diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,8170 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.9988925802879294,
+  "global_step": 1353,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.992609016999261e-05,
+      "loss": 1.0451,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.985218033998522e-05,
+      "loss": 3.8431,
+      "step": 2
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.977827050997783e-05,
+      "loss": 4.1686,
+      "step": 3
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.970436067997045e-05,
+      "loss": 4.1879,
+      "step": 4
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.963045084996306e-05,
+      "loss": 4.633,
+      "step": 5
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.955654101995566e-05,
+      "loss": 3.4155,
+      "step": 6
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.948263118994827e-05,
+      "loss": 3.3784,
+      "step": 7
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.940872135994088e-05,
+      "loss": 3.3627,
+      "step": 8
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.933481152993348e-05,
+      "loss": 3.2365,
+      "step": 9
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.926090169992609e-05,
+      "loss": 3.5945,
+      "step": 10
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.91869918699187e-05,
+      "loss": 3.4429,
+      "step": 11
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.911308203991131e-05,
+      "loss": 3.1824,
+      "step": 12
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.903917220990391e-05,
+      "loss": 3.2093,
+      "step": 13
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.896526237989652e-05,
+      "loss": 3.1471,
+      "step": 14
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.889135254988914e-05,
+      "loss": 3.0478,
+      "step": 15
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.881744271988175e-05,
+      "loss": 2.9833,
+      "step": 16
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.874353288987436e-05,
+      "loss": 2.9323,
+      "step": 17
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.866962305986696e-05,
+      "loss": 2.9048,
+      "step": 18
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.859571322985957e-05,
+      "loss": 2.8425,
+      "step": 19
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.852180339985218e-05,
+      "loss": 2.7644,
+      "step": 20
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.84478935698448e-05,
+      "loss": 2.7183,
+      "step": 21
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.837398373983741e-05,
+      "loss": 2.636,
+      "step": 22
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.830007390983002e-05,
+      "loss": 2.5524,
+      "step": 23
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.822616407982262e-05,
+      "loss": 2.4676,
+      "step": 24
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.815225424981523e-05,
+      "loss": 2.3894,
+      "step": 25
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.807834441980784e-05,
+      "loss": 2.3208,
+      "step": 26
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.800443458980046e-05,
+      "loss": 2.2373,
+      "step": 27
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.793052475979305e-05,
+      "loss": 2.1539,
+      "step": 28
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.785661492978566e-05,
+      "loss": 2.068,
+      "step": 29
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.778270509977827e-05,
+      "loss": 2.0042,
+      "step": 30
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.770879526977087e-05,
+      "loss": 1.9648,
+      "step": 31
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.76348854397635e-05,
+      "loss": 1.8853,
+      "step": 32
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.75609756097561e-05,
+      "loss": 1.8292,
+      "step": 33
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.748706577974871e-05,
+      "loss": 1.7822,
+      "step": 34
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.741315594974132e-05,
+      "loss": 1.7378,
+      "step": 35
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.733924611973393e-05,
+      "loss": 1.6707,
+      "step": 36
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.726533628972653e-05,
+      "loss": 1.6189,
+      "step": 37
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.719142645971915e-05,
+      "loss": 1.5756,
+      "step": 38
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.711751662971176e-05,
+      "loss": 1.5198,
+      "step": 39
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.704360679970437e-05,
+      "loss": 1.4909,
+      "step": 40
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.696969696969698e-05,
+      "loss": 1.4441,
+      "step": 41
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.689578713968958e-05,
+      "loss": 1.4154,
+      "step": 42
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.682187730968219e-05,
+      "loss": 1.3827,
+      "step": 43
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.674796747967481e-05,
+      "loss": 1.3536,
+      "step": 44
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.667405764966742e-05,
+      "loss": 1.3125,
+      "step": 45
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.660014781966001e-05,
+      "loss": 1.2792,
+      "step": 46
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.652623798965262e-05,
+      "loss": 1.264,
+      "step": 47
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.645232815964523e-05,
+      "loss": 1.2366,
+      "step": 48
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.637841832963785e-05,
+      "loss": 1.2136,
+      "step": 49
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.630450849963046e-05,
+      "loss": 1.2248,
+      "step": 50
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.623059866962306e-05,
+      "loss": 1.2063,
+      "step": 51
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.615668883961567e-05,
+      "loss": 1.2254,
+      "step": 52
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.608277900960828e-05,
+      "loss": 1.1853,
+      "step": 53
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.600886917960089e-05,
+      "loss": 1.1918,
+      "step": 54
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.59349593495935e-05,
+      "loss": 1.1578,
+      "step": 55
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.586104951958611e-05,
+      "loss": 1.1771,
+      "step": 56
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.578713968957872e-05,
+      "loss": 1.1823,
+      "step": 57
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.571322985957133e-05,
+      "loss": 1.1658,
+      "step": 58
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.563932002956394e-05,
+      "loss": 1.1454,
+      "step": 59
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.556541019955654e-05,
+      "loss": 1.1488,
+      "step": 60
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.549150036954916e-05,
+      "loss": 1.1033,
+      "step": 61
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.541759053954177e-05,
+      "loss": 1.1537,
+      "step": 62
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.534368070953438e-05,
+      "loss": 1.1373,
+      "step": 63
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.526977087952697e-05,
+      "loss": 1.1266,
+      "step": 64
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.519586104951958e-05,
+      "loss": 1.1476,
+      "step": 65
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.51219512195122e-05,
+      "loss": 1.124,
+      "step": 66
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.504804138950481e-05,
+      "loss": 1.1024,
+      "step": 67
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.497413155949742e-05,
+      "loss": 1.1007,
+      "step": 68
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.490022172949002e-05,
+      "loss": 1.1147,
+      "step": 69
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.482631189948263e-05,
+      "loss": 1.122,
+      "step": 70
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.475240206947524e-05,
+      "loss": 1.1051,
+      "step": 71
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.467849223946786e-05,
+      "loss": 1.1096,
+      "step": 72
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.460458240946047e-05,
+      "loss": 1.1002,
+      "step": 73
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.453067257945307e-05,
+      "loss": 1.0841,
+      "step": 74
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.445676274944568e-05,
+      "loss": 1.0895,
+      "step": 75
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.438285291943829e-05,
+      "loss": 1.0664,
+      "step": 76
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.43089430894309e-05,
+      "loss": 1.1017,
+      "step": 77
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.423503325942352e-05,
+      "loss": 1.085,
+      "step": 78
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.416112342941612e-05,
+      "loss": 1.0883,
+      "step": 79
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.408721359940873e-05,
+      "loss": 1.0971,
+      "step": 80
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.401330376940134e-05,
+      "loss": 1.0812,
+      "step": 81
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.393939393939395e-05,
+      "loss": 1.0899,
+      "step": 82
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.386548410938655e-05,
+      "loss": 1.0572,
+      "step": 83
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.379157427937916e-05,
+      "loss": 1.0804,
+      "step": 84
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.371766444937177e-05,
+      "loss": 1.0625,
+      "step": 85
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.364375461936438e-05,
+      "loss": 1.0475,
+      "step": 86
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.356984478935698e-05,
+      "loss": 1.0685,
+      "step": 87
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.349593495934959e-05,
+      "loss": 1.0713,
+      "step": 88
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.342202512934221e-05,
+      "loss": 1.0734,
+      "step": 89
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.334811529933482e-05,
+      "loss": 1.0612,
+      "step": 90
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.327420546932743e-05,
+      "loss": 1.0677,
+      "step": 91
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.320029563932003e-05,
+      "loss": 1.045,
+      "step": 92
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.312638580931264e-05,
+      "loss": 1.0542,
+      "step": 93
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.305247597930525e-05,
+      "loss": 1.0559,
+      "step": 94
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.297856614929787e-05,
+      "loss": 1.0632,
+      "step": 95
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.290465631929048e-05,
+      "loss": 1.0478,
+      "step": 96
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.283074648928308e-05,
+      "loss": 1.0505,
+      "step": 97
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.275683665927569e-05,
+      "loss": 1.0342,
+      "step": 98
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.26829268292683e-05,
+      "loss": 1.045,
+      "step": 99
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.26090169992609e-05,
+      "loss": 1.0467,
+      "step": 100
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.253510716925351e-05,
+      "loss": 1.0632,
+      "step": 101
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.246119733924612e-05,
+      "loss": 1.065,
+      "step": 102
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.238728750923873e-05,
+      "loss": 1.0673,
+      "step": 103
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.231337767923134e-05,
+      "loss": 1.0433,
+      "step": 104
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.223946784922394e-05,
+      "loss": 1.0589,
+      "step": 105
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.216555801921656e-05,
+      "loss": 1.0636,
+      "step": 106
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.209164818920917e-05,
+      "loss": 1.0308,
+      "step": 107
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.201773835920178e-05,
+      "loss": 1.0593,
+      "step": 108
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.194382852919439e-05,
+      "loss": 1.0585,
+      "step": 109
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.1869918699187e-05,
+      "loss": 1.0257,
+      "step": 110
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.17960088691796e-05,
+      "loss": 1.0379,
+      "step": 111
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.172209903917222e-05,
+      "loss": 1.0441,
+      "step": 112
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.164818920916483e-05,
+      "loss": 1.0517,
+      "step": 113
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.157427937915744e-05,
+      "loss": 1.0289,
+      "step": 114
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.150036954915004e-05,
+      "loss": 1.0441,
+      "step": 115
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.142645971914265e-05,
+      "loss": 1.0235,
+      "step": 116
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.135254988913526e-05,
+      "loss": 1.0419,
+      "step": 117
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.127864005912788e-05,
+      "loss": 1.049,
+      "step": 118
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.120473022912047e-05,
+      "loss": 1.0248,
+      "step": 119
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.113082039911308e-05,
+      "loss": 1.0314,
+      "step": 120
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.105691056910569e-05,
+      "loss": 1.0316,
+      "step": 121
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.09830007390983e-05,
+      "loss": 1.038,
+      "step": 122
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.090909090909092e-05,
+      "loss": 1.0373,
+      "step": 123
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.083518107908352e-05,
+      "loss": 1.0313,
+      "step": 124
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.076127124907613e-05,
+      "loss": 1.0317,
+      "step": 125
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.068736141906874e-05,
+      "loss": 1.0202,
+      "step": 126
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.061345158906135e-05,
+      "loss": 1.0621,
+      "step": 127
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.053954175905395e-05,
+      "loss": 1.0052,
+      "step": 128
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.046563192904656e-05,
+      "loss": 1.023,
+      "step": 129
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.039172209903918e-05,
+      "loss": 1.0052,
+      "step": 130
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.031781226903179e-05,
+      "loss": 1.0358,
+      "step": 131
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.02439024390244e-05,
+      "loss": 1.0257,
+      "step": 132
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.0169992609017e-05,
+      "loss": 1.0233,
+      "step": 133
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.009608277900961e-05,
+      "loss": 1.0122,
+      "step": 134
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.002217294900222e-05,
+      "loss": 0.9906,
+      "step": 135
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.994826311899484e-05,
+      "loss": 1.0247,
+      "step": 136
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.987435328898745e-05,
+      "loss": 1.0287,
+      "step": 137
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.980044345898004e-05,
+      "loss": 1.0192,
+      "step": 138
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.972653362897265e-05,
+      "loss": 1.007,
+      "step": 139
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.965262379896526e-05,
+      "loss": 1.0074,
+      "step": 140
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.957871396895788e-05,
+      "loss": 1.0187,
+      "step": 141
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.950480413895048e-05,
+      "loss": 1.0192,
+      "step": 142
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.943089430894309e-05,
+      "loss": 1.0086,
+      "step": 143
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.93569844789357e-05,
+      "loss": 0.9988,
+      "step": 144
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.928307464892831e-05,
+      "loss": 1.0108,
+      "step": 145
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.920916481892091e-05,
+      "loss": 1.0013,
+      "step": 146
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.913525498891354e-05,
+      "loss": 1.0042,
+      "step": 147
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.906134515890614e-05,
+      "loss": 0.993,
+      "step": 148
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.898743532889875e-05,
+      "loss": 1.0033,
+      "step": 149
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.891352549889136e-05,
+      "loss": 1.0007,
+      "step": 150
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.883961566888396e-05,
+      "loss": 0.9968,
+      "step": 151
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.876570583887657e-05,
+      "loss": 1.0201,
+      "step": 152
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.869179600886919e-05,
+      "loss": 1.0137,
+      "step": 153
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.86178861788618e-05,
+      "loss": 1.0233,
+      "step": 154
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.854397634885441e-05,
+      "loss": 0.9938,
+      "step": 155
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.8470066518847e-05,
+      "loss": 0.988,
+      "step": 156
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.839615668883961e-05,
+      "loss": 0.9947,
+      "step": 157
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.832224685883223e-05,
+      "loss": 0.9819,
+      "step": 158
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.824833702882484e-05,
+      "loss": 0.9967,
+      "step": 159
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.817442719881744e-05,
+      "loss": 1.0239,
+      "step": 160
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.810051736881005e-05,
+      "loss": 0.9696,
+      "step": 161
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.802660753880266e-05,
+      "loss": 0.9997,
+      "step": 162
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.795269770879527e-05,
+      "loss": 0.9927,
+      "step": 163
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.787878787878789e-05,
+      "loss": 0.9806,
+      "step": 164
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.78048780487805e-05,
+      "loss": 0.9724,
+      "step": 165
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.77309682187731e-05,
+      "loss": 1.0076,
+      "step": 166
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.765705838876571e-05,
+      "loss": 0.9766,
+      "step": 167
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.758314855875832e-05,
+      "loss": 1.027,
+      "step": 168
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.750923872875092e-05,
+      "loss": 0.9988,
+      "step": 169
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.743532889874355e-05,
+      "loss": 0.9907,
+      "step": 170
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.736141906873615e-05,
+      "loss": 0.9686,
+      "step": 171
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.728750923872876e-05,
+      "loss": 0.9985,
+      "step": 172
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.721359940872137e-05,
+      "loss": 0.9995,
+      "step": 173
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 8.713968957871396e-05,
+      "loss": 0.9907,
+      "step": 174
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 8.706577974870658e-05,
+      "loss": 1.0031,
+      "step": 175
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 8.699186991869919e-05,
+      "loss": 0.9838,
+      "step": 176
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 8.69179600886918e-05,
+      "loss": 1.0116,
+      "step": 177
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 8.68440502586844e-05,
+      "loss": 0.9775,
+      "step": 178
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.677014042867701e-05,
+      "loss": 0.9743,
+      "step": 179
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.669623059866962e-05,
+      "loss": 0.9861,
+      "step": 180
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.662232076866224e-05,
+      "loss": 0.9761,
+      "step": 181
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.654841093865485e-05,
+      "loss": 0.9742,
+      "step": 182
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.647450110864746e-05,
+      "loss": 0.9878,
+      "step": 183
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.640059127864006e-05,
+      "loss": 0.9682,
+      "step": 184
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.632668144863267e-05,
+      "loss": 0.9822,
+      "step": 185
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.625277161862528e-05,
+      "loss": 0.9825,
+      "step": 186
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.61788617886179e-05,
+      "loss": 0.9977,
+      "step": 187
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.61049519586105e-05,
+      "loss": 0.9965,
+      "step": 188
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.603104212860311e-05,
+      "loss": 0.9943,
+      "step": 189
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.595713229859572e-05,
+      "loss": 0.9714,
+      "step": 190
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.588322246858833e-05,
+      "loss": 0.9891,
+      "step": 191
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.580931263858094e-05,
+      "loss": 0.9951,
+      "step": 192
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.573540280857354e-05,
+      "loss": 1.0029,
+      "step": 193
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.566149297856615e-05,
+      "loss": 0.9935,
+      "step": 194
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.558758314855876e-05,
+      "loss": 0.9812,
+      "step": 195
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.551367331855137e-05,
+      "loss": 0.979,
+      "step": 196
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.543976348854397e-05,
+      "loss": 0.971,
+      "step": 197
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.53658536585366e-05,
+      "loss": 0.9928,
+      "step": 198
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.52919438285292e-05,
+      "loss": 0.9819,
+      "step": 199
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.521803399852181e-05,
+      "loss": 0.9603,
+      "step": 200
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.514412416851442e-05,
+      "loss": 0.9798,
+      "step": 201
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.507021433850702e-05,
+      "loss": 0.9626,
+      "step": 202
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.499630450849963e-05,
+      "loss": 0.9701,
+      "step": 203
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.492239467849225e-05,
+      "loss": 0.9601,
+      "step": 204
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.484848484848486e-05,
+      "loss": 0.9801,
+      "step": 205
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.477457501847747e-05,
+      "loss": 0.9757,
+      "step": 206
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.470066518847007e-05,
+      "loss": 0.9777,
+      "step": 207
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.462675535846268e-05,
+      "loss": 0.9564,
+      "step": 208
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.455284552845529e-05,
+      "loss": 0.9692,
+      "step": 209
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.447893569844791e-05,
+      "loss": 0.9658,
+      "step": 210
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.44050258684405e-05,
+      "loss": 0.9849,
+      "step": 211
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.433111603843311e-05,
+      "loss": 0.9817,
+      "step": 212
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.425720620842572e-05,
+      "loss": 0.9945,
+      "step": 213
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.418329637841833e-05,
+      "loss": 0.9493,
+      "step": 214
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.410938654841095e-05,
+      "loss": 0.9654,
+      "step": 215
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.403547671840355e-05,
+      "loss": 0.973,
+      "step": 216
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.396156688839616e-05,
+      "loss": 0.9676,
+      "step": 217
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.388765705838877e-05,
+      "loss": 0.9553,
+      "step": 218
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.381374722838138e-05,
+      "loss": 0.9632,
+      "step": 219
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.373983739837398e-05,
+      "loss": 0.9514,
+      "step": 220
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.36659275683666e-05,
+      "loss": 0.9498,
+      "step": 221
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.359201773835921e-05,
+      "loss": 0.9451,
+      "step": 222
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.351810790835182e-05,
+      "loss": 0.9893,
+      "step": 223
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.344419807834443e-05,
+      "loss": 0.9861,
+      "step": 224
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.337028824833703e-05,
+      "loss": 0.94,
+      "step": 225
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.329637841832964e-05,
+      "loss": 0.9659,
+      "step": 226
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.322246858832226e-05,
+      "loss": 0.9403,
+      "step": 227
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.314855875831487e-05,
+      "loss": 0.9683,
+      "step": 228
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.307464892830746e-05,
+      "loss": 0.9597,
+      "step": 229
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.300073909830007e-05,
+      "loss": 0.926,
+      "step": 230
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.292682926829268e-05,
+      "loss": 0.9518,
+      "step": 231
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.28529194382853e-05,
+      "loss": 0.9683,
+      "step": 232
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.27790096082779e-05,
+      "loss": 0.9516,
+      "step": 233
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.270509977827051e-05,
+      "loss": 0.9467,
+      "step": 234
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.263118994826312e-05,
+      "loss": 0.9752,
+      "step": 235
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.255728011825573e-05,
+      "loss": 0.9506,
+      "step": 236
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.248337028824834e-05,
+      "loss": 0.9516,
+      "step": 237
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.240946045824096e-05,
+      "loss": 0.9787,
+      "step": 238
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.233555062823356e-05,
+      "loss": 0.9477,
+      "step": 239
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.226164079822617e-05,
+      "loss": 0.9686,
+      "step": 240
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.218773096821878e-05,
+      "loss": 0.9642,
+      "step": 241
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.211382113821139e-05,
+      "loss": 0.9792,
+      "step": 242
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.2039911308204e-05,
+      "loss": 0.9794,
+      "step": 243
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.19660014781966e-05,
+      "loss": 0.9446,
+      "step": 244
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.189209164818922e-05,
+      "loss": 0.9402,
+      "step": 245
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.181818181818183e-05,
+      "loss": 0.9475,
+      "step": 246
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.174427198817442e-05,
+      "loss": 0.9269,
+      "step": 247
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.167036215816703e-05,
+      "loss": 0.9432,
+      "step": 248
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.159645232815965e-05,
+      "loss": 0.9557,
+      "step": 249
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.152254249815226e-05,
+      "loss": 0.9549,
+      "step": 250
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.144863266814487e-05,
+      "loss": 0.9571,
+      "step": 251
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.137472283813747e-05,
+      "loss": 0.9229,
+      "step": 252
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.130081300813008e-05,
+      "loss": 0.9485,
+      "step": 253
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.122690317812269e-05,
+      "loss": 0.936,
+      "step": 254
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.11529933481153e-05,
+      "loss": 0.9451,
+      "step": 255
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.107908351810792e-05,
+      "loss": 0.9275,
+      "step": 256
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.100517368810052e-05,
+      "loss": 0.9603,
+      "step": 257
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.093126385809313e-05,
+      "loss": 0.9744,
+      "step": 258
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.085735402808574e-05,
+      "loss": 0.9491,
+      "step": 259
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.078344419807835e-05,
+      "loss": 0.9197,
+      "step": 260
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.070953436807095e-05,
+      "loss": 0.9557,
+      "step": 261
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.063562453806357e-05,
+      "loss": 0.957,
+      "step": 262
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.056171470805618e-05,
+      "loss": 0.9301,
+      "step": 263
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.048780487804879e-05,
+      "loss": 0.9392,
+      "step": 264
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.04138950480414e-05,
+      "loss": 0.9258,
+      "step": 265
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.033998521803399e-05,
+      "loss": 0.934,
+      "step": 266
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.026607538802661e-05,
+      "loss": 0.9387,
+      "step": 267
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.019216555801922e-05,
+      "loss": 0.9501,
+      "step": 268
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.011825572801183e-05,
+      "loss": 0.9392,
+      "step": 269
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.004434589800443e-05,
+      "loss": 0.9384,
+      "step": 270
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.997043606799704e-05,
+      "loss": 0.9287,
+      "step": 271
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.989652623798965e-05,
+      "loss": 0.9518,
+      "step": 272
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.982261640798227e-05,
+      "loss": 0.9368,
+      "step": 273
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.974870657797488e-05,
+      "loss": 0.9392,
+      "step": 274
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.967479674796748e-05,
+      "loss": 0.948,
+      "step": 275
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.960088691796009e-05,
+      "loss": 0.9303,
+      "step": 276
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.95269770879527e-05,
+      "loss": 0.8942,
+      "step": 277
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.94530672579453e-05,
+      "loss": 0.9253,
+      "step": 278
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.937915742793793e-05,
+      "loss": 0.9061,
+      "step": 279
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.930524759793053e-05,
+      "loss": 0.9449,
+      "step": 280
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.923133776792314e-05,
+      "loss": 0.9148,
+      "step": 281
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.915742793791575e-05,
+      "loss": 0.9449,
+      "step": 282
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.908351810790836e-05,
+      "loss": 0.9259,
+      "step": 283
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.900960827790096e-05,
+      "loss": 0.9166,
+      "step": 284
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.893569844789357e-05,
+      "loss": 0.9077,
+      "step": 285
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.886178861788618e-05,
+      "loss": 0.9183,
+      "step": 286
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.878787878787879e-05,
+      "loss": 0.9027,
+      "step": 287
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.87139689578714e-05,
+      "loss": 0.9168,
+      "step": 288
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.8640059127864e-05,
+      "loss": 0.9315,
+      "step": 289
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.856614929785662e-05,
+      "loss": 0.9241,
+      "step": 290
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.849223946784923e-05,
+      "loss": 0.9114,
+      "step": 291
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.841832963784184e-05,
+      "loss": 0.9422,
+      "step": 292
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.834441980783444e-05,
+      "loss": 0.9066,
+      "step": 293
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.827050997782705e-05,
+      "loss": 0.9089,
+      "step": 294
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.819660014781966e-05,
+      "loss": 0.9189,
+      "step": 295
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.812269031781228e-05,
+      "loss": 0.8879,
+      "step": 296
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.804878048780489e-05,
+      "loss": 0.8975,
+      "step": 297
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.79748706577975e-05,
+      "loss": 0.9212,
+      "step": 298
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.79009608277901e-05,
+      "loss": 0.8963,
+      "step": 299
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.782705099778271e-05,
+      "loss": 0.9264,
+      "step": 300
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.775314116777532e-05,
+      "loss": 0.9348,
+      "step": 301
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.767923133776792e-05,
+      "loss": 0.9171,
+      "step": 302
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.760532150776053e-05,
+      "loss": 0.9223,
+      "step": 303
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.753141167775314e-05,
+      "loss": 0.902,
+      "step": 304
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.745750184774575e-05,
+      "loss": 0.9099,
+      "step": 305
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.738359201773835e-05,
+      "loss": 0.9211,
+      "step": 306
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.730968218773098e-05,
+      "loss": 0.9131,
+      "step": 307
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.723577235772358e-05,
+      "loss": 0.9035,
+      "step": 308
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.716186252771619e-05,
+      "loss": 0.9464,
+      "step": 309
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.70879526977088e-05,
+      "loss": 0.925,
+      "step": 310
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.70140428677014e-05,
+      "loss": 0.9286,
+      "step": 311
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.694013303769401e-05,
+      "loss": 0.9381,
+      "step": 312
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.686622320768663e-05,
+      "loss": 0.9158,
+      "step": 313
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.679231337767924e-05,
+      "loss": 0.9184,
+      "step": 314
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.671840354767185e-05,
+      "loss": 0.9114,
+      "step": 315
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.664449371766446e-05,
+      "loss": 0.8987,
+      "step": 316
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.657058388765706e-05,
+      "loss": 0.9153,
+      "step": 317
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.649667405764967e-05,
+      "loss": 0.9236,
+      "step": 318
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.642276422764229e-05,
+      "loss": 0.9105,
+      "step": 319
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.634885439763488e-05,
+      "loss": 0.9024,
+      "step": 320
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.627494456762749e-05,
+      "loss": 0.9225,
+      "step": 321
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.62010347376201e-05,
+      "loss": 0.8984,
+      "step": 322
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.612712490761271e-05,
+      "loss": 0.9116,
+      "step": 323
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.605321507760533e-05,
+      "loss": 0.8936,
+      "step": 324
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.597930524759794e-05,
+      "loss": 0.9155,
+      "step": 325
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.590539541759054e-05,
+      "loss": 0.9079,
+      "step": 326
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.583148558758315e-05,
+      "loss": 0.8987,
+      "step": 327
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.575757575757576e-05,
+      "loss": 0.9167,
+      "step": 328
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.568366592756836e-05,
+      "loss": 0.897,
+      "step": 329
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.560975609756099e-05,
+      "loss": 0.8923,
+      "step": 330
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.55358462675536e-05,
+      "loss": 0.8807,
+      "step": 331
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.54619364375462e-05,
+      "loss": 0.8959,
+      "step": 332
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.538802660753881e-05,
+      "loss": 0.8835,
+      "step": 333
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.531411677753142e-05,
+      "loss": 0.9158,
+      "step": 334
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.524020694752402e-05,
+      "loss": 0.9144,
+      "step": 335
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.516629711751664e-05,
+      "loss": 0.8882,
+      "step": 336
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.509238728750925e-05,
+      "loss": 0.9033,
+      "step": 337
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.501847745750186e-05,
+      "loss": 0.9014,
+      "step": 338
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.494456762749445e-05,
+      "loss": 0.8739,
+      "step": 339
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.487065779748706e-05,
+      "loss": 0.8935,
+      "step": 340
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.479674796747968e-05,
+      "loss": 0.8744,
+      "step": 341
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.472283813747229e-05,
+      "loss": 0.8975,
+      "step": 342
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.46489283074649e-05,
+      "loss": 0.8932,
+      "step": 343
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.45750184774575e-05,
+      "loss": 0.9007,
+      "step": 344
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.450110864745011e-05,
+      "loss": 0.8956,
+      "step": 345
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.442719881744272e-05,
+      "loss": 0.9183,
+      "step": 346
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.435328898743534e-05,
+      "loss": 0.9186,
+      "step": 347
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.427937915742795e-05,
+      "loss": 0.9008,
+      "step": 348
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.420546932742055e-05,
+      "loss": 0.9091,
+      "step": 349
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.413155949741316e-05,
+      "loss": 0.8968,
+      "step": 350
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.405764966740577e-05,
+      "loss": 0.8865,
+      "step": 351
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.398373983739838e-05,
+      "loss": 0.8842,
+      "step": 352
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.3909830007391e-05,
+      "loss": 0.8815,
+      "step": 353
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.38359201773836e-05,
+      "loss": 0.8858,
+      "step": 354
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 7.376201034737621e-05,
+      "loss": 0.8848,
+      "step": 355
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 7.368810051736882e-05,
+      "loss": 0.8863,
+      "step": 356
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 7.361419068736141e-05,
+      "loss": 0.8966,
+      "step": 357
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 7.354028085735403e-05,
+      "loss": 0.8848,
+      "step": 358
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 7.346637102734664e-05,
+      "loss": 0.8607,
+      "step": 359
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 7.339246119733925e-05,
+      "loss": 0.8998,
+      "step": 360
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 7.331855136733186e-05,
+      "loss": 0.8809,
+      "step": 361
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 7.324464153732446e-05,
+      "loss": 0.8668,
+      "step": 362
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 7.317073170731707e-05,
+      "loss": 0.8717,
+      "step": 363
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 7.309682187730969e-05,
+      "loss": 0.8684,
+      "step": 364
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 7.30229120473023e-05,
+      "loss": 0.896,
+      "step": 365
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 7.29490022172949e-05,
+      "loss": 0.8838,
+      "step": 366
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 7.287509238728751e-05,
+      "loss": 0.8756,
+      "step": 367
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.280118255728012e-05,
+      "loss": 0.8886,
+      "step": 368
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.272727272727273e-05,
+      "loss": 0.9048,
+      "step": 369
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.265336289726534e-05,
+      "loss": 0.899,
+      "step": 370
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.257945306725796e-05,
+      "loss": 0.8528,
+      "step": 371
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.250554323725056e-05,
+      "loss": 0.8946,
+      "step": 372
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.243163340724317e-05,
+      "loss": 0.8688,
+      "step": 373
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.235772357723578e-05,
+      "loss": 0.8938,
+      "step": 374
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.228381374722837e-05,
+      "loss": 0.881,
+      "step": 375
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.2209903917221e-05,
+      "loss": 0.888,
+      "step": 376
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.21359940872136e-05,
+      "loss": 0.8929,
+      "step": 377
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.206208425720621e-05,
+      "loss": 0.8655,
+      "step": 378
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.198817442719882e-05,
+      "loss": 0.8491,
+      "step": 379
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.191426459719142e-05,
+      "loss": 0.8847,
+      "step": 380
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.184035476718403e-05,
+      "loss": 0.8766,
+      "step": 381
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.176644493717665e-05,
+      "loss": 0.8869,
+      "step": 382
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.169253510716926e-05,
+      "loss": 0.8822,
+      "step": 383
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.161862527716187e-05,
+      "loss": 0.8764,
+      "step": 384
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.154471544715447e-05,
+      "loss": 0.8828,
+      "step": 385
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.147080561714708e-05,
+      "loss": 0.8862,
+      "step": 386
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.139689578713969e-05,
+      "loss": 0.8785,
+      "step": 387
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.132298595713231e-05,
+      "loss": 0.8772,
+      "step": 388
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.124907612712492e-05,
+      "loss": 0.8598,
+      "step": 389
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.117516629711752e-05,
+      "loss": 0.8865,
+      "step": 390
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.110125646711013e-05,
+      "loss": 0.8658,
+      "step": 391
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.102734663710274e-05,
+      "loss": 0.8777,
+      "step": 392
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.095343680709535e-05,
+      "loss": 0.8555,
+      "step": 393
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.087952697708795e-05,
+      "loss": 0.8594,
+      "step": 394
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.080561714708056e-05,
+      "loss": 0.8785,
+      "step": 395
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.073170731707317e-05,
+      "loss": 0.8549,
+      "step": 396
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.065779748706578e-05,
+      "loss": 0.8745,
+      "step": 397
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.058388765705838e-05,
+      "loss": 0.8816,
+      "step": 398
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.0509977827051e-05,
+      "loss": 0.8734,
+      "step": 399
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.043606799704361e-05,
+      "loss": 0.859,
+      "step": 400
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.036215816703622e-05,
+      "loss": 0.8587,
+      "step": 401
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.028824833702883e-05,
+      "loss": 0.8539,
+      "step": 402
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.021433850702143e-05,
+      "loss": 0.8734,
+      "step": 403
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.014042867701404e-05,
+      "loss": 0.8576,
+      "step": 404
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.006651884700666e-05,
+      "loss": 0.8494,
+      "step": 405
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.999260901699927e-05,
+      "loss": 0.887,
+      "step": 406
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.991869918699188e-05,
+      "loss": 0.8724,
+      "step": 407
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.984478935698448e-05,
+      "loss": 0.8618,
+      "step": 408
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 6.977087952697709e-05,
+      "loss": 0.8801,
+      "step": 409
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 6.96969696969697e-05,
+      "loss": 0.858,
+      "step": 410
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 6.962305986696232e-05,
+      "loss": 0.8715,
+      "step": 411
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 6.954915003695491e-05,
+      "loss": 0.8698,
+      "step": 412
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 6.947524020694752e-05,
+      "loss": 0.8644,
+      "step": 413
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.940133037694013e-05,
+      "loss": 0.8688,
+      "step": 414
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.932742054693274e-05,
+      "loss": 0.8771,
+      "step": 415
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.925351071692536e-05,
+      "loss": 0.8753,
+      "step": 416
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.917960088691796e-05,
+      "loss": 0.8978,
+      "step": 417
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.910569105691057e-05,
+      "loss": 0.8655,
+      "step": 418
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.903178122690318e-05,
+      "loss": 0.8651,
+      "step": 419
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.895787139689579e-05,
+      "loss": 0.8668,
+      "step": 420
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.88839615668884e-05,
+      "loss": 0.8648,
+      "step": 421
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.881005173688101e-05,
+      "loss": 0.843,
+      "step": 422
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.873614190687362e-05,
+      "loss": 0.8458,
+      "step": 423
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.866223207686623e-05,
+      "loss": 0.8617,
+      "step": 424
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.858832224685884e-05,
+      "loss": 0.8539,
+      "step": 425
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.851441241685144e-05,
+      "loss": 0.8422,
+      "step": 426
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.844050258684405e-05,
+      "loss": 0.8718,
+      "step": 427
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.836659275683667e-05,
+      "loss": 0.8288,
+      "step": 428
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.829268292682928e-05,
+      "loss": 0.8255,
+      "step": 429
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.821877309682187e-05,
+      "loss": 0.8324,
+      "step": 430
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.814486326681448e-05,
+      "loss": 0.844,
+      "step": 431
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.807095343680709e-05,
+      "loss": 0.8382,
+      "step": 432
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.799704360679971e-05,
+      "loss": 0.8692,
+      "step": 433
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.792313377679232e-05,
+      "loss": 0.8357,
+      "step": 434
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.784922394678492e-05,
+      "loss": 0.8666,
+      "step": 435
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.777531411677753e-05,
+      "loss": 0.8681,
+      "step": 436
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.770140428677014e-05,
+      "loss": 0.8415,
+      "step": 437
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.762749445676275e-05,
+      "loss": 0.8398,
+      "step": 438
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.755358462675537e-05,
+      "loss": 0.8264,
+      "step": 439
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.747967479674798e-05,
+      "loss": 0.8649,
+      "step": 440
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.740576496674058e-05,
+      "loss": 0.8357,
+      "step": 441
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.733185513673319e-05,
+      "loss": 0.8502,
+      "step": 442
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.72579453067258e-05,
+      "loss": 0.8603,
+      "step": 443
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.71840354767184e-05,
+      "loss": 0.8323,
+      "step": 444
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.711012564671103e-05,
+      "loss": 0.8238,
+      "step": 445
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.703621581670363e-05,
+      "loss": 0.8522,
+      "step": 446
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.696230598669624e-05,
+      "loss": 0.8351,
+      "step": 447
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.688839615668883e-05,
+      "loss": 0.832,
+      "step": 448
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.681448632668144e-05,
+      "loss": 0.8478,
+      "step": 449
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.674057649667406e-05,
+      "loss": 0.8497,
+      "step": 450
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.666666666666667e-05,
+      "loss": 0.8427,
+      "step": 451
+    },
+    {
+      "epoch": 1.0,
+      "eval_loss": 1.0878102779388428,
+      "eval_runtime": 495.5202,
+      "eval_samples_per_second": 32.168,
+      "eval_steps_per_second": 0.505,
+      "eval_wer": 0.8091472356907791,
+      "step": 451
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.659275683665928e-05,
+      "loss": 1.2494,
+      "step": 452
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.651884700665188e-05,
+      "loss": 0.8478,
+      "step": 453
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 6.644493717664449e-05,
+      "loss": 0.8247,
+      "step": 454
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 6.63710273466371e-05,
+      "loss": 0.8202,
+      "step": 455
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 6.629711751662972e-05,
+      "loss": 0.8152,
+      "step": 456
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 6.622320768662233e-05,
+      "loss": 0.8248,
+      "step": 457
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 6.614929785661494e-05,
+      "loss": 0.8315,
+      "step": 458
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 6.607538802660754e-05,
+      "loss": 0.8001,
+      "step": 459
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 6.600147819660015e-05,
+      "loss": 0.841,
+      "step": 460
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 6.592756836659276e-05,
+      "loss": 0.8395,
+      "step": 461
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 6.585365853658538e-05,
+      "loss": 0.8209,
+      "step": 462
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 6.577974870657799e-05,
+      "loss": 0.8413,
+      "step": 463
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 6.570583887657059e-05,
+      "loss": 0.8145,
+      "step": 464
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 6.56319290465632e-05,
+      "loss": 0.8217,
+      "step": 465
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 6.555801921655581e-05,
+      "loss": 0.8535,
+      "step": 466
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 6.548410938654842e-05,
+      "loss": 0.8173,
+      "step": 467
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 6.541019955654102e-05,
+      "loss": 0.8303,
+      "step": 468
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 6.533628972653363e-05,
+      "loss": 0.8416,
+      "step": 469
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 6.526237989652624e-05,
+      "loss": 0.8218,
+      "step": 470
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 6.518847006651884e-05,
+      "loss": 0.8257,
+      "step": 471
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 6.511456023651145e-05,
+      "loss": 0.8319,
+      "step": 472
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 6.504065040650407e-05,
+      "loss": 0.8535,
+      "step": 473
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 6.496674057649668e-05,
+      "loss": 0.8369,
+      "step": 474
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 6.489283074648929e-05,
+      "loss": 0.8115,
+      "step": 475
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 6.48189209164819e-05,
+      "loss": 0.8091,
+      "step": 476
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 6.47450110864745e-05,
+      "loss": 0.8345,
+      "step": 477
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 6.467110125646711e-05,
+      "loss": 0.8542,
+      "step": 478
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 6.459719142645973e-05,
+      "loss": 0.8158,
+      "step": 479
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 6.452328159645234e-05,
+      "loss": 0.829,
+      "step": 480
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 6.444937176644495e-05,
+      "loss": 0.8146,
+      "step": 481
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 6.437546193643755e-05,
+      "loss": 0.8232,
+      "step": 482
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 6.430155210643016e-05,
+      "loss": 0.8075,
+      "step": 483
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 6.422764227642277e-05,
+      "loss": 0.8333,
+      "step": 484
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 6.415373244641538e-05,
+      "loss": 0.8112,
+      "step": 485
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 6.407982261640798e-05,
+      "loss": 0.809,
+      "step": 486
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 6.400591278640059e-05,
+      "loss": 0.7983,
+      "step": 487
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 6.39320029563932e-05,
+      "loss": 0.7997,
+      "step": 488
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 6.38580931263858e-05,
+      "loss": 0.8087,
+      "step": 489
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 6.378418329637843e-05,
+      "loss": 0.8165,
+      "step": 490
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 6.371027346637103e-05,
+      "loss": 0.795,
+      "step": 491
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 6.363636363636364e-05,
+      "loss": 0.8302,
+      "step": 492
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 6.356245380635625e-05,
+      "loss": 0.7928,
+      "step": 493
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 6.348854397634886e-05,
+      "loss": 0.822,
+      "step": 494
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 6.341463414634146e-05,
+      "loss": 0.8243,
+      "step": 495
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 6.334072431633407e-05,
+      "loss": 0.8252,
+      "step": 496
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 6.326681448632669e-05,
+      "loss": 0.8079,
+      "step": 497
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 6.31929046563193e-05,
+      "loss": 0.8181,
+      "step": 498
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 6.31189948263119e-05,
+      "loss": 0.8216,
+      "step": 499
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 6.304508499630451e-05,
+      "loss": 0.8127,
+      "step": 500
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 6.297117516629712e-05,
+      "loss": 0.7971,
+      "step": 501
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 6.289726533628973e-05,
+      "loss": 0.8005,
+      "step": 502
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 6.282335550628234e-05,
+      "loss": 0.8065,
+      "step": 503
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 6.274944567627494e-05,
+      "loss": 0.7944,
+      "step": 504
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 6.267553584626755e-05,
+      "loss": 0.8219,
+      "step": 505
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 6.260162601626016e-05,
+      "loss": 0.8399,
+      "step": 506
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 6.252771618625277e-05,
+      "loss": 0.7951,
+      "step": 507
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 6.245380635624539e-05,
+      "loss": 0.8164,
+      "step": 508
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 6.2379896526238e-05,
+      "loss": 0.8104,
+      "step": 509
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 6.23059866962306e-05,
+      "loss": 0.8094,
+      "step": 510
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 6.223207686622321e-05,
+      "loss": 0.8018,
+      "step": 511
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 6.215816703621582e-05,
+      "loss": 0.7701,
+      "step": 512
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 6.208425720620842e-05,
+      "loss": 0.83,
+      "step": 513
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 6.201034737620104e-05,
+      "loss": 0.7991,
+      "step": 514
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 6.193643754619365e-05,
+      "loss": 0.8105,
+      "step": 515
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 6.186252771618626e-05,
+      "loss": 0.8116,
+      "step": 516
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 6.178861788617887e-05,
+      "loss": 0.8091,
+      "step": 517
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 6.171470805617147e-05,
+      "loss": 0.8257,
+      "step": 518
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 6.164079822616408e-05,
+      "loss": 0.804,
+      "step": 519
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 6.15668883961567e-05,
+      "loss": 0.8073,
+      "step": 520
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 6.149297856614931e-05,
+      "loss": 0.8012,
+      "step": 521
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 6.14190687361419e-05,
+      "loss": 0.7969,
+      "step": 522
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 6.134515890613451e-05,
+      "loss": 0.8234,
+      "step": 523
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 6.127124907612712e-05,
+      "loss": 0.8076,
+      "step": 524
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 6.119733924611974e-05,
+      "loss": 0.7882,
+      "step": 525
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 6.112342941611235e-05,
+      "loss": 0.7898,
+      "step": 526
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 6.104951958610495e-05,
+      "loss": 0.8098,
+      "step": 527
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 6.097560975609756e-05,
+      "loss": 0.8245,
+      "step": 528
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 6.0901699926090175e-05,
+      "loss": 0.8084,
+      "step": 529
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 6.082779009608278e-05,
+      "loss": 0.7964,
+      "step": 530
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 6.075388026607539e-05,
+      "loss": 0.7949,
+      "step": 531
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 6.0679970436068004e-05,
+      "loss": 0.8092,
+      "step": 532
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 6.060606060606061e-05,
+      "loss": 0.8034,
+      "step": 533
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 6.053215077605322e-05,
+      "loss": 0.8043,
+      "step": 534
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 6.045824094604583e-05,
+      "loss": 0.8164,
+      "step": 535
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 6.038433111603844e-05,
+      "loss": 0.8083,
+      "step": 536
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 6.031042128603105e-05,
+      "loss": 0.7934,
+      "step": 537
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 6.023651145602366e-05,
+      "loss": 0.7929,
+      "step": 538
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 6.016260162601627e-05,
+      "loss": 0.8063,
+      "step": 539
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 6.008869179600887e-05,
+      "loss": 0.7816,
+      "step": 540
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 6.001478196600148e-05,
+      "loss": 0.7975,
+      "step": 541
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 5.9940872135994085e-05,
+      "loss": 0.788,
+      "step": 542
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 5.98669623059867e-05,
+      "loss": 0.776,
+      "step": 543
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 5.9793052475979306e-05,
+      "loss": 0.7878,
+      "step": 544
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 5.9719142645971914e-05,
+      "loss": 0.8016,
+      "step": 545
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 5.964523281596453e-05,
+      "loss": 0.8124,
+      "step": 546
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 5.9571322985957135e-05,
+      "loss": 0.7717,
+      "step": 547
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 5.949741315594974e-05,
+      "loss": 0.8093,
+      "step": 548
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 5.942350332594236e-05,
+      "loss": 0.8056,
+      "step": 549
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 5.9349593495934964e-05,
+      "loss": 0.8196,
+      "step": 550
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 5.927568366592757e-05,
+      "loss": 0.8077,
+      "step": 551
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 5.920177383592018e-05,
+      "loss": 0.7827,
+      "step": 552
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 5.912786400591279e-05,
+      "loss": 0.8,
+      "step": 553
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 5.90539541759054e-05,
+      "loss": 0.7841,
+      "step": 554
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 5.898004434589801e-05,
+      "loss": 0.7795,
+      "step": 555
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 5.890613451589062e-05,
+      "loss": 0.7767,
+      "step": 556
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 5.883222468588323e-05,
+      "loss": 0.8054,
+      "step": 557
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 5.875831485587583e-05,
+      "loss": 0.7823,
+      "step": 558
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 5.868440502586844e-05,
+      "loss": 0.8048,
+      "step": 559
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 5.8610495195861045e-05,
+      "loss": 0.7906,
+      "step": 560
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 5.853658536585366e-05,
+      "loss": 0.7659,
+      "step": 561
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 5.8462675535846266e-05,
+      "loss": 0.7845,
+      "step": 562
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 5.8388765705838874e-05,
+      "loss": 0.7897,
+      "step": 563
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 5.831485587583149e-05,
+      "loss": 0.7927,
+      "step": 564
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 5.8240946045824095e-05,
+      "loss": 0.7902,
+      "step": 565
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 5.81670362158167e-05,
+      "loss": 0.7968,
+      "step": 566
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 5.809312638580932e-05,
+      "loss": 0.7839,
+      "step": 567
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 5.8019216555801924e-05,
+      "loss": 0.8069,
+      "step": 568
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 5.794530672579453e-05,
+      "loss": 0.8002,
+      "step": 569
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 5.7871396895787146e-05,
+      "loss": 0.7777,
+      "step": 570
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 5.779748706577975e-05,
+      "loss": 0.7824,
+      "step": 571
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 5.772357723577236e-05,
+      "loss": 0.7936,
+      "step": 572
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 5.7649667405764975e-05,
+      "loss": 0.7865,
+      "step": 573
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 5.757575757575758e-05,
+      "loss": 0.7916,
+      "step": 574
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 5.750184774575019e-05,
+      "loss": 0.7948,
+      "step": 575
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5.742793791574279e-05,
+      "loss": 0.7801,
+      "step": 576
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5.73540280857354e-05,
+      "loss": 0.816,
+      "step": 577
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5.728011825572801e-05,
+      "loss": 0.7687,
+      "step": 578
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5.720620842572062e-05,
+      "loss": 0.7912,
+      "step": 579
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.7132298595713227e-05,
+      "loss": 0.7725,
+      "step": 580
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.705838876570584e-05,
+      "loss": 0.7681,
+      "step": 581
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.698447893569845e-05,
+      "loss": 0.7738,
+      "step": 582
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.6910569105691056e-05,
+      "loss": 0.7931,
+      "step": 583
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.683665927568367e-05,
+      "loss": 0.7802,
+      "step": 584
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.676274944567628e-05,
+      "loss": 0.7733,
+      "step": 585
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.6688839615668884e-05,
+      "loss": 0.8044,
+      "step": 586
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.66149297856615e-05,
+      "loss": 0.7887,
+      "step": 587
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.6541019955654106e-05,
+      "loss": 0.7773,
+      "step": 588
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.6467110125646713e-05,
+      "loss": 0.7839,
+      "step": 589
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.639320029563933e-05,
+      "loss": 0.7642,
+      "step": 590
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.6319290465631935e-05,
+      "loss": 0.7783,
+      "step": 591
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.624538063562454e-05,
+      "loss": 0.7861,
+      "step": 592
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.6171470805617157e-05,
+      "loss": 0.7671,
+      "step": 593
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.6097560975609764e-05,
+      "loss": 0.7749,
+      "step": 594
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.6023651145602365e-05,
+      "loss": 0.7714,
+      "step": 595
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.594974131559497e-05,
+      "loss": 0.7685,
+      "step": 596
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.587583148558758e-05,
+      "loss": 0.7936,
+      "step": 597
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.5801921655580193e-05,
+      "loss": 0.7746,
+      "step": 598
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.57280118255728e-05,
+      "loss": 0.7715,
+      "step": 599
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.565410199556541e-05,
+      "loss": 0.7538,
+      "step": 600
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.558019216555802e-05,
+      "loss": 0.7734,
+      "step": 601
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.550628233555063e-05,
+      "loss": 0.7725,
+      "step": 602
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.543237250554324e-05,
+      "loss": 0.7751,
+      "step": 603
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.535846267553585e-05,
+      "loss": 0.7659,
+      "step": 604
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.528455284552846e-05,
+      "loss": 0.7691,
+      "step": 605
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.5210643015521066e-05,
+      "loss": 0.7951,
+      "step": 606
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.513673318551368e-05,
+      "loss": 0.765,
+      "step": 607
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.506282335550629e-05,
+      "loss": 0.8091,
+      "step": 608
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.4988913525498895e-05,
+      "loss": 0.7571,
+      "step": 609
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.491500369549151e-05,
+      "loss": 0.7798,
+      "step": 610
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.484109386548412e-05,
+      "loss": 0.7893,
+      "step": 611
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 5.4767184035476724e-05,
+      "loss": 0.7676,
+      "step": 612
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 5.4693274205469325e-05,
+      "loss": 0.7782,
+      "step": 613
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 5.461936437546193e-05,
+      "loss": 0.8033,
+      "step": 614
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 5.4545454545454546e-05,
+      "loss": 0.7911,
+      "step": 615
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 5.4471544715447154e-05,
+      "loss": 0.7983,
+      "step": 616
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 5.439763488543976e-05,
+      "loss": 0.7728,
+      "step": 617
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 5.4323725055432375e-05,
+      "loss": 0.7699,
+      "step": 618
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 5.424981522542498e-05,
+      "loss": 0.781,
+      "step": 619
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 5.417590539541759e-05,
+      "loss": 0.7795,
+      "step": 620
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 5.4101995565410204e-05,
+      "loss": 0.7683,
+      "step": 621
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 5.402808573540281e-05,
+      "loss": 0.7791,
+      "step": 622
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 5.395417590539542e-05,
+      "loss": 0.7517,
+      "step": 623
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 5.388026607538803e-05,
+      "loss": 0.7678,
+      "step": 624
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 5.380635624538064e-05,
+      "loss": 0.7713,
+      "step": 625
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 5.373244641537325e-05,
+      "loss": 0.7628,
+      "step": 626
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 5.365853658536586e-05,
+      "loss": 0.7781,
+      "step": 627
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 5.358462675535847e-05,
+      "loss": 0.7694,
+      "step": 628
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 5.351071692535108e-05,
+      "loss": 0.7674,
+      "step": 629
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 5.343680709534369e-05,
+      "loss": 0.7817,
+      "step": 630
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 5.3362897265336285e-05,
+      "loss": 0.7805,
+      "step": 631
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 5.32889874353289e-05,
+      "loss": 0.8098,
+      "step": 632
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 5.3215077605321506e-05,
+      "loss": 0.7793,
+      "step": 633
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.3141167775314114e-05,
+      "loss": 0.7673,
+      "step": 634
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.306725794530673e-05,
+      "loss": 0.7774,
+      "step": 635
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.2993348115299335e-05,
+      "loss": 0.7772,
+      "step": 636
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.291943828529194e-05,
+      "loss": 0.7819,
+      "step": 637
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.284552845528456e-05,
+      "loss": 0.7578,
+      "step": 638
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5.2771618625277164e-05,
+      "loss": 0.7764,
+      "step": 639
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5.269770879526977e-05,
+      "loss": 0.7864,
+      "step": 640
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5.2623798965262386e-05,
+      "loss": 0.7743,
+      "step": 641
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5.254988913525499e-05,
+      "loss": 0.7832,
+      "step": 642
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.24759793052476e-05,
+      "loss": 0.798,
+      "step": 643
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.2402069475240215e-05,
+      "loss": 0.7582,
+      "step": 644
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.232815964523282e-05,
+      "loss": 0.7563,
+      "step": 645
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.225424981522543e-05,
+      "loss": 0.7762,
+      "step": 646
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.2180339985218044e-05,
+      "loss": 0.7704,
+      "step": 647
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.210643015521065e-05,
+      "loss": 0.7543,
+      "step": 648
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.203252032520326e-05,
+      "loss": 0.7913,
+      "step": 649
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.195861049519586e-05,
+      "loss": 0.7785,
+      "step": 650
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.1884700665188466e-05,
+      "loss": 0.7571,
+      "step": 651
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.181079083518108e-05,
+      "loss": 0.7639,
+      "step": 652
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.173688100517369e-05,
+      "loss": 0.7585,
+      "step": 653
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.1662971175166295e-05,
+      "loss": 0.7729,
+      "step": 654
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.158906134515891e-05,
+      "loss": 0.7575,
+      "step": 655
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.151515151515152e-05,
+      "loss": 0.7638,
+      "step": 656
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.1441241685144124e-05,
+      "loss": 0.7424,
+      "step": 657
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.136733185513674e-05,
+      "loss": 0.7697,
+      "step": 658
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.1293422025129346e-05,
+      "loss": 0.7543,
+      "step": 659
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.121951219512195e-05,
+      "loss": 0.7695,
+      "step": 660
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.114560236511457e-05,
+      "loss": 0.7577,
+      "step": 661
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.1071692535107175e-05,
+      "loss": 0.7481,
+      "step": 662
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.099778270509978e-05,
+      "loss": 0.7727,
+      "step": 663
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.0923872875092396e-05,
+      "loss": 0.7656,
+      "step": 664
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.0849963045085004e-05,
+      "loss": 0.7512,
+      "step": 665
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.077605321507761e-05,
+      "loss": 0.7747,
+      "step": 666
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.0702143385070225e-05,
+      "loss": 0.7426,
+      "step": 667
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.062823355506282e-05,
+      "loss": 0.7453,
+      "step": 668
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.0554323725055433e-05,
+      "loss": 0.768,
+      "step": 669
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.048041389504804e-05,
+      "loss": 0.7675,
+      "step": 670
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.040650406504065e-05,
+      "loss": 0.744,
+      "step": 671
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.033259423503326e-05,
+      "loss": 0.7919,
+      "step": 672
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.025868440502587e-05,
+      "loss": 0.775,
+      "step": 673
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.018477457501848e-05,
+      "loss": 0.7365,
+      "step": 674
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.011086474501109e-05,
+      "loss": 0.7503,
+      "step": 675
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.00369549150037e-05,
+      "loss": 0.7614,
+      "step": 676
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.9963045084996306e-05,
+      "loss": 0.7528,
+      "step": 677
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.9889135254988913e-05,
+      "loss": 0.7555,
+      "step": 678
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.981522542498153e-05,
+      "loss": 0.7599,
+      "step": 679
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.9741315594974135e-05,
+      "loss": 0.7662,
+      "step": 680
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.966740576496674e-05,
+      "loss": 0.7538,
+      "step": 681
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.959349593495935e-05,
+      "loss": 0.7613,
+      "step": 682
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.951958610495196e-05,
+      "loss": 0.7519,
+      "step": 683
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.944567627494457e-05,
+      "loss": 0.7604,
+      "step": 684
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.937176644493718e-05,
+      "loss": 0.7641,
+      "step": 685
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.9297856614929786e-05,
+      "loss": 0.7631,
+      "step": 686
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.92239467849224e-05,
+      "loss": 0.7621,
+      "step": 687
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.915003695491501e-05,
+      "loss": 0.7388,
+      "step": 688
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.9076127124907615e-05,
+      "loss": 0.7679,
+      "step": 689
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.900221729490023e-05,
+      "loss": 0.767,
+      "step": 690
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.892830746489283e-05,
+      "loss": 0.7847,
+      "step": 691
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.885439763488544e-05,
+      "loss": 0.764,
+      "step": 692
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.878048780487805e-05,
+      "loss": 0.7496,
+      "step": 693
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.870657797487066e-05,
+      "loss": 0.7661,
+      "step": 694
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.8632668144863266e-05,
+      "loss": 0.7493,
+      "step": 695
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.855875831485588e-05,
+      "loss": 0.7536,
+      "step": 696
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.848484848484849e-05,
+      "loss": 0.766,
+      "step": 697
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.8410938654841095e-05,
+      "loss": 0.7424,
+      "step": 698
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.833702882483371e-05,
+      "loss": 0.7463,
+      "step": 699
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.826311899482631e-05,
+      "loss": 0.7571,
+      "step": 700
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.8189209164818924e-05,
+      "loss": 0.766,
+      "step": 701
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.811529933481153e-05,
+      "loss": 0.7507,
+      "step": 702
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.804138950480414e-05,
+      "loss": 0.7552,
+      "step": 703
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.796747967479675e-05,
+      "loss": 0.7316,
+      "step": 704
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.789356984478936e-05,
+      "loss": 0.7738,
+      "step": 705
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.781966001478197e-05,
+      "loss": 0.7618,
+      "step": 706
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.774575018477458e-05,
+      "loss": 0.767,
+      "step": 707
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.767184035476719e-05,
+      "loss": 0.7355,
+      "step": 708
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.759793052475979e-05,
+      "loss": 0.7526,
+      "step": 709
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.7524020694752404e-05,
+      "loss": 0.764,
+      "step": 710
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.745011086474501e-05,
+      "loss": 0.7423,
+      "step": 711
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.737620103473762e-05,
+      "loss": 0.7532,
+      "step": 712
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.730229120473023e-05,
+      "loss": 0.7528,
+      "step": 713
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.722838137472284e-05,
+      "loss": 0.7486,
+      "step": 714
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.715447154471545e-05,
+      "loss": 0.7676,
+      "step": 715
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.708056171470806e-05,
+      "loss": 0.7437,
+      "step": 716
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.700665188470067e-05,
+      "loss": 0.7566,
+      "step": 717
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.693274205469328e-05,
+      "loss": 0.7216,
+      "step": 718
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.6858832224685884e-05,
+      "loss": 0.7254,
+      "step": 719
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.678492239467849e-05,
+      "loss": 0.738,
+      "step": 720
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.6711012564671106e-05,
+      "loss": 0.7499,
+      "step": 721
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.663710273466371e-05,
+      "loss": 0.7488,
+      "step": 722
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.656319290465632e-05,
+      "loss": 0.7425,
+      "step": 723
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.6489283074648935e-05,
+      "loss": 0.7574,
+      "step": 724
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.641537324464154e-05,
+      "loss": 0.7394,
+      "step": 725
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.634146341463415e-05,
+      "loss": 0.742,
+      "step": 726
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.626755358462676e-05,
+      "loss": 0.7651,
+      "step": 727
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.6193643754619364e-05,
+      "loss": 0.7553,
+      "step": 728
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.611973392461197e-05,
+      "loss": 0.7625,
+      "step": 729
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.6045824094604586e-05,
+      "loss": 0.7604,
+      "step": 730
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.597191426459719e-05,
+      "loss": 0.7457,
+      "step": 731
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.58980044345898e-05,
+      "loss": 0.7268,
+      "step": 732
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.5824094604582415e-05,
+      "loss": 0.7469,
+      "step": 733
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.575018477457502e-05,
+      "loss": 0.7533,
+      "step": 734
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.567627494456763e-05,
+      "loss": 0.7617,
+      "step": 735
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.560236511456024e-05,
+      "loss": 0.7178,
+      "step": 736
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.5528455284552844e-05,
+      "loss": 0.7467,
+      "step": 737
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.545454545454546e-05,
+      "loss": 0.7467,
+      "step": 738
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.5380635624538066e-05,
+      "loss": 0.7378,
+      "step": 739
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.530672579453067e-05,
+      "loss": 0.7443,
+      "step": 740
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.523281596452328e-05,
+      "loss": 0.7225,
+      "step": 741
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.5158906134515895e-05,
+      "loss": 0.7489,
+      "step": 742
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.50849963045085e-05,
+      "loss": 0.7266,
+      "step": 743
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.501108647450111e-05,
+      "loss": 0.7338,
+      "step": 744
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.4937176644493724e-05,
+      "loss": 0.7421,
+      "step": 745
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.4863266814486324e-05,
+      "loss": 0.7834,
+      "step": 746
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.478935698447894e-05,
+      "loss": 0.7429,
+      "step": 747
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.4715447154471546e-05,
+      "loss": 0.7463,
+      "step": 748
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.4641537324464153e-05,
+      "loss": 0.756,
+      "step": 749
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.456762749445677e-05,
+      "loss": 0.7413,
+      "step": 750
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.4493717664449375e-05,
+      "loss": 0.741,
+      "step": 751
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.441980783444198e-05,
+      "loss": 0.732,
+      "step": 752
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.4345898004434597e-05,
+      "loss": 0.7492,
+      "step": 753
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.4271988174427204e-05,
+      "loss": 0.732,
+      "step": 754
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.4198078344419805e-05,
+      "loss": 0.7372,
+      "step": 755
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.412416851441242e-05,
+      "loss": 0.7509,
+      "step": 756
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.4050258684405026e-05,
+      "loss": 0.7464,
+      "step": 757
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.3976348854397633e-05,
+      "loss": 0.7616,
+      "step": 758
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.390243902439025e-05,
+      "loss": 0.7608,
+      "step": 759
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.3828529194382855e-05,
+      "loss": 0.7267,
+      "step": 760
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.375461936437546e-05,
+      "loss": 0.7436,
+      "step": 761
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.3680709534368077e-05,
+      "loss": 0.7399,
+      "step": 762
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.3606799704360684e-05,
+      "loss": 0.7488,
+      "step": 763
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.353288987435329e-05,
+      "loss": 0.7485,
+      "step": 764
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.34589800443459e-05,
+      "loss": 0.7381,
+      "step": 765
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.3385070214338506e-05,
+      "loss": 0.7301,
+      "step": 766
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.331116038433112e-05,
+      "loss": 0.7208,
+      "step": 767
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.323725055432373e-05,
+      "loss": 0.7487,
+      "step": 768
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.3163340724316335e-05,
+      "loss": 0.7508,
+      "step": 769
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.308943089430895e-05,
+      "loss": 0.7316,
+      "step": 770
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.301552106430156e-05,
+      "loss": 0.7223,
+      "step": 771
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.2941611234294164e-05,
+      "loss": 0.7169,
+      "step": 772
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.286770140428677e-05,
+      "loss": 0.7277,
+      "step": 773
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.279379157427938e-05,
+      "loss": 0.7399,
+      "step": 774
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.2719881744271986e-05,
+      "loss": 0.7319,
+      "step": 775
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.26459719142646e-05,
+      "loss": 0.7416,
+      "step": 776
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.257206208425721e-05,
+      "loss": 0.7469,
+      "step": 777
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.2498152254249815e-05,
+      "loss": 0.7492,
+      "step": 778
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.242424242424243e-05,
+      "loss": 0.7424,
+      "step": 779
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.235033259423504e-05,
+      "loss": 0.7508,
+      "step": 780
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.2276422764227644e-05,
+      "loss": 0.7464,
+      "step": 781
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.220251293422025e-05,
+      "loss": 0.7392,
+      "step": 782
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.212860310421286e-05,
+      "loss": 0.728,
+      "step": 783
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.205469327420547e-05,
+      "loss": 0.7462,
+      "step": 784
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.198078344419808e-05,
+      "loss": 0.7317,
+      "step": 785
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.190687361419069e-05,
+      "loss": 0.7147,
+      "step": 786
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.18329637841833e-05,
+      "loss": 0.7281,
+      "step": 787
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.175905395417591e-05,
+      "loss": 0.7549,
+      "step": 788
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.168514412416852e-05,
+      "loss": 0.7376,
+      "step": 789
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.161123429416113e-05,
+      "loss": 0.732,
+      "step": 790
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.153732446415373e-05,
+      "loss": 0.7325,
+      "step": 791
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.146341463414634e-05,
+      "loss": 0.7463,
+      "step": 792
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.138950480413895e-05,
+      "loss": 0.7079,
+      "step": 793
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.131559497413156e-05,
+      "loss": 0.7501,
+      "step": 794
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.124168514412417e-05,
+      "loss": 0.7268,
+      "step": 795
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.116777531411678e-05,
+      "loss": 0.7256,
+      "step": 796
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.109386548410939e-05,
+      "loss": 0.7266,
+      "step": 797
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.1019955654102e-05,
+      "loss": 0.7419,
+      "step": 798
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.094604582409461e-05,
+      "loss": 0.7396,
+      "step": 799
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.087213599408721e-05,
+      "loss": 0.7292,
+      "step": 800
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.0798226164079826e-05,
+      "loss": 0.7492,
+      "step": 801
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.072431633407243e-05,
+      "loss": 0.763,
+      "step": 802
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.065040650406504e-05,
+      "loss": 0.7275,
+      "step": 803
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.057649667405765e-05,
+      "loss": 0.7356,
+      "step": 804
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.050258684405026e-05,
+      "loss": 0.7444,
+      "step": 805
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.042867701404287e-05,
+      "loss": 0.7407,
+      "step": 806
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.035476718403548e-05,
+      "loss": 0.7187,
+      "step": 807
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.028085735402809e-05,
+      "loss": 0.733,
+      "step": 808
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.02069475240207e-05,
+      "loss": 0.748,
+      "step": 809
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.0133037694013306e-05,
+      "loss": 0.7165,
+      "step": 810
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.005912786400591e-05,
+      "loss": 0.7189,
+      "step": 811
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.998521803399852e-05,
+      "loss": 0.7429,
+      "step": 812
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.9911308203991135e-05,
+      "loss": 0.7397,
+      "step": 813
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.983739837398374e-05,
+      "loss": 0.73,
+      "step": 814
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.976348854397635e-05,
+      "loss": 0.7173,
+      "step": 815
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.9689578713968964e-05,
+      "loss": 0.7395,
+      "step": 816
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.961566888396157e-05,
+      "loss": 0.7353,
+      "step": 817
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.954175905395418e-05,
+      "loss": 0.7266,
+      "step": 818
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.9467849223946786e-05,
+      "loss": 0.7284,
+      "step": 819
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.939393939393939e-05,
+      "loss": 0.7332,
+      "step": 820
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.9320029563932e-05,
+      "loss": 0.7368,
+      "step": 821
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.9246119733924615e-05,
+      "loss": 0.7192,
+      "step": 822
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.917220990391722e-05,
+      "loss": 0.7302,
+      "step": 823
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.909830007390983e-05,
+      "loss": 0.7396,
+      "step": 824
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.9024390243902444e-05,
+      "loss": 0.7018,
+      "step": 825
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.895048041389505e-05,
+      "loss": 0.7351,
+      "step": 826
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.887657058388766e-05,
+      "loss": 0.7156,
+      "step": 827
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.8802660753880266e-05,
+      "loss": 0.7216,
+      "step": 828
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.8728750923872873e-05,
+      "loss": 0.7245,
+      "step": 829
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.865484109386549e-05,
+      "loss": 0.7329,
+      "step": 830
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.8580931263858095e-05,
+      "loss": 0.749,
+      "step": 831
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.85070214338507e-05,
+      "loss": 0.728,
+      "step": 832
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.8433111603843317e-05,
+      "loss": 0.7243,
+      "step": 833
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.8359201773835924e-05,
+      "loss": 0.7189,
+      "step": 834
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.828529194382853e-05,
+      "loss": 0.7448,
+      "step": 835
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.8211382113821145e-05,
+      "loss": 0.7482,
+      "step": 836
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.8137472283813746e-05,
+      "loss": 0.7251,
+      "step": 837
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.8063562453806354e-05,
+      "loss": 0.725,
+      "step": 838
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.798965262379897e-05,
+      "loss": 0.7204,
+      "step": 839
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.7915742793791575e-05,
+      "loss": 0.7288,
+      "step": 840
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.784183296378418e-05,
+      "loss": 0.7148,
+      "step": 841
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.77679231337768e-05,
+      "loss": 0.712,
+      "step": 842
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.7694013303769404e-05,
+      "loss": 0.7438,
+      "step": 843
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.762010347376201e-05,
+      "loss": 0.7395,
+      "step": 844
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.7546193643754626e-05,
+      "loss": 0.724,
+      "step": 845
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.7472283813747226e-05,
+      "loss": 0.7187,
+      "step": 846
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.739837398373984e-05,
+      "loss": 0.7038,
+      "step": 847
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.732446415373245e-05,
+      "loss": 0.7227,
+      "step": 848
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.7250554323725055e-05,
+      "loss": 0.7374,
+      "step": 849
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.717664449371767e-05,
+      "loss": 0.7137,
+      "step": 850
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.710273466371028e-05,
+      "loss": 0.7398,
+      "step": 851
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.7028824833702884e-05,
+      "loss": 0.7397,
+      "step": 852
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.69549150036955e-05,
+      "loss": 0.7227,
+      "step": 853
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.6881005173688106e-05,
+      "loss": 0.7066,
+      "step": 854
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.6807095343680706e-05,
+      "loss": 0.716,
+      "step": 855
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.673318551367332e-05,
+      "loss": 0.7131,
+      "step": 856
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.665927568366593e-05,
+      "loss": 0.7509,
+      "step": 857
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.6585365853658535e-05,
+      "loss": 0.7182,
+      "step": 858
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.651145602365115e-05,
+      "loss": 0.718,
+      "step": 859
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.643754619364376e-05,
+      "loss": 0.7071,
+      "step": 860
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.6363636363636364e-05,
+      "loss": 0.7157,
+      "step": 861
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.628972653362898e-05,
+      "loss": 0.7303,
+      "step": 862
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.6215816703621586e-05,
+      "loss": 0.7242,
+      "step": 863
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.6141906873614186e-05,
+      "loss": 0.7404,
+      "step": 864
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.60679970436068e-05,
+      "loss": 0.7306,
+      "step": 865
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.599408721359941e-05,
+      "loss": 0.712,
+      "step": 866
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.5920177383592015e-05,
+      "loss": 0.7281,
+      "step": 867
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.584626755358463e-05,
+      "loss": 0.7122,
+      "step": 868
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.577235772357724e-05,
+      "loss": 0.7151,
+      "step": 869
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.5698447893569844e-05,
+      "loss": 0.7353,
+      "step": 870
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.562453806356246e-05,
+      "loss": 0.7313,
+      "step": 871
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.5550628233555066e-05,
+      "loss": 0.7078,
+      "step": 872
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.547671840354767e-05,
+      "loss": 0.7207,
+      "step": 873
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.540280857354028e-05,
+      "loss": 0.7238,
+      "step": 874
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.532889874353289e-05,
+      "loss": 0.7269,
+      "step": 875
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.52549889135255e-05,
+      "loss": 0.6943,
+      "step": 876
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.518107908351811e-05,
+      "loss": 0.6894,
+      "step": 877
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.510716925351072e-05,
+      "loss": 0.7088,
+      "step": 878
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.503325942350333e-05,
+      "loss": 0.7372,
+      "step": 879
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.495934959349594e-05,
+      "loss": 0.7253,
+      "step": 880
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.4885439763488546e-05,
+      "loss": 0.7108,
+      "step": 881
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.481152993348116e-05,
+      "loss": 0.7275,
+      "step": 882
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.473762010347376e-05,
+      "loss": 0.7266,
+      "step": 883
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.466371027346637e-05,
+      "loss": 0.7009,
+      "step": 884
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.458980044345898e-05,
+      "loss": 0.7173,
+      "step": 885
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.451589061345159e-05,
+      "loss": 0.724,
+      "step": 886
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.44419807834442e-05,
+      "loss": 0.7069,
+      "step": 887
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.436807095343681e-05,
+      "loss": 0.7026,
+      "step": 888
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.429416112342942e-05,
+      "loss": 0.7246,
+      "step": 889
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.4220251293422026e-05,
+      "loss": 0.7173,
+      "step": 890
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.414634146341464e-05,
+      "loss": 0.7221,
+      "step": 891
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.407243163340724e-05,
+      "loss": 0.7281,
+      "step": 892
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.3998521803399855e-05,
+      "loss": 0.7093,
+      "step": 893
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.392461197339246e-05,
+      "loss": 0.7146,
+      "step": 894
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.385070214338507e-05,
+      "loss": 0.7276,
+      "step": 895
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.3776792313377684e-05,
+      "loss": 0.7234,
+      "step": 896
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.370288248337029e-05,
+      "loss": 0.7139,
+      "step": 897
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.36289726533629e-05,
+      "loss": 0.7339,
+      "step": 898
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.355506282335551e-05,
+      "loss": 0.7225,
+      "step": 899
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.348115299334812e-05,
+      "loss": 0.7219,
+      "step": 900
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.340724316334072e-05,
+      "loss": 0.7195,
+      "step": 901
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.3333333333333335e-05,
+      "loss": 0.722,
+      "step": 902
+    },
+    {
+      "epoch": 2.0,
+      "eval_loss": 0.9731828570365906,
+      "eval_runtime": 496.0356,
+      "eval_samples_per_second": 32.135,
+      "eval_steps_per_second": 0.504,
+      "eval_wer": 0.759285117848898,
+      "step": 902
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.325942350332594e-05,
+      "loss": 1.0519,
+      "step": 903
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.318551367331855e-05,
+      "loss": 0.7017,
+      "step": 904
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.3111603843311164e-05,
+      "loss": 0.6736,
+      "step": 905
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.303769401330377e-05,
+      "loss": 0.7095,
+      "step": 906
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.296378418329638e-05,
+      "loss": 0.708,
+      "step": 907
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.288987435328899e-05,
+      "loss": 0.6951,
+      "step": 908
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.28159645232816e-05,
+      "loss": 0.6904,
+      "step": 909
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.274205469327421e-05,
+      "loss": 0.7247,
+      "step": 910
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.2668144863266815e-05,
+      "loss": 0.6902,
+      "step": 911
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.259423503325942e-05,
+      "loss": 0.6871,
+      "step": 912
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.2520325203252037e-05,
+      "loss": 0.6836,
+      "step": 913
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.2446415373244644e-05,
+      "loss": 0.7123,
+      "step": 914
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.237250554323725e-05,
+      "loss": 0.6875,
+      "step": 915
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.2298595713229866e-05,
+      "loss": 0.6848,
+      "step": 916
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.222468588322247e-05,
+      "loss": 0.7001,
+      "step": 917
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.215077605321508e-05,
+      "loss": 0.7013,
+      "step": 918
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.207686622320769e-05,
+      "loss": 0.687,
+      "step": 919
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.2002956393200295e-05,
+      "loss": 0.7157,
+      "step": 920
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.19290465631929e-05,
+      "loss": 0.7039,
+      "step": 921
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.185513673318552e-05,
+      "loss": 0.6905,
+      "step": 922
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.1781226903178124e-05,
+      "loss": 0.7005,
+      "step": 923
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.170731707317073e-05,
+      "loss": 0.6968,
+      "step": 924
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.1633407243163346e-05,
+      "loss": 0.7053,
+      "step": 925
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.155949741315595e-05,
+      "loss": 0.7216,
+      "step": 926
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.148558758314856e-05,
+      "loss": 0.7071,
+      "step": 927
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.141167775314117e-05,
+      "loss": 0.6974,
+      "step": 928
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.1337767923133775e-05,
+      "loss": 0.7182,
+      "step": 929
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.126385809312638e-05,
+      "loss": 0.697,
+      "step": 930
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.1189948263119e-05,
+      "loss": 0.6993,
+      "step": 931
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.1116038433111604e-05,
+      "loss": 0.7236,
+      "step": 932
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.104212860310421e-05,
+      "loss": 0.7066,
+      "step": 933
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.0968218773096826e-05,
+      "loss": 0.6842,
+      "step": 934
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.089430894308943e-05,
+      "loss": 0.7084,
+      "step": 935
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.082039911308204e-05,
+      "loss": 0.6936,
+      "step": 936
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.0746489283074655e-05,
+      "loss": 0.6903,
+      "step": 937
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.0672579453067255e-05,
+      "loss": 0.6785,
+      "step": 938
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.059866962305987e-05,
+      "loss": 0.6824,
+      "step": 939
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.052475979305248e-05,
+      "loss": 0.703,
+      "step": 940
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.0450849963045088e-05,
+      "loss": 0.716,
+      "step": 941
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.0376940133037695e-05,
+      "loss": 0.7118,
+      "step": 942
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.0303030303030306e-05,
+      "loss": 0.6825,
+      "step": 943
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.0229120473022916e-05,
+      "loss": 0.679,
+      "step": 944
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.0155210643015524e-05,
+      "loss": 0.6908,
+      "step": 945
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.0081300813008135e-05,
+      "loss": 0.6697,
+      "step": 946
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.000739098300074e-05,
+      "loss": 0.6667,
+      "step": 947
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.993348115299335e-05,
+      "loss": 0.6857,
+      "step": 948
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.9859571322985957e-05,
+      "loss": 0.6816,
+      "step": 949
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.9785661492978568e-05,
+      "loss": 0.7112,
+      "step": 950
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.971175166297118e-05,
+      "loss": 0.6764,
+      "step": 951
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.9637841832963786e-05,
+      "loss": 0.6962,
+      "step": 952
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.9563932002956397e-05,
+      "loss": 0.6609,
+      "step": 953
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.9490022172949004e-05,
+      "loss": 0.6981,
+      "step": 954
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.9416112342941615e-05,
+      "loss": 0.6815,
+      "step": 955
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.934220251293422e-05,
+      "loss": 0.7157,
+      "step": 956
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.926829268292683e-05,
+      "loss": 0.6793,
+      "step": 957
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.9194382852919437e-05,
+      "loss": 0.7055,
+      "step": 958
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.9120473022912048e-05,
+      "loss": 0.7082,
+      "step": 959
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.904656319290466e-05,
+      "loss": 0.7065,
+      "step": 960
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.8972653362897266e-05,
+      "loss": 0.6811,
+      "step": 961
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.8898743532889877e-05,
+      "loss": 0.7102,
+      "step": 962
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.8824833702882487e-05,
+      "loss": 0.702,
+      "step": 963
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.8750923872875095e-05,
+      "loss": 0.6983,
+      "step": 964
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.86770140428677e-05,
+      "loss": 0.7045,
+      "step": 965
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.860310421286031e-05,
+      "loss": 0.6894,
+      "step": 966
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.852919438285292e-05,
+      "loss": 0.6839,
+      "step": 967
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.8455284552845528e-05,
+      "loss": 0.6913,
+      "step": 968
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.838137472283814e-05,
+      "loss": 0.6855,
+      "step": 969
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.830746489283075e-05,
+      "loss": 0.6843,
+      "step": 970
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.8233555062823357e-05,
+      "loss": 0.7058,
+      "step": 971
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.8159645232815967e-05,
+      "loss": 0.6956,
+      "step": 972
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.8085735402808578e-05,
+      "loss": 0.6816,
+      "step": 973
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.8011825572801182e-05,
+      "loss": 0.6936,
+      "step": 974
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.793791574279379e-05,
+      "loss": 0.6835,
+      "step": 975
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.78640059127864e-05,
+      "loss": 0.6878,
+      "step": 976
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.779009608277901e-05,
+      "loss": 0.6746,
+      "step": 977
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.771618625277162e-05,
+      "loss": 0.6847,
+      "step": 978
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.764227642276423e-05,
+      "loss": 0.6737,
+      "step": 979
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.756836659275684e-05,
+      "loss": 0.671,
+      "step": 980
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.7494456762749448e-05,
+      "loss": 0.6814,
+      "step": 981
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.742054693274206e-05,
+      "loss": 0.6643,
+      "step": 982
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.7346637102734662e-05,
+      "loss": 0.6995,
+      "step": 983
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.7272727272727273e-05,
+      "loss": 0.6728,
+      "step": 984
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.719881744271988e-05,
+      "loss": 0.6869,
+      "step": 985
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.712490761271249e-05,
+      "loss": 0.6819,
+      "step": 986
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.7050997782705102e-05,
+      "loss": 0.7077,
+      "step": 987
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.697708795269771e-05,
+      "loss": 0.6763,
+      "step": 988
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.690317812269032e-05,
+      "loss": 0.6853,
+      "step": 989
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.682926829268293e-05,
+      "loss": 0.6945,
+      "step": 990
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.675535846267554e-05,
+      "loss": 0.6916,
+      "step": 991
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.6681448632668142e-05,
+      "loss": 0.6738,
+      "step": 992
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.6607538802660753e-05,
+      "loss": 0.6954,
+      "step": 993
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.6533628972653364e-05,
+      "loss": 0.6868,
+      "step": 994
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.645971914264597e-05,
+      "loss": 0.7037,
+      "step": 995
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.6385809312638582e-05,
+      "loss": 0.7072,
+      "step": 996
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.6311899482631193e-05,
+      "loss": 0.6908,
+      "step": 997
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.62379896526238e-05,
+      "loss": 0.696,
+      "step": 998
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.616407982261641e-05,
+      "loss": 0.7197,
+      "step": 999
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.6090169992609022e-05,
+      "loss": 0.6981,
+      "step": 1000
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.601626016260163e-05,
+      "loss": 0.694,
+      "step": 1001
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.5942350332594233e-05,
+      "loss": 0.683,
+      "step": 1002
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.5868440502586844e-05,
+      "loss": 0.6763,
+      "step": 1003
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.5794530672579455e-05,
+      "loss": 0.6886,
+      "step": 1004
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.5720620842572062e-05,
+      "loss": 0.6846,
+      "step": 1005
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.5646711012564673e-05,
+      "loss": 0.6944,
+      "step": 1006
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.5572801182557284e-05,
+      "loss": 0.6795,
+      "step": 1007
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.549889135254989e-05,
+      "loss": 0.6966,
+      "step": 1008
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.5424981522542502e-05,
+      "loss": 0.6799,
+      "step": 1009
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.5351071692535113e-05,
+      "loss": 0.691,
+      "step": 1010
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.5277161862527717e-05,
+      "loss": 0.6688,
+      "step": 1011
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.5203252032520324e-05,
+      "loss": 0.678,
+      "step": 1012
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.5129342202512935e-05,
+      "loss": 0.6805,
+      "step": 1013
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.5055432372505546e-05,
+      "loss": 0.7044,
+      "step": 1014
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.4981522542498153e-05,
+      "loss": 0.6733,
+      "step": 1015
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.4907612712490764e-05,
+      "loss": 0.6841,
+      "step": 1016
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.483370288248337e-05,
+      "loss": 0.6786,
+      "step": 1017
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.475979305247598e-05,
+      "loss": 0.708,
+      "step": 1018
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.468588322246859e-05,
+      "loss": 0.6829,
+      "step": 1019
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.46119733924612e-05,
+      "loss": 0.6921,
+      "step": 1020
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.4538063562453808e-05,
+      "loss": 0.6985,
+      "step": 1021
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.4464153732446415e-05,
+      "loss": 0.6591,
+      "step": 1022
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.4390243902439026e-05,
+      "loss": 0.6993,
+      "step": 1023
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.4316334072431633e-05,
+      "loss": 0.6744,
+      "step": 1024
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.4242424242424244e-05,
+      "loss": 0.6983,
+      "step": 1025
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.4168514412416855e-05,
+      "loss": 0.6673,
+      "step": 1026
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.4094604582409462e-05,
+      "loss": 0.7011,
+      "step": 1027
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.402069475240207e-05,
+      "loss": 0.664,
+      "step": 1028
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.394678492239468e-05,
+      "loss": 0.7047,
+      "step": 1029
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.387287509238729e-05,
+      "loss": 0.6832,
+      "step": 1030
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.3798965262379895e-05,
+      "loss": 0.6799,
+      "step": 1031
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.3725055432372506e-05,
+      "loss": 0.682,
+      "step": 1032
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.3651145602365117e-05,
+      "loss": 0.6876,
+      "step": 1033
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.3577235772357724e-05,
+      "loss": 0.693,
+      "step": 1034
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.3503325942350335e-05,
+      "loss": 0.6904,
+      "step": 1035
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.3429416112342942e-05,
+      "loss": 0.6727,
+      "step": 1036
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.3355506282335553e-05,
+      "loss": 0.6891,
+      "step": 1037
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.328159645232816e-05,
+      "loss": 0.6876,
+      "step": 1038
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.320768662232077e-05,
+      "loss": 0.6763,
+      "step": 1039
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.313377679231338e-05,
+      "loss": 0.6906,
+      "step": 1040
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.3059866962305986e-05,
+      "loss": 0.6795,
+      "step": 1041
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.2985957132298597e-05,
+      "loss": 0.6989,
+      "step": 1042
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.2912047302291207e-05,
+      "loss": 0.6771,
+      "step": 1043
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.2838137472283815e-05,
+      "loss": 0.6918,
+      "step": 1044
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.2764227642276422e-05,
+      "loss": 0.7107,
+      "step": 1045
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.2690317812269033e-05,
+      "loss": 0.6802,
+      "step": 1046
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.261640798226164e-05,
+      "loss": 0.6642,
+      "step": 1047
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.254249815225425e-05,
+      "loss": 0.7026,
+      "step": 1048
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.2468588322246862e-05,
+      "loss": 0.6709,
+      "step": 1049
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.239467849223947e-05,
+      "loss": 0.682,
+      "step": 1050
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.2320768662232077e-05,
+      "loss": 0.6742,
+      "step": 1051
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.2246858832224687e-05,
+      "loss": 0.6839,
+      "step": 1052
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.2172949002217298e-05,
+      "loss": 0.6488,
+      "step": 1053
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.2099039172209902e-05,
+      "loss": 0.6798,
+      "step": 1054
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.2025129342202513e-05,
+      "loss": 0.6986,
+      "step": 1055
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.1951219512195124e-05,
+      "loss": 0.6986,
+      "step": 1056
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.187730968218773e-05,
+      "loss": 0.6855,
+      "step": 1057
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.1803399852180342e-05,
+      "loss": 0.6851,
+      "step": 1058
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.172949002217295e-05,
+      "loss": 0.6652,
+      "step": 1059
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.165558019216556e-05,
+      "loss": 0.7077,
+      "step": 1060
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.1581670362158168e-05,
+      "loss": 0.6852,
+      "step": 1061
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.150776053215078e-05,
+      "loss": 0.6733,
+      "step": 1062
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.1433850702143386e-05,
+      "loss": 0.6794,
+      "step": 1063
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.1359940872135993e-05,
+      "loss": 0.6964,
+      "step": 1064
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.1286031042128604e-05,
+      "loss": 0.6756,
+      "step": 1065
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.1212121212121215e-05,
+      "loss": 0.6911,
+      "step": 1066
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.1138211382113822e-05,
+      "loss": 0.6606,
+      "step": 1067
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.106430155210643e-05,
+      "loss": 0.697,
+      "step": 1068
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.099039172209904e-05,
+      "loss": 0.6772,
+      "step": 1069
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.091648189209165e-05,
+      "loss": 0.6729,
+      "step": 1070
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.084257206208426e-05,
+      "loss": 0.6738,
+      "step": 1071
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.0768662232076866e-05,
+      "loss": 0.6862,
+      "step": 1072
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.0694752402069477e-05,
+      "loss": 0.6868,
+      "step": 1073
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.0620842572062084e-05,
+      "loss": 0.6785,
+      "step": 1074
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.0546932742054695e-05,
+      "loss": 0.6605,
+      "step": 1075
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0473022912047306e-05,
+      "loss": 0.6947,
+      "step": 1076
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0399113082039913e-05,
+      "loss": 0.6816,
+      "step": 1077
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.032520325203252e-05,
+      "loss": 0.6899,
+      "step": 1078
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.025129342202513e-05,
+      "loss": 0.7032,
+      "step": 1079
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.017738359201774e-05,
+      "loss": 0.6956,
+      "step": 1080
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.010347376201035e-05,
+      "loss": 0.6768,
+      "step": 1081
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0029563932002957e-05,
+      "loss": 0.668,
+      "step": 1082
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9955654101995567e-05,
+      "loss": 0.6736,
+      "step": 1083
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9881744271988175e-05,
+      "loss": 0.6682,
+      "step": 1084
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9807834441980786e-05,
+      "loss": 0.6657,
+      "step": 1085
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9733924611973393e-05,
+      "loss": 0.6675,
+      "step": 1086
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9660014781966e-05,
+      "loss": 0.6985,
+      "step": 1087
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.958610495195861e-05,
+      "loss": 0.7021,
+      "step": 1088
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9512195121951222e-05,
+      "loss": 0.6778,
+      "step": 1089
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.943828529194383e-05,
+      "loss": 0.6738,
+      "step": 1090
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9364375461936437e-05,
+      "loss": 0.6746,
+      "step": 1091
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9290465631929047e-05,
+      "loss": 0.6873,
+      "step": 1092
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9216555801921658e-05,
+      "loss": 0.6645,
+      "step": 1093
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9142645971914266e-05,
+      "loss": 0.6711,
+      "step": 1094
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9068736141906873e-05,
+      "loss": 0.6691,
+      "step": 1095
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8994826311899484e-05,
+      "loss": 0.6787,
+      "step": 1096
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.892091648189209e-05,
+      "loss": 0.6857,
+      "step": 1097
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8847006651884702e-05,
+      "loss": 0.6755,
+      "step": 1098
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.8773096821877313e-05,
+      "loss": 0.663,
+      "step": 1099
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.869918699186992e-05,
+      "loss": 0.6645,
+      "step": 1100
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.8625277161862528e-05,
+      "loss": 0.6679,
+      "step": 1101
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.855136733185514e-05,
+      "loss": 0.6962,
+      "step": 1102
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.847745750184775e-05,
+      "loss": 0.6929,
+      "step": 1103
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.8403547671840353e-05,
+      "loss": 0.6926,
+      "step": 1104
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.8329637841832964e-05,
+      "loss": 0.6826,
+      "step": 1105
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.8255728011825575e-05,
+      "loss": 0.705,
+      "step": 1106
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.8181818181818182e-05,
+      "loss": 0.6744,
+      "step": 1107
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.8107908351810793e-05,
+      "loss": 0.6711,
+      "step": 1108
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.80339985218034e-05,
+      "loss": 0.6847,
+      "step": 1109
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.7960088691796008e-05,
+      "loss": 0.6809,
+      "step": 1110
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.788617886178862e-05,
+      "loss": 0.6777,
+      "step": 1111
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.781226903178123e-05,
+      "loss": 0.6767,
+      "step": 1112
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.7738359201773837e-05,
+      "loss": 0.6831,
+      "step": 1113
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.7664449371766444e-05,
+      "loss": 0.6916,
+      "step": 1114
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.7590539541759055e-05,
+      "loss": 0.6577,
+      "step": 1115
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.7516629711751666e-05,
+      "loss": 0.6701,
+      "step": 1116
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.7442719881744273e-05,
+      "loss": 0.6852,
+      "step": 1117
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.736881005173688e-05,
+      "loss": 0.6649,
+      "step": 1118
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.729490022172949e-05,
+      "loss": 0.6762,
+      "step": 1119
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.72209903917221e-05,
+      "loss": 0.6953,
+      "step": 1120
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.714708056171471e-05,
+      "loss": 0.6744,
+      "step": 1121
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.707317073170732e-05,
+      "loss": 0.676,
+      "step": 1122
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.6999260901699927e-05,
+      "loss": 0.6721,
+      "step": 1123
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.6925351071692535e-05,
+      "loss": 0.685,
+      "step": 1124
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.6851441241685146e-05,
+      "loss": 0.667,
+      "step": 1125
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.6777531411677756e-05,
+      "loss": 0.6826,
+      "step": 1126
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.670362158167036e-05,
+      "loss": 0.6829,
+      "step": 1127
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.662971175166297e-05,
+      "loss": 0.6797,
+      "step": 1128
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.6555801921655582e-05,
+      "loss": 0.6575,
+      "step": 1129
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.648189209164819e-05,
+      "loss": 0.6719,
+      "step": 1130
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.64079822616408e-05,
+      "loss": 0.6692,
+      "step": 1131
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.6334072431633408e-05,
+      "loss": 0.6697,
+      "step": 1132
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.6260162601626018e-05,
+      "loss": 0.6964,
+      "step": 1133
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.6186252771618626e-05,
+      "loss": 0.6702,
+      "step": 1134
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.6112342941611236e-05,
+      "loss": 0.6784,
+      "step": 1135
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.6038433111603844e-05,
+      "loss": 0.6816,
+      "step": 1136
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.596452328159645e-05,
+      "loss": 0.6664,
+      "step": 1137
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.5890613451589062e-05,
+      "loss": 0.6691,
+      "step": 1138
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.5816703621581673e-05,
+      "loss": 0.6831,
+      "step": 1139
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.574279379157428e-05,
+      "loss": 0.6566,
+      "step": 1140
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.5668883961566888e-05,
+      "loss": 0.6607,
+      "step": 1141
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.55949741315595e-05,
+      "loss": 0.6579,
+      "step": 1142
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.5521064301552106e-05,
+      "loss": 0.6688,
+      "step": 1143
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.5447154471544717e-05,
+      "loss": 0.6875,
+      "step": 1144
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.5373244641537327e-05,
+      "loss": 0.6742,
+      "step": 1145
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.5299334811529935e-05,
+      "loss": 0.6736,
+      "step": 1146
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.5225424981522544e-05,
+      "loss": 0.6678,
+      "step": 1147
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.5151515151515153e-05,
+      "loss": 0.6667,
+      "step": 1148
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.5077605321507762e-05,
+      "loss": 0.6491,
+      "step": 1149
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.500369549150037e-05,
+      "loss": 0.6505,
+      "step": 1150
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.4929785661492978e-05,
+      "loss": 0.6771,
+      "step": 1151
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.485587583148559e-05,
+      "loss": 0.6634,
+      "step": 1152
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.4781966001478198e-05,
+      "loss": 0.6712,
+      "step": 1153
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.4708056171470807e-05,
+      "loss": 0.6646,
+      "step": 1154
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.4634146341463415e-05,
+      "loss": 0.6992,
+      "step": 1155
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.4560236511456024e-05,
+      "loss": 0.6795,
+      "step": 1156
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.4486326681448633e-05,
+      "loss": 0.6809,
+      "step": 1157
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.4412416851441244e-05,
+      "loss": 0.6687,
+      "step": 1158
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.433850702143385e-05,
+      "loss": 0.6707,
+      "step": 1159
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.426459719142646e-05,
+      "loss": 0.6802,
+      "step": 1160
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.419068736141907e-05,
+      "loss": 0.6624,
+      "step": 1161
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.4116777531411678e-05,
+      "loss": 0.6985,
+      "step": 1162
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.4042867701404289e-05,
+      "loss": 0.6808,
+      "step": 1163
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.3968957871396895e-05,
+      "loss": 0.6802,
+      "step": 1164
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.3895048041389506e-05,
+      "loss": 0.6685,
+      "step": 1165
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.3821138211382115e-05,
+      "loss": 0.6702,
+      "step": 1166
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.3747228381374724e-05,
+      "loss": 0.6664,
+      "step": 1167
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.3673318551367331e-05,
+      "loss": 0.6597,
+      "step": 1168
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.359940872135994e-05,
+      "loss": 0.6818,
+      "step": 1169
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.3525498891352551e-05,
+      "loss": 0.6685,
+      "step": 1170
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.345158906134516e-05,
+      "loss": 0.666,
+      "step": 1171
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.337767923133777e-05,
+      "loss": 0.6882,
+      "step": 1172
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.3303769401330377e-05,
+      "loss": 0.6702,
+      "step": 1173
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.3229859571322986e-05,
+      "loss": 0.6771,
+      "step": 1174
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.3155949741315596e-05,
+      "loss": 0.6715,
+      "step": 1175
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.3082039911308206e-05,
+      "loss": 0.6538,
+      "step": 1176
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.3008130081300815e-05,
+      "loss": 0.6619,
+      "step": 1177
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.2934220251293422e-05,
+      "loss": 0.6735,
+      "step": 1178
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.2860310421286031e-05,
+      "loss": 0.6693,
+      "step": 1179
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.2786400591278642e-05,
+      "loss": 0.6474,
+      "step": 1180
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.2712490761271251e-05,
+      "loss": 0.677,
+      "step": 1181
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.2638580931263858e-05,
+      "loss": 0.6817,
+      "step": 1182
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.2564671101256467e-05,
+      "loss": 0.6704,
+      "step": 1183
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.2490761271249077e-05,
+      "loss": 0.6464,
+      "step": 1184
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.2416851441241686e-05,
+      "loss": 0.6844,
+      "step": 1185
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.2342941611234295e-05,
+      "loss": 0.681,
+      "step": 1186
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.2269031781226904e-05,
+      "loss": 0.6782,
+      "step": 1187
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.2195121951219513e-05,
+      "loss": 0.6781,
+      "step": 1188
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.2121212121212122e-05,
+      "loss": 0.6817,
+      "step": 1189
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.2047302291204731e-05,
+      "loss": 0.6717,
+      "step": 1190
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.197339246119734e-05,
+      "loss": 0.6704,
+      "step": 1191
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.1899482631189948e-05,
+      "loss": 0.6675,
+      "step": 1192
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.1825572801182558e-05,
+      "loss": 0.6586,
+      "step": 1193
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.1751662971175167e-05,
+      "loss": 0.6592,
+      "step": 1194
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.1677753141167776e-05,
+      "loss": 0.6789,
+      "step": 1195
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.1603843311160386e-05,
+      "loss": 0.6721,
+      "step": 1196
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.1529933481152993e-05,
+      "loss": 0.6675,
+      "step": 1197
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.1456023651145604e-05,
+      "loss": 0.681,
+      "step": 1198
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.1382113821138211e-05,
+      "loss": 0.6711,
+      "step": 1199
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.130820399113082e-05,
+      "loss": 0.6575,
+      "step": 1200
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.1234294161123431e-05,
+      "loss": 0.6553,
+      "step": 1201
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.1160384331116038e-05,
+      "loss": 0.6562,
+      "step": 1202
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.1086474501108649e-05,
+      "loss": 0.6706,
+      "step": 1203
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.1012564671101257e-05,
+      "loss": 0.6772,
+      "step": 1204
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.0938654841093866e-05,
+      "loss": 0.6671,
+      "step": 1205
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.0864745011086475e-05,
+      "loss": 0.6989,
+      "step": 1206
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.0790835181079084e-05,
+      "loss": 0.6861,
+      "step": 1207
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.0716925351071693e-05,
+      "loss": 0.6777,
+      "step": 1208
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.0643015521064302e-05,
+      "loss": 0.6616,
+      "step": 1209
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.0569105691056911e-05,
+      "loss": 0.6805,
+      "step": 1210
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.049519586104952e-05,
+      "loss": 0.6628,
+      "step": 1211
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.042128603104213e-05,
+      "loss": 0.6901,
+      "step": 1212
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.0347376201034738e-05,
+      "loss": 0.672,
+      "step": 1213
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.0273466371027347e-05,
+      "loss": 0.6681,
+      "step": 1214
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.0199556541019956e-05,
+      "loss": 0.6732,
+      "step": 1215
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.0125646711012566e-05,
+      "loss": 0.6462,
+      "step": 1216
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.0051736881005175e-05,
+      "loss": 0.676,
+      "step": 1217
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.977827050997784e-06,
+      "loss": 0.6782,
+      "step": 1218
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.903917220990393e-06,
+      "loss": 0.662,
+      "step": 1219
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.830007390983e-06,
+      "loss": 0.6705,
+      "step": 1220
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.756097560975611e-06,
+      "loss": 0.6712,
+      "step": 1221
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.682187730968218e-06,
+      "loss": 0.6842,
+      "step": 1222
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.608277900960829e-06,
+      "loss": 0.6746,
+      "step": 1223
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.534368070953437e-06,
+      "loss": 0.6728,
+      "step": 1224
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.460458240946046e-06,
+      "loss": 0.6767,
+      "step": 1225
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.386548410938656e-06,
+      "loss": 0.6552,
+      "step": 1226
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.312638580931264e-06,
+      "loss": 0.6695,
+      "step": 1227
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.238728750923875e-06,
+      "loss": 0.6678,
+      "step": 1228
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.164818920916482e-06,
+      "loss": 0.6719,
+      "step": 1229
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.090909090909091e-06,
+      "loss": 0.6748,
+      "step": 1230
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.0169992609017e-06,
+      "loss": 0.6622,
+      "step": 1231
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.94308943089431e-06,
+      "loss": 0.6618,
+      "step": 1232
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.869179600886918e-06,
+      "loss": 0.6543,
+      "step": 1233
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.795269770879527e-06,
+      "loss": 0.6822,
+      "step": 1234
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.721359940872136e-06,
+      "loss": 0.6577,
+      "step": 1235
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.647450110864746e-06,
+      "loss": 0.6678,
+      "step": 1236
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.573540280857355e-06,
+      "loss": 0.6461,
+      "step": 1237
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.499630450849964e-06,
+      "loss": 0.66,
+      "step": 1238
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.425720620842573e-06,
+      "loss": 0.6593,
+      "step": 1239
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.35181079083518e-06,
+      "loss": 0.6586,
+      "step": 1240
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.277900960827791e-06,
+      "loss": 0.6745,
+      "step": 1241
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.2039911308204e-06,
+      "loss": 0.6506,
+      "step": 1242
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.130081300813009e-06,
+      "loss": 0.6529,
+      "step": 1243
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.056171470805618e-06,
+      "loss": 0.6779,
+      "step": 1244
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 7.982261640798226e-06,
+      "loss": 0.64,
+      "step": 1245
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 7.908351810790836e-06,
+      "loss": 0.6603,
+      "step": 1246
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 7.834441980783444e-06,
+      "loss": 0.6954,
+      "step": 1247
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 7.760532150776053e-06,
+      "loss": 0.6671,
+      "step": 1248
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 7.686622320768664e-06,
+      "loss": 0.6412,
+      "step": 1249
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 7.612712490761272e-06,
+      "loss": 0.6763,
+      "step": 1250
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 7.538802660753881e-06,
+      "loss": 0.6633,
+      "step": 1251
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 7.464892830746489e-06,
+      "loss": 0.6496,
+      "step": 1252
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 7.390983000739099e-06,
+      "loss": 0.6481,
+      "step": 1253
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 7.317073170731707e-06,
+      "loss": 0.6718,
+      "step": 1254
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 7.2431633407243165e-06,
+      "loss": 0.6738,
+      "step": 1255
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 7.169253510716925e-06,
+      "loss": 0.6692,
+      "step": 1256
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 7.095343680709535e-06,
+      "loss": 0.6662,
+      "step": 1257
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 7.0214338507021446e-06,
+      "loss": 0.657,
+      "step": 1258
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 6.947524020694753e-06,
+      "loss": 0.646,
+      "step": 1259
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 6.873614190687362e-06,
+      "loss": 0.6506,
+      "step": 1260
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 6.79970436067997e-06,
+      "loss": 0.6588,
+      "step": 1261
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 6.72579453067258e-06,
+      "loss": 0.6906,
+      "step": 1262
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 6.651884700665188e-06,
+      "loss": 0.6626,
+      "step": 1263
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 6.577974870657798e-06,
+      "loss": 0.6795,
+      "step": 1264
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 6.504065040650407e-06,
+      "loss": 0.6471,
+      "step": 1265
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 6.4301552106430155e-06,
+      "loss": 0.6757,
+      "step": 1266
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 6.3562453806356255e-06,
+      "loss": 0.6755,
+      "step": 1267
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 6.282335550628234e-06,
+      "loss": 0.6802,
+      "step": 1268
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 6.208425720620843e-06,
+      "loss": 0.6751,
+      "step": 1269
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 6.134515890613452e-06,
+      "loss": 0.6675,
+      "step": 1270
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 6.060606060606061e-06,
+      "loss": 0.6663,
+      "step": 1271
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 5.98669623059867e-06,
+      "loss": 0.6613,
+      "step": 1272
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 5.912786400591279e-06,
+      "loss": 0.6713,
+      "step": 1273
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 5.838876570583888e-06,
+      "loss": 0.6596,
+      "step": 1274
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 5.7649667405764965e-06,
+      "loss": 0.6543,
+      "step": 1275
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 5.6910569105691056e-06,
+      "loss": 0.6693,
+      "step": 1276
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 5.6171470805617155e-06,
+      "loss": 0.6677,
+      "step": 1277
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 5.5432372505543246e-06,
+      "loss": 0.6807,
+      "step": 1278
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 5.469327420546933e-06,
+      "loss": 0.669,
+      "step": 1279
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 5.395417590539542e-06,
+      "loss": 0.6806,
+      "step": 1280
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 5.321507760532151e-06,
+      "loss": 0.6572,
+      "step": 1281
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 5.24759793052476e-06,
+      "loss": 0.659,
+      "step": 1282
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 5.173688100517369e-06,
+      "loss": 0.6614,
+      "step": 1283
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 5.099778270509978e-06,
+      "loss": 0.6713,
+      "step": 1284
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 5.025868440502587e-06,
+      "loss": 0.6608,
+      "step": 1285
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.951958610495196e-06,
+      "loss": 0.6597,
+      "step": 1286
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.8780487804878055e-06,
+      "loss": 0.661,
+      "step": 1287
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.8041389504804146e-06,
+      "loss": 0.6631,
+      "step": 1288
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.730229120473023e-06,
+      "loss": 0.6812,
+      "step": 1289
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.656319290465632e-06,
+      "loss": 0.6611,
+      "step": 1290
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.582409460458241e-06,
+      "loss": 0.6804,
+      "step": 1291
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.50849963045085e-06,
+      "loss": 0.6655,
+      "step": 1292
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.434589800443459e-06,
+      "loss": 0.6634,
+      "step": 1293
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.360679970436068e-06,
+      "loss": 0.6642,
+      "step": 1294
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.286770140428677e-06,
+      "loss": 0.6826,
+      "step": 1295
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.212860310421286e-06,
+      "loss": 0.6986,
+      "step": 1296
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.1389504804138955e-06,
+      "loss": 0.6538,
+      "step": 1297
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.0650406504065046e-06,
+      "loss": 0.6831,
+      "step": 1298
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.991130820399113e-06,
+      "loss": 0.6597,
+      "step": 1299
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.917220990391722e-06,
+      "loss": 0.6864,
+      "step": 1300
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.843311160384332e-06,
+      "loss": 0.6499,
+      "step": 1301
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.7694013303769405e-06,
+      "loss": 0.6661,
+      "step": 1302
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.6954915003695496e-06,
+      "loss": 0.6762,
+      "step": 1303
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.6215816703621582e-06,
+      "loss": 0.6672,
+      "step": 1304
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.5476718403547673e-06,
+      "loss": 0.6606,
+      "step": 1305
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.4737620103473764e-06,
+      "loss": 0.6557,
+      "step": 1306
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.399852180339985e-06,
+      "loss": 0.6763,
+      "step": 1307
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.325942350332594e-06,
+      "loss": 0.6542,
+      "step": 1308
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.2520325203252037e-06,
+      "loss": 0.6592,
+      "step": 1309
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.1781226903178127e-06,
+      "loss": 0.6636,
+      "step": 1310
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.1042128603104214e-06,
+      "loss": 0.6487,
+      "step": 1311
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.0303030303030305e-06,
+      "loss": 0.6763,
+      "step": 1312
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2.9563932002956396e-06,
+      "loss": 0.6617,
+      "step": 1313
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2.8824833702882482e-06,
+      "loss": 0.6683,
+      "step": 1314
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2.8085735402808577e-06,
+      "loss": 0.6658,
+      "step": 1315
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2.7346637102734664e-06,
+      "loss": 0.6639,
+      "step": 1316
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2.6607538802660755e-06,
+      "loss": 0.6671,
+      "step": 1317
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2.5868440502586846e-06,
+      "loss": 0.67,
+      "step": 1318
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 2.5129342202512937e-06,
+      "loss": 0.6904,
+      "step": 1319
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.4390243902439027e-06,
+      "loss": 0.6444,
+      "step": 1320
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.3651145602365114e-06,
+      "loss": 0.6626,
+      "step": 1321
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.2912047302291205e-06,
+      "loss": 0.6861,
+      "step": 1322
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.2172949002217296e-06,
+      "loss": 0.6626,
+      "step": 1323
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.1433850702143387e-06,
+      "loss": 0.6687,
+      "step": 1324
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.0694752402069477e-06,
+      "loss": 0.6723,
+      "step": 1325
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9955654101995564e-06,
+      "loss": 0.6842,
+      "step": 1326
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.921655580192166e-06,
+      "loss": 0.6391,
+      "step": 1327
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.8477457501847748e-06,
+      "loss": 0.6612,
+      "step": 1328
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.7738359201773837e-06,
+      "loss": 0.6542,
+      "step": 1329
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.6999260901699925e-06,
+      "loss": 0.671,
+      "step": 1330
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.6260162601626018e-06,
+      "loss": 0.6669,
+      "step": 1331
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.5521064301552107e-06,
+      "loss": 0.6671,
+      "step": 1332
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4781966001478198e-06,
+      "loss": 0.6375,
+      "step": 1333
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.4042867701404289e-06,
+      "loss": 0.6593,
+      "step": 1334
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.3303769401330377e-06,
+      "loss": 0.6719,
+      "step": 1335
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.2564671101256468e-06,
+      "loss": 0.6673,
+      "step": 1336
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.1825572801182557e-06,
+      "loss": 0.6624,
+      "step": 1337
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.1086474501108648e-06,
+      "loss": 0.6479,
+      "step": 1338
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.0347376201034739e-06,
+      "loss": 0.668,
+      "step": 1339
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 9.60827790096083e-07,
+      "loss": 0.6592,
+      "step": 1340
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 8.869179600886918e-07,
+      "loss": 0.658,
+      "step": 1341
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 8.130081300813009e-07,
+      "loss": 0.6538,
+      "step": 1342
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.390983000739099e-07,
+      "loss": 0.6621,
+      "step": 1343
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 6.651884700665189e-07,
+      "loss": 0.6663,
+      "step": 1344
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 5.912786400591279e-07,
+      "loss": 0.6573,
+      "step": 1345
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 5.173688100517369e-07,
+      "loss": 0.6785,
+      "step": 1346
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.434589800443459e-07,
+      "loss": 0.6408,
+      "step": 1347
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.6954915003695495e-07,
+      "loss": 0.663,
+      "step": 1348
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.956393200295639e-07,
+      "loss": 0.6562,
+      "step": 1349
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.2172949002217296e-07,
+      "loss": 0.6537,
+      "step": 1350
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.4781966001478196e-07,
+      "loss": 0.664,
+      "step": 1351
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.390983000739098e-08,
+      "loss": 0.6657,
+      "step": 1352
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0,
+      "loss": 0.6589,
+      "step": 1353
+    },
+    {
+      "epoch": 3.0,
+      "eval_loss": 0.9302364587783813,
+      "eval_runtime": 496.1152,
+      "eval_samples_per_second": 32.13,
+      "eval_steps_per_second": 0.504,
+      "eval_wer": 0.7427813602969554,
+      "step": 1353
+    },
+    {
+      "epoch": 3.0,
+      "step": 1353,
+      "total_flos": 6.225660003734846e+19,
+      "train_loss": 0.8526812995792933,
+      "train_runtime": 42752.8496,
+      "train_samples_per_second": 32.442,
+      "train_steps_per_second": 0.032
+    }
+  ],
+  "max_steps": 1353,
+  "num_train_epochs": 3,
+  "total_flos": 6.225660003734846e+19,
+  "trial_name": null,
+  "trial_params": null
+}