "step": 13003
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.355773676438789e-05,
+      "loss": 0.0757,
+      "step": 13004
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.3556823818090574e-05,
+      "loss": 0.0007,
+      "step": 13005
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.3555910824802903e-05,
+      "loss": 0.0,
+      "step": 13006
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.355499778452991e-05,
+      "loss": 0.0,
+      "step": 13007
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.355408469727659e-05,
+      "loss": 0.1147,
+      "step": 13008
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.355317156304796e-05,
+      "loss": 0.0004,
+      "step": 13009
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.3552258381849046e-05,
+      "loss": 0.0449,
+      "step": 13010
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.355134515368485e-05,
+      "loss": 0.3789,
+      "step": 13011
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.3550431878560393e-05,
+      "loss": 0.0,
+      "step": 13012
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.3549518556480688e-05,
+      "loss": 0.1147,
+      "step": 13013
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.354860518745075e-05,
+      "loss": 0.0,
+      "step": 13014
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.3547691771475603e-05,
+      "loss": 0.457,
+      "step": 13015
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.354677830856025e-05,
+      "loss": 0.0,
+      "step": 13016
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.354586479870972e-05,
+      "loss": 0.1484,
+      "step": 13017
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.3544951241929017e-05,
+      "loss": 0.0001,
+      "step": 13018
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.354403763822317e-05,
+      "loss": 0.0,
+      "step": 13019
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.354312398759719e-05,
+      "loss": 0.0649,
+      "step": 13020
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.3542210290056093e-05,
+      "loss": 0.0,
+      "step": 13021
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.35412965456049e-05,
+      "loss": 0.0,
+      "step": 13022
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.3540382754248623e-05,
+      "loss": 0.0,
+      "step": 13023
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.3539468915992288e-05,
+      "loss": 0.1162,
+      "step": 13024
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.3538555030840905e-05,
+      "loss": 0.0,
+      "step": 13025
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.3537641098799507e-05,
+      "loss": 0.0,
+      "step": 13026
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.35367271198731e-05,
+      "loss": 0.1172,
+      "step": 13027
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.3535813094066707e-05,
+      "loss": 0.1089,
+      "step": 13028
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.3534899021385344e-05,
+      "loss": 0.1426,
+      "step": 13029
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.3533984901834036e-05,
+      "loss": 0.1226,
+      "step": 13030
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.3533070735417804e-05,
+      "loss": 0.1426,
+      "step": 13031
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.353215652214166e-05,
+      "loss": 0.1436,
+      "step": 13032
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.3531242262010632e-05,
+      "loss": 0.0,
+      "step": 13033
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.3530327955029737e-05,
+      "loss": 0.1138,
+      "step": 13034
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.3529413601204e-05,
+      "loss": 0.582,
+      "step": 13035
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.3528499200538432e-05,
+      "loss": 0.4727,
+      "step": 13036
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.352758475303807e-05,
+      "loss": 0.6992,
+      "step": 13037
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.3526670258707922e-05,
+      "loss": 1.4766,
+      "step": 13038
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.3525755717553022e-05,
+      "loss": 0.625,
+      "step": 13039
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.3524841129578383e-05,
+      "loss": 0.1299,
+      "step": 13040
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.352392649478903e-05,
+      "loss": 0.3125,
+      "step": 13041
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.3523011813189988e-05,
+      "loss": 0.0059,
+      "step": 13042
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.352209708478627e-05,
+      "loss": 0.6523,
+      "step": 13043
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.3521182309582917e-05,
+      "loss": 0.0291,
+      "step": 13044
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.352026748758494e-05,
+      "loss": 1.0156,
+      "step": 13045
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.3519352618797364e-05,
+      "loss": 0.0334,
+      "step": 13046
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.3518437703225218e-05,
+      "loss": 0.1152,
+      "step": 13047
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.3517522740873522e-05,
+      "loss": 0.0265,
+      "step": 13048
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.35166077317473e-05,
+      "loss": 0.0071,
+      "step": 13049
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.351569267585158e-05,
+      "loss": 0.011,
+      "step": 13050
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.3514777573191388e-05,
+      "loss": 0.918,
+      "step": 13051
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.3513862423771747e-05,
+      "loss": 0.0001,
+      "step": 13052
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.3512947227597677e-05,
+      "loss": 0.7539,
+      "step": 13053
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.3512031984674215e-05,
+      "loss": 0.0393,
+      "step": 13054
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.351111669500638e-05,
+      "loss": 0.003,
+      "step": 13055
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.3510201358599198e-05,
+      "loss": 0.0002,
+      "step": 13056
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 2.3509285975457705e-05,
+      "loss": 0.1621,
+      "step": 13057
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.3508370545586913e-05,
+      "loss": 0.0186,
+      "step": 13058
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.350745506899186e-05,
+      "loss": 0.0076,
+      "step": 13059
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.350653954567757e-05,
+      "loss": 0.0962,
+      "step": 13060
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.3505623975649066e-05,
+      "loss": 0.6133,
+      "step": 13061
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.3504708358911388e-05,
+      "loss": 0.1543,
+      "step": 13062
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.3503792695469552e-05,
+      "loss": 0.05,
+      "step": 13063
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.3502876985328595e-05,
+      "loss": 0.0403,
+      "step": 13064
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.3501961228493537e-05,
+      "loss": 0.0,
+      "step": 13065
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.3501045424969417e-05,
+      "loss": 0.0,
+      "step": 13066
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.350012957476126e-05,
+      "loss": 0.0,
+      "step": 13067
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.3499213677874084e-05,
+      "loss": 0.0001,
+      "step": 13068
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.349829773431294e-05,
+      "loss": 0.0806,
+      "step": 13069
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.3497381744082843e-05,
+      "loss": 0.5703,
+      "step": 13070
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.349646570718883e-05,
+      "loss": 0.0,
+      "step": 13071
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.3495549623635925e-05,
+      "loss": 0.0001,
+      "step": 13072
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.3494633493429164e-05,
+      "loss": 0.0,
+      "step": 13073
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.3493717316573578e-05,
+      "loss": 0.0,
+      "step": 13074
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.3492801093074195e-05,
+      "loss": 0.3652,
+      "step": 13075
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.3491884822936053e-05,
+      "loss": 0.0,
+      "step": 13076
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.3490968506164174e-05,
+      "loss": 0.0,
+      "step": 13077
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.3490052142763594e-05,
+      "loss": 0.0,
+      "step": 13078
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.348913573273935e-05,
+      "loss": 0.0,
+      "step": 13079
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.3488219276096473e-05,
+      "loss": 0.7617,
+      "step": 13080
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.3487302772839986e-05,
+      "loss": 0.1318,
+      "step": 13081
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.3486386222974933e-05,
+      "loss": 0.168,
+      "step": 13082
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.348546962650635e-05,
+      "loss": 0.1816,
+      "step": 13083
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.348455298343926e-05,
+      "loss": 1.5859,
+      "step": 13084
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.34836362937787e-05,
+      "loss": 1.125,
+      "step": 13085
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.3482719557529707e-05,
+      "loss": 0.0259,
+      "step": 13086
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.3481802774697314e-05,
+      "loss": 0.8555,
+      "step": 13087
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.3480885945286553e-05,
+      "loss": 0.3086,
+      "step": 13088
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.347996906930246e-05,
+      "loss": 0.4648,
+      "step": 13089
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.3479052146750077e-05,
+      "loss": 0.0007,
+      "step": 13090
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.347813517763443e-05,
+      "loss": 0.6953,
+      "step": 13091
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.3477218161960562e-05,
+      "loss": 0.3809,
+      "step": 13092
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.34763010997335e-05,
+      "loss": 0.2891,
+      "step": 13093
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.347538399095829e-05,
+      "loss": 0.0171,
+      "step": 13094
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.347446683563996e-05,
+      "loss": 0.0679,
+      "step": 13095
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.347354963378355e-05,
+      "loss": 0.3828,
+      "step": 13096
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.3472632385394096e-05,
+      "loss": 0.2539,
+      "step": 13097
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.347171509047664e-05,
+      "loss": 0.4297,
+      "step": 13098
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.3470797749036213e-05,
+      "loss": 0.6016,
+      "step": 13099
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.346988036107786e-05,
+      "loss": 0.5078,
+      "step": 13100
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.3468962926606605e-05,
+      "loss": 0.7227,
+      "step": 13101
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.3468045445627502e-05,
+      "loss": 0.0039,
+      "step": 13102
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.3467127918145576e-05,
+      "loss": 0.0003,
+      "step": 13103
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.3466210344165878e-05,
+      "loss": 0.0001,
+      "step": 13104
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.346529272369344e-05,
+      "loss": 0.0006,
+      "step": 13105
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.3464375056733305e-05,
+      "loss": 0.0019,
+      "step": 13106
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.3463457343290503e-05,
+      "loss": 0.1699,
+      "step": 13107
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.3462539583370088e-05,
+      "loss": 0.0571,
+      "step": 13108
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.3461621776977088e-05,
+      "loss": 0.0075,
+      "step": 13109
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.3460703924116546e-05,
+      "loss": 0.0037,
+      "step": 13110
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.345978602479351e-05,
+      "loss": 0.0003,
+      "step": 13111
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.3458868079013008e-05,
+      "loss": 0.0012,
+      "step": 13112
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.3457950086780092e-05,
+      "loss": 0.0,
+      "step": 13113
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.3457032048099796e-05,
+      "loss": 0.0598,
+      "step": 13114
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.3456113962977168e-05,
+      "loss": 0.0002,
+      "step": 13115
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.3455195831417243e-05,
+      "loss": 0.0001,
+      "step": 13116
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.345427765342507e-05,
+      "loss": 0.0001,
+      "step": 13117
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.345335942900568e-05,
+      "loss": 0.0001,
+      "step": 13118
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.345244115816413e-05,
+      "loss": 0.0,
+      "step": 13119
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.345152284090545e-05,
+      "loss": 0.1064,
+      "step": 13120
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.345060447723469e-05,
+      "loss": 0.0,
+      "step": 13121
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.3449686067156894e-05,
+      "loss": 0.0008,
+      "step": 13122
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.3448767610677105e-05,
+      "loss": 0.0001,
+      "step": 13123
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.3447849107800362e-05,
+      "loss": 0.0,
+      "step": 13124
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 2.3446930558531712e-05,
+      "loss": 0.0,
+      "step": 13125
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.3446011962876198e-05,
+      "loss": 0.1543,
+      "step": 13126
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.3445093320838866e-05,
+      "loss": 0.0,
+      "step": 13127
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.3444174632424764e-05,
+      "loss": 0.106,
+      "step": 13128
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.344325589763893e-05,
+      "loss": 0.1133,
+      "step": 13129
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.3442337116486416e-05,
+      "loss": 0.0001,
+      "step": 13130
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.3441418288972268e-05,
+      "loss": 0.1465,
+      "step": 13131
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.3440499415101517e-05,
+      "loss": 0.1084,
+      "step": 13132
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.343958049487923e-05,
+      "loss": 0.165,
+      "step": 13133
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.343866152831044e-05,
+      "loss": 0.8828,
+      "step": 13134
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.34377425154002e-05,
+      "loss": 0.0552,
+      "step": 13135
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.3436823456153553e-05,
+      "loss": 1.0469,
+      "step": 13136
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.3435904350575547e-05,
+      "loss": 1.0078,
+      "step": 13137
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.343498519867123e-05,
+      "loss": 0.0386,
+      "step": 13138
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.3434066000445648e-05,
+      "loss": 0.4707,
+      "step": 13139
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.3433146755903847e-05,
+      "loss": 0.1157,
+      "step": 13140
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.343222746505088e-05,
+      "loss": 0.3926,
+      "step": 13141
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.3431308127891796e-05,
+      "loss": 0.6875,
+      "step": 13142
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.3430388744431635e-05,
+      "loss": 0.084,
+      "step": 13143
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.342946931467546e-05,
+      "loss": 0.1885,
+      "step": 13144
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.3428549838628305e-05,
+      "loss": 0.0913,
+      "step": 13145
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.342763031629523e-05,
+      "loss": 0.6758,
+      "step": 13146
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.3426710747681283e-05,
+      "loss": 0.0366,
+      "step": 13147
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.3425791132791507e-05,
+      "loss": 0.0334,
+      "step": 13148
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.3424871471630958e-05,
+      "loss": 0.0153,
+      "step": 13149
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.3423951764204686e-05,
+      "loss": 0.3906,
+      "step": 13150
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.3423032010517744e-05,
+      "loss": 0.0947,
+      "step": 13151
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.3422112210575172e-05,
+      "loss": 0.3887,
+      "step": 13152
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.3421192364382037e-05,
+      "loss": 0.0012,
+      "step": 13153
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.3420272471943373e-05,
+      "loss": 0.3008,
+      "step": 13154
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.341935253326425e-05,
+      "loss": 0.0001,
+      "step": 13155
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.3418432548349706e-05,
+      "loss": 0.051,
+      "step": 13156
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.34175125172048e-05,
+      "loss": 0.001,
+      "step": 13157
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.3416592439834584e-05,
+      "loss": 0.0001,
+      "step": 13158
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.3415672316244105e-05,
+      "loss": 1.4922,
+      "step": 13159
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.341475214643842e-05,
+      "loss": 0.0107,
+      "step": 13160
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.3413831930422586e-05,
+      "loss": 0.0015,
+      "step": 13161
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.341291166820165e-05,
+      "loss": 0.0337,
+      "step": 13162
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.3411991359780667e-05,
+      "loss": 0.0,
+      "step": 13163
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.3411071005164696e-05,
+      "loss": 0.0005,
+      "step": 13164
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.341015060435879e-05,
+      "loss": 0.0206,
+      "step": 13165
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.3409230157367993e-05,
+      "loss": 0.0,
+      "step": 13166
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.340830966419737e-05,
+      "loss": 0.0089,
+      "step": 13167
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.3407389124851974e-05,
+      "loss": 0.0,
+      "step": 13168
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.340646853933686e-05,
+      "loss": 0.1826,
+      "step": 13169
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.3405547907657084e-05,
+      "loss": 0.083,
+      "step": 13170
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.3404627229817698e-05,
+      "loss": 0.0244,
+      "step": 13171
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.3403706505823767e-05,
+      "loss": 0.0,
+      "step": 13172
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.3402785735680334e-05,
+      "loss": 0.0002,
+      "step": 13173
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.3401864919392467e-05,
+      "loss": 0.0,
+      "step": 13174
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.3400944056965217e-05,
+      "loss": 0.0,
+      "step": 13175
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.3400023148403638e-05,
+      "loss": 0.0001,
+      "step": 13176
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.3399102193712796e-05,
+      "loss": 0.0001,
+      "step": 13177
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.3398181192897744e-05,
+      "loss": 0.1025,
+      "step": 13178
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.339726014596354e-05,
+      "loss": 0.1064,
+      "step": 13179
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.3396339052915237e-05,
+      "loss": 0.1128,
+      "step": 13180
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.3395417913757903e-05,
+      "loss": 0.1094,
+      "step": 13181
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.3394496728496592e-05,
+      "loss": 0.1279,
+      "step": 13182
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.3393575497136353e-05,
+      "loss": 0.0,
+      "step": 13183
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.3392654219682262e-05,
+      "loss": 1.0078,
+      "step": 13184
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.339173289613937e-05,
+      "loss": 2.0,
+      "step": 13185
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.3390811526512736e-05,
+      "loss": 0.0247,
+      "step": 13186
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.338989011080742e-05,
+      "loss": 0.7109,
+      "step": 13187
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.3388968649028484e-05,
+      "loss": 0.0001,
+      "step": 13188
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.3388047141180984e-05,
+      "loss": 0.9844,
+      "step": 13189
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.3387125587269986e-05,
+      "loss": 0.5039,
+      "step": 13190
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.3386203987300548e-05,
+      "loss": 0.2812,
+      "step": 13191
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 2.3385282341277735e-05,
+      "loss": 0.0194,
+      "step": 13192
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.3384360649206596e-05,
+      "loss": 0.1709,
+      "step": 13193
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.3383438911092207e-05,
+      "loss": 0.0104,
+      "step": 13194
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.338251712693962e-05,
+      "loss": 0.1128,
+      "step": 13195
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.3381595296753903e-05,
+      "loss": 0.0004,
+      "step": 13196
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.338067342054012e-05,
+      "loss": 0.0101,
+      "step": 13197
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.3379751498303322e-05,
+      "loss": 0.0131,
+      "step": 13198
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.3378829530048587e-05,
+      "loss": 0.0003,
+      "step": 13199
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.3377907515780968e-05,
+      "loss": 0.0001,
+      "step": 13200
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.3376985455505526e-05,
+      "loss": 0.6289,
+      "step": 13201
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.3376063349227334e-05,
+      "loss": 0.1494,
+      "step": 13202
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.337514119695145e-05,
+      "loss": 0.0649,
+      "step": 13203
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.3374218998682938e-05,
+      "loss": 0.0003,
+      "step": 13204
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.3373296754426863e-05,
+      "loss": 0.7695,
+      "step": 13205
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.3372374464188294e-05,
+      "loss": 0.0036,
+      "step": 13206
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.3371452127972292e-05,
+      "loss": 3.1406,
+      "step": 13207
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.3370529745783916e-05,
+      "loss": 0.0002,
+      "step": 13208
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.3369607317628244e-05,
+      "loss": 0.4531,
+      "step": 13209
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.3368684843510332e-05,
+      "loss": 0.375,
+      "step": 13210
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.3367762323435248e-05,
+      "loss": 0.1245,
+      "step": 13211
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.3366839757408064e-05,
+      "loss": 0.0,
+      "step": 13212
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.336591714543384e-05,
+      "loss": 0.0068,
+      "step": 13213
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.3364994487517634e-05,
+      "loss": 0.0,
+      "step": 13214
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.336407178366453e-05,
+      "loss": 0.0001,
+      "step": 13215
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.336314903387959e-05,
+      "loss": 0.0,
+      "step": 13216
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.3362226238167873e-05,
+      "loss": 0.0039,
+      "step": 13217
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.3361303396534454e-05,
+      "loss": 0.002,
+      "step": 13218
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.33603805089844e-05,
+      "loss": 0.0,
+      "step": 13219
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.335945757552278e-05,
+      "loss": 0.1504,
+      "step": 13220
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.335853459615466e-05,
+      "loss": 0.0001,
+      "step": 13221
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.3357611570885114e-05,
+      "loss": 0.0,
+      "step": 13222
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.33566884997192e-05,
+      "loss": 0.0,
+      "step": 13223
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.3355765382661996e-05,
+      "loss": 0.0,
+      "step": 13224
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.3354842219718565e-05,
+      "loss": 0.0,
+      "step": 13225
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.3353919010893984e-05,
+      "loss": 0.0,
+      "step": 13226
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.335299575619332e-05,
+      "loss": 0.0,
+      "step": 13227
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.335207245562164e-05,
+      "loss": 0.1104,
+      "step": 13228
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.335114910918402e-05,
+      "loss": 0.1494,
+      "step": 13229
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.3350225716885522e-05,
+      "loss": 0.1182,
+      "step": 13230
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.334930227873123e-05,
+      "loss": 0.1035,
+      "step": 13231
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.33483787947262e-05,
+      "loss": 0.1099,
+      "step": 13232
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.3347455264875514e-05,
+      "loss": 0.1455,
+      "step": 13233
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.334653168918424e-05,
+      "loss": 0.5859,
+      "step": 13234
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.334560806765745e-05,
+      "loss": 0.5234,
+      "step": 13235
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.3344684400300215e-05,
+      "loss": 0.0014,
+      "step": 13236
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.334376068711761e-05,
+      "loss": 0.6016,
+      "step": 13237
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.334283692811471e-05,
+      "loss": 0.0118,
+      "step": 13238
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.3341913123296575e-05,
+      "loss": 0.0239,
+      "step": 13239
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.3340989272668297e-05,
+      "loss": 1.2812,
+      "step": 13240
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.3340065376234933e-05,
+      "loss": 0.4023,
+      "step": 13241
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.3339141434001566e-05,
+      "loss": 0.3887,
+      "step": 13242
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.3338217445973268e-05,
+      "loss": 0.1309,
+      "step": 13243
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.3337293412155114e-05,
+      "loss": 0.1245,
+      "step": 13244
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.333636933255217e-05,
+      "loss": 0.0317,
+      "step": 13245
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.3335445207169527e-05,
+      "loss": 0.0403,
+      "step": 13246
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.3334521036012245e-05,
+      "loss": 0.0009,
+      "step": 13247
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.3333596819085402e-05,
+      "loss": 0.2119,
+      "step": 13248
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.333267255639408e-05,
+      "loss": 0.1445,
+      "step": 13249
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.333174824794335e-05,
+      "loss": 0.032,
+      "step": 13250
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.3330823893738286e-05,
+      "loss": 0.2656,
+      "step": 13251
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.3329899493783964e-05,
+      "loss": 0.0,
+      "step": 13252
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.3328975048085472e-05,
+      "loss": 0.0004,
+      "step": 13253
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.3328050556647875e-05,
+      "loss": 0.1709,
+      "step": 13254
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.3327126019476245e-05,
+      "loss": 0.0003,
+      "step": 13255
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.3326201436575675e-05,
+      "loss": 0.0004,
+      "step": 13256
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.332527680795123e-05,
+      "loss": 0.0043,
+      "step": 13257
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.332435213360799e-05,
+      "loss": 0.0002,
+      "step": 13258
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.332342741355104e-05,
+      "loss": 0.0,
+      "step": 13259
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 2.332250264778545e-05,
+      "loss": 0.0,
+      "step": 13260
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.3321577836316297e-05,
+      "loss": 0.1064,
+      "step": 13261
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.332065297914867e-05,
+      "loss": 0.0277,
+      "step": 13262
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.3319728076287643e-05,
+      "loss": 0.1934,
+      "step": 13263
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.331880312773829e-05,
+      "loss": 0.0008,
+      "step": 13264
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.3317878133505693e-05,
+      "loss": 0.0018,
+      "step": 13265
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.3316953093594937e-05,
+      "loss": 0.1592,
+      "step": 13266
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.3316028008011097e-05,
+      "loss": 0.0986,
+      "step": 13267
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.3315102876759255e-05,
+      "loss": 0.0,
+      "step": 13268
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.331417769984449e-05,
+      "loss": 0.0,
+      "step": 13269
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.331325247727188e-05,
+      "loss": 0.0001,
+      "step": 13270
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.3312327209046513e-05,
+      "loss": 0.0,
+      "step": 13271
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.3311401895173466e-05,
+      "loss": 0.0,
+      "step": 13272
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.331047653565782e-05,
+      "loss": 0.0,
+      "step": 13273
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.3309551130504654e-05,
+      "loss": 0.2812,
+      "step": 13274
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.3308625679719055e-05,
+      "loss": 0.0,
+      "step": 13275
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.3307700183306106e-05,
+      "loss": 0.0,
+      "step": 13276
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.3306774641270888e-05,
+      "loss": 0.0981,
+      "step": 13277
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.3305849053618477e-05,
+      "loss": 0.0001,
+      "step": 13278
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.330492342035397e-05,
+      "loss": 0.1895,
+      "step": 13279
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.330399774148243e-05,
+      "loss": 0.6211,
+      "step": 13280
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.330307201700896e-05,
+      "loss": 0.1426,
+      "step": 13281
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.3302146246938636e-05,
+      "loss": 0.125,
+      "step": 13282
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.3301220431276537e-05,
+      "loss": 0.1182,
+      "step": 13283
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.3300294570027757e-05,
+      "loss": 0.4824,
+      "step": 13284
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.3299368663197373e-05,
+      "loss": 0.5859,
+      "step": 13285
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.3298442710790475e-05,
+      "loss": 1.3203,
+      "step": 13286
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.3297516712812136e-05,
+      "loss": 0.2832,
+      "step": 13287
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.329659066926746e-05,
+      "loss": 1.3594,
+      "step": 13288
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.3295664580161515e-05,
+      "loss": 0.9531,
+      "step": 13289
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.32947384454994e-05,
+      "loss": 0.0173,
+      "step": 13290
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.3293812265286193e-05,
+      "loss": 1.3438,
+      "step": 13291
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.3292886039526982e-05,
+      "loss": 0.0286,
+      "step": 13292
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.3291959768226853e-05,
+      "loss": 0.032,
+      "step": 13293
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.3291033451390896e-05,
+      "loss": 0.6172,
+      "step": 13294
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.3290107089024194e-05,
+      "loss": 0.0006,
+      "step": 13295
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.3289180681131837e-05,
+      "loss": 0.0002,
+      "step": 13296
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.3288254227718912e-05,
+      "loss": 0.0,
+      "step": 13297
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.32873277287905e-05,
+      "loss": 0.0004,
+      "step": 13298
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.3286401184351697e-05,
+      "loss": 0.0396,
+      "step": 13299
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.3285474594407588e-05,
+      "loss": 0.0009,
+      "step": 13300
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.3284547958963265e-05,
+      "loss": 0.168,
+      "step": 13301
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.328362127802381e-05,
+      "loss": 0.1611,
+      "step": 13302
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.328269455159432e-05,
+      "loss": 0.0,
+      "step": 13303
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.3281767779679877e-05,
+      "loss": 0.0003,
+      "step": 13304
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.3280840962285575e-05,
+      "loss": 0.0356,
+      "step": 13305
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.3279914099416503e-05,
+      "loss": 0.2715,
+      "step": 13306
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.3278987191077744e-05,
+      "loss": 0.0559,
+      "step": 13307
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.32780602372744e-05,
+      "loss": 0.0,
+      "step": 13308
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.3277133238011553e-05,
+      "loss": 0.0,
+      "step": 13309
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.32762061932943e-05,
+      "loss": 0.0,
+      "step": 13310
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.3275279103127725e-05,
+      "loss": 0.0007,
+      "step": 13311
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.327435196751692e-05,
+      "loss": 0.0261,
+      "step": 13312
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.3273424786466983e-05,
+      "loss": 0.0918,
+      "step": 13313
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.3272497559983002e-05,
+      "loss": 0.0022,
+      "step": 13314
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.3271570288070065e-05,
+      "loss": 0.0,
+      "step": 13315
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.327064297073327e-05,
+      "loss": 0.0002,
+      "step": 13316
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.3269715607977704e-05,
+      "loss": 0.0002,
+      "step": 13317
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.3268788199808462e-05,
+      "loss": 0.0292,
+      "step": 13318
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.3267860746230645e-05,
+      "loss": 0.0001,
+      "step": 13319
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.3266933247249337e-05,
+      "loss": 0.0001,
+      "step": 13320
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.3266005702869624e-05,
+      "loss": 0.0001,
+      "step": 13321
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.3265078113096617e-05,
+      "loss": 0.0,
+      "step": 13322
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.3264150477935403e-05,
+      "loss": 0.0001,
+      "step": 13323
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.3263222797391065e-05,
+      "loss": 0.0001,
+      "step": 13324
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.3262295071468715e-05,
+      "loss": 0.0001,
+      "step": 13325
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.326136730017344e-05,
+      "loss": 0.0001,
+      "step": 13326
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.3260439483510335e-05,
+      "loss": 0.0,
+      "step": 13327
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 2.3259511621484496e-05,
+      "loss": 0.0938,
+      "step": 13328
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.325858371410101e-05,
+      "loss": 0.1641,
+      "step": 13329
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.3257655761364987e-05,
+      "loss": 0.0957,
+      "step": 13330
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.3256727763281515e-05,
+      "loss": 0.0004,
+      "step": 13331
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.3255799719855696e-05,
+      "loss": 0.1777,
+      "step": 13332
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.3254871631092615e-05,
+      "loss": 0.1494,
+      "step": 13333
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.3253943496997373e-05,
+      "loss": 1.1016,
+      "step": 13334
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.325301531757507e-05,
+      "loss": 0.6406,
+      "step": 13335
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.3252087092830808e-05,
+      "loss": 0.5781,
+      "step": 13336
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.3251158822769675e-05,
+      "loss": 0.0194,
+      "step": 13337
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.325023050739678e-05,
+      "loss": 0.0361,
+      "step": 13338
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.3249302146717203e-05,
+      "loss": 2.5,
+      "step": 13339
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.3248373740736057e-05,
+      "loss": 0.8906,
+      "step": 13340
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.3247445289458436e-05,
+      "loss": 0.3984,
+      "step": 13341
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.324651679288944e-05,
+      "loss": 0.4961,
+      "step": 13342
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.3245588251034162e-05,
+      "loss": 0.1289,
+      "step": 13343
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.3244659663897708e-05,
+      "loss": 0.8008,
+      "step": 13344
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.3243731031485177e-05,
+      "loss": 0.498,
+      "step": 13345
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.3242802353801663e-05,
+      "loss": 0.2891,
+      "step": 13346
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.324187363085227e-05,
+      "loss": 0.1396,
+      "step": 13347
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.32409448626421e-05,
+      "loss": 0.1777,
+      "step": 13348
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.3240016049176248e-05,
+      "loss": 0.0002,
+      "step": 13349
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.3239087190459817e-05,
+      "loss": 0.0002,
+      "step": 13350
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.3238158286497916e-05,
+      "loss": 0.0576,
+      "step": 13351
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.3237229337295632e-05,
+      "loss": 0.0001,
+      "step": 13352
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.3236300342858076e-05,
+      "loss": 0.0476,
+      "step": 13353
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.3235371303190346e-05,
+      "loss": 0.9102,
+      "step": 13354
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.3234442218297542e-05,
+      "loss": 0.0,
+      "step": 13355
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.3233513088184772e-05,
+      "loss": 0.0003,
+      "step": 13356
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.3232583912857133e-05,
+      "loss": 0.0006,
+      "step": 13357
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.323165469231973e-05,
+      "loss": 0.0,
+      "step": 13358
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.3230725426577667e-05,
+      "loss": 0.004,
+      "step": 13359
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.3229796115636045e-05,
+      "loss": 0.0,
+      "step": 13360
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.322886675949997e-05,
+      "loss": 0.0063,
+      "step": 13361
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.322793735817454e-05,
+      "loss": 0.0001,
+      "step": 13362
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.3227007911664867e-05,
+      "loss": 0.3516,
+      "step": 13363
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.3226078419976046e-05,
+      "loss": 0.4316,
+      "step": 13364
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.322514888311319e-05,
+      "loss": 0.0,
+      "step": 13365
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.3224219301081394e-05,
+      "loss": 0.0103,
+      "step": 13366
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.3223289673885777e-05,
+      "loss": 0.0001,
+      "step": 13367
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.322236000153143e-05,
+      "loss": 0.0167,
+      "step": 13368
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.3221430284023466e-05,
+      "loss": 0.0005,
+      "step": 13369
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.3220500521366986e-05,
+      "loss": 0.1064,
+      "step": 13370
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.32195707135671e-05,
+      "loss": 0.0002,
+      "step": 13371
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.321864086062891e-05,
+      "loss": 0.0003,
+      "step": 13372
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.3217710962557535e-05,
+      "loss": 0.0001,
+      "step": 13373
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.3216781019358063e-05,
+      "loss": 0.0001,
+      "step": 13374
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.321585103103561e-05,
+      "loss": 0.0001,
+      "step": 13375
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.3214920997595285e-05,
+      "loss": 0.0021,
+      "step": 13376
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.321399091904219e-05,
+      "loss": 0.0003,
+      "step": 13377
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.3213060795381434e-05,
+      "loss": 0.1133,
+      "step": 13378
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.321213062661813e-05,
+      "loss": 0.123,
+      "step": 13379
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.321120041275738e-05,
+      "loss": 0.1484,
+      "step": 13380
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.3210270153804295e-05,
+      "loss": 0.1797,
+      "step": 13381
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.3209339849763983e-05,
+      "loss": 0.1855,
+      "step": 13382
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.3208409500641555e-05,
+      "loss": 0.1201,
+      "step": 13383
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.3207479106442114e-05,
+      "loss": 0.4922,
+      "step": 13384
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.320654866717078e-05,
+      "loss": 0.2715,
+      "step": 13385
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.3205618182832654e-05,
+      "loss": 0.3594,
+      "step": 13386
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.3204687653432843e-05,
+      "loss": 0.0141,
+      "step": 13387
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.3203757078976463e-05,
+      "loss": 0.9727,
+      "step": 13388
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.3202826459468623e-05,
+      "loss": 0.6875,
+      "step": 13389
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.3201895794914436e-05,
+      "loss": 1.2188,
+      "step": 13390
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.320096508531901e-05,
+      "loss": 0.2969,
+      "step": 13391
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.3200034330687457e-05,
+      "loss": 0.4141,
+      "step": 13392
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.3199103531024888e-05,
+      "loss": 0.4512,
+      "step": 13393
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.3198172686336416e-05,
+      "loss": 0.7148,
+      "step": 13394
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.319724179662715e-05,
+      "loss": 0.0022,
+      "step": 13395
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 2.31963108619022e-05,
+      "loss": 0.1089,
+      "step": 13396
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.3195379882166685e-05,
+      "loss": 0.0688,
+      "step": 13397
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.319444885742571e-05,
+      "loss": 0.9766,
+      "step": 13398
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.3193517787684395e-05,
+      "loss": 0.001,
+      "step": 13399
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.319258667294785e-05,
+      "loss": 0.7578,
+      "step": 13400
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.3191655513221187e-05,
+      "loss": 1.625,
+      "step": 13401
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.3190724308509522e-05,
+      "loss": 0.0015,
+      "step": 13402
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.3189793058817968e-05,
+      "loss": 0.6367,
+      "step": 13403
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.3188861764151638e-05,
+      "loss": 0.0001,
+      "step": 13404
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.3187930424515645e-05,
+      "loss": 0.0001,
+      "step": 13405
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.3186999039915107e-05,
+      "loss": 0.0019,
+      "step": 13406
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.3186067610355134e-05,
+      "loss": 0.0164,
+      "step": 13407
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.318513613584085e-05,
+      "loss": 0.0012,
+      "step": 13408
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.318420461637736e-05,
+      "loss": 0.0,
+      "step": 13409
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.3183273051969783e-05,
+      "loss": 0.0001,
+      "step": 13410
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.3182341442623242e-05,
+      "loss": 0.0,
+      "step": 13411
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.3181409788342837e-05,
+      "loss": 0.0001,
+      "step": 13412
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.31804780891337e-05,
+      "loss": 0.0981,
+      "step": 13413
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.317954634500094e-05,
+      "loss": 0.0,
+      "step": 13414
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.3178614555949672e-05,
+      "loss": 0.0,
+      "step": 13415
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.317768272198502e-05,
+      "loss": 0.2412,
+      "step": 13416
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.3176750843112094e-05,
+      "loss": 0.0189,
+      "step": 13417
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.3175818919336017e-05,
+      "loss": 0.0,
+      "step": 13418
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.31748869506619e-05,
+      "loss": 0.009,
+      "step": 13419
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.3173954937094873e-05,
+      "loss": 0.0038,
+      "step": 13420
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.3173022878640043e-05,
+      "loss": 0.0,
+      "step": 13421
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.317209077530253e-05,
+      "loss": 0.0,
+      "step": 13422
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.3171158627087454e-05,
+      "loss": 0.0,
+      "step": 13423
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.3170226433999938e-05,
+      "loss": 0.0,
+      "step": 13424
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.3169294196045097e-05,
+      "loss": 0.0,
+      "step": 13425
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.316836191322805e-05,
+      "loss": 0.2695,
+      "step": 13426
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.3167429585553924e-05,
+      "loss": 0.0801,
+      "step": 13427
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.3166497213027825e-05,
+      "loss": 0.1025,
+      "step": 13428
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.3165564795654882e-05,
+      "loss": 0.0962,
+      "step": 13429
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.3164632333440218e-05,
+      "loss": 0.0,
+      "step": 13430
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.3163699826388945e-05,
+      "loss": 0.103,
+      "step": 13431
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.3162767274506195e-05,
+      "loss": 0.103,
+      "step": 13432
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.316183467779708e-05,
+      "loss": 0.1396,
+      "step": 13433
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.3160902036266725e-05,
+      "loss": 1.4688,
+      "step": 13434
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.3159969349920255e-05,
+      "loss": 0.1885,
+      "step": 13435
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.3159036618762784e-05,
+      "loss": 0.7188,
+      "step": 13436
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.3158103842799442e-05,
+      "loss": 0.7852,
+      "step": 13437
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.3157171022035342e-05,
+      "loss": 0.6641,
+      "step": 13438
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.315623815647562e-05,
+      "loss": 0.0408,
+      "step": 13439
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.3155305246125384e-05,
+      "loss": 0.0474,
+      "step": 13440
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.3154372290989773e-05,
+      "loss": 0.6992,
+      "step": 13441
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.315343929107389e-05,
+      "loss": 0.8516,
+      "step": 13442
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.315250624638288e-05,
+      "loss": 0.0491,
+      "step": 13443
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.3151573156921852e-05,
+      "loss": 0.0449,
+      "step": 13444
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.3150640022695944e-05,
+      "loss": 0.0535,
+      "step": 13445
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.3149706843710267e-05,
+      "loss": 0.3164,
+      "step": 13446
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.3148773619969947e-05,
+      "loss": 0.0192,
+      "step": 13447
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.3147840351480114e-05,
+      "loss": 2.0469,
+      "step": 13448
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.3146907038245895e-05,
+      "loss": 0.0031,
+      "step": 13449
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.314597368027241e-05,
+      "loss": 0.0058,
+      "step": 13450
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.3145040277564788e-05,
+      "loss": 0.0001,
+      "step": 13451
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.3144106830128146e-05,
+      "loss": 1.3281,
+      "step": 13452
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.314317333796762e-05,
+      "loss": 0.0,
+      "step": 13453
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.3142239801088342e-05,
+      "loss": 0.0005,
+      "step": 13454
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.3141306219495423e-05,
+      "loss": 0.0352,
+      "step": 13455
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.3140372593193998e-05,
+      "loss": 0.2021,
+      "step": 13456
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.3139438922189193e-05,
+      "loss": 0.1436,
+      "step": 13457
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.3138505206486137e-05,
+      "loss": 0.0,
+      "step": 13458
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.313757144608995e-05,
+      "loss": 0.0,
+      "step": 13459
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.3136637641005772e-05,
+      "loss": 0.0,
+      "step": 13460
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.313570379123873e-05,
+      "loss": 0.1807,
+      "step": 13461
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.3134769896793937e-05,
+      "loss": 0.6016,
+      "step": 13462
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.313383595767654e-05,
+      "loss": 0.1504,
+      "step": 13463
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 2.3132901973891655e-05,
+      "loss": 0.0085,
+      "step": 13464
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.313196794544441e-05,
+      "loss": 0.0001,
+      "step": 13465
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.3131033872339947e-05,
+      "loss": 0.0,
+      "step": 13466
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.3130099754583385e-05,
+      "loss": 0.1055,
+      "step": 13467
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.3129165592179862e-05,
+      "loss": 0.0077,
+      "step": 13468
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.31282313851345e-05,
+      "loss": 0.0,
+      "step": 13469
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.3127297133452433e-05,
+      "loss": 0.002,
+      "step": 13470
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.3126362837138788e-05,
+      "loss": 0.0,
+      "step": 13471
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.31254284961987e-05,
+      "loss": 0.0001,
+      "step": 13472
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.31244941106373e-05,
+      "loss": 0.1426,
+      "step": 13473
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.3123559680459715e-05,
+      "loss": 0.0,
+      "step": 13474
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.3122625205671085e-05,
+      "loss": 0.0,
+      "step": 13475
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.312169068627653e-05,
+      "loss": 0.0216,
+      "step": 13476
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.3120756122281193e-05,
+      "loss": 0.0747,
+      "step": 13477
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.31198215136902e-05,
+      "loss": 0.0723,
+      "step": 13478
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.3118886860508684e-05,
+      "loss": 0.0,
+      "step": 13479
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.311795216274178e-05,
+      "loss": 0.1436,
+      "step": 13480
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.3117017420394614e-05,
+      "loss": 0.1079,
+      "step": 13481
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.3116082633472327e-05,
+      "loss": 0.0938,
+      "step": 13482
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.311514780198005e-05,
+      "loss": 0.1455,
+      "step": 13483
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.311421292592292e-05,
+      "loss": 0.3809,
+      "step": 13484
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.3113278005306066e-05,
+      "loss": 0.8555,
+      "step": 13485
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.3112343040134623e-05,
+      "loss": 1.1406,
+      "step": 13486
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.3111408030413723e-05,
+      "loss": 1.4766,
+      "step": 13487
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.311047297614851e-05,
+      "loss": 1.1797,
+      "step": 13488
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.310953787734411e-05,
+      "loss": 0.6523,
+      "step": 13489
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.310860273400566e-05,
+      "loss": 0.8086,
+      "step": 13490
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.3107667546138298e-05,
+      "loss": 0.1992,
+      "step": 13491
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.3106732313747158e-05,
+      "loss": 1.1797,
+      "step": 13492
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.3105797036837378e-05,
+      "loss": 0.3223,
+      "step": 13493
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.3104861715414086e-05,
+      "loss": 0.0051,
+      "step": 13494
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.3103926349482428e-05,
+      "loss": 1.2422,
+      "step": 13495
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.3102990939047537e-05,
+      "loss": 1.6797,
+      "step": 13496
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.3102055484114546e-05,
+      "loss": 1.2266,
+      "step": 13497
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.31011199846886e-05,
+      "loss": 0.1582,
+      "step": 13498
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.3100184440774837e-05,
+      "loss": 0.0337,
+      "step": 13499
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.3099248852378384e-05,
+      "loss": 0.0003,
+      "step": 13500
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.3098313219504385e-05,
+      "loss": 0.0354,
+      "step": 13501
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.3097377542157977e-05,
+      "loss": 0.4688,
+      "step": 13502
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.3096441820344305e-05,
+      "loss": 0.3262,
+      "step": 13503
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.3095506054068494e-05,
+      "loss": 0.0002,
+      "step": 13504
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.3094570243335695e-05,
+      "loss": 0.5,
+      "step": 13505
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.3093634388151044e-05,
+      "loss": 0.1309,
+      "step": 13506
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.309269848851968e-05,
+      "loss": 0.0,
+      "step": 13507
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.3091762544446735e-05,
+      "loss": 0.8086,
+      "step": 13508
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.309082655593736e-05,
+      "loss": 0.0024,
+      "step": 13509
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.3089890522996685e-05,
+      "loss": 0.0,
+      "step": 13510
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.3088954445629864e-05,
+      "loss": 0.1572,
+      "step": 13511
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.3088018323842026e-05,
+      "loss": 0.0303,
+      "step": 13512
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.3087082157638312e-05,
+      "loss": 0.0244,
+      "step": 13513
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.3086145947023868e-05,
+      "loss": 0.0094,
+      "step": 13514
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.3085209692003836e-05,
+      "loss": 0.0,
+      "step": 13515
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.3084273392583347e-05,
+      "loss": 0.0415,
+      "step": 13516
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.308333704876756e-05,
+      "loss": 0.0194,
+      "step": 13517
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.3082400660561603e-05,
+      "loss": 0.2021,
+      "step": 13518
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.308146422797062e-05,
+      "loss": 0.0001,
+      "step": 13519
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.308052775099976e-05,
+      "loss": 0.0,
+      "step": 13520
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.307959122965416e-05,
+      "loss": 0.0,
+      "step": 13521
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.3078654663938967e-05,
+      "loss": 0.0,
+      "step": 13522
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.3077718053859322e-05,
+      "loss": 0.0,
+      "step": 13523
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.307678139942037e-05,
+      "loss": 0.0,
+      "step": 13524
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.307584470062725e-05,
+      "loss": 0.0,
+      "step": 13525
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.3074907957485106e-05,
+      "loss": 0.0,
+      "step": 13526
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.3073971169999093e-05,
+      "loss": 0.167,
+      "step": 13527
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.3073034338174344e-05,
+      "loss": 0.1416,
+      "step": 13528
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.307209746201601e-05,
+      "loss": 0.0,
+      "step": 13529
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.3071160541529228e-05,
+      "loss": 0.1504,
+      "step": 13530
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.3070223576719154e-05,
+      "loss": 0.1211,
+      "step": 13531
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.3069286567590928e-05,
+      "loss": 0.1138,
+      "step": 13532
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.306834951414969e-05,
+      "loss": 0.1279,
+      "step": 13533
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.3067412416400595e-05,
+      "loss": 1.0547,
+      "step": 13534
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.306647527434879e-05,
+      "loss": 0.1235,
+      "step": 13535
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.306553808799941e-05,
+      "loss": 1.7422,
+      "step": 13536
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.3064600857357614e-05,
+      "loss": 1.0312,
+      "step": 13537
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.3063663582428538e-05,
+      "loss": 0.2734,
+      "step": 13538
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.3062726263217338e-05,
+      "loss": 0.6758,
+      "step": 13539
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.3061788899729154e-05,
+      "loss": 0.1084,
+      "step": 13540
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.3060851491969144e-05,
+      "loss": 0.0518,
+      "step": 13541
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.3059914039942446e-05,
+      "loss": 0.8711,
+      "step": 13542
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.3058976543654214e-05,
+      "loss": 0.0138,
+      "step": 13543
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.305803900310959e-05,
+      "loss": 0.0233,
+      "step": 13544
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.3057101418313725e-05,
+      "loss": 0.0371,
+      "step": 13545
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.305616378927178e-05,
+      "loss": 0.0204,
+      "step": 13546
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.305522611598888e-05,
+      "loss": 0.334,
+      "step": 13547
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.305428839847019e-05,
+      "loss": 0.1748,
+      "step": 13548
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.3053350636720865e-05,
+      "loss": 1.4297,
+      "step": 13549
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.3052412830746038e-05,
+      "loss": 0.5508,
+      "step": 13550
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.305147498055087e-05,
+      "loss": 0.0,
+      "step": 13551
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.3050537086140513e-05,
+      "loss": 0.1074,
+      "step": 13552
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.304959914752011e-05,
+      "loss": 0.0003,
+      "step": 13553
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.3048661164694815e-05,
+      "loss": 0.0281,
+      "step": 13554
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.3047723137669782e-05,
+      "loss": 0.1504,
+      "step": 13555
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.304678506645016e-05,
+      "loss": 1.2969,
+      "step": 13556
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.3045846951041097e-05,
+      "loss": 0.6211,
+      "step": 13557
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.304490879144775e-05,
+      "loss": 0.9766,
+      "step": 13558
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.3043970587675267e-05,
+      "loss": 0.0001,
+      "step": 13559
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.30430323397288e-05,
+      "loss": 0.4199,
+      "step": 13560
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.3042094047613512e-05,
+      "loss": 0.1226,
+      "step": 13561
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.304115571133454e-05,
+      "loss": 0.1787,
+      "step": 13562
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.3040217330897046e-05,
+      "loss": 0.0002,
+      "step": 13563
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.303927890630618e-05,
+      "loss": 0.0967,
+      "step": 13564
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.30383404375671e-05,
+      "loss": 0.1953,
+      "step": 13565
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.303740192468495e-05,
+      "loss": 0.1602,
+      "step": 13566
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.3036463367664894e-05,
+      "loss": 1.7578,
+      "step": 13567
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.303552476651208e-05,
+      "loss": 0.6367,
+      "step": 13568
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.3034586121231668e-05,
+      "loss": 0.2734,
+      "step": 13569
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.303364743182881e-05,
+      "loss": 0.1562,
+      "step": 13570
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.303270869830866e-05,
+      "loss": 0.0537,
+      "step": 13571
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.3031769920676367e-05,
+      "loss": 0.0408,
+      "step": 13572
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.30308310989371e-05,
+      "loss": 0.1328,
+      "step": 13573
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.302989223309601e-05,
+      "loss": 0.0493,
+      "step": 13574
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.302895332315824e-05,
+      "loss": 0.7227,
+      "step": 13575
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.3028014369128958e-05,
+      "loss": 1.0469,
+      "step": 13576
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.3027075371013326e-05,
+      "loss": 0.7891,
+      "step": 13577
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.3026136328816484e-05,
+      "loss": 0.0115,
+      "step": 13578
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.3025197242543605e-05,
+      "loss": 0.2988,
+      "step": 13579
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.3024258112199835e-05,
+      "loss": 0.0002,
+      "step": 13580
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.3023318937790342e-05,
+      "loss": 0.053,
+      "step": 13581
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.302237971932027e-05,
+      "loss": 1.6797,
+      "step": 13582
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.3021440456794788e-05,
+      "loss": 0.2773,
+      "step": 13583
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.3020501150219047e-05,
+      "loss": 0.0,
+      "step": 13584
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.3019561799598207e-05,
+      "loss": 0.1128,
+      "step": 13585
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.3018622404937426e-05,
+      "loss": 0.0004,
+      "step": 13586
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.3017682966241866e-05,
+      "loss": 0.1807,
+      "step": 13587
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.3016743483516683e-05,
+      "loss": 0.0005,
+      "step": 13588
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.3015803956767044e-05,
+      "loss": 0.0342,
+      "step": 13589
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.3014864385998095e-05,
+      "loss": 0.0001,
+      "step": 13590
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.3013924771215003e-05,
+      "loss": 0.004,
+      "step": 13591
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.301298511242293e-05,
+      "loss": 0.293,
+      "step": 13592
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.301204540962703e-05,
+      "loss": 0.0003,
+      "step": 13593
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.3011105662832474e-05,
+      "loss": 0.0001,
+      "step": 13594
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.301016587204441e-05,
+      "loss": 0.0,
+      "step": 13595
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.3009226037268007e-05,
+      "loss": 0.1973,
+      "step": 13596
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.300828615850842e-05,
+      "loss": 0.0216,
+      "step": 13597
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.300734623577082e-05,
+      "loss": 0.1416,
+      "step": 13598
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 2.3006406269060358e-05,
+      "loss": 0.0008,
+      "step": 13599
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.3005466258382202e-05,
+      "loss": 0.1689,
+      "step": 13600
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.300452620374151e-05,
+      "loss": 0.0425,
+      "step": 13601
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.3003586105143456e-05,
+      "loss": 0.0128,
+      "step": 13602
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.3002645962593187e-05,
+      "loss": 0.0635,
+      "step": 13603
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.3001705776095877e-05,
+      "loss": 0.0001,
+      "step": 13604
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.3000765545656686e-05,
+      "loss": 0.0001,
+      "step": 13605
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.299982527128077e-05,
+      "loss": 0.002,
+      "step": 13606
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.2998884952973303e-05,
+      "loss": 0.0001,
+      "step": 13607
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.2997944590739442e-05,
+      "loss": 0.0,
+      "step": 13608
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.2997004184584357e-05,
+      "loss": 0.0,
+      "step": 13609
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.2996063734513206e-05,
+      "loss": 0.0771,
+      "step": 13610
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.299512324053116e-05,
+      "loss": 0.0,
+      "step": 13611
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.2994182702643376e-05,
+      "loss": 0.1089,
+      "step": 13612
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.2993242120855024e-05,
+      "loss": 0.0001,
+      "step": 13613
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.2992301495171276e-05,
+      "loss": 0.1113,
+      "step": 13614
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.299136082559728e-05,
+      "loss": 0.1816,
+      "step": 13615
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.2990420112138218e-05,
+      "loss": 0.1021,
+      "step": 13616
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.298947935479925e-05,
+      "loss": 0.0354,
+      "step": 13617
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.2988538553585534e-05,
+      "loss": 0.2578,
+      "step": 13618
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.2987597708502252e-05,
+      "loss": 0.125,
+      "step": 13619
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.2986656819554562e-05,
+      "loss": 0.0001,
+      "step": 13620
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.2985715886747632e-05,
+      "loss": 1.2266,
+      "step": 13621
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.2984774910086628e-05,
+      "loss": 0.7031,
+      "step": 13622
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.298383388957672e-05,
+      "loss": 0.0029,
+      "step": 13623
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.2982892825223073e-05,
+      "loss": 0.2168,
+      "step": 13624
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.298195171703086e-05,
+      "loss": 0.2969,
+      "step": 13625
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.2981010565005246e-05,
+      "loss": 0.0234,
+      "step": 13626
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.2980069369151394e-05,
+      "loss": 0.0544,
+      "step": 13627
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.2979128129474486e-05,
+      "loss": 0.2217,
+      "step": 13628
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.297818684597967e-05,
+      "loss": 0.0164,
+      "step": 13629
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.2977245518672135e-05,
+      "loss": 0.9492,
+      "step": 13630
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.2976304147557046e-05,
+      "loss": 0.0077,
+      "step": 13631
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.2975362732639562e-05,
+      "loss": 0.4219,
+      "step": 13632
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.2974421273924864e-05,
+      "loss": 0.0004,
+      "step": 13633
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.297347977141812e-05,
+      "loss": 0.0515,
+      "step": 13634
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.2972538225124496e-05,
+      "loss": 0.0,
+      "step": 13635
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.2971596635049167e-05,
+      "loss": 0.0027,
+      "step": 13636
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.2970655001197302e-05,
+      "loss": 0.4395,
+      "step": 13637
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.2969713323574074e-05,
+      "loss": 0.0,
+      "step": 13638
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.2968771602184647e-05,
+      "loss": 0.0024,
+      "step": 13639
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.2967829837034204e-05,
+      "loss": 0.0045,
+      "step": 13640
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.2966888028127907e-05,
+      "loss": 0.0,
+      "step": 13641
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.2965946175470934e-05,
+      "loss": 0.0781,
+      "step": 13642
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.2965004279068454e-05,
+      "loss": 0.0,
+      "step": 13643
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.2964062338925643e-05,
+      "loss": 0.0,
+      "step": 13644
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.2963120355047668e-05,
+      "loss": 0.0,
+      "step": 13645
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.2962178327439707e-05,
+      "loss": 0.0,
+      "step": 13646
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.296123625610693e-05,
+      "loss": 0.0001,
+      "step": 13647
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.296029414105451e-05,
+      "loss": 0.0,
+      "step": 13648
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.2959351982287627e-05,
+      "loss": 0.0,
+      "step": 13649
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.2958409779811445e-05,
+      "loss": 0.0,
+      "step": 13650
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.295746753363115e-05,
+      "loss": 0.0,
+      "step": 13651
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.295652524375191e-05,
+      "loss": 0.0349,
+      "step": 13652
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.2955582910178896e-05,
+      "loss": 0.0,
+      "step": 13653
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.2954640532917286e-05,
+      "loss": 0.0,
+      "step": 13654
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.2953698111972258e-05,
+      "loss": 0.0017,
+      "step": 13655
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.2952755647348984e-05,
+      "loss": 0.0,
+      "step": 13656
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.2951813139052644e-05,
+      "loss": 0.0001,
+      "step": 13657
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.2950870587088404e-05,
+      "loss": 0.0,
+      "step": 13658
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.2949927991461452e-05,
+      "loss": 0.0001,
+      "step": 13659
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.2948985352176956e-05,
+      "loss": 0.0023,
+      "step": 13660
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.2948042669240094e-05,
+      "loss": 0.0,
+      "step": 13661
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.2947099942656048e-05,
+      "loss": 0.1172,
+      "step": 13662
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.2946157172429992e-05,
+      "loss": 0.103,
+      "step": 13663
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.2945214358567102e-05,
+      "loss": 0.1299,
+      "step": 13664
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.2944271501072552e-05,
+      "loss": 0.1436,
+      "step": 13665
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.2943328599951527e-05,
+      "loss": 0.2012,
+      "step": 13666
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 2.29423856552092e-05,
+      "loss": 0.0603,
+      "step": 13667
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.2941442666850752e-05,
+      "loss": 0.0403,
+      "step": 13668
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.2940499634881364e-05,
+      "loss": 0.1455,
+      "step": 13669
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.293955655930621e-05,
+      "loss": 0.0032,
+      "step": 13670
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.293861344013047e-05,
+      "loss": 0.1074,
+      "step": 13671
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.2937670277359323e-05,
+      "loss": 0.1338,
+      "step": 13672
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.2936727070997943e-05,
+      "loss": 0.4727,
+      "step": 13673
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.2935783821051525e-05,
+      "loss": 0.0101,
+      "step": 13674
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.2934840527525234e-05,
+      "loss": 0.4648,
+      "step": 13675
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.2933897190424258e-05,
+      "loss": 1.2891,
+      "step": 13676
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.2932953809753774e-05,
+      "loss": 0.0,
+      "step": 13677
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.2932010385518966e-05,
+      "loss": 0.9922,
+      "step": 13678
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.2931066917725008e-05,
+      "loss": 0.0776,
+      "step": 13679
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.293012340637709e-05,
+      "loss": 0.6875,
+      "step": 13680
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.292917985148039e-05,
+      "loss": 0.0086,
+      "step": 13681
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.2928236253040085e-05,
+      "loss": 0.0786,
+      "step": 13682
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.292729261106136e-05,
+      "loss": 0.9648,
+      "step": 13683
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.29263489255494e-05,
+      "loss": 0.0496,
+      "step": 13684
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.2925405196509387e-05,
+      "loss": 0.0322,
+      "step": 13685
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.2924461423946492e-05,
+      "loss": 0.0427,
+      "step": 13686
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.2923517607865915e-05,
+      "loss": 0.0063,
+      "step": 13687
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.292257374827283e-05,
+      "loss": 0.0,
+      "step": 13688
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.2921629845172426e-05,
+      "loss": 0.0085,
+      "step": 13689
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.2920685898569875e-05,
+      "loss": 0.0002,
+      "step": 13690
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.2919741908470374e-05,
+      "loss": 0.7852,
+      "step": 13691
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.2918797874879094e-05,
+      "loss": 0.0439,
+      "step": 13692
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.291785379780123e-05,
+      "loss": 0.0003,
+      "step": 13693
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.2916909677241963e-05,
+      "loss": 0.0332,
+      "step": 13694
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.2915965513206475e-05,
+      "loss": 0.0081,
+      "step": 13695
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.291502130569995e-05,
+      "loss": 0.1104,
+      "step": 13696
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.291407705472758e-05,
+      "loss": 0.0005,
+      "step": 13697
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.2913132760294548e-05,
+      "loss": 0.0,
+      "step": 13698
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.2912188422406038e-05,
+      "loss": 0.6836,
+      "step": 13699
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.2911244041067236e-05,
+      "loss": 0.0162,
+      "step": 13700
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.2910299616283324e-05,
+      "loss": 0.0,
+      "step": 13701
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.2909355148059497e-05,
+      "loss": 0.0001,
+      "step": 13702
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.290841063640094e-05,
+      "loss": 0.0001,
+      "step": 13703
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.2907466081312833e-05,
+      "loss": 0.0,
+      "step": 13704
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.2906521482800363e-05,
+      "loss": 0.0,
+      "step": 13705
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.290557684086873e-05,
+      "loss": 0.0,
+      "step": 13706
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.290463215552311e-05,
+      "loss": 0.0,
+      "step": 13707
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.290368742676869e-05,
+      "loss": 0.1245,
+      "step": 13708
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.290274265461067e-05,
+      "loss": 0.0,
+      "step": 13709
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.2901797839054226e-05,
+      "loss": 0.0,
+      "step": 13710
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.2900852980104546e-05,
+      "loss": 0.0,
+      "step": 13711
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.2899908077766833e-05,
+      "loss": 0.1348,
+      "step": 13712
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.289896313204626e-05,
+      "loss": 0.1279,
+      "step": 13713
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.289801814294802e-05,
+      "loss": 0.1436,
+      "step": 13714
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.289707311047731e-05,
+      "loss": 0.1191,
+      "step": 13715
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.2896128034639313e-05,
+      "loss": 0.165,
+      "step": 13716
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.2895182915439224e-05,
+      "loss": 0.3242,
+      "step": 13717
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.2894237752882226e-05,
+      "loss": 0.2676,
+      "step": 13718
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.2893292546973516e-05,
+      "loss": 0.7695,
+      "step": 13719
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.289234729771828e-05,
+      "loss": 0.2021,
+      "step": 13720
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.2891402005121715e-05,
+      "loss": 1.4688,
+      "step": 13721
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.289045666918901e-05,
+      "loss": 0.0012,
+      "step": 13722
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.288951128992535e-05,
+      "loss": 0.1221,
+      "step": 13723
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.2888565867335928e-05,
+      "loss": 0.0215,
+      "step": 13724
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.2887620401425945e-05,
+      "loss": 0.0019,
+      "step": 13725
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.2886674892200584e-05,
+      "loss": 0.21,
+      "step": 13726
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.288572933966504e-05,
+      "loss": 0.0312,
+      "step": 13727
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.288478374382451e-05,
+      "loss": 0.2891,
+      "step": 13728
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.2883838104684175e-05,
+      "loss": 0.0835,
+      "step": 13729
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.2882892422249237e-05,
+      "loss": 0.0002,
+      "step": 13730
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.2881946696524894e-05,
+      "loss": 0.0562,
+      "step": 13731
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.288100092751633e-05,
+      "loss": 0.334,
+      "step": 13732
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.2880055115228744e-05,
+      "loss": 0.0004,
+      "step": 13733
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.287910925966733e-05,
+      "loss": 0.0029,
+      "step": 13734
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 2.2878163360837275e-05,
+      "loss": 0.0042,
+      "step": 13735
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.2877217418743783e-05,
+      "loss": 0.0,
+      "step": 13736
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.287627143339204e-05,
+      "loss": 0.0002,
+      "step": 13737
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.2875325404787257e-05,
+      "loss": 0.0001,
+      "step": 13738
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.287437933293461e-05,
+      "loss": 0.0002,
+      "step": 13739
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.28734332178393e-05,
+      "loss": 0.3867,
+      "step": 13740
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.2872487059506527e-05,
+      "loss": 0.0042,
+      "step": 13741
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.2871540857941487e-05,
+      "loss": 0.0001,
+      "step": 13742
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.287059461314937e-05,
+      "loss": 0.0452,
+      "step": 13743
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.2869648325135376e-05,
+      "loss": 0.0,
+      "step": 13744
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.286870199390471e-05,
+      "loss": 0.0,
+      "step": 13745
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.2867755619462555e-05,
+      "loss": 0.0815,
+      "step": 13746
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.2866809201814108e-05,
+      "loss": 0.0165,
+      "step": 13747
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.286586274096458e-05,
+      "loss": 0.0,
+      "step": 13748
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.286491623691916e-05,
+      "loss": 0.0001,
+      "step": 13749
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.2863969689683044e-05,
+      "loss": 0.1504,
+      "step": 13750
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.2863023099261434e-05,
+      "loss": 0.0002,
+      "step": 13751
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.2862076465659524e-05,
+      "loss": 0.0271,
+      "step": 13752
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.2861129788882522e-05,
+      "loss": 0.0723,
+      "step": 13753
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.2860183068935612e-05,
+      "loss": 0.0082,
+      "step": 13754
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.2859236305824e-05,
+      "loss": 0.0,
+      "step": 13755
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.2858289499552898e-05,
+      "loss": 0.0,
+      "step": 13756
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.2857342650127484e-05,
+      "loss": 0.0,
+      "step": 13757
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.2856395757552966e-05,
+      "loss": 0.0,
+      "step": 13758
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.2855448821834553e-05,
+      "loss": 0.0001,
+      "step": 13759
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.2854501842977437e-05,
+      "loss": 0.0,
+      "step": 13760
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.2853554820986808e-05,
+      "loss": 0.0,
+      "step": 13761
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.2852607755867887e-05,
+      "loss": 0.0,
+      "step": 13762
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.285166064762586e-05,
+      "loss": 0.0,
+      "step": 13763
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.2850713496265938e-05,
+      "loss": 0.0004,
+      "step": 13764
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.2849766301793312e-05,
+      "loss": 0.0693,
+      "step": 13765
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.2848819064213194e-05,
+      "loss": 0.1543,
+      "step": 13766
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.284787178353078e-05,
+      "loss": 1.375,
+      "step": 13767
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.284692445975127e-05,
+      "loss": 0.3105,
+      "step": 13768
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.2845977092879875e-05,
+      "loss": 0.4824,
+      "step": 13769
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.284502968292179e-05,
+      "loss": 0.0195,
+      "step": 13770
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.284408222988222e-05,
+      "loss": 0.2832,
+      "step": 13771
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.2843134733766368e-05,
+      "loss": 0.2266,
+      "step": 13772
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.2842187194579434e-05,
+      "loss": 0.0126,
+      "step": 13773
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.2841239612326624e-05,
+      "loss": 0.0742,
+      "step": 13774
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.2840291987013145e-05,
+      "loss": 0.025,
+      "step": 13775
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.28393443186442e-05,
+      "loss": 0.003,
+      "step": 13776
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.2838396607224984e-05,
+      "loss": 0.207,
+      "step": 13777
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.2837448852760717e-05,
+      "loss": 0.8359,
+      "step": 13778
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.2836501055256587e-05,
+      "loss": 0.0214,
+      "step": 13779
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.2835553214717814e-05,
+      "loss": 0.3691,
+      "step": 13780
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.2834605331149598e-05,
+      "loss": 0.0001,
+      "step": 13781
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.2833657404557136e-05,
+      "loss": 0.0002,
+      "step": 13782
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.2832709434945642e-05,
+      "loss": 0.0085,
+      "step": 13783
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.2831761422320325e-05,
+      "loss": 0.4531,
+      "step": 13784
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.2830813366686384e-05,
+      "loss": 0.0,
+      "step": 13785
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.2829865268049025e-05,
+      "loss": 0.0024,
+      "step": 13786
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.2828917126413458e-05,
+      "loss": 0.0001,
+      "step": 13787
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.282796894178489e-05,
+      "loss": 0.0203,
+      "step": 13788
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.2827020714168526e-05,
+      "loss": 0.0004,
+      "step": 13789
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.2826072443569574e-05,
+      "loss": 0.0068,
+      "step": 13790
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.2825124129993244e-05,
+      "loss": 0.0752,
+      "step": 13791
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.2824175773444736e-05,
+      "loss": 0.0074,
+      "step": 13792
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.2823227373929267e-05,
+      "loss": 0.0128,
+      "step": 13793
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.2822278931452042e-05,
+      "loss": 0.0,
+      "step": 13794
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.282133044601827e-05,
+      "loss": 0.0001,
+      "step": 13795
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.2820381917633158e-05,
+      "loss": 0.0942,
+      "step": 13796
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.281943334630191e-05,
+      "loss": 0.0082,
+      "step": 13797
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.2818484732029747e-05,
+      "loss": 0.0,
+      "step": 13798
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.281753607482187e-05,
+      "loss": 0.1143,
+      "step": 13799
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.281658737468349e-05,
+      "loss": 0.0,
+      "step": 13800
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.281563863161982e-05,
+      "loss": 0.0,
+      "step": 13801
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.2814689845636065e-05,
+      "loss": 0.0,
+      "step": 13802
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 2.2813741016737436e-05,
+      "loss": 0.0,
+      "step": 13803
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.281279214492915e-05,
+      "loss": 0.0,
+      "step": 13804
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.281184323021641e-05,
+      "loss": 0.0,
+      "step": 13805
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.2810894272604434e-05,
+      "loss": 0.0,
+      "step": 13806
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.2809945272098424e-05,
+      "loss": 0.0009,
+      "step": 13807
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.2808996228703606e-05,
+      "loss": 0.0,
+      "step": 13808
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.2808047142425173e-05,
+      "loss": 0.0,
+      "step": 13809
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.2807098013268356e-05,
+      "loss": 0.1484,
+      "step": 13810
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.2806148841238347e-05,
+      "loss": 0.0,
+      "step": 13811
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.2805199626340377e-05,
+      "loss": 0.0776,
+      "step": 13812
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.280425036857965e-05,
+      "loss": 0.0,
+      "step": 13813
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.2803301067961375e-05,
+      "loss": 0.0,
+      "step": 13814
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.2802351724490773e-05,
+      "loss": 0.126,
+      "step": 13815
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.280140233817306e-05,
+      "loss": 0.166,
+      "step": 13816
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.280045290901344e-05,
+      "loss": 0.7969,
+      "step": 13817
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.2799503437017127e-05,
+      "loss": 0.1855,
+      "step": 13818
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.2798553922189344e-05,
+      "loss": 0.1641,
+      "step": 13819
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.27976043645353e-05,
+      "loss": 0.5,
+      "step": 13820
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.2796654764060208e-05,
+      "loss": 0.7617,
+      "step": 13821
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.2795705120769283e-05,
+      "loss": 0.0103,
+      "step": 13822
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.2794755434667748e-05,
+      "loss": 0.3125,
+      "step": 13823
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.2793805705760807e-05,
+      "loss": 0.2451,
+      "step": 13824
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.2792855934053684e-05,
+      "loss": 0.4512,
+      "step": 13825
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.2791906119551586e-05,
+      "loss": 0.0046,
+      "step": 13826
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.279095626225974e-05,
+      "loss": 0.0435,
+      "step": 13827
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.2790006362183357e-05,
+      "loss": 0.4434,
+      "step": 13828
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.2789056419327646e-05,
+      "loss": 0.1611,
+      "step": 13829
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.2788106433697836e-05,
+      "loss": 0.8047,
+      "step": 13830
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.2787156405299133e-05,
+      "loss": 0.7227,
+      "step": 13831
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.2786206334136766e-05,
+      "loss": 0.5938,
+      "step": 13832
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.278525622021594e-05,
+      "loss": 0.042,
+      "step": 13833
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.2784306063541882e-05,
+      "loss": 0.0004,
+      "step": 13834
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.2783355864119807e-05,
+      "loss": 0.0181,
+      "step": 13835
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.2782405621954935e-05,
+      "loss": 0.0001,
+      "step": 13836
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.2781455337052474e-05,
+      "loss": 0.3379,
+      "step": 13837
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.2780505009417655e-05,
+      "loss": 0.0591,
+      "step": 13838
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.277955463905569e-05,
+      "loss": 0.0043,
+      "step": 13839
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.2778604225971802e-05,
+      "loss": 0.0001,
+      "step": 13840
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.277765377017121e-05,
+      "loss": 0.0001,
+      "step": 13841
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.2776703271659133e-05,
+      "loss": 0.0,
+      "step": 13842
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.2775752730440785e-05,
+      "loss": 0.0053,
+      "step": 13843
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.277480214652139e-05,
+      "loss": 0.0,
+      "step": 13844
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.2773851519906175e-05,
+      "loss": 0.0071,
+      "step": 13845
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.2772900850600345e-05,
+      "loss": 0.0,
+      "step": 13846
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.2771950138609137e-05,
+      "loss": 0.1494,
+      "step": 13847
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.277099938393776e-05,
+      "loss": 0.0,
+      "step": 13848
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.2770048586591446e-05,
+      "loss": 0.0095,
+      "step": 13849
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.2769097746575406e-05,
+      "loss": 0.0,
+      "step": 13850
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.2768146863894868e-05,
+      "loss": 0.0,
+      "step": 13851
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.276719593855505e-05,
+      "loss": 0.0129,
+      "step": 13852
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.276624497056118e-05,
+      "loss": 0.0,
+      "step": 13853
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.2765293959918473e-05,
+      "loss": 0.0001,
+      "step": 13854
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.2764342906632154e-05,
+      "loss": 0.1318,
+      "step": 13855
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.2763391810707443e-05,
+      "loss": 0.0045,
+      "step": 13856
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.276244067214957e-05,
+      "loss": 0.0001,
+      "step": 13857
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.2761489490963754e-05,
+      "loss": 0.0,
+      "step": 13858
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.276053826715522e-05,
+      "loss": 0.0,
+      "step": 13859
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.275958700072919e-05,
+      "loss": 0.0,
+      "step": 13860
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.2758635691690888e-05,
+      "loss": 0.0001,
+      "step": 13861
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.2757684340045538e-05,
+      "loss": 0.0,
+      "step": 13862
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.275673294579837e-05,
+      "loss": 0.0559,
+      "step": 13863
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.27557815089546e-05,
+      "loss": 0.0889,
+      "step": 13864
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.2754830029519457e-05,
+      "loss": 0.1621,
+      "step": 13865
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.2753878507498166e-05,
+      "loss": 0.1299,
+      "step": 13866
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.2752926942895957e-05,
+      "loss": 0.332,
+      "step": 13867
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.2751975335718046e-05,
+      "loss": 0.4531,
+      "step": 13868
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.2751023685969665e-05,
+      "loss": 0.2949,
+      "step": 13869
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.2750071993656036e-05,
+      "loss": 0.0087,
+      "step": 13870
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 2.2749120258782392e-05,
+      "loss": 1.3828,
+      "step": 13871
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.2748168481353953e-05,
+      "loss": 0.0167,
+      "step": 13872
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.2747216661375948e-05,
+      "loss": 0.1426,
+      "step": 13873
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.274626479885361e-05,
+      "loss": 0.8945,
+      "step": 13874
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.274531289379215e-05,
+      "loss": 0.459,
+      "step": 13875
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.274436094619681e-05,
+      "loss": 1.5703,
+      "step": 13876
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.2743408956072818e-05,
+      "loss": 0.5625,
+      "step": 13877
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.2742456923425396e-05,
+      "loss": 0.3516,
+      "step": 13878
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.274150484825977e-05,
+      "loss": 0.003,
+      "step": 13879
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.274055273058118e-05,
+      "loss": 0.5,
+      "step": 13880
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.2739600570394838e-05,
+      "loss": 0.0322,
+      "step": 13881
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.2738648367705982e-05,
+      "loss": 0.0192,
+      "step": 13882
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.2737696122519843e-05,
+      "loss": 0.0349,
+      "step": 13883
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.273674383484165e-05,
+      "loss": 0.1475,
+      "step": 13884
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.2735791504676628e-05,
+      "loss": 0.0,
+      "step": 13885
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.273483913203001e-05,
+      "loss": 0.0466,
+      "step": 13886
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.273388671690702e-05,
+      "loss": 0.0248,
+      "step": 13887
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.27329342593129e-05,
+      "loss": 0.0,
+      "step": 13888
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.2731981759252876e-05,
+      "loss": 0.8516,
+      "step": 13889
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.273102921673217e-05,
+      "loss": 0.0001,
+      "step": 13890
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.2730076631756023e-05,
+      "loss": 0.0003,
+      "step": 13891
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.2729124004329668e-05,
+      "loss": 0.4297,
+      "step": 13892
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.2728171334458323e-05,
+      "loss": 0.0024,
+      "step": 13893
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.272721862214723e-05,
+      "loss": 0.7969,
+      "step": 13894
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.2726265867401617e-05,
+      "loss": 0.0952,
+      "step": 13895
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.272531307022672e-05,
+      "loss": 0.0066,
+      "step": 13896
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.272436023062777e-05,
+      "loss": 0.0014,
+      "step": 13897
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.272340734861e-05,
+      "loss": 0.0,
+      "step": 13898
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.272245442417864e-05,
+      "loss": 0.0,
+      "step": 13899
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.272150145733892e-05,
+      "loss": 0.0019,
+      "step": 13900
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.2720548448096084e-05,
+      "loss": 0.0,
+      "step": 13901
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.271959539645536e-05,
+      "loss": 0.0001,
+      "step": 13902
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.2718642302421978e-05,
+      "loss": 0.0,
+      "step": 13903
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.2717689166001172e-05,
+      "loss": 1.1719,
+      "step": 13904
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.271673598719819e-05,
+      "loss": 0.0001,
+      "step": 13905
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.2715782766018248e-05,
+      "loss": 0.0315,
+      "step": 13906
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.2714829502466586e-05,
+      "loss": 0.0327,
+      "step": 13907
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.2713876196548446e-05,
+      "loss": 0.0,
+      "step": 13908
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.2712922848269057e-05,
+      "loss": 0.0,
+      "step": 13909
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.2711969457633657e-05,
+      "loss": 0.0,
+      "step": 13910
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.271101602464748e-05,
+      "loss": 0.3926,
+      "step": 13911
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.2710062549315765e-05,
+      "loss": 0.0,
+      "step": 13912
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.2709109031643742e-05,
+      "loss": 0.0,
+      "step": 13913
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.270815547163665e-05,
+      "loss": 0.0,
+      "step": 13914
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.2707201869299727e-05,
+      "loss": 0.0,
+      "step": 13915
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.270624822463821e-05,
+      "loss": 0.1631,
+      "step": 13916
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.270529453765734e-05,
+      "loss": 0.0142,
+      "step": 13917
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.2704340808362344e-05,
+      "loss": 0.0518,
+      "step": 13918
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.270338703675846e-05,
+      "loss": 0.3887,
+      "step": 13919
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.2702433222850938e-05,
+      "loss": 0.1611,
+      "step": 13920
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.2701479366645003e-05,
+      "loss": 0.2051,
+      "step": 13921
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.27005254681459e-05,
+      "loss": 0.0001,
+      "step": 13922
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.269957152735887e-05,
+      "loss": 0.0908,
+      "step": 13923
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.2698617544289148e-05,
+      "loss": 0.0781,
+      "step": 13924
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.269766351894197e-05,
+      "loss": 1.7734,
+      "step": 13925
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.2696709451322573e-05,
+      "loss": 0.7461,
+      "step": 13926
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.2695755341436207e-05,
+      "loss": 0.4863,
+      "step": 13927
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.26948011892881e-05,
+      "loss": 0.0,
+      "step": 13928
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.2693846994883502e-05,
+      "loss": 1.8984,
+      "step": 13929
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.2692892758227647e-05,
+      "loss": 0.0422,
+      "step": 13930
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.2691938479325773e-05,
+      "loss": 0.0,
+      "step": 13931
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.2690984158183128e-05,
+      "loss": 0.0149,
+      "step": 13932
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.269002979480495e-05,
+      "loss": 0.1143,
+      "step": 13933
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.268907538919647e-05,
+      "loss": 0.4707,
+      "step": 13934
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.2688120941362946e-05,
+      "loss": 0.1094,
+      "step": 13935
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.268716645130961e-05,
+      "loss": 0.334,
+      "step": 13936
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.2686211919041705e-05,
+      "loss": 0.3066,
+      "step": 13937
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.2685257344564472e-05,
+      "loss": 0.0001,
+      "step": 13938
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 2.2684302727883148e-05,
+      "loss": 0.2852,
+      "step": 13939
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.268334806900299e-05,
+      "loss": 0.0,
+      "step": 13940
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.2682393367929227e-05,
+      "loss": 1.0703,
+      "step": 13941
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.2681438624667107e-05,
+      "loss": 0.0,
+      "step": 13942
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.2680483839221875e-05,
+      "loss": 0.0002,
+      "step": 13943
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.267952901159877e-05,
+      "loss": 0.0,
+      "step": 13944
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.2678574141803038e-05,
+      "loss": 0.0,
+      "step": 13945
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.267761922983992e-05,
+      "loss": 0.1245,
+      "step": 13946
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.2676664275714663e-05,
+      "loss": 0.0033,
+      "step": 13947
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.2675709279432512e-05,
+      "loss": 0.0,
+      "step": 13948
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.2674754240998708e-05,
+      "loss": 0.0,
+      "step": 13949
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.2673799160418496e-05,
+      "loss": 0.0072,
+      "step": 13950
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.267284403769712e-05,
+      "loss": 0.0,
+      "step": 13951
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.2671888872839835e-05,
+      "loss": 0.0,
+      "step": 13952
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.2670933665851875e-05,
+      "loss": 0.0001,
+      "step": 13953
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.2669978416738487e-05,
+      "loss": 0.0,
+      "step": 13954
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.266902312550492e-05,
+      "loss": 0.0,
+      "step": 13955
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.2668067792156424e-05,
+      "loss": 0.0,
+      "step": 13956
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.2667112416698232e-05,
+      "loss": 0.0,
+      "step": 13957
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.266615699913561e-05,
+      "loss": 0.0,
+      "step": 13958
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.2665201539473787e-05,
+      "loss": 0.6055,
+      "step": 13959
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.2664246037718013e-05,
+      "loss": 0.0,
+      "step": 13960
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.2663290493873537e-05,
+      "loss": 0.4688,
+      "step": 13961
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.266233490794562e-05,
+      "loss": 0.1553,
+      "step": 13962
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.266137927993949e-05,
+      "loss": 0.1128,
+      "step": 13963
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.26604236098604e-05,
+      "loss": 0.1191,
+      "step": 13964
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.2659467897713604e-05,
+      "loss": 0.1367,
+      "step": 13965
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.265851214350435e-05,
+      "loss": 0.103,
+      "step": 13966
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.265755634723788e-05,
+      "loss": 0.2422,
+      "step": 13967
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.2656600508919447e-05,
+      "loss": 0.4062,
+      "step": 13968
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.26556446285543e-05,
+      "loss": 0.4004,
+      "step": 13969
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.265468870614769e-05,
+      "loss": 0.0193,
+      "step": 13970
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.265373274170486e-05,
+      "loss": 0.0,
+      "step": 13971
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.265277673523107e-05,
+      "loss": 0.4062,
+      "step": 13972
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.2651820686731562e-05,
+      "loss": 0.4238,
+      "step": 13973
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.2650864596211592e-05,
+      "loss": 0.21,
+      "step": 13974
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.2649908463676403e-05,
+      "loss": 0.0781,
+      "step": 13975
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.264895228913125e-05,
+      "loss": 0.6992,
+      "step": 13976
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.2647996072581382e-05,
+      "loss": 0.1074,
+      "step": 13977
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.2647039814032047e-05,
+      "loss": 0.0258,
+      "step": 13978
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.264608351348851e-05,
+      "loss": 0.1816,
+      "step": 13979
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.264512717095601e-05,
+      "loss": 0.0018,
+      "step": 13980
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.2644170786439802e-05,
+      "loss": 0.0635,
+      "step": 13981
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.2643214359945135e-05,
+      "loss": 0.0,
+      "step": 13982
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.2642257891477268e-05,
+      "loss": 0.2656,
+      "step": 13983
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.2641301381041453e-05,
+      "loss": 1.1172,
+      "step": 13984
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.264034482864293e-05,
+      "loss": 0.0859,
+      "step": 13985
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.263938823428697e-05,
+      "loss": 0.0552,
+      "step": 13986
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.2638431597978818e-05,
+      "loss": 0.0,
+      "step": 13987
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.2637474919723726e-05,
+      "loss": 0.0028,
+      "step": 13988
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.2636518199526947e-05,
+      "loss": 0.0003,
+      "step": 13989
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.2635561437393737e-05,
+      "loss": 0.375,
+      "step": 13990
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.263460463332935e-05,
+      "loss": 0.0014,
+      "step": 13991
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.2633647787339047e-05,
+      "loss": 0.0,
+      "step": 13992
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.2632690899428068e-05,
+      "loss": 0.0,
+      "step": 13993
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.263173396960168e-05,
+      "loss": 0.0,
+      "step": 13994
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.2630776997865135e-05,
+      "loss": 0.051,
+      "step": 13995
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.2629819984223682e-05,
+      "loss": 0.0003,
+      "step": 13996
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.2628862928682582e-05,
+      "loss": 0.0,
+      "step": 13997
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.2627905831247097e-05,
+      "loss": 0.0,
+      "step": 13998
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.2626948691922475e-05,
+      "loss": 0.0003,
+      "step": 13999
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.2625991510713967e-05,
+      "loss": 0.0001,
+      "step": 14000
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.2625034287626836e-05,
+      "loss": 0.2129,
+      "step": 14001
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.2624077022666344e-05,
+      "loss": 0.0,
+      "step": 14002
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.2623119715837744e-05,
+      "loss": 0.0004,
+      "step": 14003
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.2622162367146286e-05,
+      "loss": 0.0,
+      "step": 14004
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.2621204976597235e-05,
+      "loss": 0.0,
+      "step": 14005
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 2.262024754419585e-05,
+      "loss": 0.3691,
+      "step": 14006
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.2619290069947383e-05,
+      "loss": 0.0,
+      "step": 14007
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.261833255385709e-05,
+      "loss": 0.1416,
+      "step": 14008
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.261737499593024e-05,
+      "loss": 0.1245,
+      "step": 14009
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.261641739617208e-05,
+      "loss": 0.1221,
+      "step": 14010
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.2615459754587875e-05,
+      "loss": 0.0,
+      "step": 14011
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.2614502071182884e-05,
+      "loss": 0.0,
+      "step": 14012
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.2613544345962364e-05,
+      "loss": 0.165,
+      "step": 14013
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.2612586578931576e-05,
+      "loss": 0.1533,
+      "step": 14014
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.2611628770095778e-05,
+      "loss": 0.1021,
+      "step": 14015
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.2610670919460232e-05,
+      "loss": 0.1406,
+      "step": 14016
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.26097130270302e-05,
+      "loss": 1.5,
+      "step": 14017
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.2608755092810935e-05,
+      "loss": 0.6406,
+      "step": 14018
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.2607797116807703e-05,
+      "loss": 0.0679,
+      "step": 14019
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.2606839099025763e-05,
+      "loss": 0.4785,
+      "step": 14020
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.2605881039470375e-05,
+      "loss": 0.0996,
+      "step": 14021
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.2604922938146805e-05,
+      "loss": 0.6289,
+      "step": 14022
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.260396479506031e-05,
+      "loss": 0.7031,
+      "step": 14023
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.2603006610216155e-05,
+      "loss": 0.9453,
+      "step": 14024
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.26020483836196e-05,
+      "loss": 0.2148,
+      "step": 14025
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.2601090115275904e-05,
+      "loss": 0.3535,
+      "step": 14026
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.260013180519034e-05,
+      "loss": 0.7812,
+      "step": 14027
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.2599173453368152e-05,
+      "loss": 0.0126,
+      "step": 14028
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.2598215059814618e-05,
+      "loss": 0.0374,
+      "step": 14029
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.2597256624535e-05,
+      "loss": 0.5195,
+      "step": 14030
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.259629814753456e-05,
+      "loss": 0.0228,
+      "step": 14031
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.2595339628818555e-05,
+      "loss": 0.1367,
+      "step": 14032
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.259438106839226e-05,
+      "loss": 0.0056,
+      "step": 14033
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.2593422466260924e-05,
+      "loss": 0.6562,
+      "step": 14034
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.2592463822429828e-05,
+      "loss": 0.0884,
+      "step": 14035
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.2591505136904223e-05,
+      "loss": 0.0,
+      "step": 14036
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.2590546409689378e-05,
+      "loss": 0.0001,
+      "step": 14037
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.2589587640790564e-05,
+      "loss": 0.0,
+      "step": 14038
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.2588628830213037e-05,
+      "loss": 0.0036,
+      "step": 14039
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.2587669977962075e-05,
+      "loss": 0.0,
+      "step": 14040
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.2586711084042926e-05,
+      "loss": 0.7656,
+      "step": 14041
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.2585752148460866e-05,
+      "loss": 0.0591,
+      "step": 14042
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.258479317122116e-05,
+      "loss": 0.0,
+      "step": 14043
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.2583834152329082e-05,
+      "loss": 0.2021,
+      "step": 14044
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.2582875091789883e-05,
+      "loss": 0.1465,
+      "step": 14045
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.2581915989608835e-05,
+      "loss": 0.0036,
+      "step": 14046
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.2580956845791216e-05,
+      "loss": 0.0,
+      "step": 14047
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.2579997660342277e-05,
+      "loss": 0.0962,
+      "step": 14048
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.25790384332673e-05,
+      "loss": 0.0001,
+      "step": 14049
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.257807916457154e-05,
+      "loss": 0.0046,
+      "step": 14050
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.2577119854260273e-05,
+      "loss": 0.0,
+      "step": 14051
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.2576160502338764e-05,
+      "loss": 0.0,
+      "step": 14052
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.257520110881228e-05,
+      "loss": 0.0,
+      "step": 14053
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.2574241673686096e-05,
+      "loss": 0.0,
+      "step": 14054
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.2573282196965478e-05,
+      "loss": 0.0,
+      "step": 14055
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.2572322678655693e-05,
+      "loss": 0.0,
+      "step": 14056
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.2571363118762006e-05,
+      "loss": 0.0,
+      "step": 14057
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.2570403517289694e-05,
+      "loss": 0.0197,
+      "step": 14058
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.2569443874244025e-05,
+      "loss": 0.0,
+      "step": 14059
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.2568484189630265e-05,
+      "loss": 0.1035,
+      "step": 14060
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.256752446345369e-05,
+      "loss": 0.1099,
+      "step": 14061
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.2566564695719568e-05,
+      "loss": 0.1338,
+      "step": 14062
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.256560488643317e-05,
+      "loss": 0.1406,
+      "step": 14063
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.256464503559976e-05,
+      "loss": 0.125,
+      "step": 14064
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.256368514322462e-05,
+      "loss": 0.1719,
+      "step": 14065
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.2562725209313017e-05,
+      "loss": 0.1104,
+      "step": 14066
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.2561765233870217e-05,
+      "loss": 0.7695,
+      "step": 14067
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.25608052169015e-05,
+      "loss": 0.1318,
+      "step": 14068
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.2559845158412136e-05,
+      "loss": 0.0569,
+      "step": 14069
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.255888505840739e-05,
+      "loss": 0.7891,
+      "step": 14070
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.2557924916892544e-05,
+      "loss": 0.0089,
+      "step": 14071
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.255696473387287e-05,
+      "loss": 0.1104,
+      "step": 14072
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.255600450935363e-05,
+      "loss": 0.0001,
+      "step": 14073
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 2.2555044243340115e-05,
+      "loss": 0.5078,
+      "step": 14074
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.255408393583758e-05,
+      "loss": 0.1006,
+      "step": 14075
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.2553123586851313e-05,
+      "loss": 0.2344,
+      "step": 14076
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.2552163196386584e-05,
+      "loss": 0.4062,
+      "step": 14077
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.255120276444866e-05,
+      "loss": 0.3301,
+      "step": 14078
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.255024229104282e-05,
+      "loss": 0.0262,
+      "step": 14079
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.2549281776174337e-05,
+      "loss": 0.0742,
+      "step": 14080
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.2548321219848496e-05,
+      "loss": 0.0,
+      "step": 14081
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.2547360622070555e-05,
+      "loss": 0.0,
+      "step": 14082
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.2546399982845795e-05,
+      "loss": 0.0004,
+      "step": 14083
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.25454393021795e-05,
+      "loss": 0.0063,
+      "step": 14084
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.254447858007694e-05,
+      "loss": 0.0,
+      "step": 14085
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.2543517816543385e-05,
+      "loss": 0.0004,
+      "step": 14086
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.2542557011584127e-05,
+      "loss": 0.0209,
+      "step": 14087
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.2541596165204425e-05,
+      "loss": 0.0,
+      "step": 14088
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.254063527740956e-05,
+      "loss": 0.026,
+      "step": 14089
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.2539674348204812e-05,
+      "loss": 0.0,
+      "step": 14090
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.2538713377595464e-05,
+      "loss": 0.0003,
+      "step": 14091
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.2537752365586777e-05,
+      "loss": 0.1807,
+      "step": 14092
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.253679131218404e-05,
+      "loss": 0.0001,
+      "step": 14093
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.2535830217392538e-05,
+      "loss": 0.0001,
+      "step": 14094
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.253486908121753e-05,
+      "loss": 0.0,
+      "step": 14095
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.2533907903664303e-05,
+      "loss": 0.0,
+      "step": 14096
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.2532946684738143e-05,
+      "loss": 0.0,
+      "step": 14097
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.2531985424444316e-05,
+      "loss": 0.0,
+      "step": 14098
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.253102412278811e-05,
+      "loss": 0.4141,
+      "step": 14099
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.2530062779774797e-05,
+      "loss": 0.0,
+      "step": 14100
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.252910139540966e-05,
+      "loss": 0.2471,
+      "step": 14101
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.252813996969798e-05,
+      "loss": 0.0,
+      "step": 14102
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.2527178502645034e-05,
+      "loss": 0.0,
+      "step": 14103
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.25262169942561e-05,
+      "loss": 0.0,
+      "step": 14104
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.2525255444536466e-05,
+      "loss": 0.0,
+      "step": 14105
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.2524293853491412e-05,
+      "loss": 0.0,
+      "step": 14106
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.2523332221126204e-05,
+      "loss": 0.0,
+      "step": 14107
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.252237054744614e-05,
+      "loss": 0.0,
+      "step": 14108
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.252140883245649e-05,
+      "loss": 0.0,
+      "step": 14109
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.252044707616254e-05,
+      "loss": 0.0,
+      "step": 14110
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.251948527856957e-05,
+      "loss": 0.1465,
+      "step": 14111
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.2518523439682867e-05,
+      "loss": 0.0,
+      "step": 14112
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.2517561559507708e-05,
+      "loss": 0.0,
+      "step": 14113
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.2516599638049375e-05,
+      "loss": 0.1602,
+      "step": 14114
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.2515637675313148e-05,
+      "loss": 0.0,
+      "step": 14115
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.251467567130432e-05,
+      "loss": 0.1689,
+      "step": 14116
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.251371362602816e-05,
+      "loss": 0.8633,
+      "step": 14117
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.251275153948996e-05,
+      "loss": 0.6328,
+      "step": 14118
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.2511789411695004e-05,
+      "loss": 0.6758,
+      "step": 14119
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.251082724264857e-05,
+      "loss": 0.5,
+      "step": 14120
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.2509865032355944e-05,
+      "loss": 0.1543,
+      "step": 14121
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.2508902780822417e-05,
+      "loss": 0.2451,
+      "step": 14122
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.2507940488053264e-05,
+      "loss": 0.2412,
+      "step": 14123
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.2506978154053775e-05,
+      "loss": 0.4766,
+      "step": 14124
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.250601577882923e-05,
+      "loss": 1.1172,
+      "step": 14125
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.2505053362384917e-05,
+      "loss": 0.457,
+      "step": 14126
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.250409090472612e-05,
+      "loss": 0.3867,
+      "step": 14127
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.250312840585812e-05,
+      "loss": 1.25,
+      "step": 14128
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.250216586578622e-05,
+      "loss": 0.2422,
+      "step": 14129
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.2501203284515682e-05,
+      "loss": 0.1035,
+      "step": 14130
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.2500240662051812e-05,
+      "loss": 0.2158,
+      "step": 14131
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.249927799839988e-05,
+      "loss": 0.6328,
+      "step": 14132
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.2498315293565182e-05,
+      "loss": 0.0342,
+      "step": 14133
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.2497352547553008e-05,
+      "loss": 0.3848,
+      "step": 14134
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.2496389760368636e-05,
+      "loss": 1.1562,
+      "step": 14135
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.249542693201736e-05,
+      "loss": 0.0085,
+      "step": 14136
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.2494464062504466e-05,
+      "loss": 0.0474,
+      "step": 14137
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.249350115183524e-05,
+      "loss": 0.0219,
+      "step": 14138
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.249253820001496e-05,
+      "loss": 0.0057,
+      "step": 14139
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.2491575207048932e-05,
+      "loss": 0.0,
+      "step": 14140
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.249061217294244e-05,
+      "loss": 0.3906,
+      "step": 14141
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 2.2489649097700764e-05,
+      "loss": 0.0,
+      "step": 14142
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.2488685981329204e-05,
+      "loss": 0.01,
+      "step": 14143
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.248772282383304e-05,
+      "loss": 0.0,
+      "step": 14144
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.248675962521756e-05,
+      "loss": 0.0003,
+      "step": 14145
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.2485796385488065e-05,
+      "loss": 0.0,
+      "step": 14146
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.2484833104649832e-05,
+      "loss": 0.2256,
+      "step": 14147
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.248386978270816e-05,
+      "loss": 0.032,
+      "step": 14148
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.2482906419668333e-05,
+      "loss": 0.011,
+      "step": 14149
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.2481943015535645e-05,
+      "loss": 0.0063,
+      "step": 14150
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.2480979570315384e-05,
+      "loss": 0.0,
+      "step": 14151
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.2480016084012843e-05,
+      "loss": 0.0669,
+      "step": 14152
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.2479052556633314e-05,
+      "loss": 0.0,
+      "step": 14153
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.247808898818209e-05,
+      "loss": 0.0,
+      "step": 14154
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.2477125378664453e-05,
+      "loss": 0.0,
+      "step": 14155
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.2476161728085703e-05,
+      "loss": 0.083,
+      "step": 14156
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.2475198036451127e-05,
+      "loss": 0.0,
+      "step": 14157
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.2474234303766027e-05,
+      "loss": 0.0,
+      "step": 14158
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.2473270530035683e-05,
+      "loss": 0.0,
+      "step": 14159
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.2472306715265393e-05,
+      "loss": 0.0,
+      "step": 14160
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.247134285946045e-05,
+      "loss": 0.0,
+      "step": 14161
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.247037896262615e-05,
+      "loss": 0.0,
+      "step": 14162
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.2469415024767782e-05,
+      "loss": 0.0,
+      "step": 14163
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.2468451045890642e-05,
+      "loss": 0.0,
+      "step": 14164
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.246748702600002e-05,
+      "loss": 0.0791,
+      "step": 14165
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.2466522965101215e-05,
+      "loss": 0.0698,
+      "step": 14166
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.2465558863199515e-05,
+      "loss": 0.2148,
+      "step": 14167
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.246459472030022e-05,
+      "loss": 0.2793,
+      "step": 14168
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.2463630536408623e-05,
+      "loss": 0.1514,
+      "step": 14169
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.2462666311530022e-05,
+      "loss": 0.0234,
+      "step": 14170
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.2461702045669703e-05,
+      "loss": 0.3672,
+      "step": 14171
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.2460737738832967e-05,
+      "loss": 0.2324,
+      "step": 14172
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.2459773391025114e-05,
+      "loss": 0.0325,
+      "step": 14173
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.2458809002251434e-05,
+      "loss": 0.1836,
+      "step": 14174
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.2457844572517224e-05,
+      "loss": 0.4688,
+      "step": 14175
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.245688010182778e-05,
+      "loss": 0.0942,
+      "step": 14176
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.24559155901884e-05,
+      "loss": 0.1187,
+      "step": 14177
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.245495103760438e-05,
+      "loss": 0.3711,
+      "step": 14178
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.2453986444081018e-05,
+      "loss": 0.0703,
+      "step": 14179
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.2453021809623606e-05,
+      "loss": 0.8203,
+      "step": 14180
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.245205713423745e-05,
+      "loss": 0.0139,
+      "step": 14181
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.245109241792784e-05,
+      "loss": 0.6367,
+      "step": 14182
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.2450127660700072e-05,
+      "loss": 0.0,
+      "step": 14183
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.244916286255946e-05,
+      "loss": 0.5547,
+      "step": 14184
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.244819802351128e-05,
+      "loss": 0.7305,
+      "step": 14185
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.2447233143560844e-05,
+      "loss": 0.0261,
+      "step": 14186
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.2446268222713454e-05,
+      "loss": 1.1953,
+      "step": 14187
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.24453032609744e-05,
+      "loss": 0.0211,
+      "step": 14188
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.244433825834898e-05,
+      "loss": 0.0017,
+      "step": 14189
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.24433732148425e-05,
+      "loss": 0.0004,
+      "step": 14190
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.2442408130460257e-05,
+      "loss": 0.5391,
+      "step": 14191
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.244144300520755e-05,
+      "loss": 0.0,
+      "step": 14192
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.2440477839089682e-05,
+      "loss": 0.0001,
+      "step": 14193
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.2439512632111954e-05,
+      "loss": 0.875,
+      "step": 14194
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.2438547384279662e-05,
+      "loss": 0.0007,
+      "step": 14195
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.2437582095598104e-05,
+      "loss": 0.0,
+      "step": 14196
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.2436616766072584e-05,
+      "loss": 0.0776,
+      "step": 14197
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.243565139570841e-05,
+      "loss": 0.0,
+      "step": 14198
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.2434685984510877e-05,
+      "loss": 0.0,
+      "step": 14199
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.2433720532485283e-05,
+      "loss": 0.0581,
+      "step": 14200
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.243275503963694e-05,
+      "loss": 0.0332,
+      "step": 14201
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.2431789505971144e-05,
+      "loss": 0.0,
+      "step": 14202
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.2430823931493193e-05,
+      "loss": 0.0,
+      "step": 14203
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.2429858316208402e-05,
+      "loss": 0.0008,
+      "step": 14204
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.2428892660122062e-05,
+      "loss": 0.0,
+      "step": 14205
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.2427926963239478e-05,
+      "loss": 0.0001,
+      "step": 14206
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.242696122556596e-05,
+      "loss": 0.0,
+      "step": 14207
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.24259954471068e-05,
+      "loss": 0.0001,
+      "step": 14208
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.2425029627867315e-05,
+      "loss": 0.2598,
+      "step": 14209
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 2.2424063767852795e-05,
+      "loss": 0.0,
+      "step": 14210
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.2423097867068553e-05,
+      "loss": 0.167,
+      "step": 14211
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.24221319255199e-05,
+      "loss": 0.0,
+      "step": 14212
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.2421165943212125e-05,
+      "loss": 0.1128,
+      "step": 14213
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.2420199920150535e-05,
+      "loss": 0.1846,
+      "step": 14214
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.2419233856340444e-05,
+      "loss": 0.123,
+      "step": 14215
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.2418267751787157e-05,
+      "loss": 0.1758,
+      "step": 14216
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.2417301606495972e-05,
+      "loss": 1.1484,
+      "step": 14217
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.2416335420472193e-05,
+      "loss": 0.0197,
+      "step": 14218
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.241536919372114e-05,
+      "loss": 0.0037,
+      "step": 14219
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.2414402926248108e-05,
+      "loss": 0.1523,
+      "step": 14220
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.2413436618058403e-05,
+      "loss": 0.064,
+      "step": 14221
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.2412470269157334e-05,
+      "loss": 0.8477,
+      "step": 14222
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.241150387955021e-05,
+      "loss": 0.0098,
+      "step": 14223
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.2410537449242327e-05,
+      "loss": 0.0003,
+      "step": 14224
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.2409570978239006e-05,
+      "loss": 0.832,
+      "step": 14225
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.2408604466545553e-05,
+      "loss": 0.1855,
+      "step": 14226
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.2407637914167267e-05,
+      "loss": 0.6992,
+      "step": 14227
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.2406671321109466e-05,
+      "loss": 0.0962,
+      "step": 14228
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.2405704687377447e-05,
+      "loss": 0.2373,
+      "step": 14229
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.2404738012976524e-05,
+      "loss": 0.022,
+      "step": 14230
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.240377129791201e-05,
+      "loss": 0.4141,
+      "step": 14231
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.2402804542189207e-05,
+      "loss": 0.0002,
+      "step": 14232
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.2401837745813432e-05,
+      "loss": 0.0001,
+      "step": 14233
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.2400870908789983e-05,
+      "loss": 0.0019,
+      "step": 14234
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.2399904031124175e-05,
+      "loss": 0.0,
+      "step": 14235
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.239893711282132e-05,
+      "loss": 0.1494,
+      "step": 14236
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.2397970153886726e-05,
+      "loss": 0.0011,
+      "step": 14237
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.2397003154325704e-05,
+      "loss": 0.0684,
+      "step": 14238
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.239603611414356e-05,
+      "loss": 0.0001,
+      "step": 14239
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.2395069033345604e-05,
+      "loss": 0.0,
+      "step": 14240
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.239410191193716e-05,
+      "loss": 0.0,
+      "step": 14241
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.2393134749923524e-05,
+      "loss": 0.0,
+      "step": 14242
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.2392167547310016e-05,
+      "loss": 0.0002,
+      "step": 14243
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.2391200304101945e-05,
+      "loss": 0.2129,
+      "step": 14244
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.2390233020304618e-05,
+      "loss": 0.0135,
+      "step": 14245
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.238926569592335e-05,
+      "loss": 0.0001,
+      "step": 14246
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.238829833096346e-05,
+      "loss": 0.0186,
+      "step": 14247
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.238733092543025e-05,
+      "loss": 0.0,
+      "step": 14248
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.2386363479329035e-05,
+      "loss": 0.1406,
+      "step": 14249
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.238539599266513e-05,
+      "loss": 0.0001,
+      "step": 14250
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.2384428465443854e-05,
+      "loss": 0.0,
+      "step": 14251
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.238346089767051e-05,
+      "loss": 0.0001,
+      "step": 14252
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.2382493289350414e-05,
+      "loss": 0.0,
+      "step": 14253
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.2381525640488883e-05,
+      "loss": 0.083,
+      "step": 14254
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.238055795109123e-05,
+      "loss": 0.0,
+      "step": 14255
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.2379590221162763e-05,
+      "loss": 0.0002,
+      "step": 14256
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.2378622450708806e-05,
+      "loss": 0.0,
+      "step": 14257
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.2377654639734673e-05,
+      "loss": 0.0,
+      "step": 14258
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.237668678824567e-05,
+      "loss": 0.1289,
+      "step": 14259
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.2375718896247112e-05,
+      "loss": 0.0811,
+      "step": 14260
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.2374750963744327e-05,
+      "loss": 0.0,
+      "step": 14261
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.2373782990742617e-05,
+      "loss": 0.0991,
+      "step": 14262
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.2372814977247307e-05,
+      "loss": 0.0,
+      "step": 14263
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.2371846923263708e-05,
+      "loss": 0.1099,
+      "step": 14264
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.2370878828797137e-05,
+      "loss": 0.1084,
+      "step": 14265
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.2369910693852912e-05,
+      "loss": 0.0002,
+      "step": 14266
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.2368942518436344e-05,
+      "loss": 1.3828,
+      "step": 14267
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.2367974302552758e-05,
+      "loss": 0.4199,
+      "step": 14268
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.236700604620747e-05,
+      "loss": 0.1826,
+      "step": 14269
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.236603774940579e-05,
+      "loss": 0.0723,
+      "step": 14270
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.236506941215304e-05,
+      "loss": 0.0003,
+      "step": 14271
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.2364101034454537e-05,
+      "loss": 0.957,
+      "step": 14272
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.2363132616315603e-05,
+      "loss": 0.2451,
+      "step": 14273
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.2362164157741545e-05,
+      "loss": 0.2305,
+      "step": 14274
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.2361195658737694e-05,
+      "loss": 0.0051,
+      "step": 14275
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.2360227119309365e-05,
+      "loss": 0.0908,
+      "step": 14276
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.2359258539461877e-05,
+      "loss": 0.0115,
+      "step": 14277
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.235828991920054e-05,
+      "loss": 0.0552,
+      "step": 14278
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.2357321258530682e-05,
+      "loss": 0.291,
+      "step": 14279
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.2356352557457624e-05,
+      "loss": 0.1367,
+      "step": 14280
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.235538381598668e-05,
+      "loss": 0.3418,
+      "step": 14281
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.235441503412317e-05,
+      "loss": 0.0327,
+      "step": 14282
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.2353446211872425e-05,
+      "loss": 0.167,
+      "step": 14283
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.235247734923975e-05,
+      "loss": 0.1455,
+      "step": 14284
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.2351508446230475e-05,
+      "loss": 0.0002,
+      "step": 14285
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.235053950284992e-05,
+      "loss": 0.0004,
+      "step": 14286
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.2349570519103405e-05,
+      "loss": 0.1963,
+      "step": 14287
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.2348601494996247e-05,
+      "loss": 0.0008,
+      "step": 14288
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.2347632430533772e-05,
+      "loss": 0.0012,
+      "step": 14289
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.23466633257213e-05,
+      "loss": 0.0001,
+      "step": 14290
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.2345694180564157e-05,
+      "loss": 0.0121,
+      "step": 14291
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.2344724995067657e-05,
+      "loss": 0.0018,
+      "step": 14292
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.234375576923713e-05,
+      "loss": 0.0018,
+      "step": 14293
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.2342786503077896e-05,
+      "loss": 0.0,
+      "step": 14294
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.2341817196595276e-05,
+      "loss": 0.0006,
+      "step": 14295
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.2340847849794596e-05,
+      "loss": 0.0,
+      "step": 14296
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.2339878462681176e-05,
+      "loss": 0.0,
+      "step": 14297
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.2338909035260344e-05,
+      "loss": 0.1406,
+      "step": 14298
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.2337939567537416e-05,
+      "loss": 0.0,
+      "step": 14299
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.2336970059517727e-05,
+      "loss": 0.3477,
+      "step": 14300
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.2336000511206594e-05,
+      "loss": 0.0,
+      "step": 14301
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.2335030922609337e-05,
+      "loss": 0.1504,
+      "step": 14302
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.233406129373129e-05,
+      "loss": 0.0,
+      "step": 14303
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.2333091624577772e-05,
+      "loss": 0.0,
+      "step": 14304
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.2332121915154112e-05,
+      "loss": 0.0001,
+      "step": 14305
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.2331152165465626e-05,
+      "loss": 0.0476,
+      "step": 14306
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.233018237551765e-05,
+      "loss": 0.0532,
+      "step": 14307
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.2329212545315508e-05,
+      "loss": 0.0,
+      "step": 14308
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.2328242674864525e-05,
+      "loss": 0.1689,
+      "step": 14309
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.232727276417002e-05,
+      "loss": 0.0,
+      "step": 14310
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.2326302813237328e-05,
+      "loss": 0.1504,
+      "step": 14311
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.2325332822071773e-05,
+      "loss": 0.1572,
+      "step": 14312
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.2324362790678682e-05,
+      "loss": 0.0,
+      "step": 14313
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.2323392719063378e-05,
+      "loss": 0.0001,
+      "step": 14314
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.2322422607231193e-05,
+      "loss": 0.167,
+      "step": 14315
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.2321452455187457e-05,
+      "loss": 0.1465,
+      "step": 14316
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.232048226293749e-05,
+      "loss": 0.1235,
+      "step": 14317
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.2319512030486628e-05,
+      "loss": 0.1406,
+      "step": 14318
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.231854175784019e-05,
+      "loss": 0.582,
+      "step": 14319
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.231757144500351e-05,
+      "loss": 0.4922,
+      "step": 14320
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.2316601091981915e-05,
+      "loss": 0.0092,
+      "step": 14321
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.2315630698780737e-05,
+      "loss": 0.9688,
+      "step": 14322
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.2314660265405305e-05,
+      "loss": 0.0002,
+      "step": 14323
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.2313689791860937e-05,
+      "loss": 0.9258,
+      "step": 14324
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.2312719278152978e-05,
+      "loss": 0.4805,
+      "step": 14325
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.2311748724286746e-05,
+      "loss": 0.3262,
+      "step": 14326
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.2310778130267582e-05,
+      "loss": 0.0298,
+      "step": 14327
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.23098074961008e-05,
+      "loss": 0.5977,
+      "step": 14328
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.2308836821791746e-05,
+      "loss": 0.0854,
+      "step": 14329
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.2307866107345743e-05,
+      "loss": 0.0041,
+      "step": 14330
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.2306895352768122e-05,
+      "loss": 0.001,
+      "step": 14331
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.2305924558064217e-05,
+      "loss": 0.0087,
+      "step": 14332
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.230495372323936e-05,
+      "loss": 0.0571,
+      "step": 14333
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.230398284829887e-05,
+      "loss": 0.0001,
+      "step": 14334
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.2303011933248097e-05,
+      "loss": 0.4746,
+      "step": 14335
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.2302040978092363e-05,
+      "loss": 0.0,
+      "step": 14336
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.2301069982836995e-05,
+      "loss": 0.0,
+      "step": 14337
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.230009894748734e-05,
+      "loss": 0.011,
+      "step": 14338
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.2299127872048715e-05,
+      "loss": 0.0,
+      "step": 14339
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.2298156756526466e-05,
+      "loss": 0.0413,
+      "step": 14340
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.2297185600925915e-05,
+      "loss": 0.0,
+      "step": 14341
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.22962144052524e-05,
+      "loss": 0.0603,
+      "step": 14342
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.2295243169511256e-05,
+      "loss": 0.0,
+      "step": 14343
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.2294271893707812e-05,
+      "loss": 0.0991,
+      "step": 14344
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.229330057784741e-05,
+      "loss": 0.0,
+      "step": 14345
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.229232922193537e-05,
+      "loss": 0.0,
+      "step": 14346
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.229135782597704e-05,
+      "loss": 0.0,
+      "step": 14347
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.229038638997775e-05,
+      "loss": 0.0,
+      "step": 14348
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.2289414913942833e-05,
+      "loss": 0.0,
+      "step": 14349
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.2288443397877626e-05,
+      "loss": 0.0,
+      "step": 14350
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.2287471841787462e-05,
+      "loss": 0.0,
+      "step": 14351
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.228650024567768e-05,
+      "loss": 0.0,
+      "step": 14352
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.228552860955361e-05,
+      "loss": 0.0581,
+      "step": 14353
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.228455693342059e-05,
+      "loss": 0.0,
+      "step": 14354
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.228358521728396e-05,
+      "loss": 0.0,
+      "step": 14355
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.228261346114905e-05,
+      "loss": 0.0,
+      "step": 14356
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.2281641665021202e-05,
+      "loss": 0.0,
+      "step": 14357
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.228066982890575e-05,
+      "loss": 0.0,
+      "step": 14358
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.227969795280803e-05,
+      "loss": 0.0,
+      "step": 14359
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.227872603673338e-05,
+      "loss": 0.0,
+      "step": 14360
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.2277754080687142e-05,
+      "loss": 0.0981,
+      "step": 14361
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.227678208467465e-05,
+      "loss": 0.1104,
+      "step": 14362
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.2275810048701232e-05,
+      "loss": 0.0815,
+      "step": 14363
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.2274837972772237e-05,
+      "loss": 0.1582,
+      "step": 14364
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.2273865856893005e-05,
+      "loss": 0.1328,
+      "step": 14365
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.227289370106887e-05,
+      "loss": 0.1196,
+      "step": 14366
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.2271921505305172e-05,
+      "loss": 0.4238,
+      "step": 14367
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.2270949269607248e-05,
+      "loss": 0.0119,
+      "step": 14368
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.2269976993980437e-05,
+      "loss": 0.3613,
+      "step": 14369
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.2269004678430084e-05,
+      "loss": 0.053,
+      "step": 14370
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.226803232296152e-05,
+      "loss": 1.2031,
+      "step": 14371
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.226705992758009e-05,
+      "loss": 0.0088,
+      "step": 14372
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.2266087492291132e-05,
+      "loss": 0.3516,
+      "step": 14373
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.2265115017099988e-05,
+      "loss": 0.0161,
+      "step": 14374
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.2264142502011997e-05,
+      "loss": 0.0454,
+      "step": 14375
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.22631699470325e-05,
+      "loss": 2.5938,
+      "step": 14376
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.2262197352166843e-05,
+      "loss": 0.293,
+      "step": 14377
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.2261224717420354e-05,
+      "loss": 0.0002,
+      "step": 14378
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.2260252042798386e-05,
+      "loss": 0.0001,
+      "step": 14379
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.225927932830628e-05,
+      "loss": 0.6133,
+      "step": 14380
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.225830657394937e-05,
+      "loss": 0.4902,
+      "step": 14381
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.2257333779733e-05,
+      "loss": 1.1719,
+      "step": 14382
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.225636094566252e-05,
+      "loss": 0.0454,
+      "step": 14383
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.2255388071743265e-05,
+      "loss": 0.5859,
+      "step": 14384
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.225441515798058e-05,
+      "loss": 0.0006,
+      "step": 14385
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.2253442204379805e-05,
+      "loss": 0.0001,
+      "step": 14386
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.2252469210946287e-05,
+      "loss": 0.5703,
+      "step": 14387
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.2251496177685365e-05,
+      "loss": 0.0002,
+      "step": 14388
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.225052310460239e-05,
+      "loss": 0.0003,
+      "step": 14389
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.22495499917027e-05,
+      "loss": 0.0354,
+      "step": 14390
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.2248576838991637e-05,
+      "loss": 1.8984,
+      "step": 14391
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.224760364647455e-05,
+      "loss": 0.0,
+      "step": 14392
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.2246630414156783e-05,
+      "loss": 0.3105,
+      "step": 14393
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.2245657142043676e-05,
+      "loss": 0.0369,
+      "step": 14394
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.2244683830140577e-05,
+      "loss": 0.0,
+      "step": 14395
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.2243710478452827e-05,
+      "loss": 0.0,
+      "step": 14396
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.224273708698578e-05,
+      "loss": 0.0,
+      "step": 14397
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.2241763655744776e-05,
+      "loss": 0.1064,
+      "step": 14398
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.224079018473516e-05,
+      "loss": 0.0,
+      "step": 14399
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.223981667396228e-05,
+      "loss": 0.0,
+      "step": 14400
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.223884312343148e-05,
+      "loss": 0.0001,
+      "step": 14401
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.223786953314811e-05,
+      "loss": 0.0532,
+      "step": 14402
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.223689590311751e-05,
+      "loss": 0.0,
+      "step": 14403
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.2235922233345026e-05,
+      "loss": 0.0,
+      "step": 14404
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.2234948523836018e-05,
+      "loss": 0.0,
+      "step": 14405
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.2233974774595818e-05,
+      "loss": 0.0,
+      "step": 14406
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.2233000985629786e-05,
+      "loss": 0.0,
+      "step": 14407
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.2232027156943262e-05,
+      "loss": 0.0,
+      "step": 14408
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.2231053288541597e-05,
+      "loss": 0.0,
+      "step": 14409
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.2230079380430132e-05,
+      "loss": 0.1128,
+      "step": 14410
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.222910543261422e-05,
+      "loss": 0.0913,
+      "step": 14411
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.222813144509922e-05,
+      "loss": 0.0,
+      "step": 14412
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.2227157417890462e-05,
+      "loss": 0.0,
+      "step": 14413
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.222618335099331e-05,
+      "loss": 0.1094,
+      "step": 14414
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.2225209244413105e-05,
+      "loss": 0.1328,
+      "step": 14415
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.2224235098155197e-05,
+      "loss": 0.1807,
+      "step": 14416
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.2223260912224938e-05,
+      "loss": 0.2217,
+      "step": 14417
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.222228668662768e-05,
+      "loss": 0.6172,
+      "step": 14418
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.2221312421368767e-05,
+      "loss": 0.1846,
+      "step": 14419
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.2220338116453554e-05,
+      "loss": 0.1494,
+      "step": 14420
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.2219363771887384e-05,
+      "loss": 0.5078,
+      "step": 14421
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.2218389387675617e-05,
+      "loss": 0.0035,
+      "step": 14422
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.2217414963823605e-05,
+      "loss": 0.0549,
+      "step": 14423
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.221644050033669e-05,
+      "loss": 0.0086,
+      "step": 14424
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.2215465997220226e-05,
+      "loss": 0.9297,
+      "step": 14425
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.2214491454479567e-05,
+      "loss": 0.3066,
+      "step": 14426
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.221351687212007e-05,
+      "loss": 0.3223,
+      "step": 14427
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.2212542250147074e-05,
+      "loss": 0.0608,
+      "step": 14428
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.2211567588565938e-05,
+      "loss": 0.0073,
+      "step": 14429
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.2210592887382018e-05,
+      "loss": 0.0508,
+      "step": 14430
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.2209618146600662e-05,
+      "loss": 0.0679,
+      "step": 14431
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.220864336622722e-05,
+      "loss": 1.0,
+      "step": 14432
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.220766854626706e-05,
+      "loss": 0.0708,
+      "step": 14433
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.2206693686725517e-05,
+      "loss": 0.2812,
+      "step": 14434
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.220571878760795e-05,
+      "loss": 0.2324,
+      "step": 14435
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.2204743848919715e-05,
+      "loss": 0.0674,
+      "step": 14436
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.220376887066617e-05,
+      "loss": 0.0249,
+      "step": 14437
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.2202793852852664e-05,
+      "loss": 0.3281,
+      "step": 14438
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.220181879548455e-05,
+      "loss": 0.2432,
+      "step": 14439
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.2200843698567188e-05,
+      "loss": 0.014,
+      "step": 14440
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.2199868562105924e-05,
+      "loss": 0.0002,
+      "step": 14441
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.2198893386106124e-05,
+      "loss": 0.2832,
+      "step": 14442
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.2197918170573132e-05,
+      "loss": 0.0006,
+      "step": 14443
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.219694291551231e-05,
+      "loss": 0.0006,
+      "step": 14444
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.219596762092902e-05,
+      "loss": 0.021,
+      "step": 14445
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.2194992286828607e-05,
+      "loss": 0.0099,
+      "step": 14446
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.2194016913216427e-05,
+      "loss": 0.0001,
+      "step": 14447
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.2193041500097845e-05,
+      "loss": 0.1689,
+      "step": 14448
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.2192066047478207e-05,
+      "loss": 0.0156,
+      "step": 14449
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.2191090555362882e-05,
+      "loss": 0.0933,
+      "step": 14450
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.219011502375722e-05,
+      "loss": 0.7109,
+      "step": 14451
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.2189139452666575e-05,
+      "loss": 0.0,
+      "step": 14452
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.218816384209631e-05,
+      "loss": 0.0,
+      "step": 14453
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.2187188192051782e-05,
+      "loss": 0.0,
+      "step": 14454
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.2186212502538346e-05,
+      "loss": 0.0,
+      "step": 14455
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.218523677356136e-05,
+      "loss": 0.0,
+      "step": 14456
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.2184261005126186e-05,
+      "loss": 0.0001,
+      "step": 14457
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.2183285197238182e-05,
+      "loss": 0.0,
+      "step": 14458
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.2182309349902708e-05,
+      "loss": 0.0,
+      "step": 14459
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.218133346312511e-05,
+      "loss": 0.0,
+      "step": 14460
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.2180357536910764e-05,
+      "loss": 0.0,
+      "step": 14461
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.2179381571265025e-05,
+      "loss": 0.0,
+      "step": 14462
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.2178405566193246e-05,
+      "loss": 0.0,
+      "step": 14463
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.217742952170079e-05,
+      "loss": 0.0903,
+      "step": 14464
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.217645343779302e-05,
+      "loss": 0.1768,
+      "step": 14465
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.2175477314475295e-05,
+      "loss": 0.1074,
+      "step": 14466
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.217450115175297e-05,
+      "loss": 0.3379,
+      "step": 14467
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.2173524949631415e-05,
+      "loss": 0.3145,
+      "step": 14468
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.2172548708115986e-05,
+      "loss": 0.2412,
+      "step": 14469
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.2171572427212044e-05,
+      "loss": 0.5859,
+      "step": 14470
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.217059610692495e-05,
+      "loss": 0.0005,
+      "step": 14471
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.216961974726007e-05,
+      "loss": 0.0342,
+      "step": 14472
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.2168643348222754e-05,
+      "loss": 0.0669,
+      "step": 14473
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.216766690981838e-05,
+      "loss": 0.1108,
+      "step": 14474
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.2166690432052294e-05,
+      "loss": 0.1689,
+      "step": 14475
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.2165713914929867e-05,
+      "loss": 0.3066,
+      "step": 14476
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.216473735845647e-05,
+      "loss": 0.6719,
+      "step": 14477
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.2163760762637444e-05,
+      "loss": 0.0007,
+      "step": 14478
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.2162784127478174e-05,
+      "loss": 0.0026,
+      "step": 14479
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.2161807452984012e-05,
+      "loss": 0.1875,
+      "step": 14480
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.216083073916032e-05,
+      "loss": 0.0002,
+      "step": 14481
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.215985398601247e-05,
+      "loss": 1.4297,
+      "step": 14482
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.2158877193545818e-05,
+      "loss": 0.0136,
+      "step": 14483
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.2157900361765736e-05,
+      "loss": 0.003,
+      "step": 14484
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.215692349067758e-05,
+      "loss": 0.0437,
+      "step": 14485
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.215594658028672e-05,
+      "loss": 0.0061,
+      "step": 14486
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.215496963059852e-05,
+      "loss": 0.0023,
+      "step": 14487
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.2153992641618336e-05,
+      "loss": 0.0003,
+      "step": 14488
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.2153015613351547e-05,
+      "loss": 0.0019,
+      "step": 14489
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.2152038545803516e-05,
+      "loss": 0.0,
+      "step": 14490
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.2151061438979604e-05,
+      "loss": 0.127,
+      "step": 14491
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.215008429288517e-05,
+      "loss": 0.0053,
+      "step": 14492
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.2149107107525595e-05,
+      "loss": 0.2314,
+      "step": 14493
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.2148129882906238e-05,
+      "loss": 0.0004,
+      "step": 14494
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.2147152619032463e-05,
+      "loss": 0.0026,
+      "step": 14495
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.2146175315909638e-05,
+      "loss": 0.0,
+      "step": 14496
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.214519797354314e-05,
+      "loss": 0.0093,
+      "step": 14497
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.2144220591938318e-05,
+      "loss": 0.0806,
+      "step": 14498
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.2143243171100556e-05,
+      "loss": 0.0,
+      "step": 14499
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.2142265711035208e-05,
+      "loss": 0.0,
+      "step": 14500
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 33915,
+  "num_train_epochs": 5,
+  "save_steps": 500,
+  "total_flos": 4.361911553654784e+16,
+  "trial_name": null,
+  "trial_params": null
+}