diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,7530 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "eval_steps": 500,
+  "global_step": 1250,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.263157894736842e-06,
+      "loss": 1.3743,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0526315789473684e-05,
+      "loss": 1.4711,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5789473684210526e-05,
+      "loss": 1.4956,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.105263157894737e-05,
+      "loss": 1.255,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6315789473684212e-05,
+      "loss": 1.1365,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.157894736842105e-05,
+      "loss": 0.969,
+      "step": 6
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6842105263157895e-05,
+      "loss": 0.8513,
+      "step": 7
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.210526315789474e-05,
+      "loss": 0.7272,
+      "step": 8
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.736842105263158e-05,
+      "loss": 0.5576,
+      "step": 9
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.2631578947368424e-05,
+      "loss": 0.5404,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.789473684210527e-05,
+      "loss": 0.5655,
+      "step": 11
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.31578947368421e-05,
+      "loss": 0.5956,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.842105263157895e-05,
+      "loss": 0.5265,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.368421052631579e-05,
+      "loss": 0.5235,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.894736842105263e-05,
+      "loss": 0.4948,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.421052631578948e-05,
+      "loss": 0.4261,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.947368421052632e-05,
+      "loss": 0.4719,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.473684210526316e-05,
+      "loss": 0.5953,
+      "step": 18
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001,
+      "loss": 0.5123,
+      "step": 19
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00010526315789473685,
+      "loss": 0.5325,
+      "step": 20
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001105263157894737,
+      "loss": 0.4554,
+      "step": 21
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00011578947368421053,
+      "loss": 0.484,
+      "step": 22
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00012105263157894738,
+      "loss": 0.5041,
+      "step": 23
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001263157894736842,
+      "loss": 0.535,
+      "step": 24
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00013157894736842108,
+      "loss": 0.5531,
+      "step": 25
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001368421052631579,
+      "loss": 0.5099,
+      "step": 26
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00014210526315789474,
+      "loss": 0.4426,
+      "step": 27
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00014736842105263158,
+      "loss": 0.4545,
+      "step": 28
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00015263157894736845,
+      "loss": 0.5823,
+      "step": 29
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.00015789473684210527,
+      "loss": 0.558,
+      "step": 30
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0001631578947368421,
+      "loss": 0.4563,
+      "step": 31
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00016842105263157895,
+      "loss": 0.3814,
+      "step": 32
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001736842105263158,
+      "loss": 0.4578,
+      "step": 33
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00017894736842105264,
+      "loss": 0.5195,
+      "step": 34
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018421052631578948,
+      "loss": 0.5808,
+      "step": 35
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00018947368421052632,
+      "loss": 0.3652,
+      "step": 36
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019473684210526317,
+      "loss": 0.5934,
+      "step": 37
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.5397,
+      "step": 38
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999966405802826,
+      "loss": 0.4596,
+      "step": 39
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999865623437013,
+      "loss": 0.4437,
+      "step": 40
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999697653579705,
+      "loss": 0.4137,
+      "step": 41
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.00019999462497359466,
+      "loss": 0.4119,
+      "step": 42
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001999916015635627,
+      "loss": 0.4858,
+      "step": 43
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019998790632601496,
+      "loss": 0.4374,
+      "step": 44
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019998353928577919,
+      "loss": 0.4447,
+      "step": 45
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999785004721968,
+      "loss": 0.3738,
+      "step": 46
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999727899191228,
+      "loss": 0.4658,
+      "step": 47
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019996640766492543,
+      "loss": 0.5095,
+      "step": 48
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019995935375248606,
+      "loss": 0.3847,
+      "step": 49
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019995162822919883,
+      "loss": 0.4148,
+      "step": 50
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019994323114697022,
+      "loss": 0.4282,
+      "step": 51
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019993416256221895,
+      "loss": 0.5612,
+      "step": 52
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0001999244225358753,
+      "loss": 0.5061,
+      "step": 53
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019991401113338104,
+      "loss": 0.4215,
+      "step": 54
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019990292842468868,
+      "loss": 0.4539,
+      "step": 55
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.00019989117448426108,
+      "loss": 0.4132,
+      "step": 56
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001998787493910712,
+      "loss": 0.5334,
+      "step": 57
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019986565322860115,
+      "loss": 0.3975,
+      "step": 58
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.000199851886084842,
+      "loss": 0.4914,
+      "step": 59
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019983744805229296,
+      "loss": 0.4908,
+      "step": 60
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019982233922796085,
+      "loss": 0.4647,
+      "step": 61
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019980655971335945,
+      "loss": 0.4099,
+      "step": 62
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019979010961450878,
+      "loss": 0.4475,
+      "step": 63
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019977298904193437,
+      "loss": 0.4224,
+      "step": 64
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019975519811066663,
+      "loss": 0.501,
+      "step": 65
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019973673694024,
+      "loss": 0.5064,
+      "step": 66
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0001997176056546921,
+      "loss": 0.4736,
+      "step": 67
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.00019969780438256293,
+      "loss": 0.3828,
+      "step": 68
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001996773332568941,
+      "loss": 0.4568,
+      "step": 69
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001996561924152278,
+      "loss": 0.4305,
+      "step": 70
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019963438199960599,
+      "loss": 0.3915,
+      "step": 71
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001996119021565693,
+      "loss": 0.4726,
+      "step": 72
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019958875303715615,
+      "loss": 0.5466,
+      "step": 73
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001995649347969019,
+      "loss": 0.5346,
+      "step": 74
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001995404475958373,
+      "loss": 0.4408,
+      "step": 75
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019951529159848805,
+      "loss": 0.4905,
+      "step": 76
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001994894669738732,
+      "loss": 0.3999,
+      "step": 77
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019946297389550433,
+      "loss": 0.3946,
+      "step": 78
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001994358125413841,
+      "loss": 0.4198,
+      "step": 79
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.00019940798309400526,
+      "loss": 0.4225,
+      "step": 80
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0001993794857403495,
+      "loss": 0.369,
+      "step": 81
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001993503206718859,
+      "loss": 0.5407,
+      "step": 82
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0001993204880845699,
+      "loss": 0.4247,
+      "step": 83
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019928998817884182,
+      "loss": 0.5015,
+      "step": 84
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019925882115962568,
+      "loss": 0.5137,
+      "step": 85
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019922698723632767,
+      "loss": 0.3776,
+      "step": 86
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019919448662283478,
+      "loss": 0.5028,
+      "step": 87
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019916131953751342,
+      "loss": 0.41,
+      "step": 88
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019912748620320794,
+      "loss": 0.5868,
+      "step": 89
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019909298684723904,
+      "loss": 0.4004,
+      "step": 90
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019905782170140238,
+      "loss": 0.5511,
+      "step": 91
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019902199100196697,
+      "loss": 0.4121,
+      "step": 92
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.00019898549498967343,
+      "loss": 0.4313,
+      "step": 93
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019894833390973266,
+      "loss": 0.4684,
+      "step": 94
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.000198910508011824,
+      "loss": 0.4664,
+      "step": 95
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019887201755009357,
+      "loss": 0.4033,
+      "step": 96
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019883286278315262,
+      "loss": 0.4115,
+      "step": 97
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001987930439740757,
+      "loss": 0.4507,
+      "step": 98
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019875256139039902,
+      "loss": 0.4928,
+      "step": 99
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019871141530411853,
+      "loss": 0.4194,
+      "step": 100
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019866960599168826,
+      "loss": 0.4481,
+      "step": 101
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001986271337340182,
+      "loss": 0.4171,
+      "step": 102
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001985839988164726,
+      "loss": 0.4239,
+      "step": 103
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019854020152886814,
+      "loss": 0.4402,
+      "step": 104
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.00019849574216547171,
+      "loss": 0.4233,
+      "step": 105
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0001984506210249986,
+      "loss": 0.3756,
+      "step": 106
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019840483841061058,
+      "loss": 0.4683,
+      "step": 107
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019835839462991361,
+      "loss": 0.4715,
+      "step": 108
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019831128999495606,
+      "loss": 0.3766,
+      "step": 109
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019826352482222638,
+      "loss": 0.4561,
+      "step": 110
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001982150994326511,
+      "loss": 0.4632,
+      "step": 111
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019816601415159263,
+      "loss": 0.4321,
+      "step": 112
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001981162693088471,
+      "loss": 0.518,
+      "step": 113
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001980658652386421,
+      "loss": 0.4307,
+      "step": 114
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0001980148022796345,
+      "loss": 0.3748,
+      "step": 115
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019796308077490817,
+      "loss": 0.4742,
+      "step": 116
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019791070107197153,
+      "loss": 0.3425,
+      "step": 117
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.00019785766352275542,
+      "loss": 0.3998,
+      "step": 118
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001978039684836106,
+      "loss": 0.4494,
+      "step": 119
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019774961631530545,
+      "loss": 0.4629,
+      "step": 120
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001976946073830234,
+      "loss": 0.3397,
+      "step": 121
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019763894205636072,
+      "loss": 0.4447,
+      "step": 122
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019758262070932375,
+      "loss": 0.4102,
+      "step": 123
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019752564372032657,
+      "loss": 0.43,
+      "step": 124
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019746801147218842,
+      "loss": 0.4552,
+      "step": 125
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019740972435213115,
+      "loss": 0.3853,
+      "step": 126
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019735078275177654,
+      "loss": 0.4654,
+      "step": 127
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019729118706714375,
+      "loss": 0.4187,
+      "step": 128
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019723093769864663,
+      "loss": 0.4115,
+      "step": 129
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.00019717003505109095,
+      "loss": 0.4004,
+      "step": 130
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0001971084795336719,
+      "loss": 0.4113,
+      "step": 131
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019704627155997108,
+      "loss": 0.3344,
+      "step": 132
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019698341154795389,
+      "loss": 0.4155,
+      "step": 133
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019691989991996663,
+      "loss": 0.449,
+      "step": 134
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019685573710273376,
+      "loss": 0.4322,
+      "step": 135
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0001967909235273549,
+      "loss": 0.3733,
+      "step": 136
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019672545962930215,
+      "loss": 0.4939,
+      "step": 137
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019665934584841682,
+      "loss": 0.3069,
+      "step": 138
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019659258262890683,
+      "loss": 0.4131,
+      "step": 139
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019652517041934356,
+      "loss": 0.4096,
+      "step": 140
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019645710967265882,
+      "loss": 0.3401,
+      "step": 141
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019638840084614182,
+      "loss": 0.4534,
+      "step": 142
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.00019631904440143612,
+      "loss": 0.4531,
+      "step": 143
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019624904080453655,
+      "loss": 0.391,
+      "step": 144
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019617839052578603,
+      "loss": 0.4382,
+      "step": 145
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019610709403987246,
+      "loss": 0.3724,
+      "step": 146
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001960351518258255,
+      "loss": 0.4042,
+      "step": 147
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019596256436701324,
+      "loss": 0.4238,
+      "step": 148
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019588933215113926,
+      "loss": 0.3721,
+      "step": 149
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.000195815455670239,
+      "loss": 0.4925,
+      "step": 150
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019574093542067673,
+      "loss": 0.4119,
+      "step": 151
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019566577190314197,
+      "loss": 0.3727,
+      "step": 152
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001955899656226464,
+      "loss": 0.4626,
+      "step": 153
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001955135170885202,
+      "loss": 0.4567,
+      "step": 154
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0001954364268144088,
+      "loss": 0.4549,
+      "step": 155
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.00019535869531826937,
+      "loss": 0.4744,
+      "step": 156
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019528032312236736,
+      "loss": 0.4035,
+      "step": 157
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019520131075327298,
+      "loss": 0.3822,
+      "step": 158
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019512165874185767,
+      "loss": 0.4007,
+      "step": 159
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019504136762329047,
+      "loss": 0.36,
+      "step": 160
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001949604379370345,
+      "loss": 0.3513,
+      "step": 161
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019487887022684336,
+      "loss": 0.5134,
+      "step": 162
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019479666504075736,
+      "loss": 0.3578,
+      "step": 163
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019471382293110003,
+      "loss": 0.4461,
+      "step": 164
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0001946303444544741,
+      "loss": 0.414,
+      "step": 165
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019454623017175812,
+      "loss": 0.4136,
+      "step": 166
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019446148064810242,
+      "loss": 0.4328,
+      "step": 167
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.00019437609645292546,
+      "loss": 0.424,
+      "step": 168
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019429007815990993,
+      "loss": 0.4311,
+      "step": 169
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0001942034263469989,
+      "loss": 0.4436,
+      "step": 170
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019411614159639204,
+      "loss": 0.4044,
+      "step": 171
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019402822449454153,
+      "loss": 0.451,
+      "step": 172
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019393967563214833,
+      "loss": 0.5273,
+      "step": 173
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019385049560415794,
+      "loss": 0.4633,
+      "step": 174
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019376068500975667,
+      "loss": 0.4262,
+      "step": 175
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019367024445236754,
+      "loss": 0.3665,
+      "step": 176
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.000193579174539646,
+      "loss": 0.3799,
+      "step": 177
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019348747588347637,
+      "loss": 0.4026,
+      "step": 178
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019339514909996706,
+      "loss": 0.4504,
+      "step": 179
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019330219480944694,
+      "loss": 0.4022,
+      "step": 180
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.00019320861363646095,
+      "loss": 0.4088,
+      "step": 181
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019311440620976597,
+      "loss": 0.4839,
+      "step": 182
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019301957316232658,
+      "loss": 0.3915,
+      "step": 183
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001929241151313108,
+      "loss": 0.395,
+      "step": 184
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0001928280327580858,
+      "loss": 0.4676,
+      "step": 185
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019273132668821364,
+      "loss": 0.3749,
+      "step": 186
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019263399757144683,
+      "loss": 0.5115,
+      "step": 187
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019253604606172417,
+      "loss": 0.3358,
+      "step": 188
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.000192437472817166,
+      "loss": 0.4594,
+      "step": 189
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019233827850007027,
+      "loss": 0.4467,
+      "step": 190
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019223846377690754,
+      "loss": 0.4724,
+      "step": 191
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019213802931831696,
+      "loss": 0.4222,
+      "step": 192
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.00019203697579910154,
+      "loss": 0.4325,
+      "step": 193
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019193530389822363,
+      "loss": 0.4405,
+      "step": 194
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019183301429880043,
+      "loss": 0.4196,
+      "step": 195
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019173010768809933,
+      "loss": 0.4413,
+      "step": 196
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019162658475753327,
+      "loss": 0.394,
+      "step": 197
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001915224462026563,
+      "loss": 0.4334,
+      "step": 198
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019141769272315858,
+      "loss": 0.4027,
+      "step": 199
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019131232502286188,
+      "loss": 0.4273,
+      "step": 200
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019120634380971496,
+      "loss": 0.3915,
+      "step": 201
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001910997497957885,
+      "loss": 0.4044,
+      "step": 202
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0001909925436972706,
+      "loss": 0.4601,
+      "step": 203
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019088472623446183,
+      "loss": 0.3653,
+      "step": 204
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019077629813177036,
+      "loss": 0.5381,
+      "step": 205
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.00019066726011770726,
+      "loss": 0.3353,
+      "step": 206
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019055761292488142,
+      "loss": 0.5004,
+      "step": 207
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001904473572899947,
+      "loss": 0.4432,
+      "step": 208
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019033649395383702,
+      "loss": 0.4044,
+      "step": 209
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019022502366128135,
+      "loss": 0.385,
+      "step": 210
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019011294716127867,
+      "loss": 0.3639,
+      "step": 211
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00019000026520685302,
+      "loss": 0.428,
+      "step": 212
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001898869785550963,
+      "loss": 0.406,
+      "step": 213
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0001897730879671634,
+      "loss": 0.3834,
+      "step": 214
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018965859420826684,
+      "loss": 0.4342,
+      "step": 215
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018954349804767184,
+      "loss": 0.4371,
+      "step": 216
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018942780025869098,
+      "loss": 0.4926,
+      "step": 217
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.00018931150161867916,
+      "loss": 0.3638,
+      "step": 218
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018919460290902826,
+      "loss": 0.3927,
+      "step": 219
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018907710491516199,
+      "loss": 0.4095,
+      "step": 220
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001889590084265304,
+      "loss": 0.4031,
+      "step": 221
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001888403142366049,
+      "loss": 0.4448,
+      "step": 222
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001887210231428727,
+      "loss": 0.4159,
+      "step": 223
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018860113594683148,
+      "loss": 0.575,
+      "step": 224
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001884806534539841,
+      "loss": 0.4159,
+      "step": 225
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018835957647383303,
+      "loss": 0.3819,
+      "step": 226
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0001882379058198751,
+      "loss": 0.4074,
+      "step": 227
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018811564230959588,
+      "loss": 0.4245,
+      "step": 228
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018799278676446423,
+      "loss": 0.3517,
+      "step": 229
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018786934000992688,
+      "loss": 0.4335,
+      "step": 230
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.00018774530287540278,
+      "loss": 0.4654,
+      "step": 231
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018762067619427746,
+      "loss": 0.4688,
+      "step": 232
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018749546080389757,
+      "loss": 0.454,
+      "step": 233
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018736965754556528,
+      "loss": 0.4713,
+      "step": 234
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018724326726453244,
+      "loss": 0.412,
+      "step": 235
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018711629080999504,
+      "loss": 0.4679,
+      "step": 236
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018698872903508755,
+      "loss": 0.4249,
+      "step": 237
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018686058279687698,
+      "loss": 0.4443,
+      "step": 238
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0001867318529563574,
+      "loss": 0.429,
+      "step": 239
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018660254037844388,
+      "loss": 0.4404,
+      "step": 240
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018647264593196688,
+      "loss": 0.3382,
+      "step": 241
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018634217048966637,
+      "loss": 0.3414,
+      "step": 242
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.00018621111492818585,
+      "loss": 0.5505,
+      "step": 243
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001860794801280666,
+      "loss": 0.3852,
+      "step": 244
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018594726697374175,
+      "loss": 0.3138,
+      "step": 245
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001858144763535302,
+      "loss": 0.3969,
+      "step": 246
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001856811091596308,
+      "loss": 0.4442,
+      "step": 247
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001855471662881164,
+      "loss": 0.3114,
+      "step": 248
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018541264863892754,
+      "loss": 0.4597,
+      "step": 249
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018527755711586678,
+      "loss": 0.4419,
+      "step": 250
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018514189262659235,
+      "loss": 0.4762,
+      "step": 251
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018500565608261214,
+      "loss": 0.3457,
+      "step": 252
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018486884839927768,
+      "loss": 0.3576,
+      "step": 253
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018473147049577774,
+      "loss": 0.3853,
+      "step": 254
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0001845935232951325,
+      "loss": 0.4179,
+      "step": 255
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.00018445500772418697,
+      "loss": 0.4,
+      "step": 256
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018431592471360503,
+      "loss": 0.3849,
+      "step": 257
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018417627519786315,
+      "loss": 0.4135,
+      "step": 258
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.000184036060115244,
+      "loss": 0.4207,
+      "step": 259
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018389528040783012,
+      "loss": 0.322,
+      "step": 260
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018375393702149787,
+      "loss": 0.3933,
+      "step": 261
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018361203090591071,
+      "loss": 0.4766,
+      "step": 262
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018346956301451304,
+      "loss": 0.4401,
+      "step": 263
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018332653430452376,
+      "loss": 0.3865,
+      "step": 264
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018318294573692985,
+      "loss": 0.4273,
+      "step": 265
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018303879827647975,
+      "loss": 0.4232,
+      "step": 266
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0001828940928916772,
+      "loss": 0.4601,
+      "step": 267
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.00018274883055477436,
+      "loss": 0.3942,
+      "step": 268
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018260301224176558,
+      "loss": 0.4529,
+      "step": 269
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018245663893238075,
+      "loss": 0.42,
+      "step": 270
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018230971161007853,
+      "loss": 0.4342,
+      "step": 271
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018216223126204007,
+      "loss": 0.3648,
+      "step": 272
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018201419887916214,
+      "loss": 0.455,
+      "step": 273
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018186561545605054,
+      "loss": 0.4129,
+      "step": 274
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018171648199101346,
+      "loss": 0.488,
+      "step": 275
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018156679948605467,
+      "loss": 0.4255,
+      "step": 276
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018141656894686689,
+      "loss": 0.3598,
+      "step": 277
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018126579138282503,
+      "loss": 0.4861,
+      "step": 278
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018111446780697929,
+      "loss": 0.4044,
+      "step": 279
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.0001809625992360485,
+      "loss": 0.3472,
+      "step": 280
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 0.00018081018669041324,
+      "loss": 0.3529,
+      "step": 281
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018065723119410884,
+      "loss": 0.4792,
+      "step": 282
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018050373377481878,
+      "loss": 0.4892,
+      "step": 283
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00018034969546386757,
+      "loss": 0.529,
+      "step": 284
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001801951172962139,
+      "loss": 0.4703,
+      "step": 285
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001800400003104436,
+      "loss": 0.533,
+      "step": 286
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001798843455487629,
+      "loss": 0.4317,
+      "step": 287
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017972815405699103,
+      "loss": 0.4235,
+      "step": 288
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017957142688455362,
+      "loss": 0.4492,
+      "step": 289
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017941416508447536,
+      "loss": 0.3977,
+      "step": 290
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017925636971337304,
+      "loss": 0.3422,
+      "step": 291
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0001790980418314484,
+      "loss": 0.4228,
+      "step": 292
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.00017893918250248104,
+      "loss": 0.3657,
+      "step": 293
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017877979279382135,
+      "loss": 0.4469,
+      "step": 294
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017861987377638312,
+      "loss": 0.3115,
+      "step": 295
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001784594265246366,
+      "loss": 0.4818,
+      "step": 296
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001782984521166011,
+      "loss": 0.3938,
+      "step": 297
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.0001781369516338378,
+      "loss": 0.337,
+      "step": 298
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017797492616144256,
+      "loss": 0.5138,
+      "step": 299
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017781237678803847,
+      "loss": 0.4314,
+      "step": 300
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017764930460576866,
+      "loss": 0.3882,
+      "step": 301
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.000177485710710289,
+      "loss": 0.3778,
+      "step": 302
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017732159620076053,
+      "loss": 0.4052,
+      "step": 303
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017715696217984235,
+      "loss": 0.4382,
+      "step": 304
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017699180975368396,
+      "loss": 0.3181,
+      "step": 305
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 0.00017682614003191807,
+      "loss": 0.4399,
+      "step": 306
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017665995412765285,
+      "loss": 0.3993,
+      "step": 307
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017649325315746478,
+      "loss": 0.3964,
+      "step": 308
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017632603824139085,
+      "loss": 0.339,
+      "step": 309
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001761583105029213,
+      "loss": 0.3611,
+      "step": 310
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.0001759900710689918,
+      "loss": 0.4313,
+      "step": 311
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017582132106997616,
+      "loss": 0.2978,
+      "step": 312
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017565206163967846,
+      "loss": 0.4403,
+      "step": 313
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017548229391532572,
+      "loss": 0.3423,
+      "step": 314
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017531201903755994,
+      "loss": 0.4589,
+      "step": 315
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017514123815043074,
+      "loss": 0.3753,
+      "step": 316
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017496995240138744,
+      "loss": 0.3612,
+      "step": 317
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 0.00017479816294127152,
+      "loss": 0.5044,
+      "step": 318
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017462587092430875,
+      "loss": 0.3459,
+      "step": 319
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001744530775081015,
+      "loss": 0.3828,
+      "step": 320
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017427978385362112,
+      "loss": 0.4456,
+      "step": 321
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001741059911251997,
+      "loss": 0.3541,
+      "step": 322
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001739317004905227,
+      "loss": 0.3316,
+      "step": 323
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.000173756913120621,
+      "loss": 0.4954,
+      "step": 324
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017358163018986282,
+      "loss": 0.4061,
+      "step": 325
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017340585287594604,
+      "loss": 0.3657,
+      "step": 326
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017322958235989016,
+      "loss": 0.3442,
+      "step": 327
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.0001730528198260285,
+      "loss": 0.5041,
+      "step": 328
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017287556646200018,
+      "loss": 0.3918,
+      "step": 329
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017269782345874203,
+      "loss": 0.4131,
+      "step": 330
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 0.00017251959201048083,
+      "loss": 0.3373,
+      "step": 331
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017234087331472497,
+      "loss": 0.3778,
+      "step": 332
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017216166857225674,
+      "loss": 0.392,
+      "step": 333
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017198197898712404,
+      "loss": 0.3917,
+      "step": 334
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017180180576663228,
+      "loss": 0.3563,
+      "step": 335
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017162115012133643,
+      "loss": 0.4711,
+      "step": 336
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017144001326503273,
+      "loss": 0.4436,
+      "step": 337
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017125839641475072,
+      "loss": 0.4274,
+      "step": 338
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017107630079074478,
+      "loss": 0.4549,
+      "step": 339
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017089372761648616,
+      "loss": 0.441,
+      "step": 340
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017071067811865476,
+      "loss": 0.4489,
+      "step": 341
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017052715352713075,
+      "loss": 0.3289,
+      "step": 342
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 0.00017034315507498635,
+      "loss": 0.4181,
+      "step": 343
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00017015868399847768,
+      "loss": 0.4073,
+      "step": 344
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016997374153703625,
+      "loss": 0.3102,
+      "step": 345
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016978832893326074,
+      "loss": 0.5464,
+      "step": 346
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016960244743290868,
+      "loss": 0.4358,
+      "step": 347
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016941609828488807,
+      "loss": 0.3866,
+      "step": 348
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016922928274124886,
+      "loss": 0.3649,
+      "step": 349
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001690420020571747,
+      "loss": 0.4177,
+      "step": 350
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016885425749097444,
+      "loss": 0.3419,
+      "step": 351
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001686660503040737,
+      "loss": 0.3675,
+      "step": 352
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016847738176100632,
+      "loss": 0.5156,
+      "step": 353
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.00016828825312940592,
+      "loss": 0.2901,
+      "step": 354
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001680986656799975,
+      "loss": 0.426,
+      "step": 355
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 0.0001679086206865886,
+      "loss": 0.4512,
+      "step": 356
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016771811942606108,
+      "loss": 0.3759,
+      "step": 357
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016752716317836229,
+      "loss": 0.5114,
+      "step": 358
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016733575322649657,
+      "loss": 0.4163,
+      "step": 359
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001671438908565167,
+      "loss": 0.3961,
+      "step": 360
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016695157735751513,
+      "loss": 0.4853,
+      "step": 361
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016675881402161536,
+      "loss": 0.4281,
+      "step": 362
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001665656021439633,
+      "loss": 0.4268,
+      "step": 363
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.0001663719430227186,
+      "loss": 0.4238,
+      "step": 364
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016617783795904565,
+      "loss": 0.3746,
+      "step": 365
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016598328825710533,
+      "loss": 0.4609,
+      "step": 366
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.00016578829522404583,
+      "loss": 0.435,
+      "step": 367
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 0.000165592860169994,
+      "loss": 0.3792,
+      "step": 368
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016539698440804661,
+      "loss": 0.3862,
+      "step": 369
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016520066925426144,
+      "loss": 0.4721,
+      "step": 370
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001650039160276485,
+      "loss": 0.4511,
+      "step": 371
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001648067260501611,
+      "loss": 0.4376,
+      "step": 372
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001646091006466871,
+      "loss": 0.4276,
+      "step": 373
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001644110411450398,
+      "loss": 0.4185,
+      "step": 374
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016421254887594917,
+      "loss": 0.3767,
+      "step": 375
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016401362517305296,
+      "loss": 0.4907,
+      "step": 376
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016381427137288754,
+      "loss": 0.398,
+      "step": 377
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016361448881487914,
+      "loss": 0.393,
+      "step": 378
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.0001634142788413346,
+      "loss": 0.3703,
+      "step": 379
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016321364279743266,
+      "loss": 0.3618,
+      "step": 380
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 0.00016301258203121462,
+      "loss": 0.4271,
+      "step": 381
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001628110978935756,
+      "loss": 0.3972,
+      "step": 382
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016260919173825508,
+      "loss": 0.4312,
+      "step": 383
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016240686492182804,
+      "loss": 0.441,
+      "step": 384
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016220411880369601,
+      "loss": 0.355,
+      "step": 385
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016200095474607753,
+      "loss": 0.41,
+      "step": 386
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016179737411399926,
+      "loss": 0.3869,
+      "step": 387
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016159337827528685,
+      "loss": 0.3893,
+      "step": 388
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016138896860055555,
+      "loss": 0.3417,
+      "step": 389
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001611841464632011,
+      "loss": 0.4074,
+      "step": 390
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016097891323939062,
+      "loss": 0.399,
+      "step": 391
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.0001607732703080532,
+      "loss": 0.4432,
+      "step": 392
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 0.00016056721905087056,
+      "loss": 0.4508,
+      "step": 393
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016036076085226814,
+      "loss": 0.3061,
+      "step": 394
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00016015389709940538,
+      "loss": 0.3302,
+      "step": 395
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001599466291821666,
+      "loss": 0.3062,
+      "step": 396
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001597389584931517,
+      "loss": 0.4222,
+      "step": 397
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001595308864276666,
+      "loss": 0.4475,
+      "step": 398
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001593224143837142,
+      "loss": 0.3293,
+      "step": 399
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001591135437619847,
+      "loss": 0.3547,
+      "step": 400
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015890427596584617,
+      "loss": 0.3162,
+      "step": 401
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001586946124013354,
+      "loss": 0.354,
+      "step": 402
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015848455447714822,
+      "loss": 0.4657,
+      "step": 403
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.0001582741036046301,
+      "loss": 0.3665,
+      "step": 404
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015806326119776663,
+      "loss": 0.4512,
+      "step": 405
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 0.00015785202867317407,
+      "loss": 0.3972,
+      "step": 406
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015764040745008988,
+      "loss": 0.3948,
+      "step": 407
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015742839895036305,
+      "loss": 0.5151,
+      "step": 408
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015721600459844468,
+      "loss": 0.3601,
+      "step": 409
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015700322582137827,
+      "loss": 0.4264,
+      "step": 410
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015679006404879033,
+      "loss": 0.3056,
+      "step": 411
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001565765207128805,
+      "loss": 0.2742,
+      "step": 412
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015636259724841222,
+      "loss": 0.4562,
+      "step": 413
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001561482950927029,
+      "loss": 0.3478,
+      "step": 414
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015593361568561428,
+      "loss": 0.3145,
+      "step": 415
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015571856046954285,
+      "loss": 0.3183,
+      "step": 416
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.0001555031308894101,
+      "loss": 0.4193,
+      "step": 417
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 0.00015528732839265272,
+      "loss": 0.3795,
+      "step": 418
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001550711544292131,
+      "loss": 0.4058,
+      "step": 419
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001548546104515294,
+      "loss": 0.4208,
+      "step": 420
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015463769791452574,
+      "loss": 0.3257,
+      "step": 421
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015442041827560274,
+      "loss": 0.369,
+      "step": 422
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015420277299462736,
+      "loss": 0.4066,
+      "step": 423
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015398476353392323,
+      "loss": 0.383,
+      "step": 424
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015376639135826107,
+      "loss": 0.3686,
+      "step": 425
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015354765793484834,
+      "loss": 0.3552,
+      "step": 426
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015332856473331978,
+      "loss": 0.379,
+      "step": 427
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015310911322572753,
+      "loss": 0.4177,
+      "step": 428
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.00015288930488653094,
+      "loss": 0.4321,
+      "step": 429
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.000152669141192587,
+      "loss": 0.3254,
+      "step": 430
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 0.0001524486236231402,
+      "loss": 0.3177,
+      "step": 431
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015222775365981273,
+      "loss": 0.36,
+      "step": 432
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015200653278659432,
+      "loss": 0.3668,
+      "step": 433
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015178496248983254,
+      "loss": 0.3808,
+      "step": 434
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015156304425822267,
+      "loss": 0.4583,
+      "step": 435
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015134077958279765,
+      "loss": 0.4418,
+      "step": 436
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015111816995691809,
+      "loss": 0.3838,
+      "step": 437
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015089521687626243,
+      "loss": 0.438,
+      "step": 438
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015067192183881658,
+      "loss": 0.3882,
+      "step": 439
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.000150448286344864,
+      "loss": 0.3072,
+      "step": 440
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015022431189697568,
+      "loss": 0.3415,
+      "step": 441
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.00015000000000000001,
+      "loss": 0.4306,
+      "step": 442
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 0.0001497753521610526,
+      "loss": 0.3373,
+      "step": 443
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014955036988950618,
+      "loss": 0.3842,
+      "step": 444
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014932505469698052,
+      "loss": 0.3894,
+      "step": 445
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014909940809733222,
+      "loss": 0.5026,
+      "step": 446
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001488734316066446,
+      "loss": 0.3691,
+      "step": 447
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014864712674321734,
+      "loss": 0.3581,
+      "step": 448
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001484204950275565,
+      "loss": 0.3883,
+      "step": 449
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014819353798236427,
+      "loss": 0.36,
+      "step": 450
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014796625713252848,
+      "loss": 0.3592,
+      "step": 451
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014773865400511272,
+      "loss": 0.4268,
+      "step": 452
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014751073012934587,
+      "loss": 0.3887,
+      "step": 453
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014728248703661182,
+      "loss": 0.4541,
+      "step": 454
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.0001470539262604393,
+      "loss": 0.3293,
+      "step": 455
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 0.00014682504933649144,
+      "loss": 0.2813,
+      "step": 456
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014659585780255556,
+      "loss": 0.4143,
+      "step": 457
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014636635319853275,
+      "loss": 0.3714,
+      "step": 458
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001461365370664276,
+      "loss": 0.3051,
+      "step": 459
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014590641095033787,
+      "loss": 0.3641,
+      "step": 460
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014567597639644387,
+      "loss": 0.3185,
+      "step": 461
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014544523495299842,
+      "loss": 0.3769,
+      "step": 462
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014521418817031628,
+      "loss": 0.461,
+      "step": 463
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001449828376007636,
+      "loss": 0.3757,
+      "step": 464
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.00014475118479874774,
+      "loss": 0.3758,
+      "step": 465
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001445192313207067,
+      "loss": 0.3823,
+      "step": 466
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001442869787250987,
+      "loss": 0.4141,
+      "step": 467
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 0.0001440544285723915,
+      "loss": 0.3606,
+      "step": 468
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014382158242505234,
+      "loss": 0.4119,
+      "step": 469
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014358844184753712,
+      "loss": 0.3907,
+      "step": 470
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014335500840627986,
+      "loss": 0.4131,
+      "step": 471
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014312128366968243,
+      "loss": 0.3724,
+      "step": 472
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001428872692081038,
+      "loss": 0.3609,
+      "step": 473
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014265296659384956,
+      "loss": 0.3161,
+      "step": 474
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014241837740116132,
+      "loss": 0.3643,
+      "step": 475
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014218350320620624,
+      "loss": 0.4533,
+      "step": 476
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014194834558706632,
+      "loss": 0.4384,
+      "step": 477
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001417129061237278,
+      "loss": 0.4173,
+      "step": 478
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.0001414771863980707,
+      "loss": 0.453,
+      "step": 479
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014124118799385796,
+      "loss": 0.4004,
+      "step": 480
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 0.00014100491249672498,
+      "loss": 0.3802,
+      "step": 481
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014076836149416887,
+      "loss": 0.4443,
+      "step": 482
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001405315365755379,
+      "loss": 0.4134,
+      "step": 483
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001402944393320206,
+      "loss": 0.3651,
+      "step": 484
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00014005707135663527,
+      "loss": 0.4471,
+      "step": 485
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013981943424421932,
+      "loss": 0.4264,
+      "step": 486
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013958152959141825,
+      "loss": 0.3674,
+      "step": 487
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013934335899667527,
+      "loss": 0.3872,
+      "step": 488
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013910492406022033,
+      "loss": 0.4823,
+      "step": 489
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013886622638405952,
+      "loss": 0.4459,
+      "step": 490
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.0001386272675719642,
+      "loss": 0.3766,
+      "step": 491
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013838804922946027,
+      "loss": 0.3992,
+      "step": 492
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 0.00013814857296381728,
+      "loss": 0.5281,
+      "step": 493
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013790884038403795,
+      "loss": 0.3719,
+      "step": 494
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013766885310084688,
+      "loss": 0.3139,
+      "step": 495
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013742861272668012,
+      "loss": 0.3363,
+      "step": 496
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013718812087567414,
+      "loss": 0.3636,
+      "step": 497
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013694737916365517,
+      "loss": 0.4486,
+      "step": 498
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.000136706389208128,
+      "loss": 0.4413,
+      "step": 499
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013646515262826552,
+      "loss": 0.37,
+      "step": 500
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013622367104489756,
+      "loss": 0.33,
+      "step": 501
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001359819460805001,
+      "loss": 0.3901,
+      "step": 502
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001357399793591844,
+      "loss": 0.3398,
+      "step": 503
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.0001354977725066859,
+      "loss": 0.3644,
+      "step": 504
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013525532715035366,
+      "loss": 0.3752,
+      "step": 505
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 0.00013501264491913906,
+      "loss": 0.4211,
+      "step": 506
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013476972744358507,
+      "loss": 0.3039,
+      "step": 507
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001345265763558152,
+      "loss": 0.3783,
+      "step": 508
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013428319328952253,
+      "loss": 0.3472,
+      "step": 509
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013403957987995882,
+      "loss": 0.3942,
+      "step": 510
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001337957377639235,
+      "loss": 0.4851,
+      "step": 511
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001335516685797525,
+      "loss": 0.3841,
+      "step": 512
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001333073739673076,
+      "loss": 0.4186,
+      "step": 513
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013306285556796495,
+      "loss": 0.4185,
+      "step": 514
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.0001328181150246045,
+      "loss": 0.4052,
+      "step": 515
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013257315398159864,
+      "loss": 0.4218,
+      "step": 516
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013232797408480127,
+      "loss": 0.3696,
+      "step": 517
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 0.00013208257698153677,
+      "loss": 0.3573,
+      "step": 518
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013183696432058888,
+      "loss": 0.4207,
+      "step": 519
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013159113775218964,
+      "loss": 0.3867,
+      "step": 520
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013134509892800822,
+      "loss": 0.4265,
+      "step": 521
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013109884950114007,
+      "loss": 0.3436,
+      "step": 522
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013085239112609547,
+      "loss": 0.3501,
+      "step": 523
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013060572545878875,
+      "loss": 0.2919,
+      "step": 524
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013035885415652685,
+      "loss": 0.3489,
+      "step": 525
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00013011177887799845,
+      "loss": 0.321,
+      "step": 526
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012986450128326266,
+      "loss": 0.5439,
+      "step": 527
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012961702303373795,
+      "loss": 0.3203,
+      "step": 528
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012936934579219094,
+      "loss": 0.4865,
+      "step": 529
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012912147122272523,
+      "loss": 0.343,
+      "step": 530
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 0.00012887340099077024,
+      "loss": 0.4279,
+      "step": 531
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012862513676307008,
+      "loss": 0.2932,
+      "step": 532
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001283766802076722,
+      "loss": 0.349,
+      "step": 533
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012812803299391628,
+      "loss": 0.3941,
+      "step": 534
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012787919679242306,
+      "loss": 0.3383,
+      "step": 535
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012763017327508305,
+      "loss": 0.3641,
+      "step": 536
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012738096411504522,
+      "loss": 0.4241,
+      "step": 537
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001271315709867059,
+      "loss": 0.3602,
+      "step": 538
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012688199556569753,
+      "loss": 0.3823,
+      "step": 539
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012663223952887723,
+      "loss": 0.3866,
+      "step": 540
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.0001263823045543158,
+      "loss": 0.4981,
+      "step": 541
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012613219232128608,
+      "loss": 0.4194,
+      "step": 542
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 0.00012588190451025207,
+      "loss": 0.3578,
+      "step": 543
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012563144280285741,
+      "loss": 0.3256,
+      "step": 544
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012538080888191408,
+      "loss": 0.3916,
+      "step": 545
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012513000443139112,
+      "loss": 0.4016,
+      "step": 546
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012487903113640337,
+      "loss": 0.3064,
+      "step": 547
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012462789068320017,
+      "loss": 0.3882,
+      "step": 548
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012437658475915377,
+      "loss": 0.3464,
+      "step": 549
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012412511505274844,
+      "loss": 0.4203,
+      "step": 550
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012387348325356874,
+      "loss": 0.3114,
+      "step": 551
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012362169105228826,
+      "loss": 0.4307,
+      "step": 552
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012336974014065844,
+      "loss": 0.3872,
+      "step": 553
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.000123117632211497,
+      "loss": 0.4227,
+      "step": 554
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012286536895867654,
+      "loss": 0.421,
+      "step": 555
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 0.00012261295207711346,
+      "loss": 0.3835,
+      "step": 556
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012236038326275626,
+      "loss": 0.3621,
+      "step": 557
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.0001221076642125742,
+      "loss": 0.4028,
+      "step": 558
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012185479662454595,
+      "loss": 0.3766,
+      "step": 559
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012160178219764837,
+      "loss": 0.3918,
+      "step": 560
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012134862263184467,
+      "loss": 0.3743,
+      "step": 561
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012109531962807332,
+      "loss": 0.4155,
+      "step": 562
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012084187488823657,
+      "loss": 0.2959,
+      "step": 563
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012058829011518896,
+      "loss": 0.3358,
+      "step": 564
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012033456701272576,
+      "loss": 0.3923,
+      "step": 565
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00012008070728557186,
+      "loss": 0.467,
+      "step": 566
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011982671263936995,
+      "loss": 0.3472,
+      "step": 567
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 0.00011957258478066931,
+      "loss": 0.3583,
+      "step": 568
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011931832541691418,
+      "loss": 0.3361,
+      "step": 569
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011906393625643244,
+      "loss": 0.3957,
+      "step": 570
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011880941900842397,
+      "loss": 0.3742,
+      "step": 571
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011855477538294935,
+      "loss": 0.3296,
+      "step": 572
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011830000709091815,
+      "loss": 0.4858,
+      "step": 573
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011804511584407763,
+      "loss": 0.4322,
+      "step": 574
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.0001177901033550012,
+      "loss": 0.3287,
+      "step": 575
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011753497133707679,
+      "loss": 0.3797,
+      "step": 576
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011727972150449544,
+      "loss": 0.3018,
+      "step": 577
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011702435557223987,
+      "loss": 0.3814,
+      "step": 578
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011676887525607271,
+      "loss": 0.3249,
+      "step": 579
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011651328227252517,
+      "loss": 0.3203,
+      "step": 580
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 0.00011625757833888551,
+      "loss": 0.375,
+      "step": 581
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011600176517318741,
+      "loss": 0.382,
+      "step": 582
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001157458444941984,
+      "loss": 0.4611,
+      "step": 583
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011548981802140848,
+      "loss": 0.44,
+      "step": 584
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011523368747501839,
+      "loss": 0.3354,
+      "step": 585
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011497745457592816,
+      "loss": 0.3448,
+      "step": 586
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011472112104572547,
+      "loss": 0.4108,
+      "step": 587
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011446468860667421,
+      "loss": 0.389,
+      "step": 588
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.0001142081589817027,
+      "loss": 0.3,
+      "step": 589
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011395153389439233,
+      "loss": 0.352,
+      "step": 590
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011369481506896582,
+      "loss": 0.3715,
+      "step": 591
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011343800423027582,
+      "loss": 0.374,
+      "step": 592
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 0.00011318110310379301,
+      "loss": 0.3661,
+      "step": 593
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.0001129241134155949,
+      "loss": 0.3727,
+      "step": 594
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011266703689235394,
+      "loss": 0.4159,
+      "step": 595
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011240987526132594,
+      "loss": 0.3104,
+      "step": 596
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011215263025033869,
+      "loss": 0.3427,
+      "step": 597
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011189530358778005,
+      "loss": 0.3454,
+      "step": 598
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011163789700258655,
+      "loss": 0.4066,
+      "step": 599
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011138041222423177,
+      "loss": 0.3483,
+      "step": 600
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011112285098271451,
+      "loss": 0.3454,
+      "step": 601
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011086521500854745,
+      "loss": 0.3292,
+      "step": 602
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011060750603274535,
+      "loss": 0.3044,
+      "step": 603
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011034972578681338,
+      "loss": 0.3086,
+      "step": 604
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00011009187600273566,
+      "loss": 0.3375,
+      "step": 605
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 0.00010983395841296348,
+      "loss": 0.3608,
+      "step": 606
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010957597475040373,
+      "loss": 0.3631,
+      "step": 607
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010931792674840718,
+      "loss": 0.3704,
+      "step": 608
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010905981614075693,
+      "loss": 0.3356,
+      "step": 609
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010880164466165674,
+      "loss": 0.412,
+      "step": 610
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010854341404571928,
+      "loss": 0.3183,
+      "step": 611
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010828512602795462,
+      "loss": 0.38,
+      "step": 612
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010802678234375851,
+      "loss": 0.4332,
+      "step": 613
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010776838472890065,
+      "loss": 0.3537,
+      "step": 614
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.0001075099349195131,
+      "loss": 0.4136,
+      "step": 615
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010725143465207867,
+      "loss": 0.2705,
+      "step": 616
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010699288566341914,
+      "loss": 0.3217,
+      "step": 617
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 0.00010673428969068364,
+      "loss": 0.3613,
+      "step": 618
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.000106475648471337,
+      "loss": 0.3649,
+      "step": 619
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010621696374314807,
+      "loss": 0.3684,
+      "step": 620
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010595823724417795,
+      "loss": 0.3609,
+      "step": 621
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010569947071276847,
+      "loss": 0.3229,
+      "step": 622
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010544066588753044,
+      "loss": 0.3284,
+      "step": 623
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010518182450733186,
+      "loss": 0.4019,
+      "step": 624
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010492294831128641,
+      "loss": 0.3447,
+      "step": 625
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010466403903874176,
+      "loss": 0.4262,
+      "step": 626
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010440509842926767,
+      "loss": 0.4816,
+      "step": 627
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010414612822264455,
+      "loss": 0.4194,
+      "step": 628
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010388713015885161,
+      "loss": 0.3295,
+      "step": 629
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010362810597805526,
+      "loss": 0.4148,
+      "step": 630
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 0.00010336905742059742,
+      "loss": 0.3208,
+      "step": 631
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001031099862269837,
+      "loss": 0.3423,
+      "step": 632
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001028508941378719,
+      "loss": 0.2693,
+      "step": 633
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010259178289406011,
+      "loss": 0.3091,
+      "step": 634
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010233265423647523,
+      "loss": 0.3189,
+      "step": 635
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010207350990616107,
+      "loss": 0.3388,
+      "step": 636
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010181435164426676,
+      "loss": 0.3758,
+      "step": 637
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.0001015551811920351,
+      "loss": 0.4182,
+      "step": 638
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010129600029079072,
+      "loss": 0.3962,
+      "step": 639
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010103681068192845,
+      "loss": 0.3378,
+      "step": 640
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010077761410690172,
+      "loss": 0.4199,
+      "step": 641
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010051841230721065,
+      "loss": 0.2436,
+      "step": 642
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 0.00010025920702439051,
+      "loss": 0.3342,
+      "step": 643
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 0.0001,
+      "loss": 0.3424,
+      "step": 644
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.97407929756095e-05,
+      "loss": 0.3596,
+      "step": 645
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.948158769278939e-05,
+      "loss": 0.2859,
+      "step": 646
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.92223858930983e-05,
+      "loss": 0.4127,
+      "step": 647
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.896318931807155e-05,
+      "loss": 0.4664,
+      "step": 648
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.870399970920932e-05,
+      "loss": 0.3487,
+      "step": 649
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.844481880796491e-05,
+      "loss": 0.3971,
+      "step": 650
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.818564835573323e-05,
+      "loss": 0.3814,
+      "step": 651
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.792649009383899e-05,
+      "loss": 0.3574,
+      "step": 652
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.766734576352478e-05,
+      "loss": 0.3591,
+      "step": 653
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.740821710593989e-05,
+      "loss": 0.3586,
+      "step": 654
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.714910586212816e-05,
+      "loss": 0.404,
+      "step": 655
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.689001377301633e-05,
+      "loss": 0.293,
+      "step": 656
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.663094257940258e-05,
+      "loss": 0.388,
+      "step": 657
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.637189402194476e-05,
+      "loss": 0.3348,
+      "step": 658
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.611286984114841e-05,
+      "loss": 0.3796,
+      "step": 659
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.585387177735547e-05,
+      "loss": 0.4278,
+      "step": 660
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.559490157073236e-05,
+      "loss": 0.3951,
+      "step": 661
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.533596096125825e-05,
+      "loss": 0.3523,
+      "step": 662
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.507705168871358e-05,
+      "loss": 0.3271,
+      "step": 663
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.481817549266817e-05,
+      "loss": 0.3204,
+      "step": 664
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.455933411246958e-05,
+      "loss": 0.4091,
+      "step": 665
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.430052928723153e-05,
+      "loss": 0.299,
+      "step": 666
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.404176275582208e-05,
+      "loss": 0.3839,
+      "step": 667
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.378303625685195e-05,
+      "loss": 0.3391,
+      "step": 668
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.352435152866298e-05,
+      "loss": 0.4553,
+      "step": 669
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.326571030931637e-05,
+      "loss": 0.3467,
+      "step": 670
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.300711433658087e-05,
+      "loss": 0.3471,
+      "step": 671
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.274856534792138e-05,
+      "loss": 0.4564,
+      "step": 672
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.249006508048694e-05,
+      "loss": 0.4898,
+      "step": 673
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.223161527109937e-05,
+      "loss": 0.4517,
+      "step": 674
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.197321765624152e-05,
+      "loss": 0.4256,
+      "step": 675
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.171487397204539e-05,
+      "loss": 0.2827,
+      "step": 676
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.145658595428074e-05,
+      "loss": 0.4068,
+      "step": 677
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.119835533834331e-05,
+      "loss": 0.3051,
+      "step": 678
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.09401838592431e-05,
+      "loss": 0.398,
+      "step": 679
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.068207325159284e-05,
+      "loss": 0.3798,
+      "step": 680
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.04240252495963e-05,
+      "loss": 0.3658,
+      "step": 681
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.016604158703654e-05,
+      "loss": 0.3392,
+      "step": 682
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.990812399726435e-05,
+      "loss": 0.3456,
+      "step": 683
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.965027421318665e-05,
+      "loss": 0.3149,
+      "step": 684
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.939249396725467e-05,
+      "loss": 0.3265,
+      "step": 685
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.913478499145254e-05,
+      "loss": 0.3778,
+      "step": 686
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.887714901728551e-05,
+      "loss": 0.3881,
+      "step": 687
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.861958777576827e-05,
+      "loss": 0.3836,
+      "step": 688
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.836210299741346e-05,
+      "loss": 0.4346,
+      "step": 689
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.810469641222001e-05,
+      "loss": 0.3806,
+      "step": 690
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.784736974966135e-05,
+      "loss": 0.3587,
+      "step": 691
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.759012473867407e-05,
+      "loss": 0.3003,
+      "step": 692
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.733296310764611e-05,
+      "loss": 0.4033,
+      "step": 693
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.707588658440511e-05,
+      "loss": 0.3338,
+      "step": 694
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.6818896896207e-05,
+      "loss": 0.4128,
+      "step": 695
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.656199576972423e-05,
+      "loss": 0.3772,
+      "step": 696
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.63051849310342e-05,
+      "loss": 0.3233,
+      "step": 697
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.604846610560771e-05,
+      "loss": 0.3677,
+      "step": 698
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.579184101829734e-05,
+      "loss": 0.2934,
+      "step": 699
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.553531139332582e-05,
+      "loss": 0.4106,
+      "step": 700
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.527887895427454e-05,
+      "loss": 0.3216,
+      "step": 701
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.502254542407186e-05,
+      "loss": 0.3369,
+      "step": 702
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.476631252498162e-05,
+      "loss": 0.3274,
+      "step": 703
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.451018197859153e-05,
+      "loss": 0.3047,
+      "step": 704
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.425415550580162e-05,
+      "loss": 0.3228,
+      "step": 705
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.399823482681262e-05,
+      "loss": 0.403,
+      "step": 706
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.374242166111448e-05,
+      "loss": 0.3588,
+      "step": 707
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.348671772747487e-05,
+      "loss": 0.3133,
+      "step": 708
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.323112474392731e-05,
+      "loss": 0.3715,
+      "step": 709
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.297564442776014e-05,
+      "loss": 0.3624,
+      "step": 710
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.272027849550457e-05,
+      "loss": 0.3512,
+      "step": 711
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.246502866292324e-05,
+      "loss": 0.4024,
+      "step": 712
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.220989664499878e-05,
+      "loss": 0.3788,
+      "step": 713
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.195488415592238e-05,
+      "loss": 0.329,
+      "step": 714
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.169999290908188e-05,
+      "loss": 0.3386,
+      "step": 715
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.144522461705067e-05,
+      "loss": 0.2401,
+      "step": 716
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.119058099157604e-05,
+      "loss": 0.3926,
+      "step": 717
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.093606374356759e-05,
+      "loss": 0.3315,
+      "step": 718
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.068167458308582e-05,
+      "loss": 0.4074,
+      "step": 719
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.042741521933071e-05,
+      "loss": 0.4517,
+      "step": 720
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.017328736063006e-05,
+      "loss": 0.3476,
+      "step": 721
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.991929271442817e-05,
+      "loss": 0.2977,
+      "step": 722
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.966543298727425e-05,
+      "loss": 0.3368,
+      "step": 723
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.941170988481108e-05,
+      "loss": 0.3237,
+      "step": 724
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.915812511176347e-05,
+      "loss": 0.3455,
+      "step": 725
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.89046803719267e-05,
+      "loss": 0.3412,
+      "step": 726
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.865137736815535e-05,
+      "loss": 0.4571,
+      "step": 727
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.839821780235168e-05,
+      "loss": 0.422,
+      "step": 728
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.814520337545406e-05,
+      "loss": 0.4013,
+      "step": 729
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.789233578742582e-05,
+      "loss": 0.3223,
+      "step": 730
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.763961673724379e-05,
+      "loss": 0.3291,
+      "step": 731
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.738704792288655e-05,
+      "loss": 0.3519,
+      "step": 732
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.713463104132345e-05,
+      "loss": 0.4064,
+      "step": 733
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.688236778850306e-05,
+      "loss": 0.395,
+      "step": 734
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.663025985934158e-05,
+      "loss": 0.3674,
+      "step": 735
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.637830894771175e-05,
+      "loss": 0.4724,
+      "step": 736
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.61265167464313e-05,
+      "loss": 0.3368,
+      "step": 737
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.587488494725157e-05,
+      "loss": 0.3682,
+      "step": 738
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.562341524084623e-05,
+      "loss": 0.3312,
+      "step": 739
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.537210931679987e-05,
+      "loss": 0.35,
+      "step": 740
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.512096886359664e-05,
+      "loss": 0.3069,
+      "step": 741
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.48699955686089e-05,
+      "loss": 0.3537,
+      "step": 742
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.461919111808595e-05,
+      "loss": 0.3789,
+      "step": 743
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.43685571971426e-05,
+      "loss": 0.3782,
+      "step": 744
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.411809548974792e-05,
+      "loss": 0.2462,
+      "step": 745
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.386780767871397e-05,
+      "loss": 0.3947,
+      "step": 746
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.361769544568425e-05,
+      "loss": 0.2983,
+      "step": 747
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.336776047112276e-05,
+      "loss": 0.4125,
+      "step": 748
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.311800443430251e-05,
+      "loss": 0.4534,
+      "step": 749
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.286842901329412e-05,
+      "loss": 0.3619,
+      "step": 750
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.26190358849548e-05,
+      "loss": 0.276,
+      "step": 751
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.236982672491698e-05,
+      "loss": 0.3009,
+      "step": 752
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.212080320757695e-05,
+      "loss": 0.3443,
+      "step": 753
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.187196700608373e-05,
+      "loss": 0.4037,
+      "step": 754
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.162331979232783e-05,
+      "loss": 0.3165,
+      "step": 755
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.137486323692995e-05,
+      "loss": 0.3483,
+      "step": 756
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.112659900922976e-05,
+      "loss": 0.4637,
+      "step": 757
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.087852877727481e-05,
+      "loss": 0.4686,
+      "step": 758
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.06306542078091e-05,
+      "loss": 0.3235,
+      "step": 759
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.038297696626206e-05,
+      "loss": 0.3769,
+      "step": 760
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.013549871673736e-05,
+      "loss": 0.409,
+      "step": 761
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.988822112200156e-05,
+      "loss": 0.3452,
+      "step": 762
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.964114584347316e-05,
+      "loss": 0.4128,
+      "step": 763
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.939427454121128e-05,
+      "loss": 0.3713,
+      "step": 764
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.914760887390452e-05,
+      "loss": 0.3782,
+      "step": 765
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.890115049885994e-05,
+      "loss": 0.3576,
+      "step": 766
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.865490107199181e-05,
+      "loss": 0.2946,
+      "step": 767
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.84088622478104e-05,
+      "loss": 0.3249,
+      "step": 768
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.816303567941112e-05,
+      "loss": 0.3997,
+      "step": 769
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.791742301846326e-05,
+      "loss": 0.4067,
+      "step": 770
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.767202591519875e-05,
+      "loss": 0.4097,
+      "step": 771
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.742684601840141e-05,
+      "loss": 0.3873,
+      "step": 772
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.718188497539554e-05,
+      "loss": 0.3568,
+      "step": 773
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.693714443203507e-05,
+      "loss": 0.3249,
+      "step": 774
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.669262603269246e-05,
+      "loss": 0.2828,
+      "step": 775
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.644833142024751e-05,
+      "loss": 0.3334,
+      "step": 776
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.620426223607654e-05,
+      "loss": 0.4247,
+      "step": 777
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.59604201200412e-05,
+      "loss": 0.3164,
+      "step": 778
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.571680671047749e-05,
+      "loss": 0.3579,
+      "step": 779
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.547342364418481e-05,
+      "loss": 0.3214,
+      "step": 780
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.523027255641493e-05,
+      "loss": 0.3181,
+      "step": 781
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.498735508086093e-05,
+      "loss": 0.3368,
+      "step": 782
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.474467284964634e-05,
+      "loss": 0.3751,
+      "step": 783
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.450222749331414e-05,
+      "loss": 0.3211,
+      "step": 784
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.426002064081565e-05,
+      "loss": 0.3162,
+      "step": 785
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.40180539194999e-05,
+      "loss": 0.365,
+      "step": 786
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.377632895510248e-05,
+      "loss": 0.3723,
+      "step": 787
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.35348473717345e-05,
+      "loss": 0.4449,
+      "step": 788
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.329361079187199e-05,
+      "loss": 0.355,
+      "step": 789
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.305262083634488e-05,
+      "loss": 0.4048,
+      "step": 790
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.281187912432587e-05,
+      "loss": 0.3333,
+      "step": 791
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.25713872733199e-05,
+      "loss": 0.4015,
+      "step": 792
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.233114689915316e-05,
+      "loss": 0.2927,
+      "step": 793
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.209115961596208e-05,
+      "loss": 0.3233,
+      "step": 794
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.18514270361827e-05,
+      "loss": 0.2932,
+      "step": 795
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.161195077053976e-05,
+      "loss": 0.3963,
+      "step": 796
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.13727324280358e-05,
+      "loss": 0.3269,
+      "step": 797
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.113377361594049e-05,
+      "loss": 0.3164,
+      "step": 798
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.08950759397797e-05,
+      "loss": 0.2922,
+      "step": 799
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.065664100332478e-05,
+      "loss": 0.3614,
+      "step": 800
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0418470408581774e-05,
+      "loss": 0.3039,
+      "step": 801
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.018056575578075e-05,
+      "loss": 0.3361,
+      "step": 802
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9942928643364724e-05,
+      "loss": 0.3199,
+      "step": 803
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.970556066797941e-05,
+      "loss": 0.391,
+      "step": 804
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.946846342446214e-05,
+      "loss": 0.3763,
+      "step": 805
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.923163850583113e-05,
+      "loss": 0.2789,
+      "step": 806
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.899508750327501e-05,
+      "loss": 0.3985,
+      "step": 807
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.875881200614207e-05,
+      "loss": 0.335,
+      "step": 808
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8522813601929324e-05,
+      "loss": 0.3806,
+      "step": 809
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.828709387627218e-05,
+      "loss": 0.3228,
+      "step": 810
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.80516544129337e-05,
+      "loss": 0.4046,
+      "step": 811
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.781649679379378e-05,
+      "loss": 0.3456,
+      "step": 812
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.758162259883867e-05,
+      "loss": 0.323,
+      "step": 813
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.73470334061505e-05,
+      "loss": 0.3047,
+      "step": 814
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7112730791896207e-05,
+      "loss": 0.4486,
+      "step": 815
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.687871633031754e-05,
+      "loss": 0.3545,
+      "step": 816
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.664499159372017e-05,
+      "loss": 0.4016,
+      "step": 817
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6411558152462894e-05,
+      "loss": 0.3081,
+      "step": 818
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.617841757494762e-05,
+      "loss": 0.41,
+      "step": 819
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5945571427608526e-05,
+      "loss": 0.3484,
+      "step": 820
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5713021274901335e-05,
+      "loss": 0.328,
+      "step": 821
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.54807686792933e-05,
+      "loss": 0.3437,
+      "step": 822
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.524881520125229e-05,
+      "loss": 0.4109,
+      "step": 823
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.501716239923642e-05,
+      "loss": 0.2963,
+      "step": 824
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4785811829683764e-05,
+      "loss": 0.3062,
+      "step": 825
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4554765047001613e-05,
+      "loss": 0.3305,
+      "step": 826
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.432402360355615e-05,
+      "loss": 0.3328,
+      "step": 827
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4093589049662175e-05,
+      "loss": 0.2893,
+      "step": 828
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.386346293357242e-05,
+      "loss": 0.3099,
+      "step": 829
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.363364680146725e-05,
+      "loss": 0.2945,
+      "step": 830
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3404142197444506e-05,
+      "loss": 0.2948,
+      "step": 831
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.31749506635086e-05,
+      "loss": 0.3885,
+      "step": 832
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2946073739560706e-05,
+      "loss": 0.3883,
+      "step": 833
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.271751296338823e-05,
+      "loss": 0.3405,
+      "step": 834
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.248926987065417e-05,
+      "loss": 0.3905,
+      "step": 835
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.226134599488728e-05,
+      "loss": 0.3323,
+      "step": 836
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.203374286747158e-05,
+      "loss": 0.3305,
+      "step": 837
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.180646201763577e-05,
+      "loss": 0.4387,
+      "step": 838
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.15795049724435e-05,
+      "loss": 0.4234,
+      "step": 839
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.135287325678271e-05,
+      "loss": 0.3835,
+      "step": 840
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.112656839335543e-05,
+      "loss": 0.3598,
+      "step": 841
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.090059190266779e-05,
+      "loss": 0.4039,
+      "step": 842
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0674945303019526e-05,
+      "loss": 0.3867,
+      "step": 843
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0449630110493836e-05,
+      "loss": 0.3405,
+      "step": 844
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.022464783894744e-05,
+      "loss": 0.3879,
+      "step": 845
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.000000000000002e-05,
+      "loss": 0.3996,
+      "step": 846
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.977568810302432e-05,
+      "loss": 0.3765,
+      "step": 847
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.955171365513603e-05,
+      "loss": 0.3764,
+      "step": 848
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9328078161183464e-05,
+      "loss": 0.3178,
+      "step": 849
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9104783123737566e-05,
+      "loss": 0.3766,
+      "step": 850
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.88818300430819e-05,
+      "loss": 0.2917,
+      "step": 851
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.865922041720239e-05,
+      "loss": 0.2525,
+      "step": 852
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.843695574177737e-05,
+      "loss": 0.2865,
+      "step": 853
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.821503751016746e-05,
+      "loss": 0.3114,
+      "step": 854
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7993467213405706e-05,
+      "loss": 0.3916,
+      "step": 855
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.777224634018732e-05,
+      "loss": 0.2784,
+      "step": 856
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.755137637685979e-05,
+      "loss": 0.2886,
+      "step": 857
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.733085880741301e-05,
+      "loss": 0.3188,
+      "step": 858
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7110695113469085e-05,
+      "loss": 0.4117,
+      "step": 859
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.689088677427249e-05,
+      "loss": 0.3191,
+      "step": 860
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6671435266680216e-05,
+      "loss": 0.3056,
+      "step": 861
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.645234206515171e-05,
+      "loss": 0.3189,
+      "step": 862
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.623360864173893e-05,
+      "loss": 0.295,
+      "step": 863
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6015236466076747e-05,
+      "loss": 0.4706,
+      "step": 864
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.579722700537268e-05,
+      "loss": 0.2599,
+      "step": 865
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5579581724397255e-05,
+      "loss": 0.3834,
+      "step": 866
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5362302085474254e-05,
+      "loss": 0.3784,
+      "step": 867
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.514538954847064e-05,
+      "loss": 0.3768,
+      "step": 868
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.492884557078688e-05,
+      "loss": 0.333,
+      "step": 869
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.471267160734731e-05,
+      "loss": 0.4112,
+      "step": 870
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.449686911058992e-05,
+      "loss": 0.3116,
+      "step": 871
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.428143953045717e-05,
+      "loss": 0.399,
+      "step": 872
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.406638431438576e-05,
+      "loss": 0.4131,
+      "step": 873
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.385170490729712e-05,
+      "loss": 0.4145,
+      "step": 874
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.36374027515878e-05,
+      "loss": 0.422,
+      "step": 875
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.342347928711953e-05,
+      "loss": 0.2728,
+      "step": 876
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.320993595120969e-05,
+      "loss": 0.4189,
+      "step": 877
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2996774178621736e-05,
+      "loss": 0.3222,
+      "step": 878
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.278399540155536e-05,
+      "loss": 0.3664,
+      "step": 879
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.257160104963696e-05,
+      "loss": 0.3898,
+      "step": 880
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2359592549910145e-05,
+      "loss": 0.4123,
+      "step": 881
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2147971326825966e-05,
+      "loss": 0.4391,
+      "step": 882
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.193673880223339e-05,
+      "loss": 0.3474,
+      "step": 883
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.172589639536991e-05,
+      "loss": 0.333,
+      "step": 884
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1515445522851784e-05,
+      "loss": 0.2805,
+      "step": 885
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.130538759866457e-05,
+      "loss": 0.3558,
+      "step": 886
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.109572403415386e-05,
+      "loss": 0.3208,
+      "step": 887
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.088645623801534e-05,
+      "loss": 0.2736,
+      "step": 888
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0677585616285774e-05,
+      "loss": 0.3262,
+      "step": 889
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.046911357233343e-05,
+      "loss": 0.2908,
+      "step": 890
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.026104150684835e-05,
+      "loss": 0.3344,
+      "step": 891
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.00533708178334e-05,
+      "loss": 0.2956,
+      "step": 892
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.984610290059467e-05,
+      "loss": 0.3396,
+      "step": 893
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.963923914773187e-05,
+      "loss": 0.3109,
+      "step": 894
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.943278094912946e-05,
+      "loss": 0.3538,
+      "step": 895
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.922672969194686e-05,
+      "loss": 0.3245,
+      "step": 896
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.902108676060937e-05,
+      "loss": 0.3714,
+      "step": 897
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8815853536798904e-05,
+      "loss": 0.3984,
+      "step": 898
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.861103139944449e-05,
+      "loss": 0.3998,
+      "step": 899
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.840662172471315e-05,
+      "loss": 0.4233,
+      "step": 900
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.820262588600074e-05,
+      "loss": 0.2933,
+      "step": 901
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.79990452539225e-05,
+      "loss": 0.334,
+      "step": 902
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7795881196303995e-05,
+      "loss": 0.2884,
+      "step": 903
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.759313507817196e-05,
+      "loss": 0.3179,
+      "step": 904
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.739080826174498e-05,
+      "loss": 0.3418,
+      "step": 905
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7188902106424416e-05,
+      "loss": 0.3086,
+      "step": 906
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6987417968785366e-05,
+      "loss": 0.4863,
+      "step": 907
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.678635720256737e-05,
+      "loss": 0.3178,
+      "step": 908
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.658572115866541e-05,
+      "loss": 0.3112,
+      "step": 909
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.638551118512089e-05,
+      "loss": 0.2868,
+      "step": 910
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.618572862711247e-05,
+      "loss": 0.2929,
+      "step": 911
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5986374826947066e-05,
+      "loss": 0.3695,
+      "step": 912
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.578745112405083e-05,
+      "loss": 0.2974,
+      "step": 913
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.558895885496023e-05,
+      "loss": 0.3214,
+      "step": 914
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.539089935331294e-05,
+      "loss": 0.3335,
+      "step": 915
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.519327394983888e-05,
+      "loss": 0.3607,
+      "step": 916
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4996083972351515e-05,
+      "loss": 0.2875,
+      "step": 917
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.479933074573858e-05,
+      "loss": 0.5191,
+      "step": 918
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4603015591953395e-05,
+      "loss": 0.375,
+      "step": 919
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.440713983000601e-05,
+      "loss": 0.3537,
+      "step": 920
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.421170477595419e-05,
+      "loss": 0.2802,
+      "step": 921
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.401671174289469e-05,
+      "loss": 0.3784,
+      "step": 922
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3822162040954354e-05,
+      "loss": 0.3554,
+      "step": 923
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.362805697728145e-05,
+      "loss": 0.3427,
+      "step": 924
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.34343978560367e-05,
+      "loss": 0.3137,
+      "step": 925
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.324118597838464e-05,
+      "loss": 0.3402,
+      "step": 926
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3048422642484886e-05,
+      "loss": 0.3334,
+      "step": 927
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.285610914348332e-05,
+      "loss": 0.3811,
+      "step": 928
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.266424677350346e-05,
+      "loss": 0.3654,
+      "step": 929
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2472836821637744e-05,
+      "loss": 0.3917,
+      "step": 930
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.228188057393895e-05,
+      "loss": 0.3913,
+      "step": 931
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.209137931341143e-05,
+      "loss": 0.3505,
+      "step": 932
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.190133432000252e-05,
+      "loss": 0.3261,
+      "step": 933
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1711746870594086e-05,
+      "loss": 0.2955,
+      "step": 934
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1522618238993725e-05,
+      "loss": 0.3575,
+      "step": 935
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1333949695926324e-05,
+      "loss": 0.3459,
+      "step": 936
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.114574250902558e-05,
+      "loss": 0.3576,
+      "step": 937
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0957997942825336e-05,
+      "loss": 0.3736,
+      "step": 938
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.077071725875116e-05,
+      "loss": 0.3324,
+      "step": 939
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.058390171511196e-05,
+      "loss": 0.3328,
+      "step": 940
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0397552567091337e-05,
+      "loss": 0.4085,
+      "step": 941
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.021167106673928e-05,
+      "loss": 0.4298,
+      "step": 942
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0026258462963787e-05,
+      "loss": 0.3834,
+      "step": 943
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9841316001522347e-05,
+      "loss": 0.3395,
+      "step": 944
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9656844925013637e-05,
+      "loss": 0.3696,
+      "step": 945
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9472846472869298e-05,
+      "loss": 0.3298,
+      "step": 946
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9289321881345254e-05,
+      "loss": 0.395,
+      "step": 947
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9106272383513835e-05,
+      "loss": 0.2746,
+      "step": 948
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8923699209255284e-05,
+      "loss": 0.3218,
+      "step": 949
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.874160358524931e-05,
+      "loss": 0.3004,
+      "step": 950
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8559986734967282e-05,
+      "loss": 0.321,
+      "step": 951
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8378849878663628e-05,
+      "loss": 0.3177,
+      "step": 952
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.819819423336775e-05,
+      "loss": 0.3519,
+      "step": 953
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8018021012875994e-05,
+      "loss": 0.2873,
+      "step": 954
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7838331427743282e-05,
+      "loss": 0.3627,
+      "step": 955
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7659126685275027e-05,
+      "loss": 0.3981,
+      "step": 956
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7480407989519198e-05,
+      "loss": 0.343,
+      "step": 957
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7302176541257986e-05,
+      "loss": 0.2593,
+      "step": 958
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.712443353799984e-05,
+      "loss": 0.2304,
+      "step": 959
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6947180173971508e-05,
+      "loss": 0.3479,
+      "step": 960
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.677041764010988e-05,
+      "loss": 0.2976,
+      "step": 961
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.659414712405398e-05,
+      "loss": 0.3588,
+      "step": 962
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6418369810137188e-05,
+      "loss": 0.2836,
+      "step": 963
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6243086879379e-05,
+      "loss": 0.2683,
+      "step": 964
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6068299509477266e-05,
+      "loss": 0.344,
+      "step": 965
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5894008874800325e-05,
+      "loss": 0.3694,
+      "step": 966
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5720216146378917e-05,
+      "loss": 0.3771,
+      "step": 967
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5546922491898495e-05,
+      "loss": 0.3577,
+      "step": 968
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5374129075691265e-05,
+      "loss": 0.3676,
+      "step": 969
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5201837058728505e-05,
+      "loss": 0.329,
+      "step": 970
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.503004759861258e-05,
+      "loss": 0.3751,
+      "step": 971
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.485876184956928e-05,
+      "loss": 0.3913,
+      "step": 972
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4687980962440072e-05,
+      "loss": 0.3014,
+      "step": 973
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.451770608467432e-05,
+      "loss": 0.3384,
+      "step": 974
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4347938360321566e-05,
+      "loss": 0.4399,
+      "step": 975
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.417867893002387e-05,
+      "loss": 0.4843,
+      "step": 976
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.400992893100822e-05,
+      "loss": 0.3088,
+      "step": 977
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3841689497078746e-05,
+      "loss": 0.4318,
+      "step": 978
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3673961758609152e-05,
+      "loss": 0.4099,
+      "step": 979
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3506746842535242e-05,
+      "loss": 0.3759,
+      "step": 980
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.334004587234717e-05,
+      "loss": 0.2787,
+      "step": 981
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3173859968081944e-05,
+      "loss": 0.3308,
+      "step": 982
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.300819024631603e-05,
+      "loss": 0.4735,
+      "step": 983
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2843037820157675e-05,
+      "loss": 0.3671,
+      "step": 984
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.26784037992395e-05,
+      "loss": 0.2745,
+      "step": 985
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.251428928971102e-05,
+      "loss": 0.3244,
+      "step": 986
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2350695394231345e-05,
+      "loss": 0.364,
+      "step": 987
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2187623211961562e-05,
+      "loss": 0.369,
+      "step": 988
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2025073838557454e-05,
+      "loss": 0.3967,
+      "step": 989
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1863048366162208e-05,
+      "loss": 0.2173,
+      "step": 990
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1701547883398922e-05,
+      "loss": 0.3552,
+      "step": 991
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1540573475363402e-05,
+      "loss": 0.3459,
+      "step": 992
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.138012622361689e-05,
+      "loss": 0.291,
+      "step": 993
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1220207206178688e-05,
+      "loss": 0.3307,
+      "step": 994
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.106081749751897e-05,
+      "loss": 0.3772,
+      "step": 995
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0901958168551638e-05,
+      "loss": 0.2498,
+      "step": 996
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0743630286627002e-05,
+      "loss": 0.3048,
+      "step": 997
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.058583491552465e-05,
+      "loss": 0.2953,
+      "step": 998
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0428573115446392e-05,
+      "loss": 0.2986,
+      "step": 999
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.027184594300898e-05,
+      "loss": 0.3984,
+      "step": 1000
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.011565445123711e-05,
+      "loss": 0.3912,
+      "step": 1001
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.995999968955641e-05,
+      "loss": 0.3576,
+      "step": 1002
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.980488270378612e-05,
+      "loss": 0.3282,
+      "step": 1003
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9650304536132426e-05,
+      "loss": 0.4322,
+      "step": 1004
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9496266225181248e-05,
+      "loss": 0.2882,
+      "step": 1005
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9342768805891178e-05,
+      "loss": 0.4054,
+      "step": 1006
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.918981330958678e-05,
+      "loss": 0.2423,
+      "step": 1007
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.903740076395151e-05,
+      "loss": 0.2636,
+      "step": 1008
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8885532193020704e-05,
+      "loss": 0.3519,
+      "step": 1009
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8734208617174988e-05,
+      "loss": 0.3776,
+      "step": 1010
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8583431053133127e-05,
+      "loss": 0.3947,
+      "step": 1011
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8433200513945337e-05,
+      "loss": 0.3286,
+      "step": 1012
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8283518008986567e-05,
+      "loss": 0.4002,
+      "step": 1013
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8134384543949478e-05,
+      "loss": 0.3874,
+      "step": 1014
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7985801120837865e-05,
+      "loss": 0.3938,
+      "step": 1015
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.783776873795994e-05,
+      "loss": 0.2722,
+      "step": 1016
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7690288389921493e-05,
+      "loss": 0.3343,
+      "step": 1017
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.754336106761927e-05,
+      "loss": 0.2909,
+      "step": 1018
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.739698775823442e-05,
+      "loss": 0.3441,
+      "step": 1019
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7251169445225657e-05,
+      "loss": 0.2899,
+      "step": 1020
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7105907108322816e-05,
+      "loss": 0.3915,
+      "step": 1021
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.696120172352025e-05,
+      "loss": 0.3449,
+      "step": 1022
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6817054263070174e-05,
+      "loss": 0.3066,
+      "step": 1023
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6673465695476232e-05,
+      "loss": 0.3762,
+      "step": 1024
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6530436985486996e-05,
+      "loss": 0.4618,
+      "step": 1025
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6387969094089316e-05,
+      "loss": 0.3111,
+      "step": 1026
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6246062978502164e-05,
+      "loss": 0.4103,
+      "step": 1027
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6104719592169902e-05,
+      "loss": 0.3055,
+      "step": 1028
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5963939884756042e-05,
+      "loss": 0.31,
+      "step": 1029
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5823724802136865e-05,
+      "loss": 0.2655,
+      "step": 1030
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5684075286394985e-05,
+      "loss": 0.377,
+      "step": 1031
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5544992275813053e-05,
+      "loss": 0.3571,
+      "step": 1032
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5406476704867524e-05,
+      "loss": 0.2985,
+      "step": 1033
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.526852950422226e-05,
+      "loss": 0.3773,
+      "step": 1034
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5131151600722337e-05,
+      "loss": 0.3805,
+      "step": 1035
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4994343917387854e-05,
+      "loss": 0.3597,
+      "step": 1036
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.485810737340767e-05,
+      "loss": 0.4539,
+      "step": 1037
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4722442884133214e-05,
+      "loss": 0.3775,
+      "step": 1038
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4587351361072454e-05,
+      "loss": 0.4079,
+      "step": 1039
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4452833711883628e-05,
+      "loss": 0.344,
+      "step": 1040
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4318890840369182e-05,
+      "loss": 0.2853,
+      "step": 1041
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4185523646469822e-05,
+      "loss": 0.3187,
+      "step": 1042
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4052733026258281e-05,
+      "loss": 0.383,
+      "step": 1043
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3920519871933424e-05,
+      "loss": 0.36,
+      "step": 1044
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3788885071814172e-05,
+      "loss": 0.266,
+      "step": 1045
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3657829510333654e-05,
+      "loss": 0.2738,
+      "step": 1046
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3527354068033139e-05,
+      "loss": 0.263,
+      "step": 1047
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.339745962155613e-05,
+      "loss": 0.4266,
+      "step": 1048
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.326814704364262e-05,
+      "loss": 0.3282,
+      "step": 1049
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3139417203123027e-05,
+      "loss": 0.3145,
+      "step": 1050
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3011270964912459e-05,
+      "loss": 0.3178,
+      "step": 1051
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2883709190004955e-05,
+      "loss": 0.3483,
+      "step": 1052
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.275673273546758e-05,
+      "loss": 0.3442,
+      "step": 1053
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.263034245443473e-05,
+      "loss": 0.3252,
+      "step": 1054
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2504539196102439e-05,
+      "loss": 0.3059,
+      "step": 1055
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2379323805722576e-05,
+      "loss": 0.3987,
+      "step": 1056
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2254697124597237e-05,
+      "loss": 0.3608,
+      "step": 1057
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2130659990073146e-05,
+      "loss": 0.3373,
+      "step": 1058
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2007213235535786e-05,
+      "loss": 0.2947,
+      "step": 1059
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1884357690404158e-05,
+      "loss": 0.3396,
+      "step": 1060
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.176209418012495e-05,
+      "loss": 0.3535,
+      "step": 1061
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1640423526166988e-05,
+      "loss": 0.2754,
+      "step": 1062
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1519346546015907e-05,
+      "loss": 0.3654,
+      "step": 1063
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1398864053168534e-05,
+      "loss": 0.3034,
+      "step": 1064
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1278976857127311e-05,
+      "loss": 0.3971,
+      "step": 1065
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1159685763395111e-05,
+      "loss": 0.3898,
+      "step": 1066
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1040991573469629e-05,
+      "loss": 0.3239,
+      "step": 1067
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0922895084838037e-05,
+      "loss": 0.3489,
+      "step": 1068
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0805397090971737e-05,
+      "loss": 0.3682,
+      "step": 1069
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0688498381320855e-05,
+      "loss": 0.3041,
+      "step": 1070
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.057219974130903e-05,
+      "loss": 0.2852,
+      "step": 1071
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.045650195232819e-05,
+      "loss": 0.2388,
+      "step": 1072
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0341405791733183e-05,
+      "loss": 0.2838,
+      "step": 1073
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0226912032836611e-05,
+      "loss": 0.3129,
+      "step": 1074
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0113021444903726e-05,
+      "loss": 0.3842,
+      "step": 1075
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.999734793146998e-06,
+      "loss": 0.3933,
+      "step": 1076
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.887052838721322e-06,
+      "loss": 0.3197,
+      "step": 1077
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.774976338718677e-06,
+      "loss": 0.3423,
+      "step": 1078
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.663506046162985e-06,
+      "loss": 0.3849,
+      "step": 1079
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.552642710005299e-06,
+      "loss": 0.365,
+      "step": 1080
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.44238707511862e-06,
+      "loss": 0.3233,
+      "step": 1081
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.332739882292752e-06,
+      "loss": 0.4528,
+      "step": 1082
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.22370186822965e-06,
+      "loss": 0.3937,
+      "step": 1083
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.115273765538202e-06,
+      "loss": 0.4314,
+      "step": 1084
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.0074563027294e-06,
+      "loss": 0.4369,
+      "step": 1085
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.900250204211514e-06,
+      "loss": 0.3812,
+      "step": 1086
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.79365619028507e-06,
+      "loss": 0.332,
+      "step": 1087
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.687674977138116e-06,
+      "loss": 0.3073,
+      "step": 1088
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.582307276841462e-06,
+      "loss": 0.3619,
+      "step": 1089
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.47755379734373e-06,
+      "loss": 0.3095,
+      "step": 1090
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.37341524246672e-06,
+      "loss": 0.3622,
+      "step": 1091
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.269892311900696e-06,
+      "loss": 0.3586,
+      "step": 1092
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.166985701199582e-06,
+      "loss": 0.3558,
+      "step": 1093
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.064696101776358e-06,
+      "loss": 0.2679,
+      "step": 1094
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.963024200898462e-06,
+      "loss": 0.3446,
+      "step": 1095
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.861970681683051e-06,
+      "loss": 0.2811,
+      "step": 1096
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.761536223092458e-06,
+      "loss": 0.304,
+      "step": 1097
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.661721499929753e-06,
+      "loss": 0.2873,
+      "step": 1098
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.562527182833978e-06,
+      "loss": 0.2931,
+      "step": 1099
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.463953938275858e-06,
+      "loss": 0.3258,
+      "step": 1100
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.366002428553153e-06,
+      "loss": 0.3395,
+      "step": 1101
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.2686733117863784e-06,
+      "loss": 0.3928,
+      "step": 1102
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.171967241914224e-06,
+      "loss": 0.3927,
+      "step": 1103
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.07588486868922e-06,
+      "loss": 0.3196,
+      "step": 1104
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.980426837673437e-06,
+      "loss": 0.2813,
+      "step": 1105
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.8855937902340576e-06,
+      "loss": 0.2564,
+      "step": 1106
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.791386363539065e-06,
+      "loss": 0.2908,
+      "step": 1107
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.6978051905530855e-06,
+      "loss": 0.3994,
+      "step": 1108
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.604850900032955e-06,
+      "loss": 0.3111,
+      "step": 1109
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.512524116523633e-06,
+      "loss": 0.4047,
+      "step": 1110
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.420825460353974e-06,
+      "loss": 0.4258,
+      "step": 1111
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.329755547632499e-06,
+      "loss": 0.286,
+      "step": 1112
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.239314990243339e-06,
+      "loss": 0.3303,
+      "step": 1113
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.149504395842087e-06,
+      "loss": 0.2815,
+      "step": 1114
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.0603243678516995e-06,
+      "loss": 0.2749,
+      "step": 1115
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.971775505458444e-06,
+      "loss": 0.3412,
+      "step": 1116
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.883858403607967e-06,
+      "loss": 0.3638,
+      "step": 1117
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.7965736530010916e-06,
+      "loss": 0.277,
+      "step": 1118
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.7099218400900716e-06,
+      "loss": 0.3253,
+      "step": 1119
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.623903547074549e-06,
+      "loss": 0.2732,
+      "step": 1120
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.538519351897575e-06,
+      "loss": 0.2596,
+      "step": 1121
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.453769828241872e-06,
+      "loss": 0.3846,
+      "step": 1122
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.369655545525909e-06,
+      "loss": 0.3616,
+      "step": 1123
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.286177068899989e-06,
+      "loss": 0.3023,
+      "step": 1124
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.2033349592426335e-06,
+      "loss": 0.358,
+      "step": 1125
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.121129773156663e-06,
+      "loss": 0.2486,
+      "step": 1126
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.039562062965508e-06,
+      "loss": 0.3509,
+      "step": 1127
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.95863237670956e-06,
+      "loss": 0.3041,
+      "step": 1128
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.87834125814235e-06,
+      "loss": 0.3634,
+      "step": 1129
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.798689246727006e-06,
+      "loss": 0.2367,
+      "step": 1130
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.719676877632639e-06,
+      "loss": 0.3476,
+      "step": 1131
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.641304681730641e-06,
+      "loss": 0.2854,
+      "step": 1132
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.563573185591219e-06,
+      "loss": 0.3789,
+      "step": 1133
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.486482911479839e-06,
+      "loss": 0.217,
+      "step": 1134
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4100343773536225e-06,
+      "loss": 0.289,
+      "step": 1135
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3342280968580285e-06,
+      "loss": 0.3003,
+      "step": 1136
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.259064579323302e-06,
+      "loss": 0.2561,
+      "step": 1137
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.184544329761009e-06,
+      "loss": 0.2877,
+      "step": 1138
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1106678488607495e-06,
+      "loss": 0.2936,
+      "step": 1139
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.037435632986786e-06,
+      "loss": 0.3117,
+      "step": 1140
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.964848174174541e-06,
+      "loss": 0.2497,
+      "step": 1141
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.892905960127546e-06,
+      "loss": 0.376,
+      "step": 1142
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.821609474213983e-06,
+      "loss": 0.31,
+      "step": 1143
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.750959195463466e-06,
+      "loss": 0.419,
+      "step": 1144
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6809555985639068e-06,
+      "loss": 0.4238,
+      "step": 1145
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.611599153858214e-06,
+      "loss": 0.2663,
+      "step": 1146
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5428903273411863e-06,
+      "loss": 0.3714,
+      "step": 1147
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4748295806564356e-06,
+      "loss": 0.2625,
+      "step": 1148
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.40741737109318e-06,
+      "loss": 0.3291,
+      "step": 1149
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3406541515832003e-06,
+      "loss": 0.273,
+      "step": 1150
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2745403706978872e-06,
+      "loss": 0.3681,
+      "step": 1151
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.209076472645112e-06,
+      "loss": 0.3584,
+      "step": 1152
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1442628972662704e-06,
+      "loss": 0.3623,
+      "step": 1153
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0801000800333877e-06,
+      "loss": 0.3027,
+      "step": 1154
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0165884520461316e-06,
+      "loss": 0.3221,
+      "step": 1155
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9537284400289355e-06,
+      "loss": 0.3278,
+      "step": 1156
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8915204663281013e-06,
+      "loss": 0.3158,
+      "step": 1157
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8299649489090475e-06,
+      "loss": 0.431,
+      "step": 1158
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7690623013533976e-06,
+      "loss": 0.3271,
+      "step": 1159
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.708812932856253e-06,
+      "loss": 0.3439,
+      "step": 1160
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.649217248223468e-06,
+      "loss": 0.292,
+      "step": 1161
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.590275647868867e-06,
+      "loss": 0.3679,
+      "step": 1162
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5319885278115906e-06,
+      "loss": 0.2992,
+      "step": 1163
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4743562796734622e-06,
+      "loss": 0.3715,
+      "step": 1164
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4173792906762804e-06,
+      "loss": 0.2487,
+      "step": 1165
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3610579436393e-06,
+      "loss": 0.2563,
+      "step": 1166
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3053926169765984e-06,
+      "loss": 0.2289,
+      "step": 1167
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.250383684694579e-06,
+      "loss": 0.291,
+      "step": 1168
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1960315163894075e-06,
+      "loss": 0.2936,
+      "step": 1169
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1423364772445887e-06,
+      "loss": 0.3852,
+      "step": 1170
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0892989280284823e-06,
+      "loss": 0.33,
+      "step": 1171
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.036919225091827e-06,
+      "loss": 0.3215,
+      "step": 1172
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9851977203654835e-06,
+      "loss": 0.2994,
+      "step": 1173
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9341347613579087e-06,
+      "loss": 0.3561,
+      "step": 1174
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8837306911529184e-06,
+      "loss": 0.3127,
+      "step": 1175
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8339858484073935e-06,
+      "loss": 0.2835,
+      "step": 1176
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7849005673489127e-06,
+      "loss": 0.3612,
+      "step": 1177
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7364751777736332e-06,
+      "loss": 0.3076,
+      "step": 1178
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6887100050439587e-06,
+      "loss": 0.3536,
+      "step": 1179
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6416053700863964e-06,
+      "loss": 0.3495,
+      "step": 1180
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.595161589389449e-06,
+      "loss": 0.2622,
+      "step": 1181
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5493789750014031e-06,
+      "loss": 0.3464,
+      "step": 1182
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5042578345283108e-06,
+      "loss": 0.3077,
+      "step": 1183
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.459798471131868e-06,
+      "loss": 0.2202,
+      "step": 1184
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4160011835273934e-06,
+      "loss": 0.3643,
+      "step": 1185
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3728662659818204e-06,
+      "loss": 0.4312,
+      "step": 1186
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3303940083117527e-06,
+      "loss": 0.3488,
+      "step": 1187
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2885846958814673e-06,
+      "loss": 0.3173,
+      "step": 1188
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2474386096010039e-06,
+      "loss": 0.37,
+      "step": 1189
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2069560259243328e-06,
+      "loss": 0.424,
+      "step": 1190
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1671372168474138e-06,
+      "loss": 0.3486,
+      "step": 1191
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1279824499064396e-06,
+      "loss": 0.3194,
+      "step": 1192
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.089491988176017e-06,
+      "loss": 0.3488,
+      "step": 1193
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0516660902673448e-06,
+      "loss": 0.3003,
+      "step": 1194
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.014505010326583e-06,
+      "loss": 0.3093,
+      "step": 1195
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.780089980330642e-07,
+      "loss": 0.3481,
+      "step": 1196
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.421782985976068e-07,
+      "loss": 0.2612,
+      "step": 1197
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.070131527609604e-07,
+      "loss": 0.2684,
+      "step": 1198
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.725137967920738e-07,
+      "loss": 0.3081,
+      "step": 1199
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.386804624865851e-07,
+      "loss": 0.324,
+      "step": 1200
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.055133771652345e-07,
+      "loss": 0.3859,
+      "step": 1201
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.730127636723539e-07,
+      "loss": 0.3932,
+      "step": 1202
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.411788403743237e-07,
+      "loss": 0.2652,
+      "step": 1203
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.100118211581852e-07,
+      "loss": 0.3908,
+      "step": 1204
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.7951191543012e-07,
+      "loss": 0.3773,
+      "step": 1205
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.496793281141056e-07,
+      "loss": 0.3204,
+      "step": 1206
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.205142596505176e-07,
+      "loss": 0.3642,
+      "step": 1207
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.920169059947411e-07,
+      "loss": 0.3567,
+      "step": 1208
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.64187458615939e-07,
+      "loss": 0.3758,
+      "step": 1209
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.370261044956971e-07,
+      "loss": 0.3023,
+      "step": 1210
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.105330261267916e-07,
+      "loss": 0.3484,
+      "step": 1211
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.847084015119574e-07,
+      "loss": 0.4084,
+      "step": 1212
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.5955240416271084e-07,
+      "loss": 0.3467,
+      "step": 1213
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.3506520309813947e-07,
+      "loss": 0.3125,
+      "step": 1214
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.112469628438365e-07,
+      "loss": 0.2977,
+      "step": 1215
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.8809784343072366e-07,
+      "loss": 0.423,
+      "step": 1216
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.6561800039403016e-07,
+      "loss": 0.2776,
+      "step": 1217
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.4380758477219333e-07,
+      "loss": 0.3507,
+      "step": 1218
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.2266674310589273e-07,
+      "loss": 0.3505,
+      "step": 1219
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.0219561743707326e-07,
+      "loss": 0.3179,
+      "step": 1220
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8239434530792365e-07,
+      "loss": 0.3001,
+      "step": 1221
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6326305976001055e-07,
+      "loss": 0.3569,
+      "step": 1222
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.448018893333681e-07,
+      "loss": 0.2935,
+      "step": 1223
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2701095806565432e-07,
+      "loss": 0.3251,
+      "step": 1224
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.098903854912515e-07,
+      "loss": 0.4101,
+      "step": 1225
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9344028664056713e-07,
+      "loss": 0.2809,
+      "step": 1226
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7766077203915655e-07,
+      "loss": 0.347,
+      "step": 1227
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6255194770704586e-07,
+      "loss": 0.3813,
+      "step": 1228
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.481139151579991e-07,
+      "loss": 0.2351,
+      "step": 1229
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3434677139885222e-07,
+      "loss": 0.3222,
+      "step": 1230
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2125060892881346e-07,
+      "loss": 0.34,
+      "step": 1231
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0882551573891953e-07,
+      "loss": 0.3433,
+      "step": 1232
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.707157531134713e-08,
+      "loss": 0.4656,
+      "step": 1233
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.598886661895788e-08,
+      "loss": 0.3056,
+      "step": 1234
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.557746412468758e-08,
+      "loss": 0.29,
+      "step": 1235
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.583743778106887e-08,
+      "loss": 0.2882,
+      "step": 1236
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.6768853029787184e-08,
+      "loss": 0.2813,
+      "step": 1237
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.837177080119215e-08,
+      "loss": 0.2738,
+      "step": 1238
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.064624751394242e-08,
+      "loss": 0.2525,
+      "step": 1239
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.359233507459481e-08,
+      "loss": 0.4166,
+      "step": 1240
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.7210080877237976e-08,
+      "loss": 0.3054,
+      "step": 1241
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.1499527803214846e-08,
+      "loss": 0.3561,
+      "step": 1242
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.646071422083395e-08,
+      "loss": 0.2677,
+      "step": 1243
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.209367398504746e-08,
+      "loss": 0.2838,
+      "step": 1244
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.398436437317969e-09,
+      "loss": 0.2758,
+      "step": 1245
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.375026405352035e-09,
+      "loss": 0.2639,
+      "step": 1246
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.023464202944748e-09,
+      "loss": 0.4449,
+      "step": 1247
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3437656298687097e-09,
+      "loss": 0.4056,
+      "step": 1248
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3594197175190745e-10,
+      "loss": 0.2927,
+      "step": 1249
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 0.238,
+      "step": 1250
+    },
+    {
+      "epoch": 1.0,
+      "step": 1250,
+      "total_flos": 65298262523904.0,
+      "train_loss": 0.3829174147963524,
+      "train_runtime": 23366.8469,
+      "train_samples_per_second": 3.424,
+      "train_steps_per_second": 0.053
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 1250,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 50000,
+  "total_flos": 65298262523904.0,
+  "train_batch_size": 16,
+  "trial_name": null,
+  "trial_params": null
+}