diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,6943 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.9995665366276549,
+  "global_step": 1153,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.714285714285715e-07,
+      "loss": 1.4282,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.142857142857143e-06,
+      "loss": 1.458,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.7142857142857145e-06,
+      "loss": 1.459,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.285714285714286e-06,
+      "loss": 1.4712,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8571428571428573e-06,
+      "loss": 1.4258,
+      "step": 5
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.428571428571429e-06,
+      "loss": 1.4712,
+      "step": 6
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 1.4443,
+      "step": 7
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.571428571428572e-06,
+      "loss": 1.4023,
+      "step": 8
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.142857142857142e-06,
+      "loss": 1.4141,
+      "step": 9
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.7142857142857145e-06,
+      "loss": 1.4429,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.285714285714286e-06,
+      "loss": 1.4448,
+      "step": 11
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.857142857142858e-06,
+      "loss": 1.3799,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.428571428571429e-06,
+      "loss": 1.5327,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 1.4116,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.571428571428571e-06,
+      "loss": 1.3833,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.142857142857144e-06,
+      "loss": 1.4385,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.714285714285715e-06,
+      "loss": 1.4985,
+      "step": 17
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0285714285714285e-05,
+      "loss": 1.437,
+      "step": 18
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0857142857142858e-05,
+      "loss": 1.4961,
+      "step": 19
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1428571428571429e-05,
+      "loss": 1.4189,
+      "step": 20
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2e-05,
+      "loss": 1.3662,
+      "step": 21
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2571428571428572e-05,
+      "loss": 1.4595,
+      "step": 22
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3142857142857145e-05,
+      "loss": 1.4507,
+      "step": 23
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3714285714285716e-05,
+      "loss": 1.4004,
+      "step": 24
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4285714285714287e-05,
+      "loss": 1.4243,
+      "step": 25
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4857142857142858e-05,
+      "loss": 1.416,
+      "step": 26
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.542857142857143e-05,
+      "loss": 1.4487,
+      "step": 27
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 1.3823,
+      "step": 28
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6571428571428574e-05,
+      "loss": 1.4053,
+      "step": 29
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7142857142857142e-05,
+      "loss": 1.4233,
+      "step": 30
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7714285714285717e-05,
+      "loss": 1.4067,
+      "step": 31
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8285714285714288e-05,
+      "loss": 1.4224,
+      "step": 32
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.885714285714286e-05,
+      "loss": 1.4302,
+      "step": 33
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.942857142857143e-05,
+      "loss": 1.5254,
+      "step": 34
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2e-05,
+      "loss": 1.3965,
+      "step": 35
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999960519207943e-05,
+      "loss": 1.3989,
+      "step": 36
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999984207714351e-05,
+      "loss": 1.394,
+      "step": 37
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999644674741944e-05,
+      "loss": 1.4131,
+      "step": 38
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999368313561967e-05,
+      "loss": 1.3281,
+      "step": 39
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999012995785765e-05,
+      "loss": 1.4541,
+      "step": 40
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998578724218986e-05,
+      "loss": 1.3794,
+      "step": 41
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999806550229071e-05,
+      "loss": 1.4214,
+      "step": 42
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997473334053415e-05,
+      "loss": 1.3823,
+      "step": 43
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996802224182953e-05,
+      "loss": 1.335,
+      "step": 44
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999605217797852e-05,
+      "loss": 1.4575,
+      "step": 45
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995223201362596e-05,
+      "loss": 1.3579,
+      "step": 46
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994315300880913e-05,
+      "loss": 1.3188,
+      "step": 47
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993328483702393e-05,
+      "loss": 1.3042,
+      "step": 48
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999226275761911e-05,
+      "loss": 1.3223,
+      "step": 49
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991118131046197e-05,
+      "loss": 1.3804,
+      "step": 50
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998989461302181e-05,
+      "loss": 1.3115,
+      "step": 51
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998859221320704e-05,
+      "loss": 1.2666,
+      "step": 52
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987210941885843e-05,
+      "loss": 1.3018,
+      "step": 53
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998575080996496e-05,
+      "loss": 1.2803,
+      "step": 54
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998421182897382e-05,
+      "loss": 1.3098,
+      "step": 55
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982594011064457e-05,
+      "loss": 1.334,
+      "step": 56
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980897369011424e-05,
+      "loss": 1.2168,
+      "step": 57
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979121916211674e-05,
+      "loss": 1.2656,
+      "step": 58
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997726766668446e-05,
+      "loss": 1.3188,
+      "step": 59
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975334635071234e-05,
+      "loss": 1.2012,
+      "step": 60
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9973322836635517e-05,
+      "loss": 1.1697,
+      "step": 61
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997123228726279e-05,
+      "loss": 1.228,
+      "step": 62
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9969063003460364e-05,
+      "loss": 1.2695,
+      "step": 63
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9966815002357243e-05,
+      "loss": 1.2266,
+      "step": 64
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9964488301704004e-05,
+      "loss": 1.2876,
+      "step": 65
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996208291987264e-05,
+      "loss": 1.2544,
+      "step": 66
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995959887585643e-05,
+      "loss": 1.2334,
+      "step": 67
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9957036189269777e-05,
+      "loss": 1.198,
+      "step": 68
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995439488034806e-05,
+      "loss": 1.2397,
+      "step": 69
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9951674969947477e-05,
+      "loss": 1.2383,
+      "step": 70
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9948876479544866e-05,
+      "loss": 1.1514,
+      "step": 71
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994599943123755e-05,
+      "loss": 1.2554,
+      "step": 72
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9943043847743164e-05,
+      "loss": 1.2332,
+      "step": 73
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9940009752399462e-05,
+      "loss": 1.2283,
+      "step": 74
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993689716916414e-05,
+      "loss": 1.1477,
+      "step": 75
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9933706122614645e-05,
+      "loss": 1.1663,
+      "step": 76
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993043663794799e-05,
+      "loss": 1.1292,
+      "step": 77
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992708874098054e-05,
+      "loss": 1.1929,
+      "step": 78
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992366245814783e-05,
+      "loss": 1.2341,
+      "step": 79
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992015781650432e-05,
+      "loss": 1.1909,
+      "step": 80
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991657484372322e-05,
+      "loss": 1.1677,
+      "step": 81
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9912913568096252e-05,
+      "loss": 1.1301,
+      "step": 82
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9909174018533428e-05,
+      "loss": 1.1519,
+      "step": 83
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9905356224562824e-05,
+      "loss": 1.1885,
+      "step": 84
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9901460216330344e-05,
+      "loss": 1.2166,
+      "step": 85
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989748602459949e-05,
+      "loss": 1.2358,
+      "step": 86
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9893433680751105e-05,
+      "loss": 1.1416,
+      "step": 87
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988930321678314e-05,
+      "loss": 1.105,
+      "step": 88
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988509466531039e-05,
+      "loss": 1.123,
+      "step": 89
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988080805956425e-05,
+      "loss": 1.2227,
+      "step": 90
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9876443433392433e-05,
+      "loss": 1.0977,
+      "step": 91
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9872000821258723e-05,
+      "loss": 1.1104,
+      "step": 92
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986748025824268e-05,
+      "loss": 1.1179,
+      "step": 93
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9862881780039398e-05,
+      "loss": 1.1687,
+      "step": 94
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9858205422959183e-05,
+      "loss": 1.0859,
+      "step": 95
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9853451223927293e-05,
+      "loss": 1.2053,
+      "step": 96
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9848619220483632e-05,
+      "loss": 1.1899,
+      "step": 97
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984370945078247e-05,
+      "loss": 1.0742,
+      "step": 98
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9838721953592122e-05,
+      "loss": 1.2053,
+      "step": 99
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983365676829466e-05,
+      "loss": 1.1755,
+      "step": 100
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9828513934885592e-05,
+      "loss": 1.1665,
+      "step": 101
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982329349397354e-05,
+      "loss": 1.1365,
+      "step": 102
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981799548677993e-05,
+      "loss": 1.082,
+      "step": 103
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981261995513867e-05,
+      "loss": 1.0588,
+      "step": 104
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9807166941495813e-05,
+      "loss": 1.0984,
+      "step": 105
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9801636488909208e-05,
+      "loss": 1.1648,
+      "step": 106
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9796028641048197e-05,
+      "loss": 1.074,
+      "step": 107
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979034344219323e-05,
+      "loss": 1.0605,
+      "step": 108
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9784580937235535e-05,
+      "loss": 1.1479,
+      "step": 109
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9778741171676766e-05,
+      "loss": 1.2012,
+      "step": 110
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9772824191628635e-05,
+      "loss": 1.1184,
+      "step": 111
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976683004381256e-05,
+      "loss": 1.1372,
+      "step": 112
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9760758775559275e-05,
+      "loss": 1.1753,
+      "step": 113
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9754610434808476e-05,
+      "loss": 1.1184,
+      "step": 114
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9748385070108444e-05,
+      "loss": 1.1348,
+      "step": 115
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9742082730615634e-05,
+      "loss": 1.1145,
+      "step": 116
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9735703466094326e-05,
+      "loss": 1.134,
+      "step": 117
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9729247326916194e-05,
+      "loss": 1.0896,
+      "step": 118
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9722714364059943e-05,
+      "loss": 1.1172,
+      "step": 119
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9716104629110882e-05,
+      "loss": 1.1057,
+      "step": 120
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9709418174260523e-05,
+      "loss": 1.0625,
+      "step": 121
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.970265505230617e-05,
+      "loss": 1.1138,
+      "step": 122
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9695815316650514e-05,
+      "loss": 1.1284,
+      "step": 123
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9688899021301182e-05,
+      "loss": 1.1067,
+      "step": 124
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9681906220870345e-05,
+      "loss": 1.1492,
+      "step": 125
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9674836970574253e-05,
+      "loss": 1.0776,
+      "step": 126
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966769132623284e-05,
+      "loss": 1.1738,
+      "step": 127
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9660469344269234e-05,
+      "loss": 1.1147,
+      "step": 128
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9653171081709353e-05,
+      "loss": 1.1379,
+      "step": 129
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9645796596181436e-05,
+      "loss": 1.1509,
+      "step": 130
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9638345945915588e-05,
+      "loss": 1.1729,
+      "step": 131
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963081918974332e-05,
+      "loss": 1.1013,
+      "step": 132
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9623216387097095e-05,
+      "loss": 1.1362,
+      "step": 133
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9615537598009845e-05,
+      "loss": 1.1433,
+      "step": 134
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9607782883114507e-05,
+      "loss": 1.1169,
+      "step": 135
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9599952303643538e-05,
+      "loss": 1.1633,
+      "step": 136
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9592045921428435e-05,
+      "loss": 1.1497,
+      "step": 137
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958406379889924e-05,
+      "loss": 1.1467,
+      "step": 138
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957600599908406e-05,
+      "loss": 1.0566,
+      "step": 139
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956787258560856e-05,
+      "loss": 1.145,
+      "step": 140
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955966362269546e-05,
+      "loss": 1.0774,
+      "step": 141
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955137917516403e-05,
+      "loss": 1.2251,
+      "step": 142
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954301930842958e-05,
+      "loss": 1.1072,
+      "step": 143
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9534584088502946e-05,
+      "loss": 1.0513,
+      "step": 144
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952607358198996e-05,
+      "loss": 1.083,
+      "step": 145
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9517487856090923e-05,
+      "loss": 1.0674,
+      "step": 146
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9508826978600095e-05,
+      "loss": 1.1001,
+      "step": 147
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9500091017905133e-05,
+      "loss": 1.0842,
+      "step": 148
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9491280042986566e-05,
+      "loss": 1.1316,
+      "step": 149
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9482394123417246e-05,
+      "loss": 1.1184,
+      "step": 150
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9473433329361805e-05,
+      "loss": 1.1218,
+      "step": 151
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9464397731576093e-05,
+      "loss": 1.0813,
+      "step": 152
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945528740140662e-05,
+      "loss": 1.0742,
+      "step": 153
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.944610241079e-05,
+      "loss": 1.0691,
+      "step": 154
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9436842832252363e-05,
+      "loss": 1.0923,
+      "step": 155
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9427508738908817e-05,
+      "loss": 1.1353,
+      "step": 156
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9418100204462843e-05,
+      "loss": 1.1392,
+      "step": 157
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9408617303205712e-05,
+      "loss": 1.0652,
+      "step": 158
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9399060110015918e-05,
+      "loss": 1.085,
+      "step": 159
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9389428700358575e-05,
+      "loss": 1.0779,
+      "step": 160
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9379723150284817e-05,
+      "loss": 1.1453,
+      "step": 161
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9369943536431207e-05,
+      "loss": 1.1675,
+      "step": 162
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.936008993601912e-05,
+      "loss": 1.0417,
+      "step": 163
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9350162426854152e-05,
+      "loss": 1.0901,
+      "step": 164
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9340161087325483e-05,
+      "loss": 1.1226,
+      "step": 165
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.933008599640528e-05,
+      "loss": 1.1821,
+      "step": 166
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9319937233648044e-05,
+      "loss": 1.1182,
+      "step": 167
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9309714879190032e-05,
+      "loss": 1.0193,
+      "step": 168
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.929941901374856e-05,
+      "loss": 1.0535,
+      "step": 169
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9289049718621423e-05,
+      "loss": 1.1357,
+      "step": 170
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9278607075686207e-05,
+      "loss": 1.0457,
+      "step": 171
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9268091167399682e-05,
+      "loss": 1.0725,
+      "step": 172
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9257502076797125e-05,
+      "loss": 1.1003,
+      "step": 173
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.924683988749167e-05,
+      "loss": 1.1221,
+      "step": 174
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9236104683673656e-05,
+      "loss": 1.0168,
+      "step": 175
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9225296550109948e-05,
+      "loss": 1.0496,
+      "step": 176
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.921441557214329e-05,
+      "loss": 0.9783,
+      "step": 177
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9203461835691596e-05,
+      "loss": 1.1194,
+      "step": 178
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9192435427247313e-05,
+      "loss": 1.0591,
+      "step": 179
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9181336433876704e-05,
+      "loss": 1.0767,
+      "step": 180
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9170164943219186e-05,
+      "loss": 1.1206,
+      "step": 181
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.915892104348661e-05,
+      "loss": 1.1401,
+      "step": 182
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.914760482346259e-05,
+      "loss": 1.0347,
+      "step": 183
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.91362163725018e-05,
+      "loss": 0.9954,
+      "step": 184
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9124755780529243e-05,
+      "loss": 1.1501,
+      "step": 185
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.911322313803958e-05,
+      "loss": 1.1416,
+      "step": 186
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.910161853609637e-05,
+      "loss": 1.1692,
+      "step": 187
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.90899420663314e-05,
+      "loss": 1.042,
+      "step": 188
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9078193820943916e-05,
+      "loss": 1.05,
+      "step": 189
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9066373892699933e-05,
+      "loss": 1.0447,
+      "step": 190
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.905448237493147e-05,
+      "loss": 1.0864,
+      "step": 191
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9042519361535835e-05,
+      "loss": 1.0244,
+      "step": 192
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.903048494697488e-05,
+      "loss": 1.0066,
+      "step": 193
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.901837922627425e-05,
+      "loss": 1.0938,
+      "step": 194
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9006202295022632e-05,
+      "loss": 1.0686,
+      "step": 195
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8993954249371e-05,
+      "loss": 1.0171,
+      "step": 196
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.898163518603187e-05,
+      "loss": 1.043,
+      "step": 197
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.896924520227851e-05,
+      "loss": 1.1401,
+      "step": 198
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.89567843959442e-05,
+      "loss": 1.0608,
+      "step": 199
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.894425286542144e-05,
+      "loss": 1.051,
+      "step": 200
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8931650709661178e-05,
+      "loss": 1.0613,
+      "step": 201
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8918978028172034e-05,
+      "loss": 1.1443,
+      "step": 202
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8906234921019505e-05,
+      "loss": 1.0156,
+      "step": 203
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.889342148882519e-05,
+      "loss": 1.0435,
+      "step": 204
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8880537832765976e-05,
+      "loss": 1.0657,
+      "step": 205
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.886758405457325e-05,
+      "loss": 1.1021,
+      "step": 206
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8854560256532098e-05,
+      "loss": 1.0449,
+      "step": 207
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.88414665414805e-05,
+      "loss": 0.998,
+      "step": 208
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8828303012808492e-05,
+      "loss": 1.061,
+      "step": 209
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8815069774457386e-05,
+      "loss": 1.0774,
+      "step": 210
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8801766930918934e-05,
+      "loss": 1.043,
+      "step": 211
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8788394587234486e-05,
+      "loss": 1.074,
+      "step": 212
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8774952848994195e-05,
+      "loss": 1.0732,
+      "step": 213
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8761441822336152e-05,
+      "loss": 1.1199,
+      "step": 214
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.874786161394556e-05,
+      "loss": 1.071,
+      "step": 215
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8734212331053904e-05,
+      "loss": 1.1189,
+      "step": 216
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.872049408143808e-05,
+      "loss": 1.0867,
+      "step": 217
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8706706973419555e-05,
+      "loss": 1.0535,
+      "step": 218
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8692851115863524e-05,
+      "loss": 1.0723,
+      "step": 219
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8678926618178034e-05,
+      "loss": 1.082,
+      "step": 220
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.866493359031312e-05,
+      "loss": 1.1277,
+      "step": 221
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8650872142759947e-05,
+      "loss": 1.123,
+      "step": 222
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.863674238654994e-05,
+      "loss": 1.1387,
+      "step": 223
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8622544433253883e-05,
+      "loss": 1.0393,
+      "step": 224
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8608278394981068e-05,
+      "loss": 1.0896,
+      "step": 225
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.859394438437839e-05,
+      "loss": 0.9807,
+      "step": 226
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8579542514629476e-05,
+      "loss": 1.1313,
+      "step": 227
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8565072899453763e-05,
+      "loss": 1.0581,
+      "step": 228
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8550535653105625e-05,
+      "loss": 0.9973,
+      "step": 229
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8535930890373467e-05,
+      "loss": 1.0593,
+      "step": 230
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8521258726578804e-05,
+      "loss": 1.1113,
+      "step": 231
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.850651927757537e-05,
+      "loss": 1.0317,
+      "step": 232
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8491712659748183e-05,
+      "loss": 1.0823,
+      "step": 233
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.847683899001265e-05,
+      "loss": 1.0391,
+      "step": 234
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8461898385813622e-05,
+      "loss": 1.0093,
+      "step": 235
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8446890965124472e-05,
+      "loss": 1.0989,
+      "step": 236
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8431816846446173e-05,
+      "loss": 1.0974,
+      "step": 237
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8416676148806356e-05,
+      "loss": 1.0527,
+      "step": 238
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8401468991758368e-05,
+      "loss": 1.0369,
+      "step": 239
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8386195495380324e-05,
+      "loss": 1.0649,
+      "step": 240
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8370855780274182e-05,
+      "loss": 1.0894,
+      "step": 241
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8355449967564754e-05,
+      "loss": 1.0376,
+      "step": 242
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.833997817889878e-05,
+      "loss": 1.0107,
+      "step": 243
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8324440536443956e-05,
+      "loss": 1.0593,
+      "step": 244
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8308837162887964e-05,
+      "loss": 1.0212,
+      "step": 245
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8293168181437516e-05,
+      "loss": 1.001,
+      "step": 246
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.827743371581737e-05,
+      "loss": 1.033,
+      "step": 247
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8261633890269363e-05,
+      "loss": 1.0513,
+      "step": 248
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8245768829551417e-05,
+      "loss": 1.0637,
+      "step": 249
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8229838658936566e-05,
+      "loss": 1.1067,
+      "step": 250
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.821384350421196e-05,
+      "loss": 1.051,
+      "step": 251
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8197783491677873e-05,
+      "loss": 1.1094,
+      "step": 252
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.818165874814671e-05,
+      "loss": 1.0076,
+      "step": 253
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8165469400942e-05,
+      "loss": 1.1072,
+      "step": 254
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8149215577897397e-05,
+      "loss": 1.0317,
+      "step": 255
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8132897407355657e-05,
+      "loss": 1.0293,
+      "step": 256
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8116515018167636e-05,
+      "loss": 1.0181,
+      "step": 257
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8100068539691283e-05,
+      "loss": 1.0547,
+      "step": 258
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8083558101790598e-05,
+      "loss": 1.0962,
+      "step": 259
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8066983834834607e-05,
+      "loss": 1.0444,
+      "step": 260
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.805034586969635e-05,
+      "loss": 1.0305,
+      "step": 261
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8033644337751833e-05,
+      "loss": 1.054,
+      "step": 262
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8016879370879006e-05,
+      "loss": 1.1201,
+      "step": 263
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.800005110145669e-05,
+      "loss": 1.0361,
+      "step": 264
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7983159662363582e-05,
+      "loss": 1.1047,
+      "step": 265
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7966205186977154e-05,
+      "loss": 1.0967,
+      "step": 266
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7949187809172623e-05,
+      "loss": 1.0322,
+      "step": 267
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.793210766332191e-05,
+      "loss": 1.0574,
+      "step": 268
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7914964884292543e-05,
+      "loss": 1.082,
+      "step": 269
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7897759607446627e-05,
+      "loss": 1.0273,
+      "step": 270
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.788049196863975e-05,
+      "loss": 1.0786,
+      "step": 271
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.786316210421993e-05,
+      "loss": 1.0415,
+      "step": 272
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7845770151026516e-05,
+      "loss": 1.0522,
+      "step": 273
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7828316246389125e-05,
+      "loss": 1.1172,
+      "step": 274
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7810800528126556e-05,
+      "loss": 1.0425,
+      "step": 275
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7793223134545692e-05,
+      "loss": 0.9978,
+      "step": 276
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.777558420444042e-05,
+      "loss": 1.1089,
+      "step": 277
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7757883877090526e-05,
+      "loss": 1.0491,
+      "step": 278
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7740122292260594e-05,
+      "loss": 1.0662,
+      "step": 279
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.772229959019892e-05,
+      "loss": 1.1101,
+      "step": 280
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7704415911636378e-05,
+      "loss": 0.9956,
+      "step": 281
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7686471397785322e-05,
+      "loss": 0.9792,
+      "step": 282
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7668466190338484e-05,
+      "loss": 1.0498,
+      "step": 283
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7650400431467832e-05,
+      "loss": 1.0879,
+      "step": 284
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7632274263823458e-05,
+      "loss": 1.0759,
+      "step": 285
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7614087830532453e-05,
+      "loss": 0.9521,
+      "step": 286
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7595841275197783e-05,
+      "loss": 1.074,
+      "step": 287
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7577534741897126e-05,
+      "loss": 1.0852,
+      "step": 288
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7559168375181777e-05,
+      "loss": 1.0432,
+      "step": 289
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7540742320075476e-05,
+      "loss": 1.1135,
+      "step": 290
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7522256722073273e-05,
+      "loss": 1.052,
+      "step": 291
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7503711727140384e-05,
+      "loss": 1.0938,
+      "step": 292
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7485107481711014e-05,
+      "loss": 1.0784,
+      "step": 293
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.746644413268724e-05,
+      "loss": 1.0737,
+      "step": 294
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.744772182743782e-05,
+      "loss": 1.0781,
+      "step": 295
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7428940713797044e-05,
+      "loss": 1.0542,
+      "step": 296
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.741010094006356e-05,
+      "loss": 1.0273,
+      "step": 297
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7391202654999207e-05,
+      "loss": 1.0,
+      "step": 298
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7372246007827837e-05,
+      "loss": 1.0347,
+      "step": 299
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7353231148234134e-05,
+      "loss": 1.1033,
+      "step": 300
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7334158226362446e-05,
+      "loss": 1.0828,
+      "step": 301
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7315027392815588e-05,
+      "loss": 1.0918,
+      "step": 302
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.729583879865365e-05,
+      "loss": 1.0725,
+      "step": 303
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.727659259539281e-05,
+      "loss": 1.1208,
+      "step": 304
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7257288935004134e-05,
+      "loss": 0.9946,
+      "step": 305
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7237927969912393e-05,
+      "loss": 1.1011,
+      "step": 306
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7218509852994825e-05,
+      "loss": 1.0896,
+      "step": 307
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7199034737579962e-05,
+      "loss": 1.1091,
+      "step": 308
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7179502777446395e-05,
+      "loss": 1.0332,
+      "step": 309
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.715991412682158e-05,
+      "loss": 1.0225,
+      "step": 310
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7140268940380608e-05,
+      "loss": 1.0615,
+      "step": 311
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7120567373244976e-05,
+      "loss": 1.0154,
+      "step": 312
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7100809580981384e-05,
+      "loss": 1.0151,
+      "step": 313
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.708099571960049e-05,
+      "loss": 1.0862,
+      "step": 314
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.706112594555568e-05,
+      "loss": 1.0781,
+      "step": 315
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7041200415741838e-05,
+      "loss": 1.0344,
+      "step": 316
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7021219287494103e-05,
+      "loss": 1.031,
+      "step": 317
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7001182718586625e-05,
+      "loss": 1.0293,
+      "step": 318
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6981090867231337e-05,
+      "loss": 1.0361,
+      "step": 319
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6960943892076672e-05,
+      "loss": 0.9905,
+      "step": 320
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6940741952206342e-05,
+      "loss": 1.0178,
+      "step": 321
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.692048520713806e-05,
+      "loss": 1.0215,
+      "step": 322
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.690017381682229e-05,
+      "loss": 1.0198,
+      "step": 323
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6879807941641e-05,
+      "loss": 1.0527,
+      "step": 324
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6859387742406358e-05,
+      "loss": 1.0291,
+      "step": 325
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6838913380359492e-05,
+      "loss": 1.0325,
+      "step": 326
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6818385017169213e-05,
+      "loss": 1.0532,
+      "step": 327
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6797802814930723e-05,
+      "loss": 1.1135,
+      "step": 328
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6777166936164354e-05,
+      "loss": 1.0164,
+      "step": 329
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6756477543814276e-05,
+      "loss": 1.0378,
+      "step": 330
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6735734801247204e-05,
+      "loss": 1.0581,
+      "step": 331
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.671493887225112e-05,
+      "loss": 1.0852,
+      "step": 332
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6694089921033977e-05,
+      "loss": 0.9973,
+      "step": 333
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6673188112222394e-05,
+      "loss": 1.0603,
+      "step": 334
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6652233610860368e-05,
+      "loss": 0.9949,
+      "step": 335
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6631226582407954e-05,
+      "loss": 1.0608,
+      "step": 336
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.661016719273998e-05,
+      "loss": 1.0884,
+      "step": 337
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6589055608144727e-05,
+      "loss": 1.0361,
+      "step": 338
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6567891995322605e-05,
+      "loss": 0.9966,
+      "step": 339
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6546676521384856e-05,
+      "loss": 1.0337,
+      "step": 340
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6525409353852226e-05,
+      "loss": 1.1138,
+      "step": 341
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.650409066065363e-05,
+      "loss": 1.0,
+      "step": 342
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6482720610124858e-05,
+      "loss": 1.123,
+      "step": 343
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.646129937100721e-05,
+      "loss": 1.0432,
+      "step": 344
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6439827112446177e-05,
+      "loss": 1.0002,
+      "step": 345
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.641830400399012e-05,
+      "loss": 1.0496,
+      "step": 346
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6396730215588913e-05,
+      "loss": 1.0681,
+      "step": 347
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6375105917592607e-05,
+      "loss": 1.0923,
+      "step": 348
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6353431280750083e-05,
+      "loss": 1.0159,
+      "step": 349
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6331706476207708e-05,
+      "loss": 1.1025,
+      "step": 350
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.630993167550798e-05,
+      "loss": 0.9944,
+      "step": 351
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6288107050588172e-05,
+      "loss": 1.0327,
+      "step": 352
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6266232773778984e-05,
+      "loss": 1.0791,
+      "step": 353
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.624430901780317e-05,
+      "loss": 1.0647,
+      "step": 354
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6222335955774178e-05,
+      "loss": 1.0237,
+      "step": 355
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6200313761194785e-05,
+      "loss": 1.0359,
+      "step": 356
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.617824260795573e-05,
+      "loss": 1.0112,
+      "step": 357
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.615612267033434e-05,
+      "loss": 1.0693,
+      "step": 358
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.613395412299314e-05,
+      "loss": 1.0457,
+      "step": 359
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6111737140978495e-05,
+      "loss": 0.9858,
+      "step": 360
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6089471899719213e-05,
+      "loss": 1.0354,
+      "step": 361
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6067158575025165e-05,
+      "loss": 1.0701,
+      "step": 362
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6044797343085902e-05,
+      "loss": 1.0261,
+      "step": 363
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6022388380469248e-05,
+      "loss": 0.9333,
+      "step": 364
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5999931864119923e-05,
+      "loss": 1.0261,
+      "step": 365
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.597742797135814e-05,
+      "loss": 1.01,
+      "step": 366
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.59548768798782e-05,
+      "loss": 0.9958,
+      "step": 367
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5932278767747092e-05,
+      "loss": 1.0789,
+      "step": 368
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5909633813403094e-05,
+      "loss": 1.0493,
+      "step": 369
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.588694219565435e-05,
+      "loss": 1.0264,
+      "step": 370
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5864204093677464e-05,
+      "loss": 1.0703,
+      "step": 371
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5841419687016097e-05,
+      "loss": 1.0349,
+      "step": 372
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.581858915557953e-05,
+      "loss": 0.9749,
+      "step": 373
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5795712679641257e-05,
+      "loss": 1.0083,
+      "step": 374
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5772790439837557e-05,
+      "loss": 1.0029,
+      "step": 375
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5749822617166067e-05,
+      "loss": 1.0439,
+      "step": 376
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.572680939298435e-05,
+      "loss": 0.9871,
+      "step": 377
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.570375094900847e-05,
+      "loss": 1.0352,
+      "step": 378
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.568064746731156e-05,
+      "loss": 1.0906,
+      "step": 379
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5657499130322366e-05,
+      "loss": 1.0134,
+      "step": 380
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5634306120823822e-05,
+      "loss": 1.1108,
+      "step": 381
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5611068621951603e-05,
+      "loss": 1.0271,
+      "step": 382
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.558778681719269e-05,
+      "loss": 0.9937,
+      "step": 383
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5564460890383897e-05,
+      "loss": 1.053,
+      "step": 384
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5541091025710436e-05,
+      "loss": 1.0376,
+      "step": 385
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.551767740770446e-05,
+      "loss": 1.0205,
+      "step": 386
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.549422022124361e-05,
+      "loss": 1.1287,
+      "step": 387
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5470719651549543e-05,
+      "loss": 1.0466,
+      "step": 388
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.544717588418648e-05,
+      "loss": 1.0527,
+      "step": 389
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.542358910505974e-05,
+      "loss": 1.0396,
+      "step": 390
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5399959500414264e-05,
+      "loss": 1.0273,
+      "step": 391
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5376287256833154e-05,
+      "loss": 1.0452,
+      "step": 392
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.53525725612362e-05,
+      "loss": 1.042,
+      "step": 393
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.532881560087839e-05,
+      "loss": 1.0569,
+      "step": 394
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5305016563348444e-05,
+      "loss": 1.063,
+      "step": 395
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.528117563656734e-05,
+      "loss": 1.0454,
+      "step": 396
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.525729300878681e-05,
+      "loss": 1.075,
+      "step": 397
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5233368868587865e-05,
+      "loss": 0.9866,
+      "step": 398
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5209403404879305e-05,
+      "loss": 0.9807,
+      "step": 399
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5185396806896231e-05,
+      "loss": 1.0476,
+      "step": 400
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5161349264198537e-05,
+      "loss": 0.9956,
+      "step": 401
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5137260966669436e-05,
+      "loss": 1.0674,
+      "step": 402
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.511313210451394e-05,
+      "loss": 1.1304,
+      "step": 403
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5088962868257367e-05,
+      "loss": 1.0388,
+      "step": 404
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5064753448743834e-05,
+      "loss": 0.9661,
+      "step": 405
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5040504037134751e-05,
+      "loss": 1.0066,
+      "step": 406
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5016214824907315e-05,
+      "loss": 1.0923,
+      "step": 407
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4991886003852995e-05,
+      "loss": 1.0835,
+      "step": 408
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4967517766076017e-05,
+      "loss": 1.0581,
+      "step": 409
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4943110303991844e-05,
+      "loss": 1.0725,
+      "step": 410
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.491866381032566e-05,
+      "loss": 1.0342,
+      "step": 411
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4894178478110856e-05,
+      "loss": 0.9751,
+      "step": 412
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4869654500687494e-05,
+      "loss": 1.0085,
+      "step": 413
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4845092071700786e-05,
+      "loss": 1.0464,
+      "step": 414
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4820491385099555e-05,
+      "loss": 1.0139,
+      "step": 415
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4795852635134727e-05,
+      "loss": 1.0356,
+      "step": 416
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4771176016357771e-05,
+      "loss": 0.9988,
+      "step": 417
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4746461723619181e-05,
+      "loss": 1.0818,
+      "step": 418
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4721709952066925e-05,
+      "loss": 1.0771,
+      "step": 419
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4696920897144915e-05,
+      "loss": 1.0322,
+      "step": 420
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.467209475459145e-05,
+      "loss": 0.9885,
+      "step": 421
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4647231720437687e-05,
+      "loss": 0.9424,
+      "step": 422
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4622331991006085e-05,
+      "loss": 1.0488,
+      "step": 423
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4597395762908848e-05,
+      "loss": 1.0747,
+      "step": 424
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4572423233046386e-05,
+      "loss": 1.0698,
+      "step": 425
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4547414598605752e-05,
+      "loss": 1.0366,
+      "step": 426
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4522370057059082e-05,
+      "loss": 1.0295,
+      "step": 427
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.449728980616204e-05,
+      "loss": 0.99,
+      "step": 428
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4472174043952271e-05,
+      "loss": 1.0315,
+      "step": 429
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4447022968747803e-05,
+      "loss": 1.0049,
+      "step": 430
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4421836779145513e-05,
+      "loss": 1.0503,
+      "step": 431
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4396615674019542e-05,
+      "loss": 1.0146,
+      "step": 432
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4371359852519736e-05,
+      "loss": 1.0574,
+      "step": 433
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4346069514070055e-05,
+      "loss": 1.0488,
+      "step": 434
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4320744858367026e-05,
+      "loss": 1.0781,
+      "step": 435
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4295386085378138e-05,
+      "loss": 1.0886,
+      "step": 436
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4269993395340278e-05,
+      "loss": 0.9924,
+      "step": 437
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4244566988758152e-05,
+      "loss": 1.0518,
+      "step": 438
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4219107066402694e-05,
+      "loss": 1.061,
+      "step": 439
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.419361382930948e-05,
+      "loss": 1.0229,
+      "step": 440
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4168087478777156e-05,
+      "loss": 0.9565,
+      "step": 441
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4142528216365822e-05,
+      "loss": 1.0596,
+      "step": 442
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4116936243895467e-05,
+      "loss": 1.0093,
+      "step": 443
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4091311763444358e-05,
+      "loss": 1.0178,
+      "step": 444
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4065654977347452e-05,
+      "loss": 1.0278,
+      "step": 445
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4039966088194802e-05,
+      "loss": 1.0315,
+      "step": 446
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4014245298829936e-05,
+      "loss": 1.0381,
+      "step": 447
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.398849281234829e-05,
+      "loss": 1.0103,
+      "step": 448
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.396270883209557e-05,
+      "loss": 1.0386,
+      "step": 449
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3936893561666171e-05,
+      "loss": 1.0417,
+      "step": 450
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.391104720490156e-05,
+      "loss": 1.0596,
+      "step": 451
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3885169965888667e-05,
+      "loss": 0.9983,
+      "step": 452
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3859262048958261e-05,
+      "loss": 1.0457,
+      "step": 453
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3833323658683364e-05,
+      "loss": 1.0469,
+      "step": 454
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3807354999877615e-05,
+      "loss": 1.0483,
+      "step": 455
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3781356277593659e-05,
+      "loss": 1.1204,
+      "step": 456
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3755327697121523e-05,
+      "loss": 1.0911,
+      "step": 457
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3729269463987002e-05,
+      "loss": 1.0078,
+      "step": 458
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3703181783950032e-05,
+      "loss": 1.0137,
+      "step": 459
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3677064863003069e-05,
+      "loss": 0.9763,
+      "step": 460
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3650918907369455e-05,
+      "loss": 0.9871,
+      "step": 461
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.36247441235018e-05,
+      "loss": 1.071,
+      "step": 462
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3598540718080346e-05,
+      "loss": 1.0234,
+      "step": 463
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3572308898011328e-05,
+      "loss": 1.0239,
+      "step": 464
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3546048870425356e-05,
+      "loss": 1.0051,
+      "step": 465
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3519760842675767e-05,
+      "loss": 1.1313,
+      "step": 466
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3493445022336996e-05,
+      "loss": 1.0583,
+      "step": 467
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3467101617202922e-05,
+      "loss": 1.0515,
+      "step": 468
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.344073083528525e-05,
+      "loss": 1.0442,
+      "step": 469
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3414332884811847e-05,
+      "loss": 1.042,
+      "step": 470
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3387907974225119e-05,
+      "loss": 1.0203,
+      "step": 471
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3361456312180339e-05,
+      "loss": 1.0518,
+      "step": 472
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3334978107544026e-05,
+      "loss": 0.979,
+      "step": 473
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3308473569392274e-05,
+      "loss": 0.948,
+      "step": 474
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3281942907009113e-05,
+      "loss": 1.0366,
+      "step": 475
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3255386329884867e-05,
+      "loss": 0.9912,
+      "step": 476
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3228804047714462e-05,
+      "loss": 1.0652,
+      "step": 477
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.320219627039582e-05,
+      "loss": 0.9214,
+      "step": 478
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3175563208028161e-05,
+      "loss": 1.0095,
+      "step": 479
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3148905070910363e-05,
+      "loss": 1.0063,
+      "step": 480
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.31222220695393e-05,
+      "loss": 1.0339,
+      "step": 481
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3095514414608181e-05,
+      "loss": 0.9858,
+      "step": 482
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3068782317004876e-05,
+      "loss": 1.0576,
+      "step": 483
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3042025987810267e-05,
+      "loss": 0.9526,
+      "step": 484
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3015245638296565e-05,
+      "loss": 1.0522,
+      "step": 485
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2988441479925652e-05,
+      "loss": 0.9885,
+      "step": 486
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.296161372434741e-05,
+      "loss": 1.0396,
+      "step": 487
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2934762583398043e-05,
+      "loss": 1.0791,
+      "step": 488
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2907888269098418e-05,
+      "loss": 1.0459,
+      "step": 489
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2880990993652379e-05,
+      "loss": 1.0562,
+      "step": 490
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2854070969445067e-05,
+      "loss": 0.9473,
+      "step": 491
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2827128409041263e-05,
+      "loss": 0.9646,
+      "step": 492
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2800163525183689e-05,
+      "loss": 1.0681,
+      "step": 493
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2773176530791342e-05,
+      "loss": 0.9941,
+      "step": 494
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2746167638957808e-05,
+      "loss": 1.0916,
+      "step": 495
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2719137062949569e-05,
+      "loss": 1.0344,
+      "step": 496
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2692085016204334e-05,
+      "loss": 1.0349,
+      "step": 497
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2665011712329355e-05,
+      "loss": 0.9993,
+      "step": 498
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2637917365099726e-05,
+      "loss": 1.0315,
+      "step": 499
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2610802188456708e-05,
+      "loss": 1.0107,
+      "step": 500
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2583666396506024e-05,
+      "loss": 0.968,
+      "step": 501
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.255651020351619e-05,
+      "loss": 1.0291,
+      "step": 502
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2529333823916807e-05,
+      "loss": 0.947,
+      "step": 503
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2502137472296879e-05,
+      "loss": 1.0557,
+      "step": 504
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2474921363403096e-05,
+      "loss": 1.0151,
+      "step": 505
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2447685712138175e-05,
+      "loss": 1.0352,
+      "step": 506
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2420430733559126e-05,
+      "loss": 1.0425,
+      "step": 507
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2393156642875579e-05,
+      "loss": 0.981,
+      "step": 508
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2365863655448076e-05,
+      "loss": 1.0457,
+      "step": 509
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2338551986786368e-05,
+      "loss": 1.04,
+      "step": 510
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2311221852547724e-05,
+      "loss": 1.0994,
+      "step": 511
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2283873468535201e-05,
+      "loss": 0.9519,
+      "step": 512
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2256507050695978e-05,
+      "loss": 0.9958,
+      "step": 513
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2229122815119629e-05,
+      "loss": 1.0493,
+      "step": 514
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2201720978036411e-05,
+      "loss": 1.0662,
+      "step": 515
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2174301755815572e-05,
+      "loss": 1.0515,
+      "step": 516
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2146865364963634e-05,
+      "loss": 1.0698,
+      "step": 517
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2119412022122684e-05,
+      "loss": 0.9995,
+      "step": 518
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2091941944068666e-05,
+      "loss": 0.9885,
+      "step": 519
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2064455347709672e-05,
+      "loss": 0.9866,
+      "step": 520
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2036952450084216e-05,
+      "loss": 1.0371,
+      "step": 521
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2009433468359539e-05,
+      "loss": 1.0659,
+      "step": 522
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1981898619829879e-05,
+      "loss": 0.9294,
+      "step": 523
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1954348121914758e-05,
+      "loss": 0.9871,
+      "step": 524
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1926782192157276e-05,
+      "loss": 0.9282,
+      "step": 525
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1899201048222377e-05,
+      "loss": 0.9756,
+      "step": 526
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1871604907895149e-05,
+      "loss": 1.043,
+      "step": 527
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1843993989079081e-05,
+      "loss": 0.9841,
+      "step": 528
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1816368509794365e-05,
+      "loss": 1.0935,
+      "step": 529
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1788728688176163e-05,
+      "loss": 1.021,
+      "step": 530
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1761074742472882e-05,
+      "loss": 0.9553,
+      "step": 531
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.173340689104446e-05,
+      "loss": 1.0793,
+      "step": 532
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1705725352360635e-05,
+      "loss": 0.9861,
+      "step": 533
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.167803034499922e-05,
+      "loss": 1.0327,
+      "step": 534
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.165032208764438e-05,
+      "loss": 1.0596,
+      "step": 535
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1622600799084907e-05,
+      "loss": 0.9512,
+      "step": 536
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1594866698212484e-05,
+      "loss": 1.052,
+      "step": 537
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1567120004019967e-05,
+      "loss": 1.0347,
+      "step": 538
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1539360935599646e-05,
+      "loss": 0.9924,
+      "step": 539
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1511589712141527e-05,
+      "loss": 1.0042,
+      "step": 540
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1483806552931585e-05,
+      "loss": 1.0354,
+      "step": 541
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1456011677350052e-05,
+      "loss": 1.0518,
+      "step": 542
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1428205304869662e-05,
+      "loss": 1.0835,
+      "step": 543
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1400387655053942e-05,
+      "loss": 1.0898,
+      "step": 544
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1372558947555457e-05,
+      "loss": 1.1426,
+      "step": 545
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1344719402114093e-05,
+      "loss": 1.0334,
+      "step": 546
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1316869238555312e-05,
+      "loss": 0.9934,
+      "step": 547
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1289008676788415e-05,
+      "loss": 1.0647,
+      "step": 548
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1261137936804813e-05,
+      "loss": 1.0195,
+      "step": 549
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.123325723867628e-05,
+      "loss": 0.9617,
+      "step": 550
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1205366802553231e-05,
+      "loss": 1.053,
+      "step": 551
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1177466848662965e-05,
+      "loss": 1.0945,
+      "step": 552
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1149557597307936e-05,
+      "loss": 1.0117,
+      "step": 553
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1121639268864014e-05,
+      "loss": 1.0042,
+      "step": 554
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1093712083778748e-05,
+      "loss": 1.0532,
+      "step": 555
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1065776262569605e-05,
+      "loss": 1.0527,
+      "step": 556
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1037832025822266e-05,
+      "loss": 0.9573,
+      "step": 557
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1009879594188844e-05,
+      "loss": 0.9678,
+      "step": 558
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0981919188386171e-05,
+      "loss": 1.0854,
+      "step": 559
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0953951029194039e-05,
+      "loss": 1.0378,
+      "step": 560
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0925975337453462e-05,
+      "loss": 1.0408,
+      "step": 561
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0897992334064937e-05,
+      "loss": 1.0085,
+      "step": 562
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0870002239986689e-05,
+      "loss": 1.0552,
+      "step": 563
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0842005276232932e-05,
+      "loss": 1.071,
+      "step": 564
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0814001663872133e-05,
+      "loss": 0.9241,
+      "step": 565
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0785991624025249e-05,
+      "loss": 1.0022,
+      "step": 566
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0757975377863989e-05,
+      "loss": 1.0044,
+      "step": 567
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0729953146609076e-05,
+      "loss": 1.146,
+      "step": 568
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0701925151528484e-05,
+      "loss": 0.99,
+      "step": 569
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.06738916139357e-05,
+      "loss": 1.085,
+      "step": 570
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0645852755187984e-05,
+      "loss": 1.043,
+      "step": 571
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.06178087966846e-05,
+      "loss": 1.0857,
+      "step": 572
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0589759959865087e-05,
+      "loss": 1.0271,
+      "step": 573
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0561706466207506e-05,
+      "loss": 1.0212,
+      "step": 574
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0533648537226688e-05,
+      "loss": 0.9705,
+      "step": 575
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0505586394472486e-05,
+      "loss": 1.0334,
+      "step": 576
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0477520259528023e-05,
+      "loss": 1.0024,
+      "step": 577
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0449450354007942e-05,
+      "loss": 1.0193,
+      "step": 578
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0421376899556668e-05,
+      "loss": 1.0015,
+      "step": 579
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0393300117846648e-05,
+      "loss": 0.9795,
+      "step": 580
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0365220230576592e-05,
+      "loss": 1.021,
+      "step": 581
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.033713745946974e-05,
+      "loss": 1.0593,
+      "step": 582
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0309052026272099e-05,
+      "loss": 1.0217,
+      "step": 583
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0280964152750701e-05,
+      "loss": 0.988,
+      "step": 584
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0252874060691847e-05,
+      "loss": 1.0173,
+      "step": 585
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0224781971899353e-05,
+      "loss": 1.0442,
+      "step": 586
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0196688108192798e-05,
+      "loss": 1.0532,
+      "step": 587
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0168592691405782e-05,
+      "loss": 0.9502,
+      "step": 588
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.014049594338417e-05,
+      "loss": 0.9663,
+      "step": 589
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0112398085984333e-05,
+      "loss": 1.0425,
+      "step": 590
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0084299341071404e-05,
+      "loss": 1.0115,
+      "step": 591
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0056199930517525e-05,
+      "loss": 0.9739,
+      "step": 592
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0028100076200088e-05,
+      "loss": 0.9939,
+      "step": 593
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1e-05,
+      "loss": 1.0388,
+      "step": 594
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.971899923799915e-06,
+      "loss": 1.0813,
+      "step": 595
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.943800069482478e-06,
+      "loss": 1.0505,
+      "step": 596
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.915700658928599e-06,
+      "loss": 1.0056,
+      "step": 597
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.887601914015668e-06,
+      "loss": 0.9768,
+      "step": 598
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.859504056615832e-06,
+      "loss": 1.0251,
+      "step": 599
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.831407308594216e-06,
+      "loss": 0.9771,
+      "step": 600
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.803311891807205e-06,
+      "loss": 1.0242,
+      "step": 601
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.77521802810065e-06,
+      "loss": 1.0601,
+      "step": 602
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.747125939308154e-06,
+      "loss": 1.0339,
+      "step": 603
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.719035847249304e-06,
+      "loss": 1.0107,
+      "step": 604
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.690947973727905e-06,
+      "loss": 1.0757,
+      "step": 605
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.662862540530267e-06,
+      "loss": 1.0317,
+      "step": 606
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.634779769423412e-06,
+      "loss": 0.9563,
+      "step": 607
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.606699882153357e-06,
+      "loss": 0.9792,
+      "step": 608
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.578623100443333e-06,
+      "loss": 0.9741,
+      "step": 609
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.550549645992061e-06,
+      "loss": 0.9658,
+      "step": 610
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.522479740471979e-06,
+      "loss": 1.054,
+      "step": 611
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.494413605527515e-06,
+      "loss": 1.042,
+      "step": 612
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.466351462773314e-06,
+      "loss": 0.9871,
+      "step": 613
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.438293533792496e-06,
+      "loss": 1.0281,
+      "step": 614
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.410240040134918e-06,
+      "loss": 1.0195,
+      "step": 615
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.382191203315404e-06,
+      "loss": 1.0254,
+      "step": 616
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.354147244812021e-06,
+      "loss": 0.9915,
+      "step": 617
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.326108386064303e-06,
+      "loss": 0.9995,
+      "step": 618
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.298074848471521e-06,
+      "loss": 1.0332,
+      "step": 619
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.270046853390924e-06,
+      "loss": 1.0024,
+      "step": 620
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.242024622136013e-06,
+      "loss": 0.9937,
+      "step": 621
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.214008375974751e-06,
+      "loss": 1.0186,
+      "step": 622
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.185998336127869e-06,
+      "loss": 0.9724,
+      "step": 623
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.157994723767071e-06,
+      "loss": 1.0159,
+      "step": 624
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.129997760013316e-06,
+      "loss": 1.0488,
+      "step": 625
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.102007665935068e-06,
+      "loss": 0.9922,
+      "step": 626
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.07402466254654e-06,
+      "loss": 0.9331,
+      "step": 627
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.046048970805964e-06,
+      "loss": 1.0439,
+      "step": 628
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.018080811613832e-06,
+      "loss": 1.0046,
+      "step": 629
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.99012040581116e-06,
+      "loss": 0.9846,
+      "step": 630
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.962167974177736e-06,
+      "loss": 1.0767,
+      "step": 631
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.934223737430397e-06,
+      "loss": 0.998,
+      "step": 632
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.906287916221259e-06,
+      "loss": 1.0017,
+      "step": 633
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.878360731135987e-06,
+      "loss": 1.0249,
+      "step": 634
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.850442402692068e-06,
+      "loss": 0.9893,
+      "step": 635
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.822533151337038e-06,
+      "loss": 1.0022,
+      "step": 636
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.79463319744677e-06,
+      "loss": 1.0227,
+      "step": 637
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.766742761323718e-06,
+      "loss": 1.0608,
+      "step": 638
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.73886206319519e-06,
+      "loss": 0.9885,
+      "step": 639
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.710991323211586e-06,
+      "loss": 0.9792,
+      "step": 640
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.683130761444691e-06,
+      "loss": 0.9397,
+      "step": 641
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.65528059788591e-06,
+      "loss": 1.0037,
+      "step": 642
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.627441052444545e-06,
+      "loss": 0.9985,
+      "step": 643
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.599612344946063e-06,
+      "loss": 0.9902,
+      "step": 644
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.571794695130343e-06,
+      "loss": 1.0923,
+      "step": 645
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.543988322649954e-06,
+      "loss": 0.9763,
+      "step": 646
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.516193447068415e-06,
+      "loss": 0.9783,
+      "step": 647
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.488410287858478e-06,
+      "loss": 1.02,
+      "step": 648
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.460639064400354e-06,
+      "loss": 1.0127,
+      "step": 649
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.432879995980036e-06,
+      "loss": 1.0962,
+      "step": 650
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.405133301787521e-06,
+      "loss": 1.0083,
+      "step": 651
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.377399200915098e-06,
+      "loss": 1.0405,
+      "step": 652
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.349677912355624e-06,
+      "loss": 0.9277,
+      "step": 653
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.321969655000782e-06,
+      "loss": 1.0515,
+      "step": 654
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.294274647639369e-06,
+      "loss": 1.0776,
+      "step": 655
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.266593108955541e-06,
+      "loss": 1.0229,
+      "step": 656
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.238925257527122e-06,
+      "loss": 1.0066,
+      "step": 657
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.211271311823839e-06,
+      "loss": 1.0525,
+      "step": 658
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.183631490205636e-06,
+      "loss": 1.0593,
+      "step": 659
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.156006010920919e-06,
+      "loss": 1.031,
+      "step": 660
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.128395092104855e-06,
+      "loss": 1.0059,
+      "step": 661
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.100798951777626e-06,
+      "loss": 0.9568,
+      "step": 662
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.073217807842727e-06,
+      "loss": 1.064,
+      "step": 663
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.045651878085245e-06,
+      "loss": 1.0803,
+      "step": 664
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.018101380170123e-06,
+      "loss": 1.0181,
+      "step": 665
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.990566531640463e-06,
+      "loss": 0.9819,
+      "step": 666
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.963047549915785e-06,
+      "loss": 1.0,
+      "step": 667
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.935544652290331e-06,
+      "loss": 1.0139,
+      "step": 668
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.908058055931334e-06,
+      "loss": 1.0049,
+      "step": 669
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.88058797787732e-06,
+      "loss": 1.0081,
+      "step": 670
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.853134635036371e-06,
+      "loss": 1.0356,
+      "step": 671
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.825698244184432e-06,
+      "loss": 1.0659,
+      "step": 672
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.798279021963594e-06,
+      "loss": 1.0735,
+      "step": 673
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.770877184880373e-06,
+      "loss": 0.9844,
+      "step": 674
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.743492949304023e-06,
+      "loss": 1.0593,
+      "step": 675
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.716126531464799e-06,
+      "loss": 1.0364,
+      "step": 676
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.688778147452281e-06,
+      "loss": 1.0554,
+      "step": 677
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.66144801321363e-06,
+      "loss": 0.9734,
+      "step": 678
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.634136344551926e-06,
+      "loss": 1.0322,
+      "step": 679
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.606843357124426e-06,
+      "loss": 0.9641,
+      "step": 680
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.579569266440877e-06,
+      "loss": 1.0305,
+      "step": 681
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.55231428786183e-06,
+      "loss": 0.9995,
+      "step": 682
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.525078636596906e-06,
+      "loss": 1.0486,
+      "step": 683
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.497862527703127e-06,
+      "loss": 0.9709,
+      "step": 684
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.470666176083193e-06,
+      "loss": 1.041,
+      "step": 685
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.443489796483812e-06,
+      "loss": 1.0251,
+      "step": 686
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.416333603493977e-06,
+      "loss": 1.0552,
+      "step": 687
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.389197811543296e-06,
+      "loss": 1.0415,
+      "step": 688
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.362082634900276e-06,
+      "loss": 0.9976,
+      "step": 689
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.334988287670648e-06,
+      "loss": 1.0203,
+      "step": 690
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3079149837956695e-06,
+      "loss": 1.0769,
+      "step": 691
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.280862937050435e-06,
+      "loss": 0.9834,
+      "step": 692
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2538323610421965e-06,
+      "loss": 0.9805,
+      "step": 693
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.226823469208659e-06,
+      "loss": 0.9604,
+      "step": 694
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.199836474816313e-06,
+      "loss": 0.9636,
+      "step": 695
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.172871590958739e-06,
+      "loss": 1.0474,
+      "step": 696
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.145929030554936e-06,
+      "loss": 1.0105,
+      "step": 697
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.119009006347625e-06,
+      "loss": 1.0569,
+      "step": 698
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0921117309015855e-06,
+      "loss": 1.0249,
+      "step": 699
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0652374166019625e-06,
+      "loss": 1.0552,
+      "step": 700
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.038386275652594e-06,
+      "loss": 1.0166,
+      "step": 701
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.011558520074352e-06,
+      "loss": 1.0527,
+      "step": 702
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.984754361703437e-06,
+      "loss": 1.0222,
+      "step": 703
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.957974012189734e-06,
+      "loss": 0.9578,
+      "step": 704
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.931217682995124e-06,
+      "loss": 0.9817,
+      "step": 705
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9044855853918224e-06,
+      "loss": 1.0593,
+      "step": 706
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.877777930460701e-06,
+      "loss": 1.0132,
+      "step": 707
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.85109492908964e-06,
+      "loss": 1.0442,
+      "step": 708
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8244367919718445e-06,
+      "loss": 1.0015,
+      "step": 709
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.797803729604183e-06,
+      "loss": 1.0132,
+      "step": 710
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.771195952285541e-06,
+      "loss": 1.0242,
+      "step": 711
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.744613670115137e-06,
+      "loss": 1.0457,
+      "step": 712
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.718057092990889e-06,
+      "loss": 0.9917,
+      "step": 713
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.691526430607729e-06,
+      "loss": 1.0896,
+      "step": 714
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.665021892455978e-06,
+      "loss": 1.0515,
+      "step": 715
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.638543687819663e-06,
+      "loss": 1.0149,
+      "step": 716
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.612092025774885e-06,
+      "loss": 1.0359,
+      "step": 717
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.585667115188157e-06,
+      "loss": 0.9995,
+      "step": 718
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.559269164714753e-06,
+      "loss": 1.0759,
+      "step": 719
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.532898382797082e-06,
+      "loss": 0.9622,
+      "step": 720
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.5065549776630076e-06,
+      "loss": 0.947,
+      "step": 721
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.480239157324235e-06,
+      "loss": 1.0486,
+      "step": 722
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.453951129574644e-06,
+      "loss": 0.9844,
+      "step": 723
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.427691101988673e-06,
+      "loss": 0.9631,
+      "step": 724
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.401459281919656e-06,
+      "loss": 1.0139,
+      "step": 725
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.375255876498202e-06,
+      "loss": 1.0559,
+      "step": 726
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3490810926305495e-06,
+      "loss": 1.0483,
+      "step": 727
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.322935136996935e-06,
+      "loss": 1.0129,
+      "step": 728
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.296818216049973e-06,
+      "loss": 1.0303,
+      "step": 729
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.270730536013e-06,
+      "loss": 0.9419,
+      "step": 730
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.244672302878479e-06,
+      "loss": 1.0647,
+      "step": 731
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.218643722406342e-06,
+      "loss": 0.9692,
+      "step": 732
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.192645000122386e-06,
+      "loss": 1.0874,
+      "step": 733
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1666763413166355e-06,
+      "loss": 1.0835,
+      "step": 734
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.140737951041742e-06,
+      "loss": 0.9558,
+      "step": 735
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.114830034111336e-06,
+      "loss": 0.9614,
+      "step": 736
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.088952795098442e-06,
+      "loss": 1.0605,
+      "step": 737
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.063106438333832e-06,
+      "loss": 1.0669,
+      "step": 738
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0372911679044334e-06,
+      "loss": 1.0383,
+      "step": 739
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.011507187651715e-06,
+      "loss": 1.0378,
+      "step": 740
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.985754701170066e-06,
+      "loss": 0.9375,
+      "step": 741
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.960033911805202e-06,
+      "loss": 1.0474,
+      "step": 742
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9343450226525475e-06,
+      "loss": 0.9084,
+      "step": 743
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.908688236555644e-06,
+      "loss": 1.061,
+      "step": 744
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.883063756104535e-06,
+      "loss": 1.0142,
+      "step": 745
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.857471783634181e-06,
+      "loss": 1.0049,
+      "step": 746
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.831912521222849e-06,
+      "loss": 1.0134,
+      "step": 747
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.806386170690519e-06,
+      "loss": 0.9727,
+      "step": 748
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.780892933597309e-06,
+      "loss": 1.0059,
+      "step": 749
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.755433011241851e-06,
+      "loss": 1.033,
+      "step": 750
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.730006604659726e-06,
+      "loss": 0.9565,
+      "step": 751
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.704613914621865e-06,
+      "loss": 1.0674,
+      "step": 752
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.679255141632977e-06,
+      "loss": 1.0327,
+      "step": 753
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.653930485929944e-06,
+      "loss": 1.0076,
+      "step": 754
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6286401474802675e-06,
+      "loss": 1.0049,
+      "step": 755
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.603384325980461e-06,
+      "loss": 0.959,
+      "step": 756
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.578163220854489e-06,
+      "loss": 1.0186,
+      "step": 757
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5529770312522e-06,
+      "loss": 0.9846,
+      "step": 758
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.52782595604773e-06,
+      "loss": 0.991,
+      "step": 759
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.502710193837959e-06,
+      "loss": 1.0034,
+      "step": 760
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.477629942940923e-06,
+      "loss": 0.9761,
+      "step": 761
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.452585401394254e-06,
+      "loss": 0.9968,
+      "step": 762
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.427576766953615e-06,
+      "loss": 0.9243,
+      "step": 763
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.402604237091156e-06,
+      "loss": 0.968,
+      "step": 764
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.377668008993921e-06,
+      "loss": 0.9634,
+      "step": 765
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.352768279562315e-06,
+      "loss": 0.9963,
+      "step": 766
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.327905245408556e-06,
+      "loss": 1.0508,
+      "step": 767
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.303079102855089e-06,
+      "loss": 0.9851,
+      "step": 768
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.278290047933077e-06,
+      "loss": 1.0676,
+      "step": 769
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2535382763808185e-06,
+      "loss": 1.0481,
+      "step": 770
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.228823983642228e-06,
+      "loss": 1.0837,
+      "step": 771
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.204147364865275e-06,
+      "loss": 1.0349,
+      "step": 772
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.179508614900448e-06,
+      "loss": 1.009,
+      "step": 773
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.154907928299217e-06,
+      "loss": 1.0049,
+      "step": 774
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1303454993125075e-06,
+      "loss": 1.0137,
+      "step": 775
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.105821521889147e-06,
+      "loss": 1.0244,
+      "step": 776
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.081336189674342e-06,
+      "loss": 0.9631,
+      "step": 777
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.056889696008162e-06,
+      "loss": 1.0525,
+      "step": 778
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.032482233923984e-06,
+      "loss": 1.001,
+      "step": 779
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.008113996147006e-06,
+      "loss": 1.0715,
+      "step": 780
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.983785175092684e-06,
+      "loss": 0.9241,
+      "step": 781
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.959495962865249e-06,
+      "loss": 1.0273,
+      "step": 782
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.93524655125617e-06,
+      "loss": 1.0122,
+      "step": 783
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.911037131742637e-06,
+      "loss": 0.9766,
+      "step": 784
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.886867895486065e-06,
+      "loss": 1.0039,
+      "step": 785
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.862739033330566e-06,
+      "loss": 1.0706,
+      "step": 786
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8386507358014664e-06,
+      "loss": 0.9998,
+      "step": 787
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.814603193103773e-06,
+      "loss": 1.01,
+      "step": 788
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.790596595120699e-06,
+      "loss": 1.0112,
+      "step": 789
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.766631131412136e-06,
+      "loss": 1.1001,
+      "step": 790
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.742706991213191e-06,
+      "loss": 0.9675,
+      "step": 791
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.718824363432663e-06,
+      "loss": 1.02,
+      "step": 792
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.694983436651559e-06,
+      "loss": 1.0503,
+      "step": 793
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.671184399121619e-06,
+      "loss": 1.009,
+      "step": 794
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.647427438763806e-06,
+      "loss": 0.9844,
+      "step": 795
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.623712743166851e-06,
+      "loss": 0.99,
+      "step": 796
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.60004049958574e-06,
+      "loss": 1.0264,
+      "step": 797
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.576410894940265e-06,
+      "loss": 1.0376,
+      "step": 798
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.552824115813522e-06,
+      "loss": 1.0203,
+      "step": 799
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.52928034845046e-06,
+      "loss": 0.9749,
+      "step": 800
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5057797787563905e-06,
+      "loss": 1.0251,
+      "step": 801
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.482322592295541e-06,
+      "loss": 1.0049,
+      "step": 802
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.458908974289568e-06,
+      "loss": 0.9951,
+      "step": 803
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4355391096161085e-06,
+      "loss": 1.054,
+      "step": 804
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.412213182807315e-06,
+      "loss": 1.0103,
+      "step": 805
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3889313780484e-06,
+      "loss": 0.9514,
+      "step": 806
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.365693879176186e-06,
+      "loss": 0.9802,
+      "step": 807
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.342500869677638e-06,
+      "loss": 1.0769,
+      "step": 808
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.319352532688444e-06,
+      "loss": 0.9949,
+      "step": 809
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.296249050991529e-06,
+      "loss": 0.9888,
+      "step": 810
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.273190607015652e-06,
+      "loss": 1.0615,
+      "step": 811
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.250177382833937e-06,
+      "loss": 1.0408,
+      "step": 812
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.227209560162443e-06,
+      "loss": 0.9897,
+      "step": 813
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.204287320358744e-06,
+      "loss": 1.1174,
+      "step": 814
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.181410844420473e-06,
+      "loss": 1.0632,
+      "step": 815
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.158580312983908e-06,
+      "loss": 1.0242,
+      "step": 816
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.135795906322539e-06,
+      "loss": 0.9636,
+      "step": 817
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.113057804345656e-06,
+      "loss": 1.0369,
+      "step": 818
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.090366186596907e-06,
+      "loss": 0.9707,
+      "step": 819
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.067721232252909e-06,
+      "loss": 1.0117,
+      "step": 820
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0451231201217996e-06,
+      "loss": 1.0427,
+      "step": 821
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.022572028641861e-06,
+      "loss": 0.9905,
+      "step": 822
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.00006813588008e-06,
+      "loss": 0.9509,
+      "step": 823
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.977611619530755e-06,
+      "loss": 0.9539,
+      "step": 824
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.955202656914102e-06,
+      "loss": 1.0295,
+      "step": 825
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.932841424974838e-06,
+      "loss": 0.9878,
+      "step": 826
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.910528100280793e-06,
+      "loss": 1.1384,
+      "step": 827
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.888262859021508e-06,
+      "loss": 1.04,
+      "step": 828
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.866045877006863e-06,
+      "loss": 0.9949,
+      "step": 829
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.843877329665662e-06,
+      "loss": 1.0427,
+      "step": 830
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.821757392044271e-06,
+      "loss": 1.0637,
+      "step": 831
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7996862388052192e-06,
+      "loss": 0.9573,
+      "step": 832
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.777664044225825e-06,
+      "loss": 1.0449,
+      "step": 833
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7556909821968335e-06,
+      "loss": 1.0645,
+      "step": 834
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7337672262210155e-06,
+      "loss": 1.0779,
+      "step": 835
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7118929494118284e-06,
+      "loss": 1.0088,
+      "step": 836
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.690068324492023e-06,
+      "loss": 1.0193,
+      "step": 837
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6682935237922956e-06,
+      "loss": 1.0393,
+      "step": 838
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6465687192499176e-06,
+      "loss": 0.9873,
+      "step": 839
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6248940824073955e-06,
+      "loss": 1.0222,
+      "step": 840
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6032697844110896e-06,
+      "loss": 1.0176,
+      "step": 841
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5816959960098817e-06,
+      "loss": 1.021,
+      "step": 842
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.560172887553827e-06,
+      "loss": 0.9819,
+      "step": 843
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.538700628992794e-06,
+      "loss": 0.9954,
+      "step": 844
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5172793898751446e-06,
+      "loss": 1.0154,
+      "step": 845
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4959093393463716e-06,
+      "loss": 1.0442,
+      "step": 846
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.474590646147781e-06,
+      "loss": 0.9932,
+      "step": 847
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.453323478615146e-06,
+      "loss": 0.9846,
+      "step": 848
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4321080046773993e-06,
+      "loss": 1.0425,
+      "step": 849
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.410944391855279e-06,
+      "loss": 0.9607,
+      "step": 850
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3898328072600216e-06,
+      "loss": 0.9739,
+      "step": 851
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3687734175920505e-06,
+      "loss": 1.0146,
+      "step": 852
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.347766389139635e-06,
+      "loss": 1.0002,
+      "step": 853
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.326811887777607e-06,
+      "loss": 1.0903,
+      "step": 854
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3059100789660215e-06,
+      "loss": 0.9775,
+      "step": 855
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.28506112774888e-06,
+      "loss": 1.0293,
+      "step": 856
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2642651987527984e-06,
+      "loss": 1.032,
+      "step": 857
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.243522456185728e-06,
+      "loss": 0.9722,
+      "step": 858
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2228330638356463e-06,
+      "loss": 0.9932,
+      "step": 859
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.202197185069279e-06,
+      "loss": 1.0,
+      "step": 860
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.181614982830792e-06,
+      "loss": 1.0168,
+      "step": 861
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.161086619640509e-06,
+      "loss": 1.0117,
+      "step": 862
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1406122575936447e-06,
+      "loss": 1.0879,
+      "step": 863
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1201920583590005e-06,
+      "loss": 1.0134,
+      "step": 864
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.09982618317771e-06,
+      "loss": 0.9841,
+      "step": 865
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.079514792861943e-06,
+      "loss": 1.0176,
+      "step": 866
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0592580477936606e-06,
+      "loss": 1.0562,
+      "step": 867
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0390561079233294e-06,
+      "loss": 1.0281,
+      "step": 868
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.018909132768666e-06,
+      "loss": 1.0627,
+      "step": 869
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9988172814133774e-06,
+      "loss": 1.019,
+      "step": 870
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.978780712505902e-06,
+      "loss": 1.0093,
+      "step": 871
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9587995842581673e-06,
+      "loss": 1.1165,
+      "step": 872
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.938874054444323e-06,
+      "loss": 0.9556,
+      "step": 873
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9190042803995135e-06,
+      "loss": 1.0452,
+      "step": 874
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.899190419018616e-06,
+      "loss": 0.9507,
+      "step": 875
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8794326267550243e-06,
+      "loss": 1.0295,
+      "step": 876
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.859731059619393e-06,
+      "loss": 1.0913,
+      "step": 877
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.84008587317842e-06,
+      "loss": 0.9846,
+      "step": 878
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8204972225536066e-06,
+      "loss": 0.9639,
+      "step": 879
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8009652624200436e-06,
+      "loss": 1.0225,
+      "step": 880
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7814901470051804e-06,
+      "loss": 0.9424,
+      "step": 881
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.76207203008761e-06,
+      "loss": 1.0237,
+      "step": 882
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7427110649958676e-06,
+      "loss": 1.0295,
+      "step": 883
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7234074046071933e-06,
+      "loss": 1.0701,
+      "step": 884
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7041612013463535e-06,
+      "loss": 1.05,
+      "step": 885
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6849726071844117e-06,
+      "loss": 0.9771,
+      "step": 886
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6658417736375543e-06,
+      "loss": 0.9609,
+      "step": 887
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6467688517658685e-06,
+      "loss": 0.9661,
+      "step": 888
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6277539921721663e-06,
+      "loss": 1.0137,
+      "step": 889
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6087973450007953e-06,
+      "loss": 1.032,
+      "step": 890
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5898990599364416e-06,
+      "loss": 0.9988,
+      "step": 891
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5710592862029594e-06,
+      "loss": 1.0659,
+      "step": 892
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5522781725621814e-06,
+      "loss": 1.0034,
+      "step": 893
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5335558673127647e-06,
+      "loss": 0.9753,
+      "step": 894
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.514892518288988e-06,
+      "loss": 0.9617,
+      "step": 895
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4962882728596215e-06,
+      "loss": 0.9697,
+      "step": 896
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4777432779267254e-06,
+      "loss": 0.9585,
+      "step": 897
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.459257679924524e-06,
+      "loss": 0.9829,
+      "step": 898
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.440831624818224e-06,
+      "loss": 0.9702,
+      "step": 899
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.422465258102875e-06,
+      "loss": 1.0439,
+      "step": 900
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4041587248022215e-06,
+      "loss": 1.0354,
+      "step": 901
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.385912169467547e-06,
+      "loss": 1.0894,
+      "step": 902
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.367725736176545e-06,
+      "loss": 0.9438,
+      "step": 903
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3495995685321705e-06,
+      "loss": 1.0923,
+      "step": 904
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.331533809661518e-06,
+      "loss": 0.991,
+      "step": 905
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3135286022146785e-06,
+      "loss": 0.9597,
+      "step": 906
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.295584088363627e-06,
+      "loss": 0.9294,
+      "step": 907
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2777004098010834e-06,
+      "loss": 1.0283,
+      "step": 908
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2598777077394063e-06,
+      "loss": 1.0015,
+      "step": 909
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.242116122909478e-06,
+      "loss": 1.0388,
+      "step": 910
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2244157955595835e-06,
+      "loss": 1.0493,
+      "step": 911
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.206776865454312e-06,
+      "loss": 1.0664,
+      "step": 912
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1891994718734477e-06,
+      "loss": 1.052,
+      "step": 913
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.171683753610878e-06,
+      "loss": 0.9695,
+      "step": 914
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.154229848973486e-06,
+      "loss": 0.9036,
+      "step": 915
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1368378957800716e-06,
+      "loss": 1.0188,
+      "step": 916
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1195080313602514e-06,
+      "loss": 0.9944,
+      "step": 917
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.102240392553375e-06,
+      "loss": 1.0051,
+      "step": 918
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.08503511570746e-06,
+      "loss": 1.0322,
+      "step": 919
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0678923366780934e-06,
+      "loss": 1.0974,
+      "step": 920
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0508121908273783e-06,
+      "loss": 0.9736,
+      "step": 921
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.03379481302285e-06,
+      "loss": 1.0142,
+      "step": 922
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.01684033763642e-06,
+      "loss": 1.0315,
+      "step": 923
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.999948898543308e-06,
+      "loss": 0.9631,
+      "step": 924
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.983120629120998e-06,
+      "loss": 0.9888,
+      "step": 925
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9663556622481695e-06,
+      "loss": 1.062,
+      "step": 926
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9496541303036533e-06,
+      "loss": 1.0471,
+      "step": 927
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9330161651653968e-06,
+      "loss": 1.0044,
+      "step": 928
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9164418982094036e-06,
+      "loss": 1.0527,
+      "step": 929
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.899931460308716e-06,
+      "loss": 0.947,
+      "step": 930
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.883484981832362e-06,
+      "loss": 0.9697,
+      "step": 931
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8671025926443464e-06,
+      "loss": 1.0051,
+      "step": 932
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8507844221026061e-06,
+      "loss": 1.0383,
+      "step": 933
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.834530599058001e-06,
+      "loss": 0.9875,
+      "step": 934
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8183412518532907e-06,
+      "loss": 0.9688,
+      "step": 935
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8022165083221298e-06,
+      "loss": 1.0039,
+      "step": 936
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7861564957880451e-06,
+      "loss": 0.9019,
+      "step": 937
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7701613410634367e-06,
+      "loss": 0.9614,
+      "step": 938
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.754231170448586e-06,
+      "loss": 1.0764,
+      "step": 939
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.738366109730638e-06,
+      "loss": 0.97,
+      "step": 940
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7225662841826308e-06,
+      "loss": 0.959,
+      "step": 941
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7068318185624844e-06,
+      "loss": 1.094,
+      "step": 942
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.691162837112038e-06,
+      "loss": 1.0264,
+      "step": 943
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6755594635560468e-06,
+      "loss": 1.0122,
+      "step": 944
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.660021821101222e-06,
+      "loss": 1.0432,
+      "step": 945
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6445500324352493e-06,
+      "loss": 1.0105,
+      "step": 946
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.62914421972582e-06,
+      "loss": 1.0237,
+      "step": 947
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.613804504619676e-06,
+      "loss": 1.0635,
+      "step": 948
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.598531008241635e-06,
+      "loss": 0.9927,
+      "step": 949
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5833238511936467e-06,
+      "loss": 1.0667,
+      "step": 950
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5681831535538283e-06,
+      "loss": 0.9604,
+      "step": 951
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5531090348755307e-06,
+      "loss": 1.02,
+      "step": 952
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5381016141863804e-06,
+      "loss": 1.0374,
+      "step": 953
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5231610099873505e-06,
+      "loss": 0.9695,
+      "step": 954
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5082873402518172e-06,
+      "loss": 1.0249,
+      "step": 955
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4934807224246338e-06,
+      "loss": 1.1211,
+      "step": 956
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4787412734211992e-06,
+      "loss": 0.9634,
+      "step": 957
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4640691096265358e-06,
+      "loss": 0.9915,
+      "step": 958
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4494643468943782e-06,
+      "loss": 1.0056,
+      "step": 959
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4349271005462395e-06,
+      "loss": 0.9849,
+      "step": 960
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4204574853705278e-06,
+      "loss": 1.1167,
+      "step": 961
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.40605561562161e-06,
+      "loss": 0.9712,
+      "step": 962
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3917216050189353e-06,
+      "loss": 1.0349,
+      "step": 963
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.377455566746121e-06,
+      "loss": 0.9871,
+      "step": 964
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3632576134500653e-06,
+      "loss": 1.0083,
+      "step": 965
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3491278572400558e-06,
+      "loss": 1.0356,
+      "step": 966
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3350664096868837e-06,
+      "loss": 1.0723,
+      "step": 967
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3210733818219701e-06,
+      "loss": 1.0642,
+      "step": 968
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3071488841364776e-06,
+      "loss": 0.9946,
+      "step": 969
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2932930265804478e-06,
+      "loss": 0.9568,
+      "step": 970
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.279505918561923e-06,
+      "loss": 1.0325,
+      "step": 971
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2657876689460979e-06,
+      "loss": 0.9956,
+      "step": 972
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.252138386054441e-06,
+      "loss": 1.0369,
+      "step": 973
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2385581776638512e-06,
+      "loss": 0.9453,
+      "step": 974
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.225047151005807e-06,
+      "loss": 1.0088,
+      "step": 975
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2116054127655163e-06,
+      "loss": 1.0088,
+      "step": 976
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1982330690810717e-06,
+      "loss": 1.0913,
+      "step": 977
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1849302255426154e-06,
+      "loss": 1.061,
+      "step": 978
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.171696987191513e-06,
+      "loss": 0.9951,
+      "step": 979
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.158533458519504e-06,
+      "loss": 1.0171,
+      "step": 980
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1454397434679022e-06,
+      "loss": 1.1023,
+      "step": 981
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1324159454267503e-06,
+      "loss": 1.0649,
+      "step": 982
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1194621672340266e-06,
+      "loss": 0.9475,
+      "step": 983
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1065785111748117e-06,
+      "loss": 1.0061,
+      "step": 984
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0937650789804944e-06,
+      "loss": 0.9629,
+      "step": 985
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.081021971827968e-06,
+      "loss": 1.009,
+      "step": 986
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.068349290338825e-06,
+      "loss": 1.0637,
+      "step": 987
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.055747134578563e-06,
+      "loss": 1.0322,
+      "step": 988
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0432156040558018e-06,
+      "loss": 1.0398,
+      "step": 989
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0307547977214916e-06,
+      "loss": 0.9956,
+      "step": 990
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0183648139681324e-06,
+      "loss": 1.0337,
+      "step": 991
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.006045750629001e-06,
+      "loss": 1.0061,
+      "step": 992
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.937977049773717e-07,
+      "loss": 1.0698,
+      "step": 993
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.81620773725751e-07,
+      "loss": 0.9968,
+      "step": 994
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.695150530251207e-07,
+      "loss": 0.9893,
+      "step": 995
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.574806384641666e-07,
+      "loss": 1.0132,
+      "step": 996
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.455176250685338e-07,
+      "loss": 0.9612,
+      "step": 997
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.336261073000707e-07,
+      "loss": 1.0125,
+      "step": 998
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.218061790560862e-07,
+      "loss": 0.9641,
+      "step": 999
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.100579336686033e-07,
+      "loss": 0.9607,
+      "step": 1000
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.983814639036314e-07,
+      "loss": 1.0132,
+      "step": 1001
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.867768619604255e-07,
+      "loss": 1.0193,
+      "step": 1002
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.752442194707577e-07,
+      "loss": 1.0286,
+      "step": 1003
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.637836274982048e-07,
+      "loss": 1.0042,
+      "step": 1004
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.523951765374117e-07,
+      "loss": 1.0698,
+      "step": 1005
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.410789565133948e-07,
+      "loss": 0.9492,
+      "step": 1006
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.298350567808178e-07,
+      "loss": 1.0203,
+      "step": 1007
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.186635661232966e-07,
+      "loss": 1.0742,
+      "step": 1008
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.075645727526893e-07,
+      "loss": 0.9919,
+      "step": 1009
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.965381643084069e-07,
+      "loss": 0.9929,
+      "step": 1010
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.85584427856716e-07,
+      "loss": 1.071,
+      "step": 1011
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.74703449890053e-07,
+      "loss": 1.0488,
+      "step": 1012
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.638953163263485e-07,
+      "loss": 0.9788,
+      "step": 1013
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.531601125083321e-07,
+      "loss": 1.0537,
+      "step": 1014
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.424979232028784e-07,
+      "loss": 1.0117,
+      "step": 1015
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.319088326003187e-07,
+      "loss": 1.0007,
+      "step": 1016
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.213929243137952e-07,
+      "loss": 1.0544,
+      "step": 1017
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.109502813785818e-07,
+      "loss": 1.0378,
+      "step": 1018
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.005809862514412e-07,
+      "loss": 1.0603,
+      "step": 1019
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.902851208099704e-07,
+      "loss": 1.0259,
+      "step": 1020
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.800627663519566e-07,
+      "loss": 1.0308,
+      "step": 1021
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.699140035947271e-07,
+      "loss": 0.9668,
+      "step": 1022
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.598389126745209e-07,
+      "loss": 1.0264,
+      "step": 1023
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.498375731458529e-07,
+      "loss": 1.0559,
+      "step": 1024
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.399100639808819e-07,
+      "loss": 1.0,
+      "step": 1025
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.300564635687978e-07,
+      "loss": 1.0203,
+      "step": 1026
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.202768497151845e-07,
+      "loss": 1.1013,
+      "step": 1027
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.10571299641427e-07,
+      "loss": 1.0269,
+      "step": 1028
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.009398899840835e-07,
+      "loss": 1.0271,
+      "step": 1029
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.913826967942926e-07,
+      "loss": 1.0408,
+      "step": 1030
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.81899795537163e-07,
+      "loss": 1.0208,
+      "step": 1031
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.724912610911848e-07,
+      "loss": 0.9824,
+      "step": 1032
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.6315716774764e-07,
+      "loss": 1.0017,
+      "step": 1033
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.538975892100041e-07,
+      "loss": 1.0381,
+      "step": 1034
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.44712598593381e-07,
+      "loss": 1.0071,
+      "step": 1035
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.35602268423906e-07,
+      "loss": 0.9897,
+      "step": 1036
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.265666706381944e-07,
+      "loss": 1.0293,
+      "step": 1037
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.176058765827541e-07,
+      "loss": 1.0276,
+      "step": 1038
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.087199570134371e-07,
+      "loss": 0.9592,
+      "step": 1039
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.9990898209487e-07,
+      "loss": 1.0813,
+      "step": 1040
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.911730213999078e-07,
+      "loss": 0.9724,
+      "step": 1041
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.825121439090785e-07,
+      "loss": 1.0562,
+      "step": 1042
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7392641801004336e-07,
+      "loss": 1.0535,
+      "step": 1043
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.654159114970547e-07,
+      "loss": 1.031,
+      "step": 1044
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5698069157042e-07,
+      "loss": 1.104,
+      "step": 1045
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4862082483597247e-07,
+      "loss": 1.0066,
+      "step": 1046
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4033637730454103e-07,
+      "loss": 0.9717,
+      "step": 1047
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.321274143914411e-07,
+      "loss": 0.9758,
+      "step": 1048
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2399400091594154e-07,
+      "loss": 0.9536,
+      "step": 1049
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.159362011007606e-07,
+      "loss": 0.9871,
+      "step": 1050
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.079540785715674e-07,
+      "loss": 1.0325,
+      "step": 1051
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.000476963564637e-07,
+      "loss": 1.0242,
+      "step": 1052
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9221711688549493e-07,
+      "loss": 0.9343,
+      "step": 1053
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.844624019901566e-07,
+      "loss": 1.0144,
+      "step": 1054
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.76783612902909e-07,
+      "loss": 0.97,
+      "step": 1055
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6918081025668296e-07,
+      "loss": 0.9417,
+      "step": 1056
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.616540540844171e-07,
+      "loss": 1.01,
+      "step": 1057
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.54203403818566e-07,
+      "loss": 0.9932,
+      "step": 1058
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4682891829064923e-07,
+      "loss": 1.0312,
+      "step": 1059
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.395306557307698e-07,
+      "loss": 1.0083,
+      "step": 1060
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3230867376716324e-07,
+      "loss": 0.9893,
+      "step": 1061
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2516302942574794e-07,
+      "loss": 1.0764,
+      "step": 1062
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.180937791296601e-07,
+      "loss": 0.9656,
+      "step": 1063
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1110097869882063e-07,
+      "loss": 1.0242,
+      "step": 1064
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.041846833494888e-07,
+      "loss": 0.9868,
+      "step": 1065
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.973449476938306e-07,
+      "loss": 0.9468,
+      "step": 1066
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.905818257394799e-07,
+      "loss": 1.0625,
+      "step": 1067
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.838953708891201e-07,
+      "loss": 1.0425,
+      "step": 1068
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.772856359400589e-07,
+      "loss": 1.0366,
+      "step": 1069
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7075267308380726e-07,
+      "loss": 0.9609,
+      "step": 1070
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.64296533905678e-07,
+      "loss": 0.9717,
+      "step": 1071
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.579172693843679e-07,
+      "loss": 1.0051,
+      "step": 1072
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5161492989156046e-07,
+      "loss": 0.9741,
+      "step": 1073
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.45389565191525e-07,
+      "loss": 0.9702,
+      "step": 1074
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.392412244407294e-07,
+      "loss": 1.0527,
+      "step": 1075
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3316995618744343e-07,
+      "loss": 1.02,
+      "step": 1076
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.271758083713671e-07,
+      "loss": 1.0684,
+      "step": 1077
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2125882832323865e-07,
+      "loss": 0.9988,
+      "step": 1078
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1541906276446922e-07,
+      "loss": 1.0349,
+      "step": 1079
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0965655780677441e-07,
+      "loss": 0.9827,
+      "step": 1080
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0397135895180443e-07,
+      "loss": 1.032,
+      "step": 1081
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.983635110907922e-07,
+      "loss": 1.0303,
+      "step": 1082
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9283305850419153e-07,
+      "loss": 1.0459,
+      "step": 1083
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8738004486133166e-07,
+      "loss": 1.0769,
+      "step": 1084
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8200451322007208e-07,
+      "loss": 0.9712,
+      "step": 1085
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.76706506026465e-07,
+      "loss": 1.0562,
+      "step": 1086
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7148606511441123e-07,
+      "loss": 1.0376,
+      "step": 1087
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6634323170533928e-07,
+      "loss": 0.999,
+      "step": 1088
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6127804640787781e-07,
+      "loss": 0.9407,
+      "step": 1089
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.562905492175326e-07,
+      "loss": 0.9724,
+      "step": 1090
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.513807795163702e-07,
+      "loss": 1.0083,
+      "step": 1091
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4654877607271024e-07,
+      "loss": 1.0452,
+      "step": 1092
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.417945770408191e-07,
+      "loss": 0.9324,
+      "step": 1093
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.371182199606036e-07,
+      "loss": 1.1348,
+      "step": 1094
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3251974175731986e-07,
+      "loss": 0.9617,
+      "step": 1095
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2799917874128154e-07,
+      "loss": 0.9348,
+      "step": 1096
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2355656660757e-07,
+      "loss": 1.0181,
+      "step": 1097
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1919194043575333e-07,
+      "loss": 0.9492,
+      "step": 1098
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1490533468961118e-07,
+      "loss": 0.9763,
+      "step": 1099
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1069678321686372e-07,
+      "loss": 0.998,
+      "step": 1100
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0656631924889749e-07,
+      "loss": 1.0322,
+      "step": 1101
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0251397540051334e-07,
+      "loss": 0.9492,
+      "step": 1102
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.853978366965667e-08,
+      "loss": 0.9507,
+      "step": 1103
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.464377543717651e-08,
+      "loss": 0.9966,
+      "step": 1104
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.08259814665724e-08,
+      "loss": 1.0776,
+      "step": 1105
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.7086431903749e-08,
+      "loss": 1.105,
+      "step": 1106
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.342515627678182e-08,
+      "loss": 1.0896,
+      "step": 1107
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.984218349568307e-08,
+      "loss": 0.9819,
+      "step": 1108
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.63375418521739e-08,
+      "loss": 0.9771,
+      "step": 1109
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.291125901946027e-08,
+      "loss": 1.0295,
+      "step": 1110
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.956336205201419e-08,
+      "loss": 1.012,
+      "step": 1111
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.62938773853583e-08,
+      "loss": 0.9995,
+      "step": 1112
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.310283083586499e-08,
+      "loss": 1.0044,
+      "step": 1113
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.999024760054095e-08,
+      "loss": 0.9558,
+      "step": 1114
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.695615225683737e-08,
+      "loss": 1.0264,
+      "step": 1115
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.400056876245119e-08,
+      "loss": 1.0132,
+      "step": 1116
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.1123520455137465e-08,
+      "loss": 1.0432,
+      "step": 1117
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.832503005252731e-08,
+      "loss": 1.012,
+      "step": 1118
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.5605119651941366e-08,
+      "loss": 0.9502,
+      "step": 1119
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.296381073022549e-08,
+      "loss": 1.0549,
+      "step": 1120
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.040112414357311e-08,
+      "loss": 1.0076,
+      "step": 1121
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.791708012736206e-08,
+      "loss": 1.0322,
+      "step": 1122
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.551169829599799e-08,
+      "loss": 0.9912,
+      "step": 1123
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.318499764275784e-08,
+      "loss": 1.0247,
+      "step": 1124
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.0936996539636663e-08,
+      "loss": 1.0461,
+      "step": 1125
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.876771273720991e-08,
+      "loss": 1.0229,
+      "step": 1126
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.667716336448356e-08,
+      "loss": 1.0339,
+      "step": 1127
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4665364928767586e-08,
+      "loss": 1.0017,
+      "step": 1128
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.273233331554159e-08,
+      "loss": 0.9746,
+      "step": 1129
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.087808378832934e-08,
+      "loss": 1.0156,
+      "step": 1130
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9102630988577788e-08,
+      "loss": 1.0215,
+      "step": 1131
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7405988935544906e-08,
+      "loss": 0.9927,
+      "step": 1132
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.578817102618424e-08,
+      "loss": 0.9612,
+      "step": 1133
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.424919003504166e-08,
+      "loss": 0.9956,
+      "step": 1134
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2789058114156538e-08,
+      "loss": 1.0864,
+      "step": 1135
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.140778679296073e-08,
+      "loss": 1.053,
+      "step": 1136
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0105386978191967e-08,
+      "loss": 1.1125,
+      "step": 1137
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.88186895380505e-09,
+      "loss": 0.9888,
+      "step": 1138
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.737242380891907e-09,
+      "loss": 0.9946,
+      "step": 1139
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.671516297606095e-09,
+      "loss": 0.9685,
+      "step": 1140
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.684699119089532e-09,
+      "loss": 0.9658,
+      "step": 1141
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.776798637405877e-09,
+      "loss": 1.0203,
+      "step": 1142
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.9478220214816956e-09,
+      "loss": 0.9607,
+      "step": 1143
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.1977758170476105e-09,
+      "loss": 1.0466,
+      "step": 1144
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5266659465883468e-09,
+      "loss": 1.0073,
+      "step": 1145
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9344977092927707e-09,
+      "loss": 1.041,
+      "step": 1146
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4212757810150301e-09,
+      "loss": 1.0415,
+      "step": 1147
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.870042142368085e-10,
+      "loss": 1.0293,
+      "step": 1148
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.316864380351284e-10,
+      "loss": 1.0383,
+      "step": 1149
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.553252580557054e-10,
+      "loss": 1.0464,
+      "step": 1150
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.57922856489634e-10,
+      "loss": 0.9888,
+      "step": 1151
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.948079205895461e-11,
+      "loss": 1.0371,
+      "step": 1152
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 1.0901,
+      "step": 1153
+    },
+    {
+      "epoch": 1.0,
+      "step": 1153,
+      "total_flos": 5.249372826078544e+18,
+      "train_loss": 1.0561968099522983,
+      "train_runtime": 16638.3115,
+      "train_samples_per_second": 8.873,
+      "train_steps_per_second": 0.069
+    }
+  ],
+  "max_steps": 1153,
+  "num_train_epochs": 1,
+  "total_flos": 5.249372826078544e+18,
+  "trial_name": null,
+  "trial_params": null
+}