diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,6019 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.005490779608342691,
+  "eval_steps": 500,
+  "global_step": 1000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001,
+      "loss": 3.3985,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001,
+      "loss": 5.2123,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0002,
+      "loss": 3.8735,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0002,
+      "loss": 2.6185,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001997995991983968,
+      "loss": 4.2863,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001995991983967936,
+      "loss": 1.4706,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001993987975951904,
+      "loss": 3.6384,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019919839679358718,
+      "loss": 4.1227,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019919839679358718,
+      "loss": 6.3727,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019899799599198398,
+      "loss": 3.427,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019879759519038077,
+      "loss": 3.4727,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019859719438877757,
+      "loss": 5.3318,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019839679358717434,
+      "loss": 4.8612,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019839679358717434,
+      "loss": 2.9335,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019819639278557117,
+      "loss": 3.2481,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019799599198396794,
+      "loss": 4.3807,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019779559118236474,
+      "loss": 8.8975,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019759519038076154,
+      "loss": 3.0805,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019739478957915834,
+      "loss": 5.8723,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001971943887775551,
+      "loss": 4.6326,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019699398797595193,
+      "loss": 4.7934,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001967935871743487,
+      "loss": 3.7,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001965931863727455,
+      "loss": 1.9477,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001963927855711423,
+      "loss": 7.7452,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001961923847695391,
+      "loss": 3.4931,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019599198396793587,
+      "loss": 4.2015,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019579158316633267,
+      "loss": 3.0261,
+      "step": 27
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019559118236472947,
+      "loss": 1.9957,
+      "step": 28
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019539078156312627,
+      "loss": 1.795,
+      "step": 29
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019519038076152304,
+      "loss": 1.9667,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019498997995991986,
+      "loss": 2.3988,
+      "step": 31
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019478957915831664,
+      "loss": 3.0434,
+      "step": 32
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019458917835671343,
+      "loss": 1.6954,
+      "step": 33
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019438877755511023,
+      "loss": 3.5625,
+      "step": 34
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019418837675350703,
+      "loss": 3.5388,
+      "step": 35
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019418837675350703,
+      "loss": 3.8641,
+      "step": 36
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001939879759519038,
+      "loss": 2.4463,
+      "step": 37
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001937875751503006,
+      "loss": 1.5531,
+      "step": 38
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001935871743486974,
+      "loss": 2.904,
+      "step": 39
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001933867735470942,
+      "loss": 1.9943,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.000193186372745491,
+      "loss": 8.4639,
+      "step": 41
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001929859719438878,
+      "loss": 2.1307,
+      "step": 42
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019278557114228457,
+      "loss": 2.5124,
+      "step": 43
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019258517034068137,
+      "loss": 4.3622,
+      "step": 44
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019238476953907816,
+      "loss": 2.3385,
+      "step": 45
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019218436873747496,
+      "loss": 2.2836,
+      "step": 46
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019198396793587173,
+      "loss": 2.1174,
+      "step": 47
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019178356713426856,
+      "loss": 2.8594,
+      "step": 48
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019158316633266533,
+      "loss": 4.3321,
+      "step": 49
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019138276553106213,
+      "loss": 2.3192,
+      "step": 50
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019118236472945893,
+      "loss": 0.7275,
+      "step": 51
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00019098196392785573,
+      "loss": 3.198,
+      "step": 52
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001907815631262525,
+      "loss": 3.2683,
+      "step": 53
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001905811623246493,
+      "loss": 2.1871,
+      "step": 54
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001903807615230461,
+      "loss": 1.5479,
+      "step": 55
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001901803607214429,
+      "loss": 2.793,
+      "step": 56
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00018997995991983967,
+      "loss": 2.6259,
+      "step": 57
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001897795591182365,
+      "loss": 2.7578,
+      "step": 58
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00018957915831663326,
+      "loss": 2.0832,
+      "step": 59
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00018937875751503006,
+      "loss": 4.0156,
+      "step": 60
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00018917835671342686,
+      "loss": 0.0,
+      "step": 61
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00018897795591182366,
+      "loss": 2.866,
+      "step": 62
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00018877755511022046,
+      "loss": 2.8362,
+      "step": 63
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00018857715430861726,
+      "loss": 2.138,
+      "step": 64
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00018837675350701405,
+      "loss": 1.9499,
+      "step": 65
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00018817635270541083,
+      "loss": 13.8323,
+      "step": 66
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00018797595190380762,
+      "loss": 2.2264,
+      "step": 67
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00018777555110220442,
+      "loss": 3.8097,
+      "step": 68
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00018757515030060122,
+      "loss": 3.3592,
+      "step": 69
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.000187374749498998,
+      "loss": 2.6868,
+      "step": 70
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00018717434869739482,
+      "loss": 5.958,
+      "step": 71
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001869739478957916,
+      "loss": 1.1189,
+      "step": 72
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001867735470941884,
+      "loss": 2.9349,
+      "step": 73
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001865731462925852,
+      "loss": 4.0328,
+      "step": 74
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00018637274549098199,
+      "loss": 1.6495,
+      "step": 75
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00018617234468937876,
+      "loss": 1.7545,
+      "step": 76
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00018597194388777556,
+      "loss": 1.4433,
+      "step": 77
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00018577154308617235,
+      "loss": 1.2002,
+      "step": 78
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00018557114228456915,
+      "loss": 1.7638,
+      "step": 79
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00018537074148296592,
+      "loss": 1.3996,
+      "step": 80
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00018517034068136275,
+      "loss": 2.099,
+      "step": 81
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00018496993987975952,
+      "loss": 4.7314,
+      "step": 82
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00018476953907815632,
+      "loss": 3.1169,
+      "step": 83
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00018456913827655312,
+      "loss": 4.9392,
+      "step": 84
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00018436873747494992,
+      "loss": 3.1455,
+      "step": 85
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001841683366733467,
+      "loss": 0.765,
+      "step": 86
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00018396793587174351,
+      "loss": 3.067,
+      "step": 87
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00018376753507014029,
+      "loss": 1.3356,
+      "step": 88
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00018356713426853708,
+      "loss": 4.5039,
+      "step": 89
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00018336673346693388,
+      "loss": 2.6412,
+      "step": 90
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00018316633266533068,
+      "loss": 0.5048,
+      "step": 91
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00018296593186372745,
+      "loss": 1.0112,
+      "step": 92
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00018276553106212425,
+      "loss": 1.0227,
+      "step": 93
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00018256513026052105,
+      "loss": 2.4534,
+      "step": 94
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00018236472945891785,
+      "loss": 2.7282,
+      "step": 95
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00018216432865731462,
+      "loss": 3.8706,
+      "step": 96
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00018196392785571145,
+      "loss": 4.2973,
+      "step": 97
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00018176352705410822,
+      "loss": 2.874,
+      "step": 98
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00018156312625250502,
+      "loss": 3.4857,
+      "step": 99
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00018136272545090181,
+      "loss": 2.1494,
+      "step": 100
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001811623246492986,
+      "loss": 3.3004,
+      "step": 101
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00018096192384769538,
+      "loss": 0.0,
+      "step": 102
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001807615230460922,
+      "loss": 2.0755,
+      "step": 103
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00018056112224448898,
+      "loss": 2.339,
+      "step": 104
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00018036072144288578,
+      "loss": 3.2574,
+      "step": 105
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00018016032064128258,
+      "loss": 2.515,
+      "step": 106
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00017995991983967938,
+      "loss": 2.4068,
+      "step": 107
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00017975951903807615,
+      "loss": 2.46,
+      "step": 108
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00017955911823647295,
+      "loss": 3.7409,
+      "step": 109
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00017935871743486975,
+      "loss": 3.1791,
+      "step": 110
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00017915831663326654,
+      "loss": 2.4718,
+      "step": 111
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00017895791583166331,
+      "loss": 4.7393,
+      "step": 112
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00017875751503006014,
+      "loss": 1.1858,
+      "step": 113
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001785571142284569,
+      "loss": 1.3025,
+      "step": 114
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001783567134268537,
+      "loss": 2.6866,
+      "step": 115
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001781563126252505,
+      "loss": 1.636,
+      "step": 116
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001779559118236473,
+      "loss": 3.4094,
+      "step": 117
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001777555110220441,
+      "loss": 2.6015,
+      "step": 118
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00017755511022044088,
+      "loss": 2.0884,
+      "step": 119
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001773547094188377,
+      "loss": 0.0,
+      "step": 120
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00017715430861723447,
+      "loss": 0.4421,
+      "step": 121
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00017695390781563127,
+      "loss": 2.1163,
+      "step": 122
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00017675350701402807,
+      "loss": 4.7788,
+      "step": 123
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00017655310621242487,
+      "loss": 3.7254,
+      "step": 124
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00017635270541082164,
+      "loss": 6.1688,
+      "step": 125
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00017615230460921847,
+      "loss": 2.2818,
+      "step": 126
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00017595190380761524,
+      "loss": 2.9417,
+      "step": 127
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00017575150300601204,
+      "loss": 0.0,
+      "step": 128
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00017555110220440884,
+      "loss": 2.0813,
+      "step": 129
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00017535070140280563,
+      "loss": 0.0,
+      "step": 130
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001751503006012024,
+      "loss": 2.4927,
+      "step": 131
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001749498997995992,
+      "loss": 5.145,
+      "step": 132
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.000174749498997996,
+      "loss": 0.9084,
+      "step": 133
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001745490981963928,
+      "loss": 1.7761,
+      "step": 134
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00017434869739478957,
+      "loss": 3.2621,
+      "step": 135
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001741482965931864,
+      "loss": 3.3788,
+      "step": 136
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00017394789579158317,
+      "loss": 2.54,
+      "step": 137
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00017374749498997997,
+      "loss": 2.1147,
+      "step": 138
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00017354709418837677,
+      "loss": 4.4466,
+      "step": 139
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00017334669338677357,
+      "loss": 1.6063,
+      "step": 140
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00017314629258517034,
+      "loss": 2.522,
+      "step": 141
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00017294589178356714,
+      "loss": 4.5326,
+      "step": 142
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00017274549098196393,
+      "loss": 3.0165,
+      "step": 143
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00017254509018036073,
+      "loss": 2.465,
+      "step": 144
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001723446893787575,
+      "loss": 0.914,
+      "step": 145
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00017214428857715433,
+      "loss": 3.3662,
+      "step": 146
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001719438877755511,
+      "loss": 4.086,
+      "step": 147
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001717434869739479,
+      "loss": 1.5515,
+      "step": 148
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001715430861723447,
+      "loss": 0.9641,
+      "step": 149
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001713426853707415,
+      "loss": 3.3465,
+      "step": 150
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00017114228456913827,
+      "loss": 2.2881,
+      "step": 151
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001709418837675351,
+      "loss": 2.9916,
+      "step": 152
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00017074148296593187,
+      "loss": 7.9494,
+      "step": 153
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00017054108216432866,
+      "loss": 1.7766,
+      "step": 154
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00017034068136272546,
+      "loss": 3.882,
+      "step": 155
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00017014028056112226,
+      "loss": 1.26,
+      "step": 156
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00016993987975951903,
+      "loss": 2.6855,
+      "step": 157
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00016973947895791583,
+      "loss": 2.751,
+      "step": 158
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00016953907815631263,
+      "loss": 4.4908,
+      "step": 159
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00016933867735470943,
+      "loss": 4.7742,
+      "step": 160
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001691382765531062,
+      "loss": 1.0128,
+      "step": 161
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00016893787575150303,
+      "loss": 0.7312,
+      "step": 162
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001687374749498998,
+      "loss": 2.5514,
+      "step": 163
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001685370741482966,
+      "loss": 3.6529,
+      "step": 164
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001683366733466934,
+      "loss": 2.0932,
+      "step": 165
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001681362725450902,
+      "loss": 3.7294,
+      "step": 166
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00016793587174348696,
+      "loss": 5.3816,
+      "step": 167
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001677354709418838,
+      "loss": 4.4184,
+      "step": 168
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00016753507014028056,
+      "loss": 4.5827,
+      "step": 169
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00016733466933867736,
+      "loss": 2.6326,
+      "step": 170
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00016713426853707416,
+      "loss": 2.4652,
+      "step": 171
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00016693386773547096,
+      "loss": 1.7369,
+      "step": 172
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00016673346693386773,
+      "loss": 2.3528,
+      "step": 173
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00016653306613226453,
+      "loss": 2.4526,
+      "step": 174
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00016653306613226453,
+      "loss": 1.8607,
+      "step": 175
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00016633266533066135,
+      "loss": 0.0,
+      "step": 176
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00016613226452905812,
+      "loss": 4.1204,
+      "step": 177
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00016593186372745492,
+      "loss": 2.2052,
+      "step": 178
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00016573146292585172,
+      "loss": 1.4778,
+      "step": 179
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00016553106212424852,
+      "loss": 3.0413,
+      "step": 180
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001653306613226453,
+      "loss": 1.8595,
+      "step": 181
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001651302605210421,
+      "loss": 5.2543,
+      "step": 182
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001649298597194389,
+      "loss": 5.0545,
+      "step": 183
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001647294589178357,
+      "loss": 0.6023,
+      "step": 184
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00016452905811623246,
+      "loss": 2.3329,
+      "step": 185
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00016432865731462928,
+      "loss": 1.4628,
+      "step": 186
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00016412825651302606,
+      "loss": 1.9968,
+      "step": 187
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00016392785571142285,
+      "loss": 2.8452,
+      "step": 188
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00016372745490981965,
+      "loss": 1.7714,
+      "step": 189
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00016352705410821645,
+      "loss": 1.5021,
+      "step": 190
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00016332665330661322,
+      "loss": 1.4912,
+      "step": 191
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00016312625250501005,
+      "loss": 0.3812,
+      "step": 192
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00016292585170340682,
+      "loss": 5.2674,
+      "step": 193
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00016272545090180362,
+      "loss": 0.8585,
+      "step": 194
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00016252505010020042,
+      "loss": 0.5154,
+      "step": 195
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00016232464929859721,
+      "loss": 1.394,
+      "step": 196
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00016212424849699399,
+      "loss": 1.8194,
+      "step": 197
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00016192384769539078,
+      "loss": 2.0729,
+      "step": 198
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00016172344689378758,
+      "loss": 3.8512,
+      "step": 199
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00016152304609218438,
+      "loss": 2.1289,
+      "step": 200
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00016132264529058115,
+      "loss": 4.1014,
+      "step": 201
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00016112224448897798,
+      "loss": 2.5436,
+      "step": 202
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00016092184368737475,
+      "loss": 3.015,
+      "step": 203
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00016072144288577155,
+      "loss": 0.463,
+      "step": 204
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00016052104208416835,
+      "loss": 2.2814,
+      "step": 205
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00016032064128256515,
+      "loss": 3.4467,
+      "step": 206
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00016012024048096192,
+      "loss": 3.9723,
+      "step": 207
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00015991983967935872,
+      "loss": 3.7387,
+      "step": 208
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00015971943887775551,
+      "loss": 3.9553,
+      "step": 209
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001595190380761523,
+      "loss": 4.4628,
+      "step": 210
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001593186372745491,
+      "loss": 2.0064,
+      "step": 211
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001591182364729459,
+      "loss": 2.5059,
+      "step": 212
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00015891783567134268,
+      "loss": 2.3503,
+      "step": 213
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00015871743486973948,
+      "loss": 2.9199,
+      "step": 214
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00015851703406813628,
+      "loss": 2.8497,
+      "step": 215
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00015831663326653308,
+      "loss": 1.9105,
+      "step": 216
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00015811623246492985,
+      "loss": 0.0,
+      "step": 217
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00015791583166332667,
+      "loss": 3.1217,
+      "step": 218
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00015771543086172345,
+      "loss": 3.2305,
+      "step": 219
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00015751503006012024,
+      "loss": 2.5012,
+      "step": 220
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00015731462925851704,
+      "loss": 2.7481,
+      "step": 221
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00015711422845691384,
+      "loss": 2.8643,
+      "step": 222
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001569138276553106,
+      "loss": 0.0,
+      "step": 223
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001567134268537074,
+      "loss": 2.2062,
+      "step": 224
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001565130260521042,
+      "loss": 2.913,
+      "step": 225
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.000156312625250501,
+      "loss": 7.0962,
+      "step": 226
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00015611222444889778,
+      "loss": 3.1977,
+      "step": 227
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001559118236472946,
+      "loss": 0.4999,
+      "step": 228
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00015571142284569138,
+      "loss": 1.128,
+      "step": 229
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00015551102204408818,
+      "loss": 1.7394,
+      "step": 230
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.000155310621242485,
+      "loss": 2.6111,
+      "step": 231
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00015511022044088177,
+      "loss": 1.8315,
+      "step": 232
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00015490981963927857,
+      "loss": 1.1312,
+      "step": 233
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00015470941883767537,
+      "loss": 1.6202,
+      "step": 234
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00015450901803607217,
+      "loss": 0.6677,
+      "step": 235
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00015430861723446894,
+      "loss": 2.2805,
+      "step": 236
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00015410821643286574,
+      "loss": 2.1234,
+      "step": 237
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00015390781563126254,
+      "loss": 1.4029,
+      "step": 238
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00015370741482965934,
+      "loss": 1.8568,
+      "step": 239
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001535070140280561,
+      "loss": 2.4828,
+      "step": 240
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00015330661322645293,
+      "loss": 2.0113,
+      "step": 241
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001531062124248497,
+      "loss": 0.8511,
+      "step": 242
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001529058116232465,
+      "loss": 1.4795,
+      "step": 243
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001527054108216433,
+      "loss": 3.0378,
+      "step": 244
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001525050100200401,
+      "loss": 1.9997,
+      "step": 245
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00015230460921843687,
+      "loss": 1.7813,
+      "step": 246
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00015210420841683367,
+      "loss": 0.0,
+      "step": 247
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00015190380761523047,
+      "loss": 2.7432,
+      "step": 248
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00015170340681362727,
+      "loss": 1.6375,
+      "step": 249
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00015150300601202404,
+      "loss": 1.6442,
+      "step": 250
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00015130260521042086,
+      "loss": 3.8107,
+      "step": 251
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00015110220440881764,
+      "loss": 2.3884,
+      "step": 252
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00015090180360721443,
+      "loss": 1.256,
+      "step": 253
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00015070140280561123,
+      "loss": 2.6774,
+      "step": 254
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00015050100200400803,
+      "loss": 4.4725,
+      "step": 255
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001503006012024048,
+      "loss": 1.8188,
+      "step": 256
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00015010020040080163,
+      "loss": 2.8957,
+      "step": 257
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001498997995991984,
+      "loss": 4.4914,
+      "step": 258
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001496993987975952,
+      "loss": 1.7564,
+      "step": 259
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.000149498997995992,
+      "loss": 2.175,
+      "step": 260
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001492985971943888,
+      "loss": 0.4855,
+      "step": 261
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00014909819639278557,
+      "loss": 0.7024,
+      "step": 262
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00014889779559118237,
+      "loss": 2.941,
+      "step": 263
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00014869739478957916,
+      "loss": 2.8929,
+      "step": 264
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00014849699398797596,
+      "loss": 4.9599,
+      "step": 265
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00014829659318637273,
+      "loss": 3.2604,
+      "step": 266
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00014809619238476956,
+      "loss": 3.0037,
+      "step": 267
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00014789579158316633,
+      "loss": 2.6023,
+      "step": 268
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00014769539078156313,
+      "loss": 0.0,
+      "step": 269
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00014749498997995993,
+      "loss": 4.7965,
+      "step": 270
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00014729458917835673,
+      "loss": 0.8059,
+      "step": 271
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001470941883767535,
+      "loss": 0.7748,
+      "step": 272
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001468937875751503,
+      "loss": 2.2642,
+      "step": 273
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001466933867735471,
+      "loss": 2.9185,
+      "step": 274
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001464929859719439,
+      "loss": 2.4913,
+      "step": 275
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001462925851703407,
+      "loss": 3.3304,
+      "step": 276
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001460921843687375,
+      "loss": 1.5842,
+      "step": 277
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00014589178356713426,
+      "loss": 2.2768,
+      "step": 278
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00014569138276553106,
+      "loss": 2.129,
+      "step": 279
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00014549098196392786,
+      "loss": 1.7883,
+      "step": 280
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00014529058116232466,
+      "loss": 2.8672,
+      "step": 281
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00014509018036072143,
+      "loss": 1.6781,
+      "step": 282
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00014488977955911825,
+      "loss": 2.321,
+      "step": 283
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00014468937875751503,
+      "loss": 1.2013,
+      "step": 284
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00014448897795591182,
+      "loss": 2.5649,
+      "step": 285
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00014428857715430862,
+      "loss": 1.9453,
+      "step": 286
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00014408817635270542,
+      "loss": 0.32,
+      "step": 287
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00014388777555110222,
+      "loss": 3.1301,
+      "step": 288
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.000143687374749499,
+      "loss": 0.4875,
+      "step": 289
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00014348697394789582,
+      "loss": 2.9447,
+      "step": 290
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001432865731462926,
+      "loss": 0.8736,
+      "step": 291
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001430861723446894,
+      "loss": 0.5231,
+      "step": 292
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00014288577154308619,
+      "loss": 2.8954,
+      "step": 293
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00014268537074148298,
+      "loss": 1.6437,
+      "step": 294
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00014248496993987976,
+      "loss": 7.0932,
+      "step": 295
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00014228456913827658,
+      "loss": 0.2355,
+      "step": 296
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00014208416833667335,
+      "loss": 2.0368,
+      "step": 297
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00014188376753507015,
+      "loss": 8.2179,
+      "step": 298
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00014168336673346695,
+      "loss": 4.559,
+      "step": 299
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00014148296593186375,
+      "loss": 2.3565,
+      "step": 300
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00014128256513026052,
+      "loss": 3.1562,
+      "step": 301
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00014108216432865732,
+      "loss": 0.9561,
+      "step": 302
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00014088176352705412,
+      "loss": 2.3135,
+      "step": 303
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00014068136272545092,
+      "loss": 4.5708,
+      "step": 304
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001404809619238477,
+      "loss": 0.6331,
+      "step": 305
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001402805611222445,
+      "loss": 1.2219,
+      "step": 306
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00014008016032064128,
+      "loss": 2.4206,
+      "step": 307
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00013987975951903808,
+      "loss": 1.0911,
+      "step": 308
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00013967935871743488,
+      "loss": 3.7003,
+      "step": 309
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00013947895791583168,
+      "loss": 1.7502,
+      "step": 310
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00013927855711422845,
+      "loss": 1.6085,
+      "step": 311
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00013907815631262525,
+      "loss": 0.0,
+      "step": 312
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00013887775551102205,
+      "loss": 3.5848,
+      "step": 313
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00013867735470941885,
+      "loss": 4.0407,
+      "step": 314
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00013847695390781562,
+      "loss": 2.6365,
+      "step": 315
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00013827655310621244,
+      "loss": 2.0709,
+      "step": 316
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00013807615230460922,
+      "loss": 2.6688,
+      "step": 317
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00013787575150300601,
+      "loss": 1.7493,
+      "step": 318
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001376753507014028,
+      "loss": 2.1693,
+      "step": 319
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001374749498997996,
+      "loss": 5.6072,
+      "step": 320
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00013727454909819638,
+      "loss": 2.1042,
+      "step": 321
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001370741482965932,
+      "loss": 0.7863,
+      "step": 322
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00013687374749498998,
+      "loss": 2.1696,
+      "step": 323
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00013667334669338678,
+      "loss": 2.7226,
+      "step": 324
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00013647294589178358,
+      "loss": 2.0276,
+      "step": 325
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00013627254509018038,
+      "loss": 2.3375,
+      "step": 326
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00013607214428857715,
+      "loss": 1.1399,
+      "step": 327
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00013587174348697395,
+      "loss": 1.5908,
+      "step": 328
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00013567134268537074,
+      "loss": 2.9017,
+      "step": 329
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00013547094188376754,
+      "loss": 2.0847,
+      "step": 330
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00013527054108216431,
+      "loss": 2.0466,
+      "step": 331
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00013507014028056114,
+      "loss": 3.1415,
+      "step": 332
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001348697394789579,
+      "loss": 1.3235,
+      "step": 333
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001346693386773547,
+      "loss": 2.4393,
+      "step": 334
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001344689378757515,
+      "loss": 0.7366,
+      "step": 335
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001342685370741483,
+      "loss": 1.5512,
+      "step": 336
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00013406813627254508,
+      "loss": 0.9189,
+      "step": 337
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001338677354709419,
+      "loss": 0.6407,
+      "step": 338
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00013366733466933868,
+      "loss": 2.403,
+      "step": 339
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00013346693386773547,
+      "loss": 1.9463,
+      "step": 340
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00013326653306613227,
+      "loss": 3.963,
+      "step": 341
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00013306613226452907,
+      "loss": 1.1293,
+      "step": 342
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00013286573146292587,
+      "loss": 2.3127,
+      "step": 343
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00013266533066132264,
+      "loss": 1.5872,
+      "step": 344
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00013246492985971947,
+      "loss": 2.3619,
+      "step": 345
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00013226452905811624,
+      "loss": 2.153,
+      "step": 346
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00013206412825651304,
+      "loss": 0.8323,
+      "step": 347
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00013186372745490984,
+      "loss": 0.0,
+      "step": 348
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00013166332665330663,
+      "loss": 2.4433,
+      "step": 349
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001314629258517034,
+      "loss": 2.5312,
+      "step": 350
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001312625250501002,
+      "loss": 1.7814,
+      "step": 351
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.000131062124248497,
+      "loss": 2.5383,
+      "step": 352
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001308617234468938,
+      "loss": 1.2516,
+      "step": 353
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00013066132264529057,
+      "loss": 6.2541,
+      "step": 354
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001304609218436874,
+      "loss": 1.3022,
+      "step": 355
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00013026052104208417,
+      "loss": 2.8096,
+      "step": 356
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00013006012024048097,
+      "loss": 2.0949,
+      "step": 357
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00012985971943887777,
+      "loss": 3.4624,
+      "step": 358
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00012965931863727457,
+      "loss": 2.9193,
+      "step": 359
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00012945891783567134,
+      "loss": 3.0689,
+      "step": 360
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00012925851703406816,
+      "loss": 2.0878,
+      "step": 361
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00012905811623246493,
+      "loss": 2.718,
+      "step": 362
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00012885771543086173,
+      "loss": 0.9596,
+      "step": 363
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00012865731462925853,
+      "loss": 1.3671,
+      "step": 364
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00012845691382765533,
+      "loss": 3.5474,
+      "step": 365
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001282565130260521,
+      "loss": 3.185,
+      "step": 366
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001280561122244489,
+      "loss": 1.195,
+      "step": 367
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001278557114228457,
+      "loss": 2.8428,
+      "step": 368
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001276553106212425,
+      "loss": 1.6981,
+      "step": 369
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00012745490981963927,
+      "loss": 1.6565,
+      "step": 370
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001272545090180361,
+      "loss": 3.5342,
+      "step": 371
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00012705410821643286,
+      "loss": 1.9243,
+      "step": 372
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00012685370741482966,
+      "loss": 2.4072,
+      "step": 373
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00012665330661322646,
+      "loss": 2.0842,
+      "step": 374
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00012645290581162326,
+      "loss": 1.2166,
+      "step": 375
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00012625250501002003,
+      "loss": 0.0,
+      "step": 376
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00012605210420841683,
+      "loss": 2.2686,
+      "step": 377
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00012585170340681363,
+      "loss": 2.9105,
+      "step": 378
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00012565130260521043,
+      "loss": 1.7398,
+      "step": 379
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001254509018036072,
+      "loss": 1.2882,
+      "step": 380
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00012525050100200402,
+      "loss": 4.7763,
+      "step": 381
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001250501002004008,
+      "loss": 1.5265,
+      "step": 382
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001248496993987976,
+      "loss": 1.154,
+      "step": 383
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001246492985971944,
+      "loss": 2.6431,
+      "step": 384
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001244488977955912,
+      "loss": 4.251,
+      "step": 385
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00012424849699398796,
+      "loss": 1.8364,
+      "step": 386
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001240480961923848,
+      "loss": 1.7506,
+      "step": 387
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00012384769539078156,
+      "loss": 1.3577,
+      "step": 388
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00012364729458917836,
+      "loss": 5.6636,
+      "step": 389
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00012344689378757516,
+      "loss": 4.8805,
+      "step": 390
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00012324649298597196,
+      "loss": 1.8715,
+      "step": 391
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00012304609218436873,
+      "loss": 1.4509,
+      "step": 392
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00012284569138276553,
+      "loss": 2.9912,
+      "step": 393
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00012264529058116232,
+      "loss": 0.7702,
+      "step": 394
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00012244488977955912,
+      "loss": 5.1749,
+      "step": 395
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001222444889779559,
+      "loss": 1.7929,
+      "step": 396
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001220440881763527,
+      "loss": 3.7588,
+      "step": 397
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00012184368737474952,
+      "loss": 1.5476,
+      "step": 398
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001216432865731463,
+      "loss": 0.9982,
+      "step": 399
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001214428857715431,
+      "loss": 0.9221,
+      "step": 400
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00012124248496993989,
+      "loss": 1.7868,
+      "step": 401
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00012104208416833669,
+      "loss": 1.0556,
+      "step": 402
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00012084168336673347,
+      "loss": 3.1609,
+      "step": 403
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00012064128256513027,
+      "loss": 2.0551,
+      "step": 404
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00012044088176352705,
+      "loss": 2.2839,
+      "step": 405
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00012024048096192387,
+      "loss": 2.6195,
+      "step": 406
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00012004008016032065,
+      "loss": 2.4274,
+      "step": 407
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00011983967935871745,
+      "loss": 3.065,
+      "step": 408
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00011963927855711424,
+      "loss": 2.9908,
+      "step": 409
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00011943887775551103,
+      "loss": 2.584,
+      "step": 410
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00011923847695390782,
+      "loss": 6.5406,
+      "step": 411
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00011903807615230462,
+      "loss": 2.2507,
+      "step": 412
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001188376753507014,
+      "loss": 1.9925,
+      "step": 413
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00011863727454909821,
+      "loss": 2.1832,
+      "step": 414
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00011843687374749499,
+      "loss": 2.9761,
+      "step": 415
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001182364729458918,
+      "loss": 3.3134,
+      "step": 416
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00011803607214428858,
+      "loss": 3.6563,
+      "step": 417
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00011783567134268538,
+      "loss": 0.0,
+      "step": 418
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00011763527054108217,
+      "loss": 2.8706,
+      "step": 419
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00011743486973947896,
+      "loss": 1.4058,
+      "step": 420
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00011723446893787575,
+      "loss": 3.0397,
+      "step": 421
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00011703406813627256,
+      "loss": 0.4962,
+      "step": 422
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00011683366733466933,
+      "loss": 1.8918,
+      "step": 423
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00011663326653306615,
+      "loss": 1.6106,
+      "step": 424
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00011643286573146293,
+      "loss": 1.6681,
+      "step": 425
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00011623246492985973,
+      "loss": 2.438,
+      "step": 426
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00011603206412825651,
+      "loss": 1.8013,
+      "step": 427
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00011583166332665331,
+      "loss": 1.6599,
+      "step": 428
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001156312625250501,
+      "loss": 2.4527,
+      "step": 429
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00011543086172344691,
+      "loss": 2.6056,
+      "step": 430
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00011523046092184368,
+      "loss": 2.3085,
+      "step": 431
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00011503006012024049,
+      "loss": 3.2212,
+      "step": 432
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00011482965931863728,
+      "loss": 4.1987,
+      "step": 433
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00011462925851703408,
+      "loss": 0.0,
+      "step": 434
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00011442885771543086,
+      "loss": 2.9687,
+      "step": 435
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00011422845691382766,
+      "loss": 2.8237,
+      "step": 436
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00011402805611222445,
+      "loss": 1.7948,
+      "step": 437
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00011382765531062126,
+      "loss": 0.863,
+      "step": 438
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00011362725450901803,
+      "loss": 2.4803,
+      "step": 439
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00011342685370741484,
+      "loss": 2.0552,
+      "step": 440
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00011322645290581163,
+      "loss": 1.6845,
+      "step": 441
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00011302605210420842,
+      "loss": 1.8881,
+      "step": 442
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00011282565130260521,
+      "loss": 1.3237,
+      "step": 443
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00011262525050100201,
+      "loss": 1.1419,
+      "step": 444
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00011242484969939879,
+      "loss": 2.5045,
+      "step": 445
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00011222444889779559,
+      "loss": 1.9609,
+      "step": 446
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00011202404809619238,
+      "loss": 0.6873,
+      "step": 447
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00011182364729458919,
+      "loss": 2.2377,
+      "step": 448
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00011162324649298596,
+      "loss": 2.5531,
+      "step": 449
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00011142284569138277,
+      "loss": 1.0334,
+      "step": 450
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00011122244488977956,
+      "loss": 1.8675,
+      "step": 451
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00011102204408817636,
+      "loss": 1.9346,
+      "step": 452
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00011082164328657317,
+      "loss": 3.4389,
+      "step": 453
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00011062124248496994,
+      "loss": 2.9012,
+      "step": 454
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00011042084168336675,
+      "loss": 1.0503,
+      "step": 455
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00011022044088176354,
+      "loss": 2.3487,
+      "step": 456
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00011002004008016033,
+      "loss": 1.5353,
+      "step": 457
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00010981963927855712,
+      "loss": 2.6252,
+      "step": 458
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00010961923847695392,
+      "loss": 1.8809,
+      "step": 459
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001094188376753507,
+      "loss": 0.2392,
+      "step": 460
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00010921843687374752,
+      "loss": 2.9035,
+      "step": 461
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00010901803607214429,
+      "loss": 1.3167,
+      "step": 462
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001088176352705411,
+      "loss": 1.9923,
+      "step": 463
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00010861723446893788,
+      "loss": 2.0446,
+      "step": 464
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00010841683366733468,
+      "loss": 4.3443,
+      "step": 465
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00010821643286573147,
+      "loss": 2.5066,
+      "step": 466
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00010801603206412827,
+      "loss": 2.0386,
+      "step": 467
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00010781563126252505,
+      "loss": 1.5045,
+      "step": 468
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00010761523046092185,
+      "loss": 2.1218,
+      "step": 469
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00010741482965931863,
+      "loss": 3.0638,
+      "step": 470
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00010721442885771545,
+      "loss": 2.0932,
+      "step": 471
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00010701402805611223,
+      "loss": 1.5917,
+      "step": 472
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00010681362725450903,
+      "loss": 1.4797,
+      "step": 473
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00010661322645290582,
+      "loss": 2.1382,
+      "step": 474
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00010641282565130261,
+      "loss": 3.9274,
+      "step": 475
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001062124248496994,
+      "loss": 2.8679,
+      "step": 476
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001060120240480962,
+      "loss": 2.3778,
+      "step": 477
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00010581162324649298,
+      "loss": 1.6275,
+      "step": 478
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001056112224448898,
+      "loss": 2.3191,
+      "step": 479
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00010541082164328657,
+      "loss": 2.7971,
+      "step": 480
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00010521042084168338,
+      "loss": 1.9046,
+      "step": 481
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00010501002004008016,
+      "loss": 1.7607,
+      "step": 482
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00010480961923847696,
+      "loss": 2.2909,
+      "step": 483
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00010460921843687375,
+      "loss": 1.6937,
+      "step": 484
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00010440881763527055,
+      "loss": 2.3124,
+      "step": 485
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00010420841683366733,
+      "loss": 1.0625,
+      "step": 486
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00010400801603206414,
+      "loss": 1.8987,
+      "step": 487
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00010380761523046091,
+      "loss": 0.7515,
+      "step": 488
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00010360721442885773,
+      "loss": 1.4015,
+      "step": 489
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00010340681362725451,
+      "loss": 0.0,
+      "step": 490
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00010320641282565131,
+      "loss": 2.2455,
+      "step": 491
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001030060120240481,
+      "loss": 1.5684,
+      "step": 492
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00010280561122244489,
+      "loss": 0.4304,
+      "step": 493
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00010260521042084168,
+      "loss": 1.1205,
+      "step": 494
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00010240480961923849,
+      "loss": 5.8101,
+      "step": 495
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00010220440881763526,
+      "loss": 2.3707,
+      "step": 496
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00010200400801603207,
+      "loss": 4.4205,
+      "step": 497
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00010180360721442886,
+      "loss": 1.9965,
+      "step": 498
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00010160320641282566,
+      "loss": 2.1804,
+      "step": 499
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00010140280561122244,
+      "loss": 2.7072,
+      "step": 500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00010120240480961924,
+      "loss": 0.3631,
+      "step": 501
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00010100200400801603,
+      "loss": 1.192,
+      "step": 502
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00010080160320641284,
+      "loss": 1.4428,
+      "step": 503
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00010060120240480961,
+      "loss": 1.4366,
+      "step": 504
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.00010040080160320642,
+      "loss": 2.4546,
+      "step": 505
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001002004008016032,
+      "loss": 1.6439,
+      "step": 506
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0001,
+      "loss": 1.9219,
+      "step": 507
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.97995991983968e-05,
+      "loss": 1.5569,
+      "step": 508
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.959919839679359e-05,
+      "loss": 3.5738,
+      "step": 509
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.939879759519039e-05,
+      "loss": 6.3032,
+      "step": 510
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.919839679358717e-05,
+      "loss": 0.0,
+      "step": 511
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.899799599198397e-05,
+      "loss": 2.6412,
+      "step": 512
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.879759519038077e-05,
+      "loss": 2.0273,
+      "step": 513
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.859719438877755e-05,
+      "loss": 0.0,
+      "step": 514
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.839679358717435e-05,
+      "loss": 1.99,
+      "step": 515
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.819639278557115e-05,
+      "loss": 0.9392,
+      "step": 516
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.799599198396794e-05,
+      "loss": 1.0164,
+      "step": 517
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.779559118236473e-05,
+      "loss": 2.5598,
+      "step": 518
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.759519038076152e-05,
+      "loss": 5.3794,
+      "step": 519
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.739478957915832e-05,
+      "loss": 2.5614,
+      "step": 520
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.719438877755512e-05,
+      "loss": 0.7221,
+      "step": 521
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.69939879759519e-05,
+      "loss": 1.4231,
+      "step": 522
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.67935871743487e-05,
+      "loss": 1.3288,
+      "step": 523
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.65931863727455e-05,
+      "loss": 5.2266,
+      "step": 524
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.639278557114228e-05,
+      "loss": 2.4147,
+      "step": 525
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.619238476953908e-05,
+      "loss": 1.1246,
+      "step": 526
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.599198396793587e-05,
+      "loss": 0.0,
+      "step": 527
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.579158316633267e-05,
+      "loss": 4.951,
+      "step": 528
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.559118236472946e-05,
+      "loss": 2.6102,
+      "step": 529
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.539078156312625e-05,
+      "loss": 1.9989,
+      "step": 530
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.519038076152305e-05,
+      "loss": 2.2009,
+      "step": 531
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.498997995991983e-05,
+      "loss": 2.7023,
+      "step": 532
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.478957915831663e-05,
+      "loss": 2.3955,
+      "step": 533
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.458917835671343e-05,
+      "loss": 2.43,
+      "step": 534
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.438877755511023e-05,
+      "loss": 1.8758,
+      "step": 535
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.418837675350703e-05,
+      "loss": 1.45,
+      "step": 536
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.398797595190381e-05,
+      "loss": 3.1603,
+      "step": 537
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.378757515030061e-05,
+      "loss": 4.395,
+      "step": 538
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.358717434869741e-05,
+      "loss": 1.6075,
+      "step": 539
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.33867735470942e-05,
+      "loss": 0.9112,
+      "step": 540
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.318637274549099e-05,
+      "loss": 3.9706,
+      "step": 541
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.298597194388778e-05,
+      "loss": 1.1322,
+      "step": 542
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.278557114228458e-05,
+      "loss": 4.9835,
+      "step": 543
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.258517034068137e-05,
+      "loss": 2.3816,
+      "step": 544
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.238476953907816e-05,
+      "loss": 1.7184,
+      "step": 545
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.218436873747496e-05,
+      "loss": 1.9735,
+      "step": 546
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.198396793587176e-05,
+      "loss": 2.2694,
+      "step": 547
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.178356713426854e-05,
+      "loss": 2.1059,
+      "step": 548
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.158316633266534e-05,
+      "loss": 1.2872,
+      "step": 549
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.138276553106213e-05,
+      "loss": 1.6778,
+      "step": 550
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.118236472945892e-05,
+      "loss": 3.3389,
+      "step": 551
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.098196392785572e-05,
+      "loss": 1.7216,
+      "step": 552
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.078156312625251e-05,
+      "loss": 1.596,
+      "step": 553
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.05811623246493e-05,
+      "loss": 1.4005,
+      "step": 554
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.03807615230461e-05,
+      "loss": 2.4382,
+      "step": 555
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.018036072144289e-05,
+      "loss": 4.276,
+      "step": 556
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.997995991983969e-05,
+      "loss": 2.6499,
+      "step": 557
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.977955911823647e-05,
+      "loss": 3.9086,
+      "step": 558
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.957915831663327e-05,
+      "loss": 3.3864,
+      "step": 559
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.937875751503007e-05,
+      "loss": 7.2937,
+      "step": 560
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.917835671342686e-05,
+      "loss": 1.8136,
+      "step": 561
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.897795591182365e-05,
+      "loss": 2.7177,
+      "step": 562
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.877755511022044e-05,
+      "loss": 3.3096,
+      "step": 563
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.857715430861724e-05,
+      "loss": 3.5284,
+      "step": 564
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.837675350701404e-05,
+      "loss": 1.853,
+      "step": 565
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.817635270541082e-05,
+      "loss": 3.3489,
+      "step": 566
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.797595190380762e-05,
+      "loss": 2.4069,
+      "step": 567
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.777555110220442e-05,
+      "loss": 1.3353,
+      "step": 568
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.75751503006012e-05,
+      "loss": 1.1441,
+      "step": 569
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.7374749498998e-05,
+      "loss": 1.2785,
+      "step": 570
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.717434869739479e-05,
+      "loss": 1.0496,
+      "step": 571
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.697394789579159e-05,
+      "loss": 1.8968,
+      "step": 572
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.677354709418838e-05,
+      "loss": 2.4824,
+      "step": 573
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.657314629258517e-05,
+      "loss": 1.5688,
+      "step": 574
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.637274549098197e-05,
+      "loss": 2.0568,
+      "step": 575
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.617234468937875e-05,
+      "loss": 3.695,
+      "step": 576
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.597194388777555e-05,
+      "loss": 1.44,
+      "step": 577
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.577154308617235e-05,
+      "loss": 0.394,
+      "step": 578
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.557114228456913e-05,
+      "loss": 3.4525,
+      "step": 579
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.537074148296593e-05,
+      "loss": 2.219,
+      "step": 580
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.517034068136273e-05,
+      "loss": 2.9851,
+      "step": 581
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.496993987975952e-05,
+      "loss": 1.6133,
+      "step": 582
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.476953907815631e-05,
+      "loss": 0.7803,
+      "step": 583
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.45691382765531e-05,
+      "loss": 1.9138,
+      "step": 584
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.43687374749499e-05,
+      "loss": 5.0276,
+      "step": 585
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.41683366733467e-05,
+      "loss": 5.7651,
+      "step": 586
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.396793587174348e-05,
+      "loss": 0.2584,
+      "step": 587
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.376753507014028e-05,
+      "loss": 2.1676,
+      "step": 588
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.356713426853708e-05,
+      "loss": 1.5227,
+      "step": 589
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.336673346693386e-05,
+      "loss": 1.3924,
+      "step": 590
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.316633266533068e-05,
+      "loss": 2.7769,
+      "step": 591
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.296593186372746e-05,
+      "loss": 3.3983,
+      "step": 592
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.276553106212426e-05,
+      "loss": 3.5176,
+      "step": 593
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.256513026052104e-05,
+      "loss": 0.0,
+      "step": 594
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.236472945891784e-05,
+      "loss": 1.3284,
+      "step": 595
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.216432865731464e-05,
+      "loss": 2.8176,
+      "step": 596
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.196392785571143e-05,
+      "loss": 1.4708,
+      "step": 597
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.176352705410823e-05,
+      "loss": 2.7816,
+      "step": 598
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.156312625250502e-05,
+      "loss": 1.8782,
+      "step": 599
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.136272545090181e-05,
+      "loss": 1.5401,
+      "step": 600
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.116232464929861e-05,
+      "loss": 1.4211,
+      "step": 601
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.096192384769539e-05,
+      "loss": 0.7955,
+      "step": 602
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.076152304609219e-05,
+      "loss": 2.4868,
+      "step": 603
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.056112224448899e-05,
+      "loss": 0.8562,
+      "step": 604
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.036072144288577e-05,
+      "loss": 2.3006,
+      "step": 605
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.016032064128257e-05,
+      "loss": 1.9321,
+      "step": 606
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.995991983967936e-05,
+      "loss": 2.6157,
+      "step": 607
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.975951903807616e-05,
+      "loss": 2.0561,
+      "step": 608
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.955911823647296e-05,
+      "loss": 3.1193,
+      "step": 609
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.935871743486974e-05,
+      "loss": 1.6484,
+      "step": 610
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.915831663326654e-05,
+      "loss": 1.5268,
+      "step": 611
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.895791583166334e-05,
+      "loss": 0.0,
+      "step": 612
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.875751503006012e-05,
+      "loss": 0.492,
+      "step": 613
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.855711422845692e-05,
+      "loss": 4.4678,
+      "step": 614
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.83567134268537e-05,
+      "loss": 1.9929,
+      "step": 615
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.81563126252505e-05,
+      "loss": 0.7015,
+      "step": 616
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.79559118236473e-05,
+      "loss": 1.5474,
+      "step": 617
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.775551102204409e-05,
+      "loss": 1.8292,
+      "step": 618
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.755511022044089e-05,
+      "loss": 2.8704,
+      "step": 619
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.735470941883769e-05,
+      "loss": 2.0216,
+      "step": 620
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.715430861723447e-05,
+      "loss": 2.8592,
+      "step": 621
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.695390781563127e-05,
+      "loss": 2.3869,
+      "step": 622
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.675350701402805e-05,
+      "loss": 3.5467,
+      "step": 623
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.655310621242485e-05,
+      "loss": 0.785,
+      "step": 624
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.635270541082165e-05,
+      "loss": 1.8183,
+      "step": 625
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.615230460921844e-05,
+      "loss": 0.5593,
+      "step": 626
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.595190380761523e-05,
+      "loss": 3.5061,
+      "step": 627
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.575150300601202e-05,
+      "loss": 6.0013,
+      "step": 628
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.555110220440882e-05,
+      "loss": 3.9287,
+      "step": 629
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.535070140280562e-05,
+      "loss": 1.1422,
+      "step": 630
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.51503006012024e-05,
+      "loss": 0.0,
+      "step": 631
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.49498997995992e-05,
+      "loss": 2.8392,
+      "step": 632
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.4749498997996e-05,
+      "loss": 0.0,
+      "step": 633
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.454909819639278e-05,
+      "loss": 2.363,
+      "step": 634
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.434869739478958e-05,
+      "loss": 1.4281,
+      "step": 635
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.414829659318637e-05,
+      "loss": 1.4313,
+      "step": 636
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.394789579158317e-05,
+      "loss": 2.8012,
+      "step": 637
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.374749498997996e-05,
+      "loss": 2.175,
+      "step": 638
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.354709418837675e-05,
+      "loss": 3.2143,
+      "step": 639
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.334669338677355e-05,
+      "loss": 0.6305,
+      "step": 640
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.314629258517035e-05,
+      "loss": 2.0879,
+      "step": 641
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.294589178356713e-05,
+      "loss": 3.7997,
+      "step": 642
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.274549098196393e-05,
+      "loss": 1.5878,
+      "step": 643
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.254509018036071e-05,
+      "loss": 3.5472,
+      "step": 644
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.234468937875751e-05,
+      "loss": 2.5241,
+      "step": 645
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.214428857715431e-05,
+      "loss": 2.3167,
+      "step": 646
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.194388777555111e-05,
+      "loss": 0.6522,
+      "step": 647
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.174348697394791e-05,
+      "loss": 1.3162,
+      "step": 648
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.15430861723447e-05,
+      "loss": 3.8872,
+      "step": 649
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.134268537074149e-05,
+      "loss": 1.4756,
+      "step": 650
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.114228456913829e-05,
+      "loss": 2.367,
+      "step": 651
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.094188376753508e-05,
+      "loss": 1.5718,
+      "step": 652
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.074148296593187e-05,
+      "loss": 2.8435,
+      "step": 653
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.054108216432866e-05,
+      "loss": 2.7527,
+      "step": 654
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.034068136272546e-05,
+      "loss": 4.4464,
+      "step": 655
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.014028056112226e-05,
+      "loss": 1.5668,
+      "step": 656
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.993987975951904e-05,
+      "loss": 2.2184,
+      "step": 657
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.973947895791584e-05,
+      "loss": 1.2717,
+      "step": 658
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.953907815631263e-05,
+      "loss": 3.3642,
+      "step": 659
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.933867735470942e-05,
+      "loss": 0.6904,
+      "step": 660
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.913827655310622e-05,
+      "loss": 0.0,
+      "step": 661
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.893787575150301e-05,
+      "loss": 1.6057,
+      "step": 662
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.87374749498998e-05,
+      "loss": 1.9233,
+      "step": 663
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.85370741482966e-05,
+      "loss": 1.5996,
+      "step": 664
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.833667334669339e-05,
+      "loss": 2.1426,
+      "step": 665
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.813627254509019e-05,
+      "loss": 2.533,
+      "step": 666
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.793587174348697e-05,
+      "loss": 2.3267,
+      "step": 667
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.773547094188377e-05,
+      "loss": 4.0984,
+      "step": 668
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.753507014028057e-05,
+      "loss": 0.9157,
+      "step": 669
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.733466933867735e-05,
+      "loss": 2.0127,
+      "step": 670
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.713426853707415e-05,
+      "loss": 1.7905,
+      "step": 671
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.693386773547095e-05,
+      "loss": 0.6195,
+      "step": 672
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.673346693386774e-05,
+      "loss": 0.9984,
+      "step": 673
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.653306613226454e-05,
+      "loss": 2.7152,
+      "step": 674
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.633266533066132e-05,
+      "loss": 2.2094,
+      "step": 675
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.613226452905812e-05,
+      "loss": 2.0554,
+      "step": 676
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.593186372745492e-05,
+      "loss": 3.2084,
+      "step": 677
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.57314629258517e-05,
+      "loss": 1.3133,
+      "step": 678
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.55310621242485e-05,
+      "loss": 3.631,
+      "step": 679
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.533066132264529e-05,
+      "loss": 3.5943,
+      "step": 680
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.513026052104208e-05,
+      "loss": 1.6873,
+      "step": 681
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.492985971943888e-05,
+      "loss": 3.4164,
+      "step": 682
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.472945891783567e-05,
+      "loss": 1.4744,
+      "step": 683
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.452905811623247e-05,
+      "loss": 1.6474,
+      "step": 684
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.432865731462927e-05,
+      "loss": 2.2888,
+      "step": 685
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.412825651302605e-05,
+      "loss": 0.947,
+      "step": 686
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.392785571142285e-05,
+      "loss": 1.4766,
+      "step": 687
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.372745490981963e-05,
+      "loss": 3.1128,
+      "step": 688
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.352705410821643e-05,
+      "loss": 2.7201,
+      "step": 689
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.332665330661323e-05,
+      "loss": 1.0741,
+      "step": 690
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.312625250501002e-05,
+      "loss": 1.5721,
+      "step": 691
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.292585170340681e-05,
+      "loss": 4.1374,
+      "step": 692
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.27254509018036e-05,
+      "loss": 0.0,
+      "step": 693
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.25250501002004e-05,
+      "loss": 5.0251,
+      "step": 694
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.23246492985972e-05,
+      "loss": 4.394,
+      "step": 695
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.212424849699398e-05,
+      "loss": 2.5211,
+      "step": 696
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.192384769539078e-05,
+      "loss": 0.6668,
+      "step": 697
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.172344689378758e-05,
+      "loss": 3.7969,
+      "step": 698
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.152304609218436e-05,
+      "loss": 2.1938,
+      "step": 699
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.132264529058116e-05,
+      "loss": 2.3569,
+      "step": 700
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.112224448897795e-05,
+      "loss": 2.2737,
+      "step": 701
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.092184368737476e-05,
+      "loss": 2.2821,
+      "step": 702
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.072144288577155e-05,
+      "loss": 2.2069,
+      "step": 703
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.052104208416834e-05,
+      "loss": 2.4519,
+      "step": 704
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.0320641282565135e-05,
+      "loss": 1.6573,
+      "step": 705
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.012024048096193e-05,
+      "loss": 2.6574,
+      "step": 706
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.9919839679358725e-05,
+      "loss": 3.376,
+      "step": 707
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.971943887775552e-05,
+      "loss": 1.1742,
+      "step": 708
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.951903807615231e-05,
+      "loss": 0.0,
+      "step": 709
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.931863727454911e-05,
+      "loss": 1.5307,
+      "step": 710
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.91182364729459e-05,
+      "loss": 2.9334,
+      "step": 711
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.891783567134269e-05,
+      "loss": 0.0683,
+      "step": 712
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.871743486973948e-05,
+      "loss": 1.3237,
+      "step": 713
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.851703406813628e-05,
+      "loss": 0.8188,
+      "step": 714
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.831663326653307e-05,
+      "loss": 5.0694,
+      "step": 715
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.8116232464929865e-05,
+      "loss": 0.4038,
+      "step": 716
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.7915831663326656e-05,
+      "loss": 1.5647,
+      "step": 717
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.7715430861723455e-05,
+      "loss": 1.4984,
+      "step": 718
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.7515030060120247e-05,
+      "loss": 4.388,
+      "step": 719
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.731462925851704e-05,
+      "loss": 3.747,
+      "step": 720
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.711422845691383e-05,
+      "loss": 3.0854,
+      "step": 721
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.691382765531063e-05,
+      "loss": 1.4225,
+      "step": 722
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.671342685370742e-05,
+      "loss": 1.3527,
+      "step": 723
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.651302605210421e-05,
+      "loss": 1.6069,
+      "step": 724
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.6312625250501004e-05,
+      "loss": 2.4684,
+      "step": 725
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.6112224448897796e-05,
+      "loss": 3.3059,
+      "step": 726
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.5911823647294594e-05,
+      "loss": 1.2761,
+      "step": 727
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.5711422845691386e-05,
+      "loss": 4.7155,
+      "step": 728
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.551102204408818e-05,
+      "loss": 0.8618,
+      "step": 729
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.531062124248497e-05,
+      "loss": 0.1951,
+      "step": 730
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.511022044088177e-05,
+      "loss": 0.8723,
+      "step": 731
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.490981963927856e-05,
+      "loss": 2.6863,
+      "step": 732
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.470941883767535e-05,
+      "loss": 6.9487,
+      "step": 733
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.4509018036072143e-05,
+      "loss": 1.7146,
+      "step": 734
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.430861723446894e-05,
+      "loss": 4.26,
+      "step": 735
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.4108216432865734e-05,
+      "loss": 2.187,
+      "step": 736
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.3907815631262526e-05,
+      "loss": 2.7236,
+      "step": 737
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.370741482965932e-05,
+      "loss": 4.0348,
+      "step": 738
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.3507014028056116e-05,
+      "loss": 1.4374,
+      "step": 739
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.330661322645291e-05,
+      "loss": 1.0688,
+      "step": 740
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.31062124248497e-05,
+      "loss": 2.0594,
+      "step": 741
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.290581162324649e-05,
+      "loss": 0.8456,
+      "step": 742
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.270541082164328e-05,
+      "loss": 0.913,
+      "step": 743
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.250501002004008e-05,
+      "loss": 3.1388,
+      "step": 744
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.230460921843687e-05,
+      "loss": 2.5727,
+      "step": 745
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.2104208416833665e-05,
+      "loss": 1.4594,
+      "step": 746
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.190380761523046e-05,
+      "loss": 3.9008,
+      "step": 747
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.1703406813627255e-05,
+      "loss": 2.31,
+      "step": 748
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.150300601202405e-05,
+      "loss": 1.7205,
+      "step": 749
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.130260521042084e-05,
+      "loss": 2.2148,
+      "step": 750
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.110220440881763e-05,
+      "loss": 3.1292,
+      "step": 751
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.090180360721443e-05,
+      "loss": 2.3089,
+      "step": 752
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.070140280561122e-05,
+      "loss": 1.6105,
+      "step": 753
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.050100200400801e-05,
+      "loss": 3.2778,
+      "step": 754
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.0300601202404805e-05,
+      "loss": 0.0,
+      "step": 755
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.01002004008016e-05,
+      "loss": 1.9581,
+      "step": 756
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.98997995991984e-05,
+      "loss": 1.2599,
+      "step": 757
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9699398797595193e-05,
+      "loss": 1.5371,
+      "step": 758
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9498997995991985e-05,
+      "loss": 2.8294,
+      "step": 759
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.929859719438878e-05,
+      "loss": 2.2045,
+      "step": 760
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9098196392785576e-05,
+      "loss": 2.1054,
+      "step": 761
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.889779559118237e-05,
+      "loss": 1.8911,
+      "step": 762
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.869739478957916e-05,
+      "loss": 0.0,
+      "step": 763
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.849699398797595e-05,
+      "loss": 1.362,
+      "step": 764
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.829659318637275e-05,
+      "loss": 2.2082,
+      "step": 765
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.809619238476954e-05,
+      "loss": 1.4097,
+      "step": 766
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.789579158316633e-05,
+      "loss": 2.1663,
+      "step": 767
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.7695390781563125e-05,
+      "loss": 1.7452,
+      "step": 768
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.7494989979959916e-05,
+      "loss": 1.8244,
+      "step": 769
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.7294589178356715e-05,
+      "loss": 4.3847,
+      "step": 770
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.7094188376753514e-05,
+      "loss": 1.3235,
+      "step": 771
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.6893787575150305e-05,
+      "loss": 1.6868,
+      "step": 772
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.66933867735471e-05,
+      "loss": 0.4138,
+      "step": 773
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.649298597194389e-05,
+      "loss": 0.1626,
+      "step": 774
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.629258517034069e-05,
+      "loss": 1.6324,
+      "step": 775
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.609218436873748e-05,
+      "loss": 1.064,
+      "step": 776
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.589178356713427e-05,
+      "loss": 1.1306,
+      "step": 777
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.569138276553106e-05,
+      "loss": 9.5001,
+      "step": 778
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.549098196392786e-05,
+      "loss": 2.5549,
+      "step": 779
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.529058116232465e-05,
+      "loss": 1.4219,
+      "step": 780
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.5090180360721445e-05,
+      "loss": 0.9293,
+      "step": 781
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.488977955911824e-05,
+      "loss": 2.9165,
+      "step": 782
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.4689378757515035e-05,
+      "loss": 5.4542,
+      "step": 783
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.448897795591183e-05,
+      "loss": 1.6391,
+      "step": 784
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.428857715430862e-05,
+      "loss": 1.6708,
+      "step": 785
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.408817635270541e-05,
+      "loss": 1.7247,
+      "step": 786
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.388777555110221e-05,
+      "loss": 4.8019,
+      "step": 787
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.3687374749499e-05,
+      "loss": 2.9598,
+      "step": 788
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.348697394789579e-05,
+      "loss": 2.1231,
+      "step": 789
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.3286573146292584e-05,
+      "loss": 2.9594,
+      "step": 790
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.3086172344689376e-05,
+      "loss": 2.5611,
+      "step": 791
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.2885771543086175e-05,
+      "loss": 4.429,
+      "step": 792
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.2685370741482966e-05,
+      "loss": 4.2864,
+      "step": 793
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.248496993987976e-05,
+      "loss": 6.1905,
+      "step": 794
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.228456913827655e-05,
+      "loss": 5.4012,
+      "step": 795
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.208416833667335e-05,
+      "loss": 2.7522,
+      "step": 796
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.188376753507014e-05,
+      "loss": 1.9901,
+      "step": 797
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.168336673346693e-05,
+      "loss": 1.7421,
+      "step": 798
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.148296593186373e-05,
+      "loss": 0.7346,
+      "step": 799
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.128256513026052e-05,
+      "loss": 0.4529,
+      "step": 800
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.108216432865732e-05,
+      "loss": 1.7953,
+      "step": 801
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.088176352705411e-05,
+      "loss": 3.1981,
+      "step": 802
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.0681362725450904e-05,
+      "loss": 3.32,
+      "step": 803
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.0480961923847696e-05,
+      "loss": 4.134,
+      "step": 804
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.0280561122244495e-05,
+      "loss": 2.0281,
+      "step": 805
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.0080160320641287e-05,
+      "loss": 1.8402,
+      "step": 806
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.987975951903808e-05,
+      "loss": 3.2845,
+      "step": 807
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.967935871743487e-05,
+      "loss": 1.6719,
+      "step": 808
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.947895791583167e-05,
+      "loss": 0.0,
+      "step": 809
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.927855711422846e-05,
+      "loss": 1.9383,
+      "step": 810
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.907815631262525e-05,
+      "loss": 2.1896,
+      "step": 811
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.8877755511022044e-05,
+      "loss": 1.1527,
+      "step": 812
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.867735470941884e-05,
+      "loss": 5.3477,
+      "step": 813
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.8476953907815634e-05,
+      "loss": 1.5018,
+      "step": 814
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.8276553106212426e-05,
+      "loss": 1.241,
+      "step": 815
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.807615230460922e-05,
+      "loss": 2.193,
+      "step": 816
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.787575150300601e-05,
+      "loss": 2.021,
+      "step": 817
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.767535070140281e-05,
+      "loss": 0.6531,
+      "step": 818
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.74749498997996e-05,
+      "loss": 0.6331,
+      "step": 819
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.727454909819639e-05,
+      "loss": 0.0,
+      "step": 820
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.7074148296593183e-05,
+      "loss": 0.8708,
+      "step": 821
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.687374749498998e-05,
+      "loss": 2.0937,
+      "step": 822
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.6673346693386774e-05,
+      "loss": 2.5359,
+      "step": 823
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.6472945891783566e-05,
+      "loss": 1.7331,
+      "step": 824
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.627254509018036e-05,
+      "loss": 1.1601,
+      "step": 825
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.6072144288577156e-05,
+      "loss": 1.8112,
+      "step": 826
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.5871743486973954e-05,
+      "loss": 1.9119,
+      "step": 827
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.5671342685370746e-05,
+      "loss": 2.1902,
+      "step": 828
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.547094188376754e-05,
+      "loss": 0.5517,
+      "step": 829
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.527054108216433e-05,
+      "loss": 0.4456,
+      "step": 830
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.507014028056113e-05,
+      "loss": 5.7706,
+      "step": 831
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.486973947895792e-05,
+      "loss": 1.3838,
+      "step": 832
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.466933867735471e-05,
+      "loss": 1.9657,
+      "step": 833
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.4468937875751504e-05,
+      "loss": 1.5892,
+      "step": 834
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.42685370741483e-05,
+      "loss": 1.0192,
+      "step": 835
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.4068136272545094e-05,
+      "loss": 1.8128,
+      "step": 836
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.3867735470941886e-05,
+      "loss": 1.2665,
+      "step": 837
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.366733466933868e-05,
+      "loss": 1.9121,
+      "step": 838
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.3466933867735476e-05,
+      "loss": 1.7453,
+      "step": 839
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.326653306613227e-05,
+      "loss": 3.9064,
+      "step": 840
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.306613226452906e-05,
+      "loss": 0.9258,
+      "step": 841
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.286573146292585e-05,
+      "loss": 2.3671,
+      "step": 842
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.266533066132264e-05,
+      "loss": 1.9065,
+      "step": 843
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.246492985971944e-05,
+      "loss": 1.869,
+      "step": 844
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.2264529058116233e-05,
+      "loss": 0.7601,
+      "step": 845
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.2064128256513025e-05,
+      "loss": 0.4456,
+      "step": 846
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.186372745490982e-05,
+      "loss": 2.6767,
+      "step": 847
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.1663326653306616e-05,
+      "loss": 0.7504,
+      "step": 848
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.146292585170341e-05,
+      "loss": 5.5667,
+      "step": 849
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.12625250501002e-05,
+      "loss": 3.0322,
+      "step": 850
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.106212424849699e-05,
+      "loss": 1.73,
+      "step": 851
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.086172344689379e-05,
+      "loss": 1.1359,
+      "step": 852
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.066132264529058e-05,
+      "loss": 2.1959,
+      "step": 853
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.046092184368738e-05,
+      "loss": 2.5073,
+      "step": 854
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.026052104208417e-05,
+      "loss": 2.8468,
+      "step": 855
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.0060120240480967e-05,
+      "loss": 1.3095,
+      "step": 856
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.985971943887776e-05,
+      "loss": 3.6701,
+      "step": 857
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9659318637274554e-05,
+      "loss": 1.4558,
+      "step": 858
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9458917835671345e-05,
+      "loss": 2.5426,
+      "step": 859
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.925851703406814e-05,
+      "loss": 2.4209,
+      "step": 860
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9058116232464932e-05,
+      "loss": 0.5248,
+      "step": 861
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8857715430861727e-05,
+      "loss": 3.4234,
+      "step": 862
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.865731462925852e-05,
+      "loss": 2.9914,
+      "step": 863
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8456913827655314e-05,
+      "loss": 2.631,
+      "step": 864
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8256513026052106e-05,
+      "loss": 9.4896,
+      "step": 865
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8056112224448898e-05,
+      "loss": 1.3135,
+      "step": 866
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7855711422845693e-05,
+      "loss": 1.7307,
+      "step": 867
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7655310621242485e-05,
+      "loss": 6.0387,
+      "step": 868
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.745490981963928e-05,
+      "loss": 3.1468,
+      "step": 869
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7254509018036072e-05,
+      "loss": 2.8258,
+      "step": 870
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7054108216432867e-05,
+      "loss": 2.0105,
+      "step": 871
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.685370741482966e-05,
+      "loss": 1.774,
+      "step": 872
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6653306613226454e-05,
+      "loss": 2.0636,
+      "step": 873
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6452905811623246e-05,
+      "loss": 1.6851,
+      "step": 874
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.625250501002004e-05,
+      "loss": 1.1166,
+      "step": 875
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6052104208416833e-05,
+      "loss": 1.4389,
+      "step": 876
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.5851703406813628e-05,
+      "loss": 0.9514,
+      "step": 877
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.565130260521042e-05,
+      "loss": 2.6993,
+      "step": 878
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.5450901803607215e-05,
+      "loss": 2.7322,
+      "step": 879
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.5250501002004006e-05,
+      "loss": 6.489,
+      "step": 880
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.50501002004008e-05,
+      "loss": 0.8815,
+      "step": 881
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4849699398797597e-05,
+      "loss": 0.9959,
+      "step": 882
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.464929859719439e-05,
+      "loss": 0.0,
+      "step": 883
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4448897795591184e-05,
+      "loss": 2.0596,
+      "step": 884
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4248496993987975e-05,
+      "loss": 2.9538,
+      "step": 885
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.404809619238477e-05,
+      "loss": 0.7011,
+      "step": 886
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3847695390781562e-05,
+      "loss": 2.6931,
+      "step": 887
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3647294589178358e-05,
+      "loss": 3.6907,
+      "step": 888
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3446893787575153e-05,
+      "loss": 1.0606,
+      "step": 889
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3246492985971944e-05,
+      "loss": 1.1309,
+      "step": 890
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.304609218436874e-05,
+      "loss": 2.0832,
+      "step": 891
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.284569138276553e-05,
+      "loss": 1.9598,
+      "step": 892
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2645290581162327e-05,
+      "loss": 5.0853,
+      "step": 893
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.244488977955912e-05,
+      "loss": 1.3508,
+      "step": 894
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2244488977955913e-05,
+      "loss": 3.2122,
+      "step": 895
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2044088176352705e-05,
+      "loss": 1.5134,
+      "step": 896
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.18436873747495e-05,
+      "loss": 1.3606,
+      "step": 897
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1643286573146292e-05,
+      "loss": 0.9683,
+      "step": 898
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1442885771543087e-05,
+      "loss": 1.33,
+      "step": 899
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.124248496993988e-05,
+      "loss": 3.1403,
+      "step": 900
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1042084168336674e-05,
+      "loss": 3.3669,
+      "step": 901
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0841683366733466e-05,
+      "loss": 0.0,
+      "step": 902
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.064128256513026e-05,
+      "loss": 3.7623,
+      "step": 903
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0440881763527056e-05,
+      "loss": 0.1433,
+      "step": 904
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0240480961923848e-05,
+      "loss": 0.8717,
+      "step": 905
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0040080160320643e-05,
+      "loss": 2.4617,
+      "step": 906
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9839679358717435e-05,
+      "loss": 1.3359,
+      "step": 907
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.963927855711423e-05,
+      "loss": 0.9416,
+      "step": 908
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9438877755511022e-05,
+      "loss": 0.6057,
+      "step": 909
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9238476953907817e-05,
+      "loss": 2.5017,
+      "step": 910
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.903807615230461e-05,
+      "loss": 1.331,
+      "step": 911
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8837675350701404e-05,
+      "loss": 2.2103,
+      "step": 912
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8637274549098196e-05,
+      "loss": 1.4674,
+      "step": 913
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.843687374749499e-05,
+      "loss": 3.1923,
+      "step": 914
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8236472945891783e-05,
+      "loss": 1.8469,
+      "step": 915
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8036072144288578e-05,
+      "loss": 2.0753,
+      "step": 916
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7835671342685373e-05,
+      "loss": 2.9633,
+      "step": 917
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7635270541082165e-05,
+      "loss": 0.2268,
+      "step": 918
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.743486973947896e-05,
+      "loss": 2.7156,
+      "step": 919
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7234468937875752e-05,
+      "loss": 2.0701,
+      "step": 920
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7034068136272547e-05,
+      "loss": 1.3708,
+      "step": 921
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.683366733466934e-05,
+      "loss": 1.6181,
+      "step": 922
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6633266533066134e-05,
+      "loss": 1.5673,
+      "step": 923
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6432865731462926e-05,
+      "loss": 0.7278,
+      "step": 924
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.623246492985972e-05,
+      "loss": 1.1777,
+      "step": 925
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6032064128256513e-05,
+      "loss": 1.3947,
+      "step": 926
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5831663326653308e-05,
+      "loss": 1.3677,
+      "step": 927
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.56312625250501e-05,
+      "loss": 0.8851,
+      "step": 928
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5430861723446895e-05,
+      "loss": 3.3576,
+      "step": 929
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.523046092184369e-05,
+      "loss": 1.4984,
+      "step": 930
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5030060120240483e-05,
+      "loss": 0.8138,
+      "step": 931
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4829659318637277e-05,
+      "loss": 1.1379,
+      "step": 932
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.462925851703407e-05,
+      "loss": 1.8896,
+      "step": 933
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4428857715430864e-05,
+      "loss": 2.0607,
+      "step": 934
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4228456913827657e-05,
+      "loss": 4.4332,
+      "step": 935
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4028056112224449e-05,
+      "loss": 0.0,
+      "step": 936
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3827655310621242e-05,
+      "loss": 2.1725,
+      "step": 937
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3627254509018036e-05,
+      "loss": 2.1489,
+      "step": 938
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.342685370741483e-05,
+      "loss": 1.7468,
+      "step": 939
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3226452905811623e-05,
+      "loss": 1.3146,
+      "step": 940
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3026052104208416e-05,
+      "loss": 2.112,
+      "step": 941
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.282565130260521e-05,
+      "loss": 0.0,
+      "step": 942
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2625250501002003e-05,
+      "loss": 1.8155,
+      "step": 943
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2424849699398798e-05,
+      "loss": 1.6048,
+      "step": 944
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2224448897795592e-05,
+      "loss": 0.2253,
+      "step": 945
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2024048096192385e-05,
+      "loss": 4.1294,
+      "step": 946
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1823647294589179e-05,
+      "loss": 2.5192,
+      "step": 947
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1623246492985972e-05,
+      "loss": 0.52,
+      "step": 948
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1422845691382766e-05,
+      "loss": 0.0,
+      "step": 949
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.122244488977956e-05,
+      "loss": 8.9787,
+      "step": 950
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1022044088176353e-05,
+      "loss": 0.9179,
+      "step": 951
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0821643286573146e-05,
+      "loss": 1.3446,
+      "step": 952
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.062124248496994e-05,
+      "loss": 2.1705,
+      "step": 953
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0420841683366733e-05,
+      "loss": 2.7145,
+      "step": 954
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0220440881763528e-05,
+      "loss": 2.4458,
+      "step": 955
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0020040080160322e-05,
+      "loss": 1.471,
+      "step": 956
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.819639278557115e-06,
+      "loss": 0.0,
+      "step": 957
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.619238476953909e-06,
+      "loss": 2.9764,
+      "step": 958
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.418837675350702e-06,
+      "loss": 1.7468,
+      "step": 959
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.218436873747496e-06,
+      "loss": 2.3666,
+      "step": 960
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.018036072144289e-06,
+      "loss": 2.5105,
+      "step": 961
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.817635270541082e-06,
+      "loss": 2.2997,
+      "step": 962
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.617234468937876e-06,
+      "loss": 1.9821,
+      "step": 963
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.41683366733467e-06,
+      "loss": 0.8583,
+      "step": 964
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.216432865731463e-06,
+      "loss": 2.9195,
+      "step": 965
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.016032064128256e-06,
+      "loss": 3.1603,
+      "step": 966
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.81563126252505e-06,
+      "loss": 0.8383,
+      "step": 967
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.615230460921845e-06,
+      "loss": 2.7378,
+      "step": 968
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.414829659318638e-06,
+      "loss": 2.9844,
+      "step": 969
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.214428857715432e-06,
+      "loss": 1.8668,
+      "step": 970
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.0140280561122245e-06,
+      "loss": 1.6183,
+      "step": 971
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.813627254509018e-06,
+      "loss": 4.8758,
+      "step": 972
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.613226452905811e-06,
+      "loss": 2.247,
+      "step": 973
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.412825651302605e-06,
+      "loss": 1.6818,
+      "step": 974
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.212424849699399e-06,
+      "loss": 2.971,
+      "step": 975
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.012024048096193e-06,
+      "loss": 0.0,
+      "step": 976
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.811623246492986e-06,
+      "loss": 0.0,
+      "step": 977
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.61122244488978e-06,
+      "loss": 1.3841,
+      "step": 978
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.410821643286573e-06,
+      "loss": 1.3487,
+      "step": 979
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.2104208416833665e-06,
+      "loss": 2.0792,
+      "step": 980
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.010020040080161e-06,
+      "loss": 2.5128,
+      "step": 981
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.809619238476954e-06,
+      "loss": 4.2402,
+      "step": 982
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.609218436873748e-06,
+      "loss": 0.4601,
+      "step": 983
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.408817635270541e-06,
+      "loss": 2.5609,
+      "step": 984
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.208416833667335e-06,
+      "loss": 2.3339,
+      "step": 985
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.008016032064128e-06,
+      "loss": 0.9409,
+      "step": 986
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8076152304609225e-06,
+      "loss": 5.0006,
+      "step": 987
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.607214428857716e-06,
+      "loss": 3.4097,
+      "step": 988
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.406813627254509e-06,
+      "loss": 2.0473,
+      "step": 989
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2064128256513024e-06,
+      "loss": 1.3606,
+      "step": 990
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0060120240480963e-06,
+      "loss": 1.5403,
+      "step": 991
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.80561122244489e-06,
+      "loss": 1.9095,
+      "step": 992
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6052104208416833e-06,
+      "loss": 0.0,
+      "step": 993
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.404809619238477e-06,
+      "loss": 2.7986,
+      "step": 994
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2044088176352706e-06,
+      "loss": 1.1042,
+      "step": 995
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.004008016032064e-06,
+      "loss": 0.983,
+      "step": 996
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.803607214428858e-06,
+      "loss": 2.6895,
+      "step": 997
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6032064128256512e-06,
+      "loss": 1.5531,
+      "step": 998
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.402805611222445e-06,
+      "loss": 1.7102,
+      "step": 999
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2024048096192386e-06,
+      "loss": 6.9092,
+      "step": 1000
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 1000,
+  "num_train_epochs": 1,
+  "save_steps": 50,
+  "total_flos": 116009623166976.0,
+  "trial_name": null,
+  "trial_params": null
+}