diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,92442 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 3.0,
+  "eval_steps": 500,
+  "global_step": 15402,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.319654427645789e-08,
+      "loss": 0.7817,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.639308855291578e-08,
+      "loss": 0.6472,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2958963282937366e-07,
+      "loss": 0.534,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.7278617710583156e-07,
+      "loss": 0.5013,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1598272138228942e-07,
+      "loss": 0.4758,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.591792656587473e-07,
+      "loss": 0.4611,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.0237580993520524e-07,
+      "loss": 0.4274,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.455723542116631e-07,
+      "loss": 0.3732,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.88768898488121e-07,
+      "loss": 0.4325,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.3196544276457885e-07,
+      "loss": 0.376,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.7516198704103677e-07,
+      "loss": 0.3792,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.183585313174946e-07,
+      "loss": 0.399,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.615550755939525e-07,
+      "loss": 0.4054,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.047516198704105e-07,
+      "loss": 0.4164,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.479481641468684e-07,
+      "loss": 0.3744,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.911447084233262e-07,
+      "loss": 0.3496,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.343412526997841e-07,
+      "loss": 0.4041,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.77537796976242e-07,
+      "loss": 0.3426,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.207343412526998e-07,
+      "loss": 0.39,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.639308855291577e-07,
+      "loss": 0.3843,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.071274298056157e-07,
+      "loss": 0.3037,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.503239740820735e-07,
+      "loss": 0.368,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.935205183585315e-07,
+      "loss": 0.3507,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0367170626349893e-06,
+      "loss": 0.3277,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.079913606911447e-06,
+      "loss": 0.3311,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.123110151187905e-06,
+      "loss": 0.3931,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.166306695464363e-06,
+      "loss": 0.3509,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.209503239740821e-06,
+      "loss": 0.3034,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2526997840172786e-06,
+      "loss": 0.323,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2958963282937368e-06,
+      "loss": 0.3772,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3390928725701946e-06,
+      "loss": 0.3438,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3822894168466524e-06,
+      "loss": 0.3216,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4254859611231103e-06,
+      "loss": 0.3622,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4686825053995683e-06,
+      "loss": 0.3426,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.511879049676026e-06,
+      "loss": 0.356,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.555075593952484e-06,
+      "loss": 0.3577,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5982721382289417e-06,
+      "loss": 0.3522,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6414686825053995e-06,
+      "loss": 0.3165,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6846652267818576e-06,
+      "loss": 0.3446,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7278617710583154e-06,
+      "loss": 0.3403,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7710583153347732e-06,
+      "loss": 0.3417,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8142548596112314e-06,
+      "loss": 0.3586,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8574514038876893e-06,
+      "loss": 0.3649,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.900647948164147e-06,
+      "loss": 0.3682,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.943844492440605e-06,
+      "loss": 0.3238,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.987041036717063e-06,
+      "loss": 0.3444,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0302375809935205e-06,
+      "loss": 0.3014,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0734341252699786e-06,
+      "loss": 0.2819,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1166306695464366e-06,
+      "loss": 0.3206,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.159827213822894e-06,
+      "loss": 0.3094,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2030237580993522e-06,
+      "loss": 0.3106,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.24622030237581e-06,
+      "loss": 0.29,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.289416846652268e-06,
+      "loss": 0.3214,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.332613390928726e-06,
+      "loss": 0.3122,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.375809935205184e-06,
+      "loss": 0.345,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.419006479481642e-06,
+      "loss": 0.3356,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4622030237580995e-06,
+      "loss": 0.3549,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.505399568034557e-06,
+      "loss": 0.3457,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.548596112311015e-06,
+      "loss": 0.3655,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5917926565874736e-06,
+      "loss": 0.3447,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6349892008639312e-06,
+      "loss": 0.3347,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6781857451403893e-06,
+      "loss": 0.3676,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.721382289416847e-06,
+      "loss": 0.3032,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.764578833693305e-06,
+      "loss": 0.3679,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8077753779697625e-06,
+      "loss": 0.3176,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8509719222462205e-06,
+      "loss": 0.3194,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.894168466522678e-06,
+      "loss": 0.3447,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9373650107991366e-06,
+      "loss": 0.3086,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.980561555075594e-06,
+      "loss": 0.3345,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.023758099352052e-06,
+      "loss": 0.3221,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.06695464362851e-06,
+      "loss": 0.3264,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.110151187904968e-06,
+      "loss": 0.3237,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.153347732181426e-06,
+      "loss": 0.3725,
+      "step": 73
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.1965442764578835e-06,
+      "loss": 0.2866,
+      "step": 74
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.239740820734342e-06,
+      "loss": 0.3353,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.282937365010799e-06,
+      "loss": 0.3475,
+      "step": 76
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3261339092872576e-06,
+      "loss": 0.3258,
+      "step": 77
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.369330453563715e-06,
+      "loss": 0.3519,
+      "step": 78
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.412526997840173e-06,
+      "loss": 0.3393,
+      "step": 79
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.4557235421166308e-06,
+      "loss": 0.3243,
+      "step": 80
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.498920086393089e-06,
+      "loss": 0.3198,
+      "step": 81
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.5421166306695464e-06,
+      "loss": 0.294,
+      "step": 82
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.5853131749460044e-06,
+      "loss": 0.3316,
+      "step": 83
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.628509719222463e-06,
+      "loss": 0.3644,
+      "step": 84
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.6717062634989205e-06,
+      "loss": 0.3306,
+      "step": 85
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7149028077753785e-06,
+      "loss": 0.3218,
+      "step": 86
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.758099352051836e-06,
+      "loss": 0.3217,
+      "step": 87
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.801295896328294e-06,
+      "loss": 0.3188,
+      "step": 88
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.844492440604752e-06,
+      "loss": 0.3358,
+      "step": 89
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.88768898488121e-06,
+      "loss": 0.3308,
+      "step": 90
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.930885529157668e-06,
+      "loss": 0.3169,
+      "step": 91
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.974082073434126e-06,
+      "loss": 0.3215,
+      "step": 92
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.017278617710583e-06,
+      "loss": 0.2837,
+      "step": 93
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.060475161987041e-06,
+      "loss": 0.3134,
+      "step": 94
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.1036717062635e-06,
+      "loss": 0.3576,
+      "step": 95
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.146868250539957e-06,
+      "loss": 0.3251,
+      "step": 96
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.190064794816415e-06,
+      "loss": 0.3421,
+      "step": 97
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.233261339092873e-06,
+      "loss": 0.2955,
+      "step": 98
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.276457883369331e-06,
+      "loss": 0.3367,
+      "step": 99
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.319654427645788e-06,
+      "loss": 0.2994,
+      "step": 100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.362850971922246e-06,
+      "loss": 0.307,
+      "step": 101
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.4060475161987044e-06,
+      "loss": 0.3285,
+      "step": 102
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.4492440604751625e-06,
+      "loss": 0.3367,
+      "step": 103
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.49244060475162e-06,
+      "loss": 0.3768,
+      "step": 104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.5356371490280785e-06,
+      "loss": 0.2951,
+      "step": 105
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.578833693304536e-06,
+      "loss": 0.3225,
+      "step": 106
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.622030237580994e-06,
+      "loss": 0.339,
+      "step": 107
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.665226781857452e-06,
+      "loss": 0.2694,
+      "step": 108
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.70842332613391e-06,
+      "loss": 0.3058,
+      "step": 109
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.751619870410368e-06,
+      "loss": 0.3174,
+      "step": 110
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.794816414686825e-06,
+      "loss": 0.3176,
+      "step": 111
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.838012958963284e-06,
+      "loss": 0.2937,
+      "step": 112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.881209503239741e-06,
+      "loss": 0.3074,
+      "step": 113
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.924406047516199e-06,
+      "loss": 0.3729,
+      "step": 114
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.967602591792657e-06,
+      "loss": 0.2815,
+      "step": 115
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.010799136069114e-06,
+      "loss": 0.3263,
+      "step": 116
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.053995680345572e-06,
+      "loss": 0.2937,
+      "step": 117
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.09719222462203e-06,
+      "loss": 0.301,
+      "step": 118
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.140388768898489e-06,
+      "loss": 0.3267,
+      "step": 119
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.183585313174947e-06,
+      "loss": 0.2676,
+      "step": 120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.2267818574514036e-06,
+      "loss": 0.3228,
+      "step": 121
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.2699784017278624e-06,
+      "loss": 0.2977,
+      "step": 122
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.3131749460043205e-06,
+      "loss": 0.2886,
+      "step": 123
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.3563714902807785e-06,
+      "loss": 0.3396,
+      "step": 124
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.399568034557236e-06,
+      "loss": 0.3103,
+      "step": 125
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.442764578833694e-06,
+      "loss": 0.4062,
+      "step": 126
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.485961123110152e-06,
+      "loss": 0.3093,
+      "step": 127
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.52915766738661e-06,
+      "loss": 0.3341,
+      "step": 128
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.572354211663067e-06,
+      "loss": 0.3015,
+      "step": 129
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.615550755939525e-06,
+      "loss": 0.3229,
+      "step": 130
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.658747300215983e-06,
+      "loss": 0.3025,
+      "step": 131
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.701943844492441e-06,
+      "loss": 0.2994,
+      "step": 132
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.745140388768899e-06,
+      "loss": 0.3041,
+      "step": 133
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.788336933045356e-06,
+      "loss": 0.314,
+      "step": 134
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.831533477321814e-06,
+      "loss": 0.3247,
+      "step": 135
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.874730021598273e-06,
+      "loss": 0.2851,
+      "step": 136
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.917926565874731e-06,
+      "loss": 0.3209,
+      "step": 137
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.961123110151188e-06,
+      "loss": 0.3213,
+      "step": 138
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.004319654427646e-06,
+      "loss": 0.3225,
+      "step": 139
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.047516198704104e-06,
+      "loss": 0.3311,
+      "step": 140
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.0907127429805624e-06,
+      "loss": 0.2988,
+      "step": 141
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.13390928725702e-06,
+      "loss": 0.3058,
+      "step": 142
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.177105831533478e-06,
+      "loss": 0.3044,
+      "step": 143
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.220302375809936e-06,
+      "loss": 0.3452,
+      "step": 144
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.263498920086394e-06,
+      "loss": 0.3167,
+      "step": 145
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.306695464362852e-06,
+      "loss": 0.3098,
+      "step": 146
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.349892008639309e-06,
+      "loss": 0.2989,
+      "step": 147
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.393088552915767e-06,
+      "loss": 0.2975,
+      "step": 148
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.436285097192225e-06,
+      "loss": 0.2885,
+      "step": 149
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.479481641468684e-06,
+      "loss": 0.3099,
+      "step": 150
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.52267818574514e-06,
+      "loss": 0.3489,
+      "step": 151
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.565874730021598e-06,
+      "loss": 0.3066,
+      "step": 152
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.609071274298057e-06,
+      "loss": 0.3409,
+      "step": 153
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.652267818574515e-06,
+      "loss": 0.311,
+      "step": 154
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.695464362850973e-06,
+      "loss": 0.3112,
+      "step": 155
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.73866090712743e-06,
+      "loss": 0.3491,
+      "step": 156
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.781857451403888e-06,
+      "loss": 0.2901,
+      "step": 157
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.825053995680346e-06,
+      "loss": 0.2969,
+      "step": 158
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.868250539956804e-06,
+      "loss": 0.3094,
+      "step": 159
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.9114470842332616e-06,
+      "loss": 0.2784,
+      "step": 160
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.95464362850972e-06,
+      "loss": 0.3741,
+      "step": 161
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.997840172786178e-06,
+      "loss": 0.2905,
+      "step": 162
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.041036717062636e-06,
+      "loss": 0.3183,
+      "step": 163
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.084233261339093e-06,
+      "loss": 0.2872,
+      "step": 164
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.127429805615551e-06,
+      "loss": 0.3014,
+      "step": 165
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.170626349892009e-06,
+      "loss": 0.3001,
+      "step": 166
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.213822894168468e-06,
+      "loss": 0.3329,
+      "step": 167
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.257019438444926e-06,
+      "loss": 0.3164,
+      "step": 168
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.300215982721382e-06,
+      "loss": 0.3256,
+      "step": 169
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.343412526997841e-06,
+      "loss": 0.3584,
+      "step": 170
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.386609071274299e-06,
+      "loss": 0.3068,
+      "step": 171
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.429805615550757e-06,
+      "loss": 0.2938,
+      "step": 172
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.473002159827214e-06,
+      "loss": 0.2716,
+      "step": 173
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.516198704103672e-06,
+      "loss": 0.3357,
+      "step": 174
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.55939524838013e-06,
+      "loss": 0.3441,
+      "step": 175
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.602591792656588e-06,
+      "loss": 0.3367,
+      "step": 176
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.645788336933045e-06,
+      "loss": 0.314,
+      "step": 177
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.688984881209504e-06,
+      "loss": 0.3094,
+      "step": 178
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.732181425485962e-06,
+      "loss": 0.3085,
+      "step": 179
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.77537796976242e-06,
+      "loss": 0.3284,
+      "step": 180
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.818574514038878e-06,
+      "loss": 0.3088,
+      "step": 181
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.861771058315336e-06,
+      "loss": 0.322,
+      "step": 182
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.904967602591793e-06,
+      "loss": 0.3284,
+      "step": 183
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.948164146868252e-06,
+      "loss": 0.3285,
+      "step": 184
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.991360691144709e-06,
+      "loss": 0.3179,
+      "step": 185
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.034557235421166e-06,
+      "loss": 0.3294,
+      "step": 186
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.077753779697625e-06,
+      "loss": 0.3133,
+      "step": 187
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.120950323974082e-06,
+      "loss": 0.2745,
+      "step": 188
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.164146868250541e-06,
+      "loss": 0.3154,
+      "step": 189
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.207343412527e-06,
+      "loss": 0.3037,
+      "step": 190
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.250539956803455e-06,
+      "loss": 0.3129,
+      "step": 191
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.293736501079914e-06,
+      "loss": 0.305,
+      "step": 192
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.336933045356373e-06,
+      "loss": 0.3164,
+      "step": 193
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.38012958963283e-06,
+      "loss": 0.2995,
+      "step": 194
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.423326133909287e-06,
+      "loss": 0.2995,
+      "step": 195
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.466522678185746e-06,
+      "loss": 0.2821,
+      "step": 196
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.509719222462204e-06,
+      "loss": 0.2831,
+      "step": 197
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.552915766738662e-06,
+      "loss": 0.3216,
+      "step": 198
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.59611231101512e-06,
+      "loss": 0.3141,
+      "step": 199
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.639308855291577e-06,
+      "loss": 0.3051,
+      "step": 200
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.682505399568036e-06,
+      "loss": 0.2946,
+      "step": 201
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.725701943844493e-06,
+      "loss": 0.3043,
+      "step": 202
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.768898488120952e-06,
+      "loss": 0.312,
+      "step": 203
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.812095032397409e-06,
+      "loss": 0.2796,
+      "step": 204
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.855291576673866e-06,
+      "loss": 0.3252,
+      "step": 205
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.898488120950325e-06,
+      "loss": 0.3391,
+      "step": 206
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.941684665226784e-06,
+      "loss": 0.2762,
+      "step": 207
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.98488120950324e-06,
+      "loss": 0.3063,
+      "step": 208
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.028077753779698e-06,
+      "loss": 0.2839,
+      "step": 209
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.071274298056157e-06,
+      "loss": 0.2944,
+      "step": 210
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.114470842332614e-06,
+      "loss": 0.3131,
+      "step": 211
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.157667386609071e-06,
+      "loss": 0.3219,
+      "step": 212
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.20086393088553e-06,
+      "loss": 0.3002,
+      "step": 213
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.244060475161987e-06,
+      "loss": 0.2829,
+      "step": 214
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.287257019438446e-06,
+      "loss": 0.3453,
+      "step": 215
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.330453563714904e-06,
+      "loss": 0.3116,
+      "step": 216
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.37365010799136e-06,
+      "loss": 0.3037,
+      "step": 217
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.41684665226782e-06,
+      "loss": 0.2663,
+      "step": 218
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.460043196544277e-06,
+      "loss": 0.3065,
+      "step": 219
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.503239740820736e-06,
+      "loss": 0.3006,
+      "step": 220
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.546436285097193e-06,
+      "loss": 0.262,
+      "step": 221
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.58963282937365e-06,
+      "loss": 0.3112,
+      "step": 222
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.632829373650109e-06,
+      "loss": 0.3159,
+      "step": 223
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.676025917926568e-06,
+      "loss": 0.3083,
+      "step": 224
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.719222462203023e-06,
+      "loss": 0.3036,
+      "step": 225
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.762419006479482e-06,
+      "loss": 0.2834,
+      "step": 226
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.805615550755941e-06,
+      "loss": 0.3187,
+      "step": 227
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.848812095032398e-06,
+      "loss": 0.3251,
+      "step": 228
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.892008639308857e-06,
+      "loss": 0.3199,
+      "step": 229
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.935205183585314e-06,
+      "loss": 0.2926,
+      "step": 230
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.978401727861771e-06,
+      "loss": 0.2825,
+      "step": 231
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0021598272138229e-05,
+      "loss": 0.3089,
+      "step": 232
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0064794816414687e-05,
+      "loss": 0.3024,
+      "step": 233
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0107991360691145e-05,
+      "loss": 0.2955,
+      "step": 234
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0151187904967603e-05,
+      "loss": 0.2929,
+      "step": 235
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.019438444924406e-05,
+      "loss": 0.2869,
+      "step": 236
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.023758099352052e-05,
+      "loss": 0.2716,
+      "step": 237
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0280777537796978e-05,
+      "loss": 0.2952,
+      "step": 238
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0323974082073436e-05,
+      "loss": 0.2799,
+      "step": 239
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0367170626349894e-05,
+      "loss": 0.3039,
+      "step": 240
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.041036717062635e-05,
+      "loss": 0.2786,
+      "step": 241
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0453563714902807e-05,
+      "loss": 0.28,
+      "step": 242
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0496760259179266e-05,
+      "loss": 0.2918,
+      "step": 243
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0539956803455725e-05,
+      "loss": 0.2745,
+      "step": 244
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0583153347732182e-05,
+      "loss": 0.2572,
+      "step": 245
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0626349892008641e-05,
+      "loss": 0.2675,
+      "step": 246
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0669546436285098e-05,
+      "loss": 0.2793,
+      "step": 247
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0712742980561557e-05,
+      "loss": 0.3024,
+      "step": 248
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0755939524838012e-05,
+      "loss": 0.2937,
+      "step": 249
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0799136069114471e-05,
+      "loss": 0.2935,
+      "step": 250
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0842332613390929e-05,
+      "loss": 0.3682,
+      "step": 251
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0885529157667387e-05,
+      "loss": 0.3166,
+      "step": 252
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0928725701943845e-05,
+      "loss": 0.3059,
+      "step": 253
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0971922246220303e-05,
+      "loss": 0.3218,
+      "step": 254
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1015118790496762e-05,
+      "loss": 0.2855,
+      "step": 255
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.105831533477322e-05,
+      "loss": 0.3287,
+      "step": 256
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1101511879049678e-05,
+      "loss": 0.2929,
+      "step": 257
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1144708423326134e-05,
+      "loss": 0.2709,
+      "step": 258
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1187904967602593e-05,
+      "loss": 0.3358,
+      "step": 259
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.123110151187905e-05,
+      "loss": 0.2706,
+      "step": 260
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1274298056155509e-05,
+      "loss": 0.2785,
+      "step": 261
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1317494600431966e-05,
+      "loss": 0.2831,
+      "step": 262
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1360691144708425e-05,
+      "loss": 0.3577,
+      "step": 263
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1403887688984882e-05,
+      "loss": 0.3083,
+      "step": 264
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1447084233261341e-05,
+      "loss": 0.2966,
+      "step": 265
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1490280777537798e-05,
+      "loss": 0.3,
+      "step": 266
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1533477321814255e-05,
+      "loss": 0.2887,
+      "step": 267
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1576673866090712e-05,
+      "loss": 0.326,
+      "step": 268
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1619870410367171e-05,
+      "loss": 0.2643,
+      "step": 269
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1663066954643629e-05,
+      "loss": 0.3332,
+      "step": 270
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1706263498920087e-05,
+      "loss": 0.2994,
+      "step": 271
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1749460043196546e-05,
+      "loss": 0.2852,
+      "step": 272
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1792656587473003e-05,
+      "loss": 0.3015,
+      "step": 273
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1835853131749462e-05,
+      "loss": 0.2877,
+      "step": 274
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.187904967602592e-05,
+      "loss": 0.2737,
+      "step": 275
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1922246220302377e-05,
+      "loss": 0.3526,
+      "step": 276
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1965442764578834e-05,
+      "loss": 0.3189,
+      "step": 277
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2008639308855293e-05,
+      "loss": 0.2495,
+      "step": 278
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.205183585313175e-05,
+      "loss": 0.3219,
+      "step": 279
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2095032397408209e-05,
+      "loss": 0.2957,
+      "step": 280
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2138228941684666e-05,
+      "loss": 0.3454,
+      "step": 281
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2181425485961125e-05,
+      "loss": 0.3066,
+      "step": 282
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2224622030237582e-05,
+      "loss": 0.3013,
+      "step": 283
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.226781857451404e-05,
+      "loss": 0.2626,
+      "step": 284
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2311015118790496e-05,
+      "loss": 0.2938,
+      "step": 285
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2354211663066955e-05,
+      "loss": 0.3177,
+      "step": 286
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2397408207343412e-05,
+      "loss": 0.3097,
+      "step": 287
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2440604751619871e-05,
+      "loss": 0.289,
+      "step": 288
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.248380129589633e-05,
+      "loss": 0.268,
+      "step": 289
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2526997840172787e-05,
+      "loss": 0.3312,
+      "step": 290
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2570194384449246e-05,
+      "loss": 0.298,
+      "step": 291
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2613390928725703e-05,
+      "loss": 0.2601,
+      "step": 292
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.265658747300216e-05,
+      "loss": 0.2905,
+      "step": 293
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2699784017278618e-05,
+      "loss": 0.3092,
+      "step": 294
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2742980561555077e-05,
+      "loss": 0.292,
+      "step": 295
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2786177105831534e-05,
+      "loss": 0.2784,
+      "step": 296
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2829373650107993e-05,
+      "loss": 0.2662,
+      "step": 297
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.287257019438445e-05,
+      "loss": 0.2764,
+      "step": 298
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2915766738660909e-05,
+      "loss": 0.3297,
+      "step": 299
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2958963282937368e-05,
+      "loss": 0.2949,
+      "step": 300
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3002159827213825e-05,
+      "loss": 0.2868,
+      "step": 301
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.304535637149028e-05,
+      "loss": 0.2561,
+      "step": 302
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.308855291576674e-05,
+      "loss": 0.2789,
+      "step": 303
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3131749460043196e-05,
+      "loss": 0.2751,
+      "step": 304
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3174946004319655e-05,
+      "loss": 0.3088,
+      "step": 305
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3218142548596114e-05,
+      "loss": 0.3,
+      "step": 306
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3261339092872571e-05,
+      "loss": 0.2618,
+      "step": 307
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.330453563714903e-05,
+      "loss": 0.3125,
+      "step": 308
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3347732181425487e-05,
+      "loss": 0.2985,
+      "step": 309
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3390928725701946e-05,
+      "loss": 0.3029,
+      "step": 310
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3434125269978402e-05,
+      "loss": 0.2932,
+      "step": 311
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.347732181425486e-05,
+      "loss": 0.2743,
+      "step": 312
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3520518358531318e-05,
+      "loss": 0.3046,
+      "step": 313
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3563714902807777e-05,
+      "loss": 0.2901,
+      "step": 314
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3606911447084234e-05,
+      "loss": 0.2712,
+      "step": 315
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3650107991360693e-05,
+      "loss": 0.339,
+      "step": 316
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3693304535637152e-05,
+      "loss": 0.2526,
+      "step": 317
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3736501079913609e-05,
+      "loss": 0.288,
+      "step": 318
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3779697624190064e-05,
+      "loss": 0.2575,
+      "step": 319
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3822894168466523e-05,
+      "loss": 0.2993,
+      "step": 320
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.386609071274298e-05,
+      "loss": 0.289,
+      "step": 321
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.390928725701944e-05,
+      "loss": 0.2911,
+      "step": 322
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3952483801295898e-05,
+      "loss": 0.2639,
+      "step": 323
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3995680345572355e-05,
+      "loss": 0.2654,
+      "step": 324
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4038876889848814e-05,
+      "loss": 0.3159,
+      "step": 325
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4082073434125271e-05,
+      "loss": 0.3099,
+      "step": 326
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.412526997840173e-05,
+      "loss": 0.2742,
+      "step": 327
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4168466522678186e-05,
+      "loss": 0.2773,
+      "step": 328
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4211663066954645e-05,
+      "loss": 0.2928,
+      "step": 329
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4254859611231102e-05,
+      "loss": 0.3213,
+      "step": 330
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.429805615550756e-05,
+      "loss": 0.2797,
+      "step": 331
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4341252699784018e-05,
+      "loss": 0.2669,
+      "step": 332
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4384449244060477e-05,
+      "loss": 0.269,
+      "step": 333
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4427645788336936e-05,
+      "loss": 0.2949,
+      "step": 334
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4470842332613393e-05,
+      "loss": 0.344,
+      "step": 335
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4514038876889852e-05,
+      "loss": 0.2785,
+      "step": 336
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4557235421166307e-05,
+      "loss": 0.2694,
+      "step": 337
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4600431965442764e-05,
+      "loss": 0.2615,
+      "step": 338
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4643628509719223e-05,
+      "loss": 0.2826,
+      "step": 339
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4686825053995682e-05,
+      "loss": 0.293,
+      "step": 340
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.473002159827214e-05,
+      "loss": 0.3136,
+      "step": 341
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4773218142548598e-05,
+      "loss": 0.3034,
+      "step": 342
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4816414686825055e-05,
+      "loss": 0.297,
+      "step": 343
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4859611231101514e-05,
+      "loss": 0.3158,
+      "step": 344
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4902807775377971e-05,
+      "loss": 0.3037,
+      "step": 345
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4946004319654428e-05,
+      "loss": 0.2953,
+      "step": 346
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4989200863930886e-05,
+      "loss": 0.3034,
+      "step": 347
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5032397408207345e-05,
+      "loss": 0.3029,
+      "step": 348
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5075593952483802e-05,
+      "loss": 0.2722,
+      "step": 349
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.511879049676026e-05,
+      "loss": 0.2676,
+      "step": 350
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.516198704103672e-05,
+      "loss": 0.2868,
+      "step": 351
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5205183585313177e-05,
+      "loss": 0.3088,
+      "step": 352
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5248380129589636e-05,
+      "loss": 0.2886,
+      "step": 353
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.529157667386609e-05,
+      "loss": 0.2521,
+      "step": 354
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5334773218142548e-05,
+      "loss": 0.3161,
+      "step": 355
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5377969762419007e-05,
+      "loss": 0.2687,
+      "step": 356
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5421166306695466e-05,
+      "loss": 0.2923,
+      "step": 357
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5464362850971925e-05,
+      "loss": 0.2857,
+      "step": 358
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.550755939524838e-05,
+      "loss": 0.226,
+      "step": 359
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.555075593952484e-05,
+      "loss": 0.2972,
+      "step": 360
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5593952483801298e-05,
+      "loss": 0.2985,
+      "step": 361
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5637149028077757e-05,
+      "loss": 0.2576,
+      "step": 362
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5680345572354212e-05,
+      "loss": 0.2487,
+      "step": 363
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.572354211663067e-05,
+      "loss": 0.2907,
+      "step": 364
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5766738660907127e-05,
+      "loss": 0.3517,
+      "step": 365
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5809935205183586e-05,
+      "loss": 0.3293,
+      "step": 366
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5853131749460045e-05,
+      "loss": 0.245,
+      "step": 367
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5896328293736503e-05,
+      "loss": 0.2713,
+      "step": 368
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5939524838012962e-05,
+      "loss": 0.2851,
+      "step": 369
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5982721382289418e-05,
+      "loss": 0.3024,
+      "step": 370
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6025917926565877e-05,
+      "loss": 0.286,
+      "step": 371
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6069114470842332e-05,
+      "loss": 0.2863,
+      "step": 372
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.611231101511879e-05,
+      "loss": 0.3043,
+      "step": 373
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.615550755939525e-05,
+      "loss": 0.2873,
+      "step": 374
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.619870410367171e-05,
+      "loss": 0.3015,
+      "step": 375
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6241900647948164e-05,
+      "loss": 0.2741,
+      "step": 376
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6285097192224623e-05,
+      "loss": 0.2918,
+      "step": 377
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6328293736501082e-05,
+      "loss": 0.2763,
+      "step": 378
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.637149028077754e-05,
+      "loss": 0.2771,
+      "step": 379
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6414686825054e-05,
+      "loss": 0.2957,
+      "step": 380
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6457883369330455e-05,
+      "loss": 0.2559,
+      "step": 381
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.650107991360691e-05,
+      "loss": 0.2503,
+      "step": 382
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.654427645788337e-05,
+      "loss": 0.2842,
+      "step": 383
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.658747300215983e-05,
+      "loss": 0.2628,
+      "step": 384
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6630669546436287e-05,
+      "loss": 0.2786,
+      "step": 385
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6673866090712746e-05,
+      "loss": 0.2908,
+      "step": 386
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.67170626349892e-05,
+      "loss": 0.2881,
+      "step": 387
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.676025917926566e-05,
+      "loss": 0.2696,
+      "step": 388
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6803455723542116e-05,
+      "loss": 0.3137,
+      "step": 389
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6846652267818575e-05,
+      "loss": 0.2904,
+      "step": 390
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6889848812095034e-05,
+      "loss": 0.2809,
+      "step": 391
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6933045356371493e-05,
+      "loss": 0.2981,
+      "step": 392
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6976241900647948e-05,
+      "loss": 0.2816,
+      "step": 393
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7019438444924407e-05,
+      "loss": 0.3113,
+      "step": 394
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7062634989200866e-05,
+      "loss": 0.3186,
+      "step": 395
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7105831533477325e-05,
+      "loss": 0.3065,
+      "step": 396
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7149028077753784e-05,
+      "loss": 0.2722,
+      "step": 397
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.719222462203024e-05,
+      "loss": 0.2597,
+      "step": 398
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7235421166306695e-05,
+      "loss": 0.2561,
+      "step": 399
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7278617710583154e-05,
+      "loss": 0.2533,
+      "step": 400
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7321814254859612e-05,
+      "loss": 0.2665,
+      "step": 401
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.736501079913607e-05,
+      "loss": 0.2861,
+      "step": 402
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.740820734341253e-05,
+      "loss": 0.2734,
+      "step": 403
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7451403887688986e-05,
+      "loss": 0.2683,
+      "step": 404
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7494600431965444e-05,
+      "loss": 0.2596,
+      "step": 405
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7537796976241903e-05,
+      "loss": 0.2883,
+      "step": 406
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.758099352051836e-05,
+      "loss": 0.2774,
+      "step": 407
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7624190064794818e-05,
+      "loss": 0.3136,
+      "step": 408
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7667386609071277e-05,
+      "loss": 0.2812,
+      "step": 409
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7710583153347732e-05,
+      "loss": 0.2782,
+      "step": 410
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.775377969762419e-05,
+      "loss": 0.2782,
+      "step": 411
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.779697624190065e-05,
+      "loss": 0.3086,
+      "step": 412
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.784017278617711e-05,
+      "loss": 0.3047,
+      "step": 413
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7883369330453568e-05,
+      "loss": 0.2695,
+      "step": 414
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7926565874730023e-05,
+      "loss": 0.27,
+      "step": 415
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.796976241900648e-05,
+      "loss": 0.2682,
+      "step": 416
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8012958963282937e-05,
+      "loss": 0.2871,
+      "step": 417
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8056155507559396e-05,
+      "loss": 0.2925,
+      "step": 418
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8099352051835855e-05,
+      "loss": 0.2625,
+      "step": 419
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8142548596112314e-05,
+      "loss": 0.2649,
+      "step": 420
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.818574514038877e-05,
+      "loss": 0.3087,
+      "step": 421
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.822894168466523e-05,
+      "loss": 0.2883,
+      "step": 422
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8272138228941687e-05,
+      "loss": 0.2599,
+      "step": 423
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8315334773218143e-05,
+      "loss": 0.261,
+      "step": 424
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.83585313174946e-05,
+      "loss": 0.3148,
+      "step": 425
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.840172786177106e-05,
+      "loss": 0.323,
+      "step": 426
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8444924406047516e-05,
+      "loss": 0.2816,
+      "step": 427
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8488120950323975e-05,
+      "loss": 0.2764,
+      "step": 428
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8531317494600434e-05,
+      "loss": 0.2986,
+      "step": 429
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8574514038876893e-05,
+      "loss": 0.2745,
+      "step": 430
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.861771058315335e-05,
+      "loss": 0.2726,
+      "step": 431
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8660907127429807e-05,
+      "loss": 0.2727,
+      "step": 432
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8704103671706262e-05,
+      "loss": 0.2845,
+      "step": 433
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.874730021598272e-05,
+      "loss": 0.3042,
+      "step": 434
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.879049676025918e-05,
+      "loss": 0.2994,
+      "step": 435
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.883369330453564e-05,
+      "loss": 0.2651,
+      "step": 436
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8876889848812098e-05,
+      "loss": 0.2465,
+      "step": 437
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8920086393088553e-05,
+      "loss": 0.2794,
+      "step": 438
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8963282937365012e-05,
+      "loss": 0.2927,
+      "step": 439
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.900647948164147e-05,
+      "loss": 0.2737,
+      "step": 440
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.904967602591793e-05,
+      "loss": 0.2803,
+      "step": 441
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9092872570194386e-05,
+      "loss": 0.2718,
+      "step": 442
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9136069114470844e-05,
+      "loss": 0.2854,
+      "step": 443
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.91792656587473e-05,
+      "loss": 0.2675,
+      "step": 444
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.922246220302376e-05,
+      "loss": 0.271,
+      "step": 445
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9265658747300218e-05,
+      "loss": 0.3384,
+      "step": 446
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9308855291576677e-05,
+      "loss": 0.2861,
+      "step": 447
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9352051835853135e-05,
+      "loss": 0.2762,
+      "step": 448
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.939524838012959e-05,
+      "loss": 0.257,
+      "step": 449
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9438444924406046e-05,
+      "loss": 0.2927,
+      "step": 450
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9481641468682505e-05,
+      "loss": 0.262,
+      "step": 451
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9524838012958964e-05,
+      "loss": 0.2996,
+      "step": 452
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9568034557235423e-05,
+      "loss": 0.3103,
+      "step": 453
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9611231101511882e-05,
+      "loss": 0.301,
+      "step": 454
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9654427645788337e-05,
+      "loss": 0.2852,
+      "step": 455
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9697624190064796e-05,
+      "loss": 0.2738,
+      "step": 456
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9740820734341255e-05,
+      "loss": 0.3244,
+      "step": 457
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9784017278617714e-05,
+      "loss": 0.2591,
+      "step": 458
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982721382289417e-05,
+      "loss": 0.3096,
+      "step": 459
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.987041036717063e-05,
+      "loss": 0.2971,
+      "step": 460
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9913606911447084e-05,
+      "loss": 0.2874,
+      "step": 461
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9956803455723543e-05,
+      "loss": 0.2745,
+      "step": 462
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2e-05,
+      "loss": 0.2762,
+      "step": 463
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999999977888068e-05,
+      "loss": 0.2773,
+      "step": 464
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999999115522723e-05,
+      "loss": 0.2835,
+      "step": 465
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999998009926165e-05,
+      "loss": 0.3011,
+      "step": 466
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999999646209105e-05,
+      "loss": 0.2703,
+      "step": 467
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999994472017447e-05,
+      "loss": 0.2714,
+      "step": 468
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999992039705445e-05,
+      "loss": 0.2885,
+      "step": 469
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999989165155153e-05,
+      "loss": 0.2621,
+      "step": 470
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999985848366693e-05,
+      "loss": 0.2632,
+      "step": 471
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999998208934022e-05,
+      "loss": 0.3058,
+      "step": 472
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999977888075894e-05,
+      "loss": 0.3227,
+      "step": 473
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.99999732445739e-05,
+      "loss": 0.2594,
+      "step": 474
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999996815883445e-05,
+      "loss": 0.3044,
+      "step": 475
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999996263085776e-05,
+      "loss": 0.2319,
+      "step": 476
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999995666064408e-05,
+      "loss": 0.2905,
+      "step": 477
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999950248193672e-05,
+      "loss": 0.2739,
+      "step": 478
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999943393506824e-05,
+      "loss": 0.2917,
+      "step": 479
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999936096583836e-05,
+      "loss": 0.3011,
+      "step": 480
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999928357425032e-05,
+      "loss": 0.2688,
+      "step": 481
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999992017603075e-05,
+      "loss": 0.2712,
+      "step": 482
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999991155240135e-05,
+      "loss": 0.2956,
+      "step": 483
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999902486537226e-05,
+      "loss": 0.2982,
+      "step": 484
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999892978438765e-05,
+      "loss": 0.2924,
+      "step": 485
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999883028106398e-05,
+      "loss": 0.2565,
+      "step": 486
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999872635540555e-05,
+      "loss": 0.2731,
+      "step": 487
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999861800741707e-05,
+      "loss": 0.2965,
+      "step": 488
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999850523710325e-05,
+      "loss": 0.2611,
+      "step": 489
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999983880444691e-05,
+      "loss": 0.2668,
+      "step": 490
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999982664295198e-05,
+      "loss": 0.28,
+      "step": 491
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999981403922607e-05,
+      "loss": 0.2771,
+      "step": 492
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999980099326974e-05,
+      "loss": 0.2975,
+      "step": 493
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999787505083574e-05,
+      "loss": 0.2955,
+      "step": 494
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999977357466816e-05,
+      "loss": 0.2926,
+      "step": 495
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999759202024113e-05,
+      "loss": 0.3222,
+      "step": 496
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999974438715207e-05,
+      "loss": 0.2596,
+      "step": 497
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999972913005269e-05,
+      "loss": 0.267,
+      "step": 498
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999713430726645e-05,
+      "loss": 0.3068,
+      "step": 499
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999697289174632e-05,
+      "loss": 0.3161,
+      "step": 500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999968070539736e-05,
+      "loss": 0.2957,
+      "step": 501
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999663679395565e-05,
+      "loss": 0.2655,
+      "step": 502
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999964621117e-05,
+      "loss": 0.286,
+      "step": 503
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999628300721437e-05,
+      "loss": 0.2877,
+      "step": 504
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999960994805067e-05,
+      "loss": 0.2628,
+      "step": 505
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999591153158508e-05,
+      "loss": 0.277,
+      "step": 506
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999571916045785e-05,
+      "loss": 0.2836,
+      "step": 507
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999552236713348e-05,
+      "loss": 0.2517,
+      "step": 508
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999953211516207e-05,
+      "loss": 0.2514,
+      "step": 509
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999951155139284e-05,
+      "loss": 0.3114,
+      "step": 510
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999490545406564e-05,
+      "loss": 0.3253,
+      "step": 511
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999469097204182e-05,
+      "loss": 0.2816,
+      "step": 512
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999944720678663e-05,
+      "loss": 0.2388,
+      "step": 513
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999942487415488e-05,
+      "loss": 0.2822,
+      "step": 514
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999402099309923e-05,
+      "loss": 0.3385,
+      "step": 515
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999378882252764e-05,
+      "loss": 0.2489,
+      "step": 516
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999355222984433e-05,
+      "loss": 0.3019,
+      "step": 517
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999331121505966e-05,
+      "loss": 0.2947,
+      "step": 518
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999930657781844e-05,
+      "loss": 0.2718,
+      "step": 519
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999281591922935e-05,
+      "loss": 0.2503,
+      "step": 520
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999925616382056e-05,
+      "loss": 0.3194,
+      "step": 521
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999230293512432e-05,
+      "loss": 0.283,
+      "step": 522
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999203980999703e-05,
+      "loss": 0.2745,
+      "step": 523
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999177226283534e-05,
+      "loss": 0.2552,
+      "step": 524
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999150029365107e-05,
+      "loss": 0.2709,
+      "step": 525
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999122390245626e-05,
+      "loss": 0.2928,
+      "step": 526
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999094308926312e-05,
+      "loss": 0.3182,
+      "step": 527
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999065785408407e-05,
+      "loss": 0.2789,
+      "step": 528
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999036819693175e-05,
+      "loss": 0.2466,
+      "step": 529
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999900741178189e-05,
+      "loss": 0.2646,
+      "step": 530
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999897756167586e-05,
+      "loss": 0.26,
+      "step": 531
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9998947269376407e-05,
+      "loss": 0.2851,
+      "step": 532
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9998916534884863e-05,
+      "loss": 0.2677,
+      "step": 533
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999888535820259e-05,
+      "loss": 0.2735,
+      "step": 534
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999885373933097e-05,
+      "loss": 0.2672,
+      "step": 535
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9998821678271394e-05,
+      "loss": 0.2494,
+      "step": 536
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999878917502529e-05,
+      "loss": 0.2886,
+      "step": 537
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9998756229594086e-05,
+      "loss": 0.3152,
+      "step": 538
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9998722841979243e-05,
+      "loss": 0.2959,
+      "step": 539
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998689012182237e-05,
+      "loss": 0.3039,
+      "step": 540
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998654740204565e-05,
+      "loss": 0.28,
+      "step": 541
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999862002604774e-05,
+      "loss": 0.31,
+      "step": 542
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998584869713303e-05,
+      "loss": 0.2875,
+      "step": 543
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.99985492712028e-05,
+      "loss": 0.2562,
+      "step": 544
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999851323051782e-05,
+      "loss": 0.2714,
+      "step": 545
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999847674765994e-05,
+      "loss": 0.2623,
+      "step": 546
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999843982263078e-05,
+      "loss": 0.2484,
+      "step": 547
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998402455431974e-05,
+      "loss": 0.28,
+      "step": 548
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998364646065175e-05,
+      "loss": 0.2784,
+      "step": 549
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998326394532056e-05,
+      "loss": 0.3179,
+      "step": 550
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998287700834304e-05,
+      "loss": 0.2816,
+      "step": 551
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998248564973635e-05,
+      "loss": 0.2918,
+      "step": 552
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999820898695178e-05,
+      "loss": 0.2316,
+      "step": 553
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999816896677048e-05,
+      "loss": 0.293,
+      "step": 554
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998128504431516e-05,
+      "loss": 0.2594,
+      "step": 555
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999808759993667e-05,
+      "loss": 0.2849,
+      "step": 556
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999804625328776e-05,
+      "loss": 0.2844,
+      "step": 557
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998004464486605e-05,
+      "loss": 0.2816,
+      "step": 558
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997962233535057e-05,
+      "loss": 0.2827,
+      "step": 559
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997919560434985e-05,
+      "loss": 0.2945,
+      "step": 560
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997876445188274e-05,
+      "loss": 0.2648,
+      "step": 561
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999783288779683e-05,
+      "loss": 0.2856,
+      "step": 562
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997788888262578e-05,
+      "loss": 0.2701,
+      "step": 563
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997744446587473e-05,
+      "loss": 0.3009,
+      "step": 564
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997699562773468e-05,
+      "loss": 0.2931,
+      "step": 565
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997654236822556e-05,
+      "loss": 0.3159,
+      "step": 566
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999760846873674e-05,
+      "loss": 0.2896,
+      "step": 567
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997562258518043e-05,
+      "loss": 0.274,
+      "step": 568
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997515606168508e-05,
+      "loss": 0.2799,
+      "step": 569
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.99974685116902e-05,
+      "loss": 0.2693,
+      "step": 570
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997420975085197e-05,
+      "loss": 0.2791,
+      "step": 571
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999737299635561e-05,
+      "loss": 0.309,
+      "step": 572
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999732457550355e-05,
+      "loss": 0.2446,
+      "step": 573
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999727571253117e-05,
+      "loss": 0.239,
+      "step": 574
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997226407440622e-05,
+      "loss": 0.2815,
+      "step": 575
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999717666023409e-05,
+      "loss": 0.2484,
+      "step": 576
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997126470913774e-05,
+      "loss": 0.2759,
+      "step": 577
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999707583948189e-05,
+      "loss": 0.2785,
+      "step": 578
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997024765940684e-05,
+      "loss": 0.296,
+      "step": 579
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999697325029241e-05,
+      "loss": 0.2924,
+      "step": 580
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9996921292539344e-05,
+      "loss": 0.2822,
+      "step": 581
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9996868892683792e-05,
+      "loss": 0.2583,
+      "step": 582
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999681605072806e-05,
+      "loss": 0.3098,
+      "step": 583
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9996762766674495e-05,
+      "loss": 0.2755,
+      "step": 584
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999670904052545e-05,
+      "loss": 0.2647,
+      "step": 585
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.99966548722833e-05,
+      "loss": 0.3086,
+      "step": 586
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9996600261950442e-05,
+      "loss": 0.3171,
+      "step": 587
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9996545209529287e-05,
+      "loss": 0.2807,
+      "step": 588
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9996489715022274e-05,
+      "loss": 0.2713,
+      "step": 589
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9996433778431855e-05,
+      "loss": 0.2685,
+      "step": 590
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9996377399760504e-05,
+      "loss": 0.2693,
+      "step": 591
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9996320579010718e-05,
+      "loss": 0.3154,
+      "step": 592
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9996263316185004e-05,
+      "loss": 0.2896,
+      "step": 593
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9996205611285898e-05,
+      "loss": 0.2834,
+      "step": 594
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999614746431595e-05,
+      "loss": 0.3076,
+      "step": 595
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9996088875277732e-05,
+      "loss": 0.2885,
+      "step": 596
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9996029844173837e-05,
+      "loss": 0.2498,
+      "step": 597
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995970371006873e-05,
+      "loss": 0.2765,
+      "step": 598
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995910455779474e-05,
+      "loss": 0.2624,
+      "step": 599
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995850098494285e-05,
+      "loss": 0.3009,
+      "step": 600
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995789299153977e-05,
+      "loss": 0.2922,
+      "step": 601
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999572805776124e-05,
+      "loss": 0.2661,
+      "step": 602
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999566637431878e-05,
+      "loss": 0.2966,
+      "step": 603
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995604248829325e-05,
+      "loss": 0.3289,
+      "step": 604
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995541681295627e-05,
+      "loss": 0.2864,
+      "step": 605
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995478671720446e-05,
+      "loss": 0.3033,
+      "step": 606
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995415220106573e-05,
+      "loss": 0.2876,
+      "step": 607
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995351326456816e-05,
+      "loss": 0.27,
+      "step": 608
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995286990773994e-05,
+      "loss": 0.2628,
+      "step": 609
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995222213060958e-05,
+      "loss": 0.2545,
+      "step": 610
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999515699332057e-05,
+      "loss": 0.2839,
+      "step": 611
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995091331555716e-05,
+      "loss": 0.2859,
+      "step": 612
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995025227769297e-05,
+      "loss": 0.2959,
+      "step": 613
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999495868196424e-05,
+      "loss": 0.2593,
+      "step": 614
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994891694143486e-05,
+      "loss": 0.2975,
+      "step": 615
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994824264309997e-05,
+      "loss": 0.2715,
+      "step": 616
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994756392466753e-05,
+      "loss": 0.2681,
+      "step": 617
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999468807861676e-05,
+      "loss": 0.2915,
+      "step": 618
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994619322763038e-05,
+      "loss": 0.3219,
+      "step": 619
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994550124908625e-05,
+      "loss": 0.3086,
+      "step": 620
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994480485056583e-05,
+      "loss": 0.2616,
+      "step": 621
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994410403209997e-05,
+      "loss": 0.3354,
+      "step": 622
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994339879371956e-05,
+      "loss": 0.3149,
+      "step": 623
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994268913545584e-05,
+      "loss": 0.3061,
+      "step": 624
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999419750573402e-05,
+      "loss": 0.2673,
+      "step": 625
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999412565594042e-05,
+      "loss": 0.2724,
+      "step": 626
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994053364167965e-05,
+      "loss": 0.2916,
+      "step": 627
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999398063041985e-05,
+      "loss": 0.3407,
+      "step": 628
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993907454699288e-05,
+      "loss": 0.2835,
+      "step": 629
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999383383700952e-05,
+      "loss": 0.292,
+      "step": 630
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.99937597773538e-05,
+      "loss": 0.2765,
+      "step": 631
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993685275735405e-05,
+      "loss": 0.3066,
+      "step": 632
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993610332157626e-05,
+      "loss": 0.2656,
+      "step": 633
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993534946623782e-05,
+      "loss": 0.2589,
+      "step": 634
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.99934591191372e-05,
+      "loss": 0.2719,
+      "step": 635
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993382849701238e-05,
+      "loss": 0.2967,
+      "step": 636
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999330613831927e-05,
+      "loss": 0.2788,
+      "step": 637
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993228984994686e-05,
+      "loss": 0.2776,
+      "step": 638
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.99931513897309e-05,
+      "loss": 0.2653,
+      "step": 639
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999307335253134e-05,
+      "loss": 0.2732,
+      "step": 640
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9992994873399463e-05,
+      "loss": 0.3139,
+      "step": 641
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9992915952338737e-05,
+      "loss": 0.2224,
+      "step": 642
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9992836589352648e-05,
+      "loss": 0.2725,
+      "step": 643
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9992756784444713e-05,
+      "loss": 0.2821,
+      "step": 644
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9992676537618453e-05,
+      "loss": 0.2835,
+      "step": 645
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9992595848877424e-05,
+      "loss": 0.2727,
+      "step": 646
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999251471822519e-05,
+      "loss": 0.2793,
+      "step": 647
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9992433145665342e-05,
+      "loss": 0.2874,
+      "step": 648
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9992351131201485e-05,
+      "loss": 0.2963,
+      "step": 649
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999226867483725e-05,
+      "loss": 0.2653,
+      "step": 650
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9992185776576276e-05,
+      "loss": 0.2884,
+      "step": 651
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9992102436422234e-05,
+      "loss": 0.2572,
+      "step": 652
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999201865437881e-05,
+      "loss": 0.2922,
+      "step": 653
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999193443044971e-05,
+      "loss": 0.2862,
+      "step": 654
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9991849764638655e-05,
+      "loss": 0.24,
+      "step": 655
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9991764656949395e-05,
+      "loss": 0.2833,
+      "step": 656
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9991679107385688e-05,
+      "loss": 0.2581,
+      "step": 657
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9991593115951317e-05,
+      "loss": 0.2641,
+      "step": 658
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999150668265009e-05,
+      "loss": 0.3006,
+      "step": 659
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9991419807485828e-05,
+      "loss": 0.3226,
+      "step": 660
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999133249046237e-05,
+      "loss": 0.2981,
+      "step": 661
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999124473158358e-05,
+      "loss": 0.2699,
+      "step": 662
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9991156530853338e-05,
+      "loss": 0.2907,
+      "step": 663
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9991067888275543e-05,
+      "loss": 0.2977,
+      "step": 664
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9990978803854117e-05,
+      "loss": 0.2627,
+      "step": 665
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9990889277593e-05,
+      "loss": 0.2702,
+      "step": 666
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9990799309496155e-05,
+      "loss": 0.2799,
+      "step": 667
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9990708899567554e-05,
+      "loss": 0.2713,
+      "step": 668
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9990618047811195e-05,
+      "loss": 0.2506,
+      "step": 669
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.99905267542311e-05,
+      "loss": 0.2746,
+      "step": 670
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9990435018831308e-05,
+      "loss": 0.2661,
+      "step": 671
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9990342841615872e-05,
+      "loss": 0.2968,
+      "step": 672
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9990250222588868e-05,
+      "loss": 0.2642,
+      "step": 673
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9990157161754395e-05,
+      "loss": 0.2462,
+      "step": 674
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9990063659116563e-05,
+      "loss": 0.2995,
+      "step": 675
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9989969714679514e-05,
+      "loss": 0.3229,
+      "step": 676
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9989875328447398e-05,
+      "loss": 0.271,
+      "step": 677
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9989780500424394e-05,
+      "loss": 0.3133,
+      "step": 678
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.998968523061469e-05,
+      "loss": 0.2524,
+      "step": 679
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9989589519022497e-05,
+      "loss": 0.2505,
+      "step": 680
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9989493365652058e-05,
+      "loss": 0.2598,
+      "step": 681
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9989396770507618e-05,
+      "loss": 0.2772,
+      "step": 682
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.998929973359345e-05,
+      "loss": 0.2722,
+      "step": 683
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9989202254913844e-05,
+      "loss": 0.2508,
+      "step": 684
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9989104334473117e-05,
+      "loss": 0.2889,
+      "step": 685
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.998900597227559e-05,
+      "loss": 0.2674,
+      "step": 686
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9988907168325616e-05,
+      "loss": 0.2847,
+      "step": 687
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.998880792262757e-05,
+      "loss": 0.2951,
+      "step": 688
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.998870823518584e-05,
+      "loss": 0.2262,
+      "step": 689
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.998860810600483e-05,
+      "loss": 0.2811,
+      "step": 690
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9988507535088967e-05,
+      "loss": 0.2708,
+      "step": 691
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9988406522442705e-05,
+      "loss": 0.2573,
+      "step": 692
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9988305068070505e-05,
+      "loss": 0.2664,
+      "step": 693
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998820317197686e-05,
+      "loss": 0.3063,
+      "step": 694
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9988100834166272e-05,
+      "loss": 0.2277,
+      "step": 695
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9987998054643265e-05,
+      "loss": 0.28,
+      "step": 696
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9987894833412392e-05,
+      "loss": 0.2934,
+      "step": 697
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998779117047821e-05,
+      "loss": 0.2602,
+      "step": 698
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.99876870658453e-05,
+      "loss": 0.2502,
+      "step": 699
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998758251951828e-05,
+      "loss": 0.2306,
+      "step": 700
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9987477531501765e-05,
+      "loss": 0.2627,
+      "step": 701
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9987372101800395e-05,
+      "loss": 0.2601,
+      "step": 702
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9987266230418835e-05,
+      "loss": 0.2514,
+      "step": 703
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9987159917361773e-05,
+      "loss": 0.2371,
+      "step": 704
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.99870531626339e-05,
+      "loss": 0.2578,
+      "step": 705
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998694596623995e-05,
+      "loss": 0.266,
+      "step": 706
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9986838328184647e-05,
+      "loss": 0.2559,
+      "step": 707
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9986730248472766e-05,
+      "loss": 0.2902,
+      "step": 708
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9986621727109077e-05,
+      "loss": 0.2686,
+      "step": 709
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9986512764098388e-05,
+      "loss": 0.2607,
+      "step": 710
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9986403359445513e-05,
+      "loss": 0.2896,
+      "step": 711
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998629351315529e-05,
+      "loss": 0.2654,
+      "step": 712
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9986183225232575e-05,
+      "loss": 0.2746,
+      "step": 713
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9986072495682253e-05,
+      "loss": 0.2837,
+      "step": 714
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998596132450921e-05,
+      "loss": 0.2814,
+      "step": 715
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9985849711718372e-05,
+      "loss": 0.2669,
+      "step": 716
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9985737657314666e-05,
+      "loss": 0.2956,
+      "step": 717
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9985625161303055e-05,
+      "loss": 0.2934,
+      "step": 718
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998551222368851e-05,
+      "loss": 0.2754,
+      "step": 719
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998539884447603e-05,
+      "loss": 0.3186,
+      "step": 720
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9985285023670623e-05,
+      "loss": 0.2932,
+      "step": 721
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9985170761277327e-05,
+      "loss": 0.3014,
+      "step": 722
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9985056057301192e-05,
+      "loss": 0.2745,
+      "step": 723
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9984940911747296e-05,
+      "loss": 0.2614,
+      "step": 724
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9984825324620726e-05,
+      "loss": 0.2535,
+      "step": 725
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9984709295926593e-05,
+      "loss": 0.246,
+      "step": 726
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9984592825670034e-05,
+      "loss": 0.2774,
+      "step": 727
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9984475913856193e-05,
+      "loss": 0.2873,
+      "step": 728
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9984358560490246e-05,
+      "loss": 0.2904,
+      "step": 729
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9984240765577377e-05,
+      "loss": 0.255,
+      "step": 730
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.99841225291228e-05,
+      "loss": 0.2733,
+      "step": 731
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9984003851131745e-05,
+      "loss": 0.2717,
+      "step": 732
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9983884731609458e-05,
+      "loss": 0.2556,
+      "step": 733
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9983765170561208e-05,
+      "loss": 0.2712,
+      "step": 734
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9983645167992278e-05,
+      "loss": 0.2726,
+      "step": 735
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998352472390798e-05,
+      "loss": 0.3131,
+      "step": 736
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9983403838313635e-05,
+      "loss": 0.2481,
+      "step": 737
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9983282511214595e-05,
+      "loss": 0.2878,
+      "step": 738
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9983160742616227e-05,
+      "loss": 0.2678,
+      "step": 739
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9983038532523907e-05,
+      "loss": 0.2662,
+      "step": 740
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998291588094305e-05,
+      "loss": 0.2458,
+      "step": 741
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9982792787879073e-05,
+      "loss": 0.2732,
+      "step": 742
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9982669253337423e-05,
+      "loss": 0.2742,
+      "step": 743
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998254527732356e-05,
+      "loss": 0.2421,
+      "step": 744
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.998242085984297e-05,
+      "loss": 0.3035,
+      "step": 745
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9982296000901154e-05,
+      "loss": 0.249,
+      "step": 746
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9982170700503635e-05,
+      "loss": 0.2412,
+      "step": 747
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.998204495865595e-05,
+      "loss": 0.2906,
+      "step": 748
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9981918775363665e-05,
+      "loss": 0.2732,
+      "step": 749
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.998179215063236e-05,
+      "loss": 0.2983,
+      "step": 750
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9981665084467626e-05,
+      "loss": 0.2718,
+      "step": 751
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9981537576875095e-05,
+      "loss": 0.2317,
+      "step": 752
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9981409627860397e-05,
+      "loss": 0.2595,
+      "step": 753
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9981281237429196e-05,
+      "loss": 0.2926,
+      "step": 754
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9981152405587168e-05,
+      "loss": 0.275,
+      "step": 755
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.998102313234001e-05,
+      "loss": 0.2578,
+      "step": 756
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9980893417693435e-05,
+      "loss": 0.2577,
+      "step": 757
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9980763261653186e-05,
+      "loss": 0.2757,
+      "step": 758
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9980632664225017e-05,
+      "loss": 0.2578,
+      "step": 759
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9980501625414704e-05,
+      "loss": 0.2669,
+      "step": 760
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9980370145228038e-05,
+      "loss": 0.2724,
+      "step": 761
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9980238223670837e-05,
+      "loss": 0.2963,
+      "step": 762
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9980105860748934e-05,
+      "loss": 0.2708,
+      "step": 763
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9979973056468188e-05,
+      "loss": 0.2532,
+      "step": 764
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9979839810834464e-05,
+      "loss": 0.2839,
+      "step": 765
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9979706123853657e-05,
+      "loss": 0.2925,
+      "step": 766
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997957199553168e-05,
+      "loss": 0.249,
+      "step": 767
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9979437425874466e-05,
+      "loss": 0.2833,
+      "step": 768
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9979302414887965e-05,
+      "loss": 0.2523,
+      "step": 769
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997916696257815e-05,
+      "loss": 0.2741,
+      "step": 770
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9979031068951004e-05,
+      "loss": 0.3167,
+      "step": 771
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997889473401254e-05,
+      "loss": 0.2485,
+      "step": 772
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9978757957768795e-05,
+      "loss": 0.2462,
+      "step": 773
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997862074022581e-05,
+      "loss": 0.268,
+      "step": 774
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9978483081389653e-05,
+      "loss": 0.303,
+      "step": 775
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9978344981266414e-05,
+      "loss": 0.3009,
+      "step": 776
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9978206439862205e-05,
+      "loss": 0.3137,
+      "step": 777
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997806745718314e-05,
+      "loss": 0.2738,
+      "step": 778
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997792803323538e-05,
+      "loss": 0.2328,
+      "step": 779
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997778816802508e-05,
+      "loss": 0.252,
+      "step": 780
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9977647861558433e-05,
+      "loss": 0.2625,
+      "step": 781
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9977507113841637e-05,
+      "loss": 0.2757,
+      "step": 782
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9977365924880924e-05,
+      "loss": 0.259,
+      "step": 783
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9977224294682527e-05,
+      "loss": 0.2922,
+      "step": 784
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997708222325272e-05,
+      "loss": 0.2736,
+      "step": 785
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9976939710597785e-05,
+      "loss": 0.2762,
+      "step": 786
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9976796756724016e-05,
+      "loss": 0.2796,
+      "step": 787
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9976653361637745e-05,
+      "loss": 0.254,
+      "step": 788
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9976509525345306e-05,
+      "loss": 0.2619,
+      "step": 789
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9976365247853064e-05,
+      "loss": 0.3005,
+      "step": 790
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.99762205291674e-05,
+      "loss": 0.2861,
+      "step": 791
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997607536929471e-05,
+      "loss": 0.2524,
+      "step": 792
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9975929768241416e-05,
+      "loss": 0.3052,
+      "step": 793
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997578372601396e-05,
+      "loss": 0.2618,
+      "step": 794
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9975637242618798e-05,
+      "loss": 0.2949,
+      "step": 795
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9975490318062408e-05,
+      "loss": 0.2641,
+      "step": 796
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9975342952351286e-05,
+      "loss": 0.2778,
+      "step": 797
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9975195145491947e-05,
+      "loss": 0.2473,
+      "step": 798
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9975046897490934e-05,
+      "loss": 0.291,
+      "step": 799
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.99748982083548e-05,
+      "loss": 0.2595,
+      "step": 800
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997474907809012e-05,
+      "loss": 0.2887,
+      "step": 801
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9974599506703492e-05,
+      "loss": 0.2516,
+      "step": 802
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9974449494201526e-05,
+      "loss": 0.2812,
+      "step": 803
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997429904059086e-05,
+      "loss": 0.2723,
+      "step": 804
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9974148145878146e-05,
+      "loss": 0.2442,
+      "step": 805
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9973996810070057e-05,
+      "loss": 0.2626,
+      "step": 806
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9973845033173288e-05,
+      "loss": 0.2905,
+      "step": 807
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9973692815194544e-05,
+      "loss": 0.2808,
+      "step": 808
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9973540156140567e-05,
+      "loss": 0.2784,
+      "step": 809
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9973387056018102e-05,
+      "loss": 0.2697,
+      "step": 810
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9973233514833917e-05,
+      "loss": 0.2505,
+      "step": 811
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997307953259481e-05,
+      "loss": 0.2374,
+      "step": 812
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9972925109307584e-05,
+      "loss": 0.297,
+      "step": 813
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9972770244979073e-05,
+      "loss": 0.2763,
+      "step": 814
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9972614939616123e-05,
+      "loss": 0.2769,
+      "step": 815
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.99724591932256e-05,
+      "loss": 0.3247,
+      "step": 816
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9972303005814397e-05,
+      "loss": 0.2954,
+      "step": 817
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9972146377389418e-05,
+      "loss": 0.2233,
+      "step": 818
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997198930795759e-05,
+      "loss": 0.2372,
+      "step": 819
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997183179752586e-05,
+      "loss": 0.2871,
+      "step": 820
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997167384610119e-05,
+      "loss": 0.3178,
+      "step": 821
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9971515453690573e-05,
+      "loss": 0.2478,
+      "step": 822
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9971356620301007e-05,
+      "loss": 0.2788,
+      "step": 823
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9971197345939522e-05,
+      "loss": 0.3023,
+      "step": 824
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9971037630613156e-05,
+      "loss": 0.2737,
+      "step": 825
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997087747432897e-05,
+      "loss": 0.2851,
+      "step": 826
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9970716877094057e-05,
+      "loss": 0.2456,
+      "step": 827
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997055583891551e-05,
+      "loss": 0.2625,
+      "step": 828
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9970394359800454e-05,
+      "loss": 0.2943,
+      "step": 829
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9970232439756035e-05,
+      "loss": 0.2208,
+      "step": 830
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9970070078789404e-05,
+      "loss": 0.3412,
+      "step": 831
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9969907276907748e-05,
+      "loss": 0.2901,
+      "step": 832
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9969744034118263e-05,
+      "loss": 0.2994,
+      "step": 833
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9969580350428174e-05,
+      "loss": 0.3079,
+      "step": 834
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9969416225844713e-05,
+      "loss": 0.2847,
+      "step": 835
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.996925166037514e-05,
+      "loss": 0.2809,
+      "step": 836
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9969086654026736e-05,
+      "loss": 0.2533,
+      "step": 837
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9968921206806796e-05,
+      "loss": 0.266,
+      "step": 838
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9968755318722636e-05,
+      "loss": 0.3233,
+      "step": 839
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9968588989781595e-05,
+      "loss": 0.2424,
+      "step": 840
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9968422219991025e-05,
+      "loss": 0.2753,
+      "step": 841
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.99682550093583e-05,
+      "loss": 0.2762,
+      "step": 842
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9968087357890825e-05,
+      "loss": 0.2728,
+      "step": 843
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9967919265596004e-05,
+      "loss": 0.2455,
+      "step": 844
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.996775073248127e-05,
+      "loss": 0.2804,
+      "step": 845
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.996758175855408e-05,
+      "loss": 0.2809,
+      "step": 846
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.996741234382191e-05,
+      "loss": 0.264,
+      "step": 847
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9967242488292247e-05,
+      "loss": 0.2831,
+      "step": 848
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.99670721919726e-05,
+      "loss": 0.2694,
+      "step": 849
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9966901454870512e-05,
+      "loss": 0.2732,
+      "step": 850
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996673027699352e-05,
+      "loss": 0.2691,
+      "step": 851
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9966558658349204e-05,
+      "loss": 0.2954,
+      "step": 852
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996638659894515e-05,
+      "loss": 0.2423,
+      "step": 853
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9966214098788965e-05,
+      "loss": 0.2698,
+      "step": 854
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996604115788828e-05,
+      "loss": 0.277,
+      "step": 855
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9965867776250745e-05,
+      "loss": 0.2863,
+      "step": 856
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996569395388402e-05,
+      "loss": 0.2147,
+      "step": 857
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.99655196907958e-05,
+      "loss": 0.2931,
+      "step": 858
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9965344986993792e-05,
+      "loss": 0.2842,
+      "step": 859
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9965169842485714e-05,
+      "loss": 0.2635,
+      "step": 860
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9964994257279322e-05,
+      "loss": 0.2782,
+      "step": 861
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996481823138237e-05,
+      "loss": 0.2498,
+      "step": 862
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9964641764802654e-05,
+      "loss": 0.2531,
+      "step": 863
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9964464857547967e-05,
+      "loss": 0.2419,
+      "step": 864
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996428750962614e-05,
+      "loss": 0.2633,
+      "step": 865
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9964109721045015e-05,
+      "loss": 0.277,
+      "step": 866
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996393149181245e-05,
+      "loss": 0.2809,
+      "step": 867
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9963752821936336e-05,
+      "loss": 0.2804,
+      "step": 868
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996357371142456e-05,
+      "loss": 0.2645,
+      "step": 869
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996339416028506e-05,
+      "loss": 0.2474,
+      "step": 870
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9963214168525767e-05,
+      "loss": 0.3183,
+      "step": 871
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996303373615464e-05,
+      "loss": 0.2486,
+      "step": 872
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996285286317966e-05,
+      "loss": 0.2524,
+      "step": 873
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9962671549608826e-05,
+      "loss": 0.2558,
+      "step": 874
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996248979545016e-05,
+      "loss": 0.2671,
+      "step": 875
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9962307600711693e-05,
+      "loss": 0.2516,
+      "step": 876
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9962124965401487e-05,
+      "loss": 0.2662,
+      "step": 877
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996194188952762e-05,
+      "loss": 0.2855,
+      "step": 878
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9961758373098184e-05,
+      "loss": 0.2743,
+      "step": 879
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9961574416121298e-05,
+      "loss": 0.269,
+      "step": 880
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9961390018605097e-05,
+      "loss": 0.2702,
+      "step": 881
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996120518055773e-05,
+      "loss": 0.2853,
+      "step": 882
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996101990198738e-05,
+      "loss": 0.2354,
+      "step": 883
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996083418290224e-05,
+      "loss": 0.2708,
+      "step": 884
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9960648023310517e-05,
+      "loss": 0.2396,
+      "step": 885
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996046142322045e-05,
+      "loss": 0.2896,
+      "step": 886
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996027438264028e-05,
+      "loss": 0.2614,
+      "step": 887
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9960086901578295e-05,
+      "loss": 0.2448,
+      "step": 888
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9959898980042775e-05,
+      "loss": 0.2375,
+      "step": 889
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9959710618042034e-05,
+      "loss": 0.2934,
+      "step": 890
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.99595218155844e-05,
+      "loss": 0.2452,
+      "step": 891
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9959332572678228e-05,
+      "loss": 0.2985,
+      "step": 892
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.995914288933188e-05,
+      "loss": 0.2161,
+      "step": 893
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9958952765553746e-05,
+      "loss": 0.2838,
+      "step": 894
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.995876220135224e-05,
+      "loss": 0.2928,
+      "step": 895
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9958571196735782e-05,
+      "loss": 0.2193,
+      "step": 896
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9958379751712827e-05,
+      "loss": 0.2535,
+      "step": 897
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9958187866291833e-05,
+      "loss": 0.2679,
+      "step": 898
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9957995540481292e-05,
+      "loss": 0.3293,
+      "step": 899
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9957802774289706e-05,
+      "loss": 0.2449,
+      "step": 900
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9957609567725596e-05,
+      "loss": 0.251,
+      "step": 901
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9957415920797517e-05,
+      "loss": 0.2514,
+      "step": 902
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9957221833514024e-05,
+      "loss": 0.229,
+      "step": 903
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9957027305883705e-05,
+      "loss": 0.2488,
+      "step": 904
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995683233791516e-05,
+      "loss": 0.2722,
+      "step": 905
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9956636929617016e-05,
+      "loss": 0.275,
+      "step": 906
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9956441080997905e-05,
+      "loss": 0.2543,
+      "step": 907
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.99562447920665e-05,
+      "loss": 0.2545,
+      "step": 908
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9956048062831472e-05,
+      "loss": 0.2661,
+      "step": 909
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9955850893301525e-05,
+      "loss": 0.2842,
+      "step": 910
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995565328348538e-05,
+      "loss": 0.2781,
+      "step": 911
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9955455233391773e-05,
+      "loss": 0.2656,
+      "step": 912
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9955256743029466e-05,
+      "loss": 0.2834,
+      "step": 913
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9955057812407234e-05,
+      "loss": 0.2571,
+      "step": 914
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9954858441533877e-05,
+      "loss": 0.2719,
+      "step": 915
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9954658630418212e-05,
+      "loss": 0.2596,
+      "step": 916
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995445837906907e-05,
+      "loss": 0.2493,
+      "step": 917
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9954257687495314e-05,
+      "loss": 0.252,
+      "step": 918
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9954056555705813e-05,
+      "loss": 0.2611,
+      "step": 919
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9953854983709472e-05,
+      "loss": 0.2329,
+      "step": 920
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9953652971515193e-05,
+      "loss": 0.2497,
+      "step": 921
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9953450519131915e-05,
+      "loss": 0.2786,
+      "step": 922
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9953247626568593e-05,
+      "loss": 0.2792,
+      "step": 923
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9953044293834195e-05,
+      "loss": 0.3169,
+      "step": 924
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995284052093772e-05,
+      "loss": 0.2693,
+      "step": 925
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9952636307888173e-05,
+      "loss": 0.2619,
+      "step": 926
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995243165469459e-05,
+      "loss": 0.2316,
+      "step": 927
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995222656136602e-05,
+      "loss": 0.2664,
+      "step": 928
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995202102791153e-05,
+      "loss": 0.2706,
+      "step": 929
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9951815054340215e-05,
+      "loss": 0.2269,
+      "step": 930
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995160864066118e-05,
+      "loss": 0.2647,
+      "step": 931
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995140178688355e-05,
+      "loss": 0.2645,
+      "step": 932
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9951194493016482e-05,
+      "loss": 0.2989,
+      "step": 933
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995098675906914e-05,
+      "loss": 0.2997,
+      "step": 934
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9950778585050704e-05,
+      "loss": 0.2714,
+      "step": 935
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995056997097039e-05,
+      "loss": 0.2463,
+      "step": 936
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995036091683742e-05,
+      "loss": 0.2693,
+      "step": 937
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9950151422661035e-05,
+      "loss": 0.2575,
+      "step": 938
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9949941488450503e-05,
+      "loss": 0.2808,
+      "step": 939
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9949731114215113e-05,
+      "loss": 0.2572,
+      "step": 940
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.994952029996416e-05,
+      "loss": 0.2634,
+      "step": 941
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9949309045706972e-05,
+      "loss": 0.2615,
+      "step": 942
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.994909735145289e-05,
+      "loss": 0.2685,
+      "step": 943
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.994888521721128e-05,
+      "loss": 0.2665,
+      "step": 944
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9948672642991514e-05,
+      "loss": 0.2657,
+      "step": 945
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9948459628802998e-05,
+      "loss": 0.2596,
+      "step": 946
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9948246174655157e-05,
+      "loss": 0.2772,
+      "step": 947
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9948032280557424e-05,
+      "loss": 0.2697,
+      "step": 948
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9947817946519263e-05,
+      "loss": 0.3234,
+      "step": 949
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994760317255015e-05,
+      "loss": 0.2635,
+      "step": 950
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9947387958659583e-05,
+      "loss": 0.2528,
+      "step": 951
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994717230485708e-05,
+      "loss": 0.2468,
+      "step": 952
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9946956211152178e-05,
+      "loss": 0.262,
+      "step": 953
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9946739677554438e-05,
+      "loss": 0.2897,
+      "step": 954
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9946522704073426e-05,
+      "loss": 0.2586,
+      "step": 955
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9946305290718746e-05,
+      "loss": 0.2752,
+      "step": 956
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9946087437500008e-05,
+      "loss": 0.2775,
+      "step": 957
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994586914442685e-05,
+      "loss": 0.2743,
+      "step": 958
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9945650411508922e-05,
+      "loss": 0.2796,
+      "step": 959
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9945431238755903e-05,
+      "loss": 0.3405,
+      "step": 960
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9945211626177482e-05,
+      "loss": 0.2686,
+      "step": 961
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994499157378337e-05,
+      "loss": 0.2981,
+      "step": 962
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.99447710815833e-05,
+      "loss": 0.2868,
+      "step": 963
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994455014958702e-05,
+      "loss": 0.2276,
+      "step": 964
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9944328777804302e-05,
+      "loss": 0.2836,
+      "step": 965
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994410696624494e-05,
+      "loss": 0.2923,
+      "step": 966
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994388471491874e-05,
+      "loss": 0.2572,
+      "step": 967
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994366202383553e-05,
+      "loss": 0.2682,
+      "step": 968
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9943438893005163e-05,
+      "loss": 0.2898,
+      "step": 969
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9943215322437498e-05,
+      "loss": 0.2534,
+      "step": 970
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9942991312142433e-05,
+      "loss": 0.2374,
+      "step": 971
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9942766862129868e-05,
+      "loss": 0.2519,
+      "step": 972
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9942541972409727e-05,
+      "loss": 0.2927,
+      "step": 973
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9942316642991957e-05,
+      "loss": 0.2923,
+      "step": 974
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994209087388653e-05,
+      "loss": 0.2746,
+      "step": 975
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994186466510342e-05,
+      "loss": 0.2971,
+      "step": 976
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994163801665264e-05,
+      "loss": 0.2833,
+      "step": 977
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9941410928544208e-05,
+      "loss": 0.2531,
+      "step": 978
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9941183400788162e-05,
+      "loss": 0.2502,
+      "step": 979
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9940955433394577e-05,
+      "loss": 0.2438,
+      "step": 980
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9940727026373524e-05,
+      "loss": 0.2725,
+      "step": 981
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9940498179735107e-05,
+      "loss": 0.2766,
+      "step": 982
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9940268893489447e-05,
+      "loss": 0.2621,
+      "step": 983
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9940039167646684e-05,
+      "loss": 0.2858,
+      "step": 984
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.993980900221698e-05,
+      "loss": 0.2904,
+      "step": 985
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9939578397210505e-05,
+      "loss": 0.2615,
+      "step": 986
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9939347352637467e-05,
+      "loss": 0.2642,
+      "step": 987
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.993911586850808e-05,
+      "loss": 0.2506,
+      "step": 988
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9938883944832584e-05,
+      "loss": 0.2202,
+      "step": 989
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9938651581621226e-05,
+      "loss": 0.274,
+      "step": 990
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9938418778884293e-05,
+      "loss": 0.2827,
+      "step": 991
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9938185536632076e-05,
+      "loss": 0.2465,
+      "step": 992
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9937951854874887e-05,
+      "loss": 0.2854,
+      "step": 993
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9937717733623066e-05,
+      "loss": 0.2606,
+      "step": 994
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9937483172886963e-05,
+      "loss": 0.2689,
+      "step": 995
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9937248172676954e-05,
+      "loss": 0.2534,
+      "step": 996
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9937012733003426e-05,
+      "loss": 0.2695,
+      "step": 997
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9936776853876797e-05,
+      "loss": 0.2644,
+      "step": 998
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9936540535307496e-05,
+      "loss": 0.2414,
+      "step": 999
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9936303777305973e-05,
+      "loss": 0.2606,
+      "step": 1000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9936066579882704e-05,
+      "loss": 0.2633,
+      "step": 1001
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993582894304817e-05,
+      "loss": 0.2625,
+      "step": 1002
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9935590866812886e-05,
+      "loss": 0.2541,
+      "step": 1003
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993535235118738e-05,
+      "loss": 0.298,
+      "step": 1004
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9935113396182196e-05,
+      "loss": 0.3062,
+      "step": 1005
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993487400180791e-05,
+      "loss": 0.2921,
+      "step": 1006
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.99346341680751e-05,
+      "loss": 0.2827,
+      "step": 1007
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9934393894994376e-05,
+      "loss": 0.2611,
+      "step": 1008
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9934153182576367e-05,
+      "loss": 0.2574,
+      "step": 1009
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9933912030831713e-05,
+      "loss": 0.2765,
+      "step": 1010
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993367043977108e-05,
+      "loss": 0.2964,
+      "step": 1011
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9933428409405153e-05,
+      "loss": 0.2594,
+      "step": 1012
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993318593974464e-05,
+      "loss": 0.3315,
+      "step": 1013
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9932943030800254e-05,
+      "loss": 0.2347,
+      "step": 1014
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9932699682582744e-05,
+      "loss": 0.2251,
+      "step": 1015
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9932455895102873e-05,
+      "loss": 0.2477,
+      "step": 1016
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9932211668371417e-05,
+      "loss": 0.2607,
+      "step": 1017
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9931967002399178e-05,
+      "loss": 0.2317,
+      "step": 1018
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993172189719698e-05,
+      "loss": 0.266,
+      "step": 1019
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993147635277566e-05,
+      "loss": 0.2619,
+      "step": 1020
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993123036914608e-05,
+      "loss": 0.2695,
+      "step": 1021
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9930983946319113e-05,
+      "loss": 0.2969,
+      "step": 1022
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9930737084305657e-05,
+      "loss": 0.2858,
+      "step": 1023
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9930489783116632e-05,
+      "loss": 0.2858,
+      "step": 1024
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9930242042762976e-05,
+      "loss": 0.2811,
+      "step": 1025
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9929993863255642e-05,
+      "loss": 0.2703,
+      "step": 1026
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.992974524460561e-05,
+      "loss": 0.2402,
+      "step": 1027
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9929496186823862e-05,
+      "loss": 0.2685,
+      "step": 1028
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9929246689921425e-05,
+      "loss": 0.2462,
+      "step": 1029
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9928996753909333e-05,
+      "loss": 0.2636,
+      "step": 1030
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9928746378798633e-05,
+      "loss": 0.2333,
+      "step": 1031
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.99284955646004e-05,
+      "loss": 0.244,
+      "step": 1032
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9928244311325727e-05,
+      "loss": 0.2538,
+      "step": 1033
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9927992618985726e-05,
+      "loss": 0.2766,
+      "step": 1034
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9927740487591523e-05,
+      "loss": 0.2907,
+      "step": 1035
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9927487917154268e-05,
+      "loss": 0.2797,
+      "step": 1036
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.992723490768514e-05,
+      "loss": 0.2584,
+      "step": 1037
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.992698145919532e-05,
+      "loss": 0.2605,
+      "step": 1038
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.992672757169602e-05,
+      "loss": 0.2943,
+      "step": 1039
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9926473245198463e-05,
+      "loss": 0.2616,
+      "step": 1040
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9926218479713903e-05,
+      "loss": 0.2339,
+      "step": 1041
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9925963275253603e-05,
+      "loss": 0.2645,
+      "step": 1042
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.992570763182885e-05,
+      "loss": 0.2551,
+      "step": 1043
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9925451549450944e-05,
+      "loss": 0.2803,
+      "step": 1044
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9925195028131218e-05,
+      "loss": 0.283,
+      "step": 1045
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9924938067881014e-05,
+      "loss": 0.241,
+      "step": 1046
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9924680668711696e-05,
+      "loss": 0.2508,
+      "step": 1047
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9924422830634646e-05,
+      "loss": 0.2535,
+      "step": 1048
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9924164553661266e-05,
+      "loss": 0.2242,
+      "step": 1049
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.992390583780298e-05,
+      "loss": 0.282,
+      "step": 1050
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9923646683071227e-05,
+      "loss": 0.2933,
+      "step": 1051
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.992338708947747e-05,
+      "loss": 0.2315,
+      "step": 1052
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9923127057033188e-05,
+      "loss": 0.2621,
+      "step": 1053
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.992286658574988e-05,
+      "loss": 0.2643,
+      "step": 1054
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.992260567563907e-05,
+      "loss": 0.2896,
+      "step": 1055
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.992234432671229e-05,
+      "loss": 0.2728,
+      "step": 1056
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.99220825389811e-05,
+      "loss": 0.2655,
+      "step": 1057
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.992182031245708e-05,
+      "loss": 0.2359,
+      "step": 1058
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.992155764715182e-05,
+      "loss": 0.2638,
+      "step": 1059
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9921294543076945e-05,
+      "loss": 0.3073,
+      "step": 1060
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9921031000244084e-05,
+      "loss": 0.2779,
+      "step": 1061
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9920767018664893e-05,
+      "loss": 0.2564,
+      "step": 1062
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.992050259835105e-05,
+      "loss": 0.2291,
+      "step": 1063
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.992023773931424e-05,
+      "loss": 0.2399,
+      "step": 1064
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9919972441566186e-05,
+      "loss": 0.2657,
+      "step": 1065
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9919706705118616e-05,
+      "loss": 0.2389,
+      "step": 1066
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9919440529983284e-05,
+      "loss": 0.2386,
+      "step": 1067
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9919173916171957e-05,
+      "loss": 0.2545,
+      "step": 1068
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991890686369643e-05,
+      "loss": 0.2392,
+      "step": 1069
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991863937256851e-05,
+      "loss": 0.2696,
+      "step": 1070
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9918371442800025e-05,
+      "loss": 0.2915,
+      "step": 1071
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9918103074402832e-05,
+      "loss": 0.2458,
+      "step": 1072
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9917834267388793e-05,
+      "loss": 0.2645,
+      "step": 1073
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9917565021769795e-05,
+      "loss": 0.2831,
+      "step": 1074
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9917295337557747e-05,
+      "loss": 0.2573,
+      "step": 1075
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9917025214764575e-05,
+      "loss": 0.2399,
+      "step": 1076
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9916754653402225e-05,
+      "loss": 0.2302,
+      "step": 1077
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9916483653482664e-05,
+      "loss": 0.264,
+      "step": 1078
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9916212215017877e-05,
+      "loss": 0.2817,
+      "step": 1079
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9915940338019858e-05,
+      "loss": 0.2676,
+      "step": 1080
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9915668022500646e-05,
+      "loss": 0.2248,
+      "step": 1081
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9915395268472272e-05,
+      "loss": 0.2514,
+      "step": 1082
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9915122075946806e-05,
+      "loss": 0.2734,
+      "step": 1083
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9914848444936327e-05,
+      "loss": 0.2964,
+      "step": 1084
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991457437545293e-05,
+      "loss": 0.256,
+      "step": 1085
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9914299867508746e-05,
+      "loss": 0.2628,
+      "step": 1086
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9914024921115907e-05,
+      "loss": 0.3025,
+      "step": 1087
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9913749536286574e-05,
+      "loss": 0.2654,
+      "step": 1088
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991347371303293e-05,
+      "loss": 0.266,
+      "step": 1089
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9913197451367167e-05,
+      "loss": 0.2654,
+      "step": 1090
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9912920751301502e-05,
+      "loss": 0.271,
+      "step": 1091
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991264361284818e-05,
+      "loss": 0.2681,
+      "step": 1092
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9912366036019446e-05,
+      "loss": 0.2392,
+      "step": 1093
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9912088020827584e-05,
+      "loss": 0.2234,
+      "step": 1094
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9911809567284886e-05,
+      "loss": 0.2975,
+      "step": 1095
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9911530675403665e-05,
+      "loss": 0.2534,
+      "step": 1096
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9911251345196256e-05,
+      "loss": 0.2608,
+      "step": 1097
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991097157667501e-05,
+      "loss": 0.2244,
+      "step": 1098
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9910691369852302e-05,
+      "loss": 0.2817,
+      "step": 1099
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9910410724740528e-05,
+      "loss": 0.2532,
+      "step": 1100
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991012964135209e-05,
+      "loss": 0.2457,
+      "step": 1101
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9909848119699425e-05,
+      "loss": 0.2638,
+      "step": 1102
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9909566159794978e-05,
+      "loss": 0.2522,
+      "step": 1103
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9909283761651226e-05,
+      "loss": 0.2472,
+      "step": 1104
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9909000925280647e-05,
+      "loss": 0.284,
+      "step": 1105
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9908717650695763e-05,
+      "loss": 0.3033,
+      "step": 1106
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9908433937909087e-05,
+      "loss": 0.2726,
+      "step": 1107
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9908149786933175e-05,
+      "loss": 0.2497,
+      "step": 1108
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990786519778059e-05,
+      "loss": 0.3403,
+      "step": 1109
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990758017046392e-05,
+      "loss": 0.296,
+      "step": 1110
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9907294704995766e-05,
+      "loss": 0.242,
+      "step": 1111
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990700880138876e-05,
+      "loss": 0.2538,
+      "step": 1112
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9906722459655538e-05,
+      "loss": 0.2624,
+      "step": 1113
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990643567980876e-05,
+      "loss": 0.2558,
+      "step": 1114
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9906148461861124e-05,
+      "loss": 0.2648,
+      "step": 1115
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9905860805825315e-05,
+      "loss": 0.271,
+      "step": 1116
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9905572711714065e-05,
+      "loss": 0.265,
+      "step": 1117
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990528417954011e-05,
+      "loss": 0.2853,
+      "step": 1118
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990499520931621e-05,
+      "loss": 0.2888,
+      "step": 1119
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9904705801055145e-05,
+      "loss": 0.2315,
+      "step": 1120
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9904415954769717e-05,
+      "loss": 0.2591,
+      "step": 1121
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9904125670472744e-05,
+      "loss": 0.2896,
+      "step": 1122
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9903834948177054e-05,
+      "loss": 0.2677,
+      "step": 1123
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9903543787895514e-05,
+      "loss": 0.2451,
+      "step": 1124
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9903252189640995e-05,
+      "loss": 0.3065,
+      "step": 1125
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9902960153426398e-05,
+      "loss": 0.2578,
+      "step": 1126
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9902667679264633e-05,
+      "loss": 0.2785,
+      "step": 1127
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9902374767168633e-05,
+      "loss": 0.2593,
+      "step": 1128
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9902081417151358e-05,
+      "loss": 0.2739,
+      "step": 1129
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9901787629225775e-05,
+      "loss": 0.2661,
+      "step": 1130
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9901493403404882e-05,
+      "loss": 0.3014,
+      "step": 1131
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9901198739701685e-05,
+      "loss": 0.2719,
+      "step": 1132
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9900903638129218e-05,
+      "loss": 0.2784,
+      "step": 1133
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9900608098700534e-05,
+      "loss": 0.2919,
+      "step": 1134
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9900312121428698e-05,
+      "loss": 0.2789,
+      "step": 1135
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9900015706326802e-05,
+      "loss": 0.2976,
+      "step": 1136
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9899718853407956e-05,
+      "loss": 0.2382,
+      "step": 1137
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9899421562685287e-05,
+      "loss": 0.2386,
+      "step": 1138
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9899123834171938e-05,
+      "loss": 0.2652,
+      "step": 1139
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9898825667881083e-05,
+      "loss": 0.2859,
+      "step": 1140
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9898527063825904e-05,
+      "loss": 0.2426,
+      "step": 1141
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9898228022019603e-05,
+      "loss": 0.3062,
+      "step": 1142
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9897928542475412e-05,
+      "loss": 0.3027,
+      "step": 1143
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9897628625206573e-05,
+      "loss": 0.2817,
+      "step": 1144
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9897328270226343e-05,
+      "loss": 0.2395,
+      "step": 1145
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9897027477548014e-05,
+      "loss": 0.2637,
+      "step": 1146
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9896726247184885e-05,
+      "loss": 0.2505,
+      "step": 1147
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9896424579150274e-05,
+      "loss": 0.2951,
+      "step": 1148
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9896122473457527e-05,
+      "loss": 0.2266,
+      "step": 1149
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.989581993012e-05,
+      "loss": 0.2561,
+      "step": 1150
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.989551694915108e-05,
+      "loss": 0.2701,
+      "step": 1151
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9895213530564156e-05,
+      "loss": 0.2389,
+      "step": 1152
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.989490967437265e-05,
+      "loss": 0.2876,
+      "step": 1153
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9894605380590004e-05,
+      "loss": 0.2719,
+      "step": 1154
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.989430064922967e-05,
+      "loss": 0.2762,
+      "step": 1155
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.989399548030513e-05,
+      "loss": 0.2478,
+      "step": 1156
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.989368987382987e-05,
+      "loss": 0.2894,
+      "step": 1157
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9893383829817414e-05,
+      "loss": 0.2871,
+      "step": 1158
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9893077348281294e-05,
+      "loss": 0.2487,
+      "step": 1159
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9892770429235062e-05,
+      "loss": 0.3073,
+      "step": 1160
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9892463072692295e-05,
+      "loss": 0.2492,
+      "step": 1161
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9892155278666578e-05,
+      "loss": 0.2597,
+      "step": 1162
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.989184704717153e-05,
+      "loss": 0.2496,
+      "step": 1163
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9891538378220785e-05,
+      "loss": 0.2552,
+      "step": 1164
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9891229271827983e-05,
+      "loss": 0.281,
+      "step": 1165
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.98909197280068e-05,
+      "loss": 0.2579,
+      "step": 1166
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9890609746770925e-05,
+      "loss": 0.2838,
+      "step": 1167
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9890299328134068e-05,
+      "loss": 0.2707,
+      "step": 1168
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9889988472109952e-05,
+      "loss": 0.2213,
+      "step": 1169
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988967717871233e-05,
+      "loss": 0.2099,
+      "step": 1170
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9889365447954967e-05,
+      "loss": 0.265,
+      "step": 1171
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9889053279851643e-05,
+      "loss": 0.2501,
+      "step": 1172
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9888740674416175e-05,
+      "loss": 0.2556,
+      "step": 1173
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9888427631662374e-05,
+      "loss": 0.2668,
+      "step": 1174
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9888114151604093e-05,
+      "loss": 0.2891,
+      "step": 1175
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9887800234255198e-05,
+      "loss": 0.24,
+      "step": 1176
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988748587962956e-05,
+      "loss": 0.2783,
+      "step": 1177
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988717108774109e-05,
+      "loss": 0.2342,
+      "step": 1178
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9886855858603707e-05,
+      "loss": 0.278,
+      "step": 1179
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9886540192231352e-05,
+      "loss": 0.2828,
+      "step": 1180
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9886224088637985e-05,
+      "loss": 0.2295,
+      "step": 1181
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988590754783758e-05,
+      "loss": 0.2497,
+      "step": 1182
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9885590569844147e-05,
+      "loss": 0.2744,
+      "step": 1183
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9885273154671696e-05,
+      "loss": 0.2551,
+      "step": 1184
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9884955302334266e-05,
+      "loss": 0.2646,
+      "step": 1185
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9884637012845912e-05,
+      "loss": 0.249,
+      "step": 1186
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9884318286220715e-05,
+      "loss": 0.2536,
+      "step": 1187
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9883999122472764e-05,
+      "loss": 0.2639,
+      "step": 1188
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9883679521616174e-05,
+      "loss": 0.2663,
+      "step": 1189
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9883359483665084e-05,
+      "loss": 0.2616,
+      "step": 1190
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9883039008633647e-05,
+      "loss": 0.2456,
+      "step": 1191
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988271809653603e-05,
+      "loss": 0.2884,
+      "step": 1192
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9882396747386433e-05,
+      "loss": 0.232,
+      "step": 1193
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988207496119906e-05,
+      "loss": 0.2707,
+      "step": 1194
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9881752737988143e-05,
+      "loss": 0.2629,
+      "step": 1195
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9881430077767936e-05,
+      "loss": 0.2402,
+      "step": 1196
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9881106980552704e-05,
+      "loss": 0.2439,
+      "step": 1197
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988078344635674e-05,
+      "loss": 0.2874,
+      "step": 1198
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988045947519434e-05,
+      "loss": 0.332,
+      "step": 1199
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988013506707985e-05,
+      "loss": 0.2538,
+      "step": 1200
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9879810222027603e-05,
+      "loss": 0.2603,
+      "step": 1201
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9879484940051972e-05,
+      "loss": 0.2463,
+      "step": 1202
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9879159221167335e-05,
+      "loss": 0.256,
+      "step": 1203
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9878833065388105e-05,
+      "loss": 0.275,
+      "step": 1204
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.98785064727287e-05,
+      "loss": 0.2179,
+      "step": 1205
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9878179443203564e-05,
+      "loss": 0.279,
+      "step": 1206
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9877851976827157e-05,
+      "loss": 0.2852,
+      "step": 1207
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.987752407361397e-05,
+      "loss": 0.2577,
+      "step": 1208
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9877195733578494e-05,
+      "loss": 0.2541,
+      "step": 1209
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9876866956735257e-05,
+      "loss": 0.2708,
+      "step": 1210
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9876537743098792e-05,
+      "loss": 0.2809,
+      "step": 1211
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9876208092683663e-05,
+      "loss": 0.2899,
+      "step": 1212
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.987587800550445e-05,
+      "loss": 0.265,
+      "step": 1213
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9875547481575746e-05,
+      "loss": 0.2799,
+      "step": 1214
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9875216520912167e-05,
+      "loss": 0.2415,
+      "step": 1215
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9874885123528355e-05,
+      "loss": 0.2862,
+      "step": 1216
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.987455328943896e-05,
+      "loss": 0.2305,
+      "step": 1217
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9874221018658663e-05,
+      "loss": 0.2321,
+      "step": 1218
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.987388831120215e-05,
+      "loss": 0.2679,
+      "step": 1219
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9873555167084146e-05,
+      "loss": 0.2348,
+      "step": 1220
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9873221586319376e-05,
+      "loss": 0.2793,
+      "step": 1221
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9872887568922593e-05,
+      "loss": 0.2468,
+      "step": 1222
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.987255311490857e-05,
+      "loss": 0.2628,
+      "step": 1223
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9872218224292094e-05,
+      "loss": 0.2733,
+      "step": 1224
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9871882897087983e-05,
+      "loss": 0.2807,
+      "step": 1225
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.987154713331106e-05,
+      "loss": 0.2494,
+      "step": 1226
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9871210932976174e-05,
+      "loss": 0.2734,
+      "step": 1227
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9870874296098197e-05,
+      "loss": 0.2325,
+      "step": 1228
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.987053722269201e-05,
+      "loss": 0.2863,
+      "step": 1229
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9870199712772525e-05,
+      "loss": 0.2747,
+      "step": 1230
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9869861766354673e-05,
+      "loss": 0.3094,
+      "step": 1231
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9869523383453387e-05,
+      "loss": 0.2525,
+      "step": 1232
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.986918456408364e-05,
+      "loss": 0.2317,
+      "step": 1233
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9868845308260414e-05,
+      "loss": 0.281,
+      "step": 1234
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.986850561599871e-05,
+      "loss": 0.2252,
+      "step": 1235
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9868165487313553e-05,
+      "loss": 0.242,
+      "step": 1236
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.986782492221998e-05,
+      "loss": 0.2601,
+      "step": 1237
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9867483920733063e-05,
+      "loss": 0.2649,
+      "step": 1238
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9867142482867876e-05,
+      "loss": 0.3105,
+      "step": 1239
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9866800608639513e-05,
+      "loss": 0.2211,
+      "step": 1240
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9866458298063097e-05,
+      "loss": 0.2696,
+      "step": 1241
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.986611555115377e-05,
+      "loss": 0.2662,
+      "step": 1242
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.986577236792669e-05,
+      "loss": 0.246,
+      "step": 1243
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9865428748397032e-05,
+      "loss": 0.2685,
+      "step": 1244
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9865084692579986e-05,
+      "loss": 0.2379,
+      "step": 1245
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9864740200490777e-05,
+      "loss": 0.2655,
+      "step": 1246
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.986439527214463e-05,
+      "loss": 0.2921,
+      "step": 1247
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.986404990755681e-05,
+      "loss": 0.2609,
+      "step": 1248
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9863704106742582e-05,
+      "loss": 0.2542,
+      "step": 1249
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9863357869717244e-05,
+      "loss": 0.2763,
+      "step": 1250
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9863011196496102e-05,
+      "loss": 0.2578,
+      "step": 1251
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9862664087094493e-05,
+      "loss": 0.2638,
+      "step": 1252
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9862316541527765e-05,
+      "loss": 0.2509,
+      "step": 1253
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9861968559811288e-05,
+      "loss": 0.2861,
+      "step": 1254
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9861620141960453e-05,
+      "loss": 0.2684,
+      "step": 1255
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9861271287990663e-05,
+      "loss": 0.2369,
+      "step": 1256
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.986092199791735e-05,
+      "loss": 0.2907,
+      "step": 1257
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9860572271755963e-05,
+      "loss": 0.2622,
+      "step": 1258
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9860222109521963e-05,
+      "loss": 0.2368,
+      "step": 1259
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985987151123084e-05,
+      "loss": 0.2795,
+      "step": 1260
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9859520476898092e-05,
+      "loss": 0.3071,
+      "step": 1261
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9859169006539253e-05,
+      "loss": 0.2294,
+      "step": 1262
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985881710016986e-05,
+      "loss": 0.2605,
+      "step": 1263
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9858464757805475e-05,
+      "loss": 0.2723,
+      "step": 1264
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9858111979461685e-05,
+      "loss": 0.2372,
+      "step": 1265
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9857758765154084e-05,
+      "loss": 0.2729,
+      "step": 1266
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.98574051148983e-05,
+      "loss": 0.2676,
+      "step": 1267
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9857051028709968e-05,
+      "loss": 0.2489,
+      "step": 1268
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9856696506604746e-05,
+      "loss": 0.2564,
+      "step": 1269
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985634154859832e-05,
+      "loss": 0.2632,
+      "step": 1270
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9855986154706376e-05,
+      "loss": 0.233,
+      "step": 1271
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9855630324944643e-05,
+      "loss": 0.2681,
+      "step": 1272
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985527405932885e-05,
+      "loss": 0.2699,
+      "step": 1273
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9854917357874753e-05,
+      "loss": 0.2566,
+      "step": 1274
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9854560220598125e-05,
+      "loss": 0.2877,
+      "step": 1275
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9854202647514763e-05,
+      "loss": 0.2576,
+      "step": 1276
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9853844638640484e-05,
+      "loss": 0.2535,
+      "step": 1277
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9853486193991115e-05,
+      "loss": 0.2765,
+      "step": 1278
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985312731358251e-05,
+      "loss": 0.254,
+      "step": 1279
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9852767997430536e-05,
+      "loss": 0.2455,
+      "step": 1280
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9852408245551086e-05,
+      "loss": 0.2188,
+      "step": 1281
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9852048057960073e-05,
+      "loss": 0.2467,
+      "step": 1282
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9851687434673425e-05,
+      "loss": 0.2386,
+      "step": 1283
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9851326375707083e-05,
+      "loss": 0.2266,
+      "step": 1284
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9850964881077023e-05,
+      "loss": 0.2631,
+      "step": 1285
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9850602950799228e-05,
+      "loss": 0.273,
+      "step": 1286
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9850240584889706e-05,
+      "loss": 0.3353,
+      "step": 1287
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9849877783364474e-05,
+      "loss": 0.2802,
+      "step": 1288
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9849514546239586e-05,
+      "loss": 0.2568,
+      "step": 1289
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9849150873531105e-05,
+      "loss": 0.2624,
+      "step": 1290
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.984878676525511e-05,
+      "loss": 0.2745,
+      "step": 1291
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.984842222142771e-05,
+      "loss": 0.256,
+      "step": 1292
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9848057242065014e-05,
+      "loss": 0.2185,
+      "step": 1293
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9847691827183176e-05,
+      "loss": 0.2879,
+      "step": 1294
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9847325976798348e-05,
+      "loss": 0.2795,
+      "step": 1295
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9846959690926712e-05,
+      "loss": 0.2164,
+      "step": 1296
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9846592969584465e-05,
+      "loss": 0.2624,
+      "step": 1297
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.984622581278783e-05,
+      "loss": 0.2384,
+      "step": 1298
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9845858220553035e-05,
+      "loss": 0.2511,
+      "step": 1299
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.984549019289635e-05,
+      "loss": 0.2529,
+      "step": 1300
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9845121729834035e-05,
+      "loss": 0.2739,
+      "step": 1301
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9844752831382393e-05,
+      "loss": 0.2622,
+      "step": 1302
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9844383497557737e-05,
+      "loss": 0.2567,
+      "step": 1303
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9844013728376402e-05,
+      "loss": 0.2601,
+      "step": 1304
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9843643523854738e-05,
+      "loss": 0.2775,
+      "step": 1305
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.984327288400912e-05,
+      "loss": 0.2597,
+      "step": 1306
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9842901808855934e-05,
+      "loss": 0.2279,
+      "step": 1307
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.98425302984116e-05,
+      "loss": 0.3136,
+      "step": 1308
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.984215835269253e-05,
+      "loss": 0.2202,
+      "step": 1309
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9841785971715193e-05,
+      "loss": 0.2268,
+      "step": 1310
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9841413155496047e-05,
+      "loss": 0.241,
+      "step": 1311
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.984103990405158e-05,
+      "loss": 0.2374,
+      "step": 1312
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.98406662173983e-05,
+      "loss": 0.2454,
+      "step": 1313
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9840292095552728e-05,
+      "loss": 0.2653,
+      "step": 1314
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9839917538531418e-05,
+      "loss": 0.2816,
+      "step": 1315
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.983954254635093e-05,
+      "loss": 0.2578,
+      "step": 1316
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9839167119027842e-05,
+      "loss": 0.2557,
+      "step": 1317
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9838791256578764e-05,
+      "loss": 0.2699,
+      "step": 1318
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9838414959020316e-05,
+      "loss": 0.2589,
+      "step": 1319
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.983803822636914e-05,
+      "loss": 0.2564,
+      "step": 1320
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.98376610586419e-05,
+      "loss": 0.2714,
+      "step": 1321
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9837283455855264e-05,
+      "loss": 0.2709,
+      "step": 1322
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9836905418025945e-05,
+      "loss": 0.282,
+      "step": 1323
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9836526945170652e-05,
+      "loss": 0.2419,
+      "step": 1324
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9836148037306126e-05,
+      "loss": 0.2587,
+      "step": 1325
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9835768694449123e-05,
+      "loss": 0.239,
+      "step": 1326
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9835388916616423e-05,
+      "loss": 0.2331,
+      "step": 1327
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9835008703824813e-05,
+      "loss": 0.295,
+      "step": 1328
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9834628056091113e-05,
+      "loss": 0.2381,
+      "step": 1329
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.983424697343216e-05,
+      "loss": 0.2769,
+      "step": 1330
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.98338654558648e-05,
+      "loss": 0.3135,
+      "step": 1331
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9833483503405907e-05,
+      "loss": 0.2322,
+      "step": 1332
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9833101116072375e-05,
+      "loss": 0.2911,
+      "step": 1333
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9832718293881113e-05,
+      "loss": 0.2296,
+      "step": 1334
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.983233503684905e-05,
+      "loss": 0.2332,
+      "step": 1335
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9831951344993137e-05,
+      "loss": 0.2499,
+      "step": 1336
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.983156721833034e-05,
+      "loss": 0.265,
+      "step": 1337
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.983118265687765e-05,
+      "loss": 0.291,
+      "step": 1338
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.983079766065207e-05,
+      "loss": 0.2793,
+      "step": 1339
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9830412229670627e-05,
+      "loss": 0.2932,
+      "step": 1340
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.983002636395037e-05,
+      "loss": 0.2469,
+      "step": 1341
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.982964006350836e-05,
+      "loss": 0.2884,
+      "step": 1342
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9829253328361678e-05,
+      "loss": 0.2959,
+      "step": 1343
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9828866158527433e-05,
+      "loss": 0.2413,
+      "step": 1344
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9828478554022744e-05,
+      "loss": 0.2476,
+      "step": 1345
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9828090514864752e-05,
+      "loss": 0.2721,
+      "step": 1346
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9827702041070617e-05,
+      "loss": 0.2562,
+      "step": 1347
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.982731313265752e-05,
+      "loss": 0.2301,
+      "step": 1348
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.982692378964266e-05,
+      "loss": 0.2444,
+      "step": 1349
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.982653401204326e-05,
+      "loss": 0.2389,
+      "step": 1350
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.982614379987655e-05,
+      "loss": 0.2579,
+      "step": 1351
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9825753153159787e-05,
+      "loss": 0.2936,
+      "step": 1352
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9825362071910252e-05,
+      "loss": 0.2652,
+      "step": 1353
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9824970556145237e-05,
+      "loss": 0.2369,
+      "step": 1354
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9824578605882057e-05,
+      "loss": 0.2951,
+      "step": 1355
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9824186221138047e-05,
+      "loss": 0.2443,
+      "step": 1356
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9823793401930556e-05,
+      "loss": 0.2537,
+      "step": 1357
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9823400148276957e-05,
+      "loss": 0.2849,
+      "step": 1358
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9823006460194643e-05,
+      "loss": 0.2571,
+      "step": 1359
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9822612337701023e-05,
+      "loss": 0.2783,
+      "step": 1360
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9822217780813532e-05,
+      "loss": 0.2641,
+      "step": 1361
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.982182278954961e-05,
+      "loss": 0.2536,
+      "step": 1362
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.982142736392673e-05,
+      "loss": 0.2468,
+      "step": 1363
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9821031503962378e-05,
+      "loss": 0.2522,
+      "step": 1364
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9820635209674065e-05,
+      "loss": 0.2395,
+      "step": 1365
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.982023848107931e-05,
+      "loss": 0.3112,
+      "step": 1366
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9819841318195657e-05,
+      "loss": 0.2548,
+      "step": 1367
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9819443721040675e-05,
+      "loss": 0.2385,
+      "step": 1368
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.981904568963195e-05,
+      "loss": 0.282,
+      "step": 1369
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9818647223987077e-05,
+      "loss": 0.2578,
+      "step": 1370
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.981824832412368e-05,
+      "loss": 0.2605,
+      "step": 1371
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9817848990059404e-05,
+      "loss": 0.3074,
+      "step": 1372
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9817449221811906e-05,
+      "loss": 0.2659,
+      "step": 1373
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9817049019398865e-05,
+      "loss": 0.2787,
+      "step": 1374
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9816648382837978e-05,
+      "loss": 0.2427,
+      "step": 1375
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9816247312146966e-05,
+      "loss": 0.254,
+      "step": 1376
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9815845807343562e-05,
+      "loss": 0.2334,
+      "step": 1377
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9815443868445528e-05,
+      "loss": 0.2415,
+      "step": 1378
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9815041495470636e-05,
+      "loss": 0.2467,
+      "step": 1379
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9814638688436676e-05,
+      "loss": 0.3,
+      "step": 1380
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.981423544736147e-05,
+      "loss": 0.2549,
+      "step": 1381
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9813831772262843e-05,
+      "loss": 0.3071,
+      "step": 1382
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9813427663158652e-05,
+      "loss": 0.2652,
+      "step": 1383
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.981302312006677e-05,
+      "loss": 0.2844,
+      "step": 1384
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9812618143005084e-05,
+      "loss": 0.2213,
+      "step": 1385
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9812212731991503e-05,
+      "loss": 0.2324,
+      "step": 1386
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9811806887043957e-05,
+      "loss": 0.2381,
+      "step": 1387
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9811400608180395e-05,
+      "loss": 0.2504,
+      "step": 1388
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.981099389541878e-05,
+      "loss": 0.2586,
+      "step": 1389
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9810586748777106e-05,
+      "loss": 0.26,
+      "step": 1390
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9810179168273373e-05,
+      "loss": 0.2888,
+      "step": 1391
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9809771153925607e-05,
+      "loss": 0.2523,
+      "step": 1392
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9809362705751853e-05,
+      "loss": 0.2636,
+      "step": 1393
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.980895382377017e-05,
+      "loss": 0.2489,
+      "step": 1394
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.980854450799865e-05,
+      "loss": 0.2482,
+      "step": 1395
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9808134758455385e-05,
+      "loss": 0.2713,
+      "step": 1396
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9807724575158498e-05,
+      "loss": 0.2531,
+      "step": 1397
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.980731395812613e-05,
+      "loss": 0.2758,
+      "step": 1398
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.980690290737644e-05,
+      "loss": 0.2131,
+      "step": 1399
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9806491422927606e-05,
+      "loss": 0.2702,
+      "step": 1400
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9806079504797826e-05,
+      "loss": 0.2794,
+      "step": 1401
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9805667153005313e-05,
+      "loss": 0.216,
+      "step": 1402
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9805254367568313e-05,
+      "loss": 0.2946,
+      "step": 1403
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9804841148505068e-05,
+      "loss": 0.2769,
+      "step": 1404
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.980442749583386e-05,
+      "loss": 0.2586,
+      "step": 1405
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.980401340957298e-05,
+      "loss": 0.3112,
+      "step": 1406
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.980359888974074e-05,
+      "loss": 0.3019,
+      "step": 1407
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9803183936355476e-05,
+      "loss": 0.2743,
+      "step": 1408
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9802768549435533e-05,
+      "loss": 0.2785,
+      "step": 1409
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9802352728999283e-05,
+      "loss": 0.2414,
+      "step": 1410
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9801936475065116e-05,
+      "loss": 0.2582,
+      "step": 1411
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9801519787651437e-05,
+      "loss": 0.2526,
+      "step": 1412
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.980110266677668e-05,
+      "loss": 0.2735,
+      "step": 1413
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9800685112459288e-05,
+      "loss": 0.2032,
+      "step": 1414
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9800267124717725e-05,
+      "loss": 0.2545,
+      "step": 1415
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.979984870357048e-05,
+      "loss": 0.2208,
+      "step": 1416
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9799429849036053e-05,
+      "loss": 0.2449,
+      "step": 1417
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9799010561132973e-05,
+      "loss": 0.2324,
+      "step": 1418
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9798590839879778e-05,
+      "loss": 0.2878,
+      "step": 1419
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.979817068529503e-05,
+      "loss": 0.243,
+      "step": 1420
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.979775009739731e-05,
+      "loss": 0.2507,
+      "step": 1421
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.979732907620522e-05,
+      "loss": 0.2863,
+      "step": 1422
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.979690762173738e-05,
+      "loss": 0.2345,
+      "step": 1423
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.979648573401242e-05,
+      "loss": 0.2493,
+      "step": 1424
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.979606341304901e-05,
+      "loss": 0.2315,
+      "step": 1425
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9795640658865818e-05,
+      "loss": 0.2702,
+      "step": 1426
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9795217471481543e-05,
+      "loss": 0.2204,
+      "step": 1427
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.97947938509149e-05,
+      "loss": 0.2381,
+      "step": 1428
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9794369797184618e-05,
+      "loss": 0.2478,
+      "step": 1429
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9793945310309456e-05,
+      "loss": 0.2374,
+      "step": 1430
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.979352039030819e-05,
+      "loss": 0.2571,
+      "step": 1431
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9793095037199605e-05,
+      "loss": 0.2656,
+      "step": 1432
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9792669251002506e-05,
+      "loss": 0.2465,
+      "step": 1433
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9792243031735737e-05,
+      "loss": 0.2594,
+      "step": 1434
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9791816379418142e-05,
+      "loss": 0.2208,
+      "step": 1435
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9791389294068587e-05,
+      "loss": 0.2255,
+      "step": 1436
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9790961775705957e-05,
+      "loss": 0.2496,
+      "step": 1437
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9790533824349164e-05,
+      "loss": 0.2857,
+      "step": 1438
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.979010544001713e-05,
+      "loss": 0.2707,
+      "step": 1439
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9789676622728803e-05,
+      "loss": 0.2271,
+      "step": 1440
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.978924737250314e-05,
+      "loss": 0.2163,
+      "step": 1441
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9788817689359132e-05,
+      "loss": 0.2398,
+      "step": 1442
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9788387573315778e-05,
+      "loss": 0.2655,
+      "step": 1443
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9787957024392103e-05,
+      "loss": 0.2758,
+      "step": 1444
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.978752604260714e-05,
+      "loss": 0.2464,
+      "step": 1445
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9787094627979954e-05,
+      "loss": 0.2774,
+      "step": 1446
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9786662780529623e-05,
+      "loss": 0.3297,
+      "step": 1447
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9786230500275246e-05,
+      "loss": 0.2477,
+      "step": 1448
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9785797787235936e-05,
+      "loss": 0.23,
+      "step": 1449
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9785364641430834e-05,
+      "loss": 0.266,
+      "step": 1450
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9784931062879097e-05,
+      "loss": 0.2502,
+      "step": 1451
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.978449705159989e-05,
+      "loss": 0.2112,
+      "step": 1452
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9784062607612412e-05,
+      "loss": 0.2597,
+      "step": 1453
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9783627730935883e-05,
+      "loss": 0.2218,
+      "step": 1454
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9783192421589523e-05,
+      "loss": 0.2586,
+      "step": 1455
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.978275667959259e-05,
+      "loss": 0.2495,
+      "step": 1456
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9782320504964353e-05,
+      "loss": 0.2442,
+      "step": 1457
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9781883897724103e-05,
+      "loss": 0.2895,
+      "step": 1458
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.978144685789114e-05,
+      "loss": 0.284,
+      "step": 1459
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9781009385484803e-05,
+      "loss": 0.2333,
+      "step": 1460
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9780571480524433e-05,
+      "loss": 0.2277,
+      "step": 1461
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9780133143029395e-05,
+      "loss": 0.2664,
+      "step": 1462
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.977969437301908e-05,
+      "loss": 0.2385,
+      "step": 1463
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.977925517051288e-05,
+      "loss": 0.2521,
+      "step": 1464
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.977881553553023e-05,
+      "loss": 0.2472,
+      "step": 1465
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.977837546809057e-05,
+      "loss": 0.2542,
+      "step": 1466
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.977793496821336e-05,
+      "loss": 0.2226,
+      "step": 1467
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9777494035918078e-05,
+      "loss": 0.2661,
+      "step": 1468
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9777052671224225e-05,
+      "loss": 0.2435,
+      "step": 1469
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9776610874151324e-05,
+      "loss": 0.2745,
+      "step": 1470
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.977616864471891e-05,
+      "loss": 0.2545,
+      "step": 1471
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9775725982946535e-05,
+      "loss": 0.2467,
+      "step": 1472
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9775282888853784e-05,
+      "loss": 0.244,
+      "step": 1473
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9774839362460247e-05,
+      "loss": 0.2694,
+      "step": 1474
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9774395403785544e-05,
+      "loss": 0.2246,
+      "step": 1475
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.97739510128493e-05,
+      "loss": 0.2645,
+      "step": 1476
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.977350618967117e-05,
+      "loss": 0.2407,
+      "step": 1477
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9773060934270835e-05,
+      "loss": 0.2815,
+      "step": 1478
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9772615246667977e-05,
+      "loss": 0.2279,
+      "step": 1479
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9772169126882302e-05,
+      "loss": 0.2615,
+      "step": 1480
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9771722574933553e-05,
+      "loss": 0.2068,
+      "step": 1481
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9771275590841466e-05,
+      "loss": 0.2266,
+      "step": 1482
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9770828174625814e-05,
+      "loss": 0.2749,
+      "step": 1483
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9770380326306383e-05,
+      "loss": 0.3071,
+      "step": 1484
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9769932045902975e-05,
+      "loss": 0.266,
+      "step": 1485
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.976948333343542e-05,
+      "loss": 0.2589,
+      "step": 1486
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9769034188923556e-05,
+      "loss": 0.257,
+      "step": 1487
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9768584612387255e-05,
+      "loss": 0.2206,
+      "step": 1488
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.976813460384639e-05,
+      "loss": 0.2753,
+      "step": 1489
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.976768416332086e-05,
+      "loss": 0.28,
+      "step": 1490
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.97672332908306e-05,
+      "loss": 0.2745,
+      "step": 1491
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9766781986395532e-05,
+      "loss": 0.2776,
+      "step": 1492
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9766330250035625e-05,
+      "loss": 0.2528,
+      "step": 1493
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9765878081770856e-05,
+      "loss": 0.226,
+      "step": 1494
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.976542548162122e-05,
+      "loss": 0.2624,
+      "step": 1495
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.976497244960673e-05,
+      "loss": 0.2888,
+      "step": 1496
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9764518985747423e-05,
+      "loss": 0.2604,
+      "step": 1497
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.976406509006335e-05,
+      "loss": 0.254,
+      "step": 1498
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.976361076257459e-05,
+      "loss": 0.2498,
+      "step": 1499
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.976315600330123e-05,
+      "loss": 0.2173,
+      "step": 1500
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9762700812263386e-05,
+      "loss": 0.277,
+      "step": 1501
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9762245189481187e-05,
+      "loss": 0.2557,
+      "step": 1502
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.976178913497478e-05,
+      "loss": 0.2659,
+      "step": 1503
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.976133264876433e-05,
+      "loss": 0.2481,
+      "step": 1504
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9760875730870033e-05,
+      "loss": 0.2516,
+      "step": 1505
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.976041838131209e-05,
+      "loss": 0.2389,
+      "step": 1506
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9759960600110732e-05,
+      "loss": 0.257,
+      "step": 1507
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9759502387286196e-05,
+      "loss": 0.2377,
+      "step": 1508
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9759043742858753e-05,
+      "loss": 0.2348,
+      "step": 1509
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.975858466684868e-05,
+      "loss": 0.2406,
+      "step": 1510
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9758125159276284e-05,
+      "loss": 0.2298,
+      "step": 1511
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9757665220161884e-05,
+      "loss": 0.2671,
+      "step": 1512
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9757204849525825e-05,
+      "loss": 0.2716,
+      "step": 1513
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9756744047388458e-05,
+      "loss": 0.2485,
+      "step": 1514
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.975628281377017e-05,
+      "loss": 0.2445,
+      "step": 1515
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.975582114869135e-05,
+      "loss": 0.3167,
+      "step": 1516
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.975535905217242e-05,
+      "loss": 0.2393,
+      "step": 1517
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9754896524233812e-05,
+      "loss": 0.2222,
+      "step": 1518
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.975443356489599e-05,
+      "loss": 0.2829,
+      "step": 1519
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9753970174179416e-05,
+      "loss": 0.2285,
+      "step": 1520
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.975350635210459e-05,
+      "loss": 0.2322,
+      "step": 1521
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.975304209869202e-05,
+      "loss": 0.2816,
+      "step": 1522
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9752577413962242e-05,
+      "loss": 0.242,
+      "step": 1523
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9752112297935802e-05,
+      "loss": 0.2777,
+      "step": 1524
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9751646750633268e-05,
+      "loss": 0.2594,
+      "step": 1525
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9751180772075234e-05,
+      "loss": 0.2961,
+      "step": 1526
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9750714362282302e-05,
+      "loss": 0.2461,
+      "step": 1527
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9750247521275105e-05,
+      "loss": 0.2963,
+      "step": 1528
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.974978024907428e-05,
+      "loss": 0.2316,
+      "step": 1529
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.97493125457005e-05,
+      "loss": 0.2585,
+      "step": 1530
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9748844411174438e-05,
+      "loss": 0.252,
+      "step": 1531
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.974837584551681e-05,
+      "loss": 0.2934,
+      "step": 1532
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9747906848748325e-05,
+      "loss": 0.2633,
+      "step": 1533
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9747437420889734e-05,
+      "loss": 0.258,
+      "step": 1534
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9746967561961792e-05,
+      "loss": 0.259,
+      "step": 1535
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9746497271985278e-05,
+      "loss": 0.2656,
+      "step": 1536
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.974602655098099e-05,
+      "loss": 0.2229,
+      "step": 1537
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.974555539896975e-05,
+      "loss": 0.2273,
+      "step": 1538
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9745083815972387e-05,
+      "loss": 0.2896,
+      "step": 1539
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9744611802009755e-05,
+      "loss": 0.2117,
+      "step": 1540
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9744139357102737e-05,
+      "loss": 0.2596,
+      "step": 1541
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.974366648127222e-05,
+      "loss": 0.2569,
+      "step": 1542
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9743193174539123e-05,
+      "loss": 0.2454,
+      "step": 1543
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9742719436924367e-05,
+      "loss": 0.2347,
+      "step": 1544
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.974224526844891e-05,
+      "loss": 0.2612,
+      "step": 1545
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.974177066913372e-05,
+      "loss": 0.2449,
+      "step": 1546
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9741295638999785e-05,
+      "loss": 0.2881,
+      "step": 1547
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9740820178068115e-05,
+      "loss": 0.2619,
+      "step": 1548
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.974034428635973e-05,
+      "loss": 0.2614,
+      "step": 1549
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9739867963895688e-05,
+      "loss": 0.2116,
+      "step": 1550
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.973939121069704e-05,
+      "loss": 0.2278,
+      "step": 1551
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.973891402678488e-05,
+      "loss": 0.2268,
+      "step": 1552
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9738436412180306e-05,
+      "loss": 0.2326,
+      "step": 1553
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.973795836690444e-05,
+      "loss": 0.2394,
+      "step": 1554
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9737479890978424e-05,
+      "loss": 0.2587,
+      "step": 1555
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9737000984423417e-05,
+      "loss": 0.2506,
+      "step": 1556
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9736521647260603e-05,
+      "loss": 0.2692,
+      "step": 1557
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9736041879511173e-05,
+      "loss": 0.2565,
+      "step": 1558
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9735561681196352e-05,
+      "loss": 0.262,
+      "step": 1559
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9735081052337368e-05,
+      "loss": 0.2569,
+      "step": 1560
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.973459999295548e-05,
+      "loss": 0.269,
+      "step": 1561
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9734118503071963e-05,
+      "loss": 0.2451,
+      "step": 1562
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.973363658270811e-05,
+      "loss": 0.2666,
+      "step": 1563
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9733154231885234e-05,
+      "loss": 0.2164,
+      "step": 1564
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9732671450624664e-05,
+      "loss": 0.2639,
+      "step": 1565
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.973218823894775e-05,
+      "loss": 0.2606,
+      "step": 1566
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9731704596875868e-05,
+      "loss": 0.2657,
+      "step": 1567
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.97312205244304e-05,
+      "loss": 0.2452,
+      "step": 1568
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9730736021632756e-05,
+      "loss": 0.2681,
+      "step": 1569
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.973025108850436e-05,
+      "loss": 0.2456,
+      "step": 1570
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9729765725066664e-05,
+      "loss": 0.2405,
+      "step": 1571
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.972927993134112e-05,
+      "loss": 0.2771,
+      "step": 1572
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9728793707349233e-05,
+      "loss": 0.237,
+      "step": 1573
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9728307053112484e-05,
+      "loss": 0.2601,
+      "step": 1574
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9727819968652404e-05,
+      "loss": 0.2517,
+      "step": 1575
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9727332453990535e-05,
+      "loss": 0.2463,
+      "step": 1576
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9726844509148434e-05,
+      "loss": 0.2604,
+      "step": 1577
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9726356134147682e-05,
+      "loss": 0.2609,
+      "step": 1578
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9725867329009874e-05,
+      "loss": 0.2452,
+      "step": 1579
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.972537809375663e-05,
+      "loss": 0.2356,
+      "step": 1580
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9724888428409582e-05,
+      "loss": 0.2792,
+      "step": 1581
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.972439833299039e-05,
+      "loss": 0.248,
+      "step": 1582
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9723907807520722e-05,
+      "loss": 0.2578,
+      "step": 1583
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.972341685202228e-05,
+      "loss": 0.2457,
+      "step": 1584
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9722925466516764e-05,
+      "loss": 0.2499,
+      "step": 1585
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9722433651025915e-05,
+      "loss": 0.2361,
+      "step": 1586
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9721941405571478e-05,
+      "loss": 0.2515,
+      "step": 1587
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9721448730175223e-05,
+      "loss": 0.2486,
+      "step": 1588
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.972095562485894e-05,
+      "loss": 0.2935,
+      "step": 1589
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.972046208964443e-05,
+      "loss": 0.2509,
+      "step": 1590
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9719968124553525e-05,
+      "loss": 0.2436,
+      "step": 1591
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9719473729608068e-05,
+      "loss": 0.2482,
+      "step": 1592
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.971897890482992e-05,
+      "loss": 0.2571,
+      "step": 1593
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9718483650240974e-05,
+      "loss": 0.2886,
+      "step": 1594
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9717987965863123e-05,
+      "loss": 0.2241,
+      "step": 1595
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9717491851718287e-05,
+      "loss": 0.2472,
+      "step": 1596
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9716995307828413e-05,
+      "loss": 0.236,
+      "step": 1597
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9716498334215455e-05,
+      "loss": 0.2182,
+      "step": 1598
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.971600093090139e-05,
+      "loss": 0.2454,
+      "step": 1599
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9715503097908223e-05,
+      "loss": 0.2962,
+      "step": 1600
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9715004835257964e-05,
+      "loss": 0.2751,
+      "step": 1601
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9714506142972648e-05,
+      "loss": 0.2515,
+      "step": 1602
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.971400702107433e-05,
+      "loss": 0.2378,
+      "step": 1603
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9713507469585084e-05,
+      "loss": 0.2842,
+      "step": 1604
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9713007488527e-05,
+      "loss": 0.2748,
+      "step": 1605
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9712507077922187e-05,
+      "loss": 0.2554,
+      "step": 1606
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9712006237792784e-05,
+      "loss": 0.2503,
+      "step": 1607
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.971150496816093e-05,
+      "loss": 0.2575,
+      "step": 1608
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.97110032690488e-05,
+      "loss": 0.2677,
+      "step": 1609
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.971050114047858e-05,
+      "loss": 0.252,
+      "step": 1610
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9709998582472473e-05,
+      "loss": 0.2773,
+      "step": 1611
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.970949559505271e-05,
+      "loss": 0.2334,
+      "step": 1612
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9708992178241525e-05,
+      "loss": 0.264,
+      "step": 1613
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9708488332061188e-05,
+      "loss": 0.2838,
+      "step": 1614
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9707984056533982e-05,
+      "loss": 0.2736,
+      "step": 1615
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9707479351682205e-05,
+      "loss": 0.2187,
+      "step": 1616
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9706974217528176e-05,
+      "loss": 0.3087,
+      "step": 1617
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.970646865409424e-05,
+      "loss": 0.2509,
+      "step": 1618
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.970596266140275e-05,
+      "loss": 0.2426,
+      "step": 1619
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9705456239476078e-05,
+      "loss": 0.2648,
+      "step": 1620
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9704949388336633e-05,
+      "loss": 0.2171,
+      "step": 1621
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.970444210800682e-05,
+      "loss": 0.2607,
+      "step": 1622
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.970393439850908e-05,
+      "loss": 0.2798,
+      "step": 1623
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9703426259865855e-05,
+      "loss": 0.2521,
+      "step": 1624
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9702917692099627e-05,
+      "loss": 0.2358,
+      "step": 1625
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9702408695232878e-05,
+      "loss": 0.2687,
+      "step": 1626
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9701899269288128e-05,
+      "loss": 0.2314,
+      "step": 1627
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9701389414287898e-05,
+      "loss": 0.2215,
+      "step": 1628
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.970087913025474e-05,
+      "loss": 0.2424,
+      "step": 1629
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.970036841721122e-05,
+      "loss": 0.2466,
+      "step": 1630
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.969985727517992e-05,
+      "loss": 0.2588,
+      "step": 1631
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.969934570418345e-05,
+      "loss": 0.2555,
+      "step": 1632
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.969883370424443e-05,
+      "loss": 0.2543,
+      "step": 1633
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9698321275385507e-05,
+      "loss": 0.2778,
+      "step": 1634
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.969780841762934e-05,
+      "loss": 0.2476,
+      "step": 1635
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9697295130998604e-05,
+      "loss": 0.2406,
+      "step": 1636
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9696781415516006e-05,
+      "loss": 0.2235,
+      "step": 1637
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9696267271204262e-05,
+      "loss": 0.251,
+      "step": 1638
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9695752698086107e-05,
+      "loss": 0.2628,
+      "step": 1639
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9695237696184303e-05,
+      "loss": 0.2683,
+      "step": 1640
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9694722265521623e-05,
+      "loss": 0.2348,
+      "step": 1641
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9694206406120858e-05,
+      "loss": 0.2411,
+      "step": 1642
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9693690118004825e-05,
+      "loss": 0.2815,
+      "step": 1643
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9693173401196355e-05,
+      "loss": 0.292,
+      "step": 1644
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.96926562557183e-05,
+      "loss": 0.271,
+      "step": 1645
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.969213868159353e-05,
+      "loss": 0.2293,
+      "step": 1646
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.969162067884493e-05,
+      "loss": 0.2512,
+      "step": 1647
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9691102247495415e-05,
+      "loss": 0.2403,
+      "step": 1648
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.969058338756791e-05,
+      "loss": 0.2525,
+      "step": 1649
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9690064099085356e-05,
+      "loss": 0.2456,
+      "step": 1650
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9689544382070723e-05,
+      "loss": 0.2769,
+      "step": 1651
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9689024236546995e-05,
+      "loss": 0.2699,
+      "step": 1652
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9688503662537172e-05,
+      "loss": 0.2526,
+      "step": 1653
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9687982660064278e-05,
+      "loss": 0.2522,
+      "step": 1654
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9687461229151354e-05,
+      "loss": 0.2878,
+      "step": 1655
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9686939369821455e-05,
+      "loss": 0.2419,
+      "step": 1656
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9686417082097667e-05,
+      "loss": 0.2679,
+      "step": 1657
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9685894366003083e-05,
+      "loss": 0.2264,
+      "step": 1658
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9685371221560817e-05,
+      "loss": 0.2657,
+      "step": 1659
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9684847648794014e-05,
+      "loss": 0.2399,
+      "step": 1660
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.968432364772582e-05,
+      "loss": 0.2232,
+      "step": 1661
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.968379921837941e-05,
+      "loss": 0.2351,
+      "step": 1662
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9683274360777974e-05,
+      "loss": 0.2366,
+      "step": 1663
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.968274907494473e-05,
+      "loss": 0.2679,
+      "step": 1664
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9682223360902903e-05,
+      "loss": 0.2446,
+      "step": 1665
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9681697218675745e-05,
+      "loss": 0.2638,
+      "step": 1666
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9681170648286524e-05,
+      "loss": 0.1969,
+      "step": 1667
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.968064364975852e-05,
+      "loss": 0.2484,
+      "step": 1668
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.968011622311505e-05,
+      "loss": 0.2412,
+      "step": 1669
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9679588368379427e-05,
+      "loss": 0.2636,
+      "step": 1670
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9679060085575007e-05,
+      "loss": 0.254,
+      "step": 1671
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9678531374725143e-05,
+      "loss": 0.2809,
+      "step": 1672
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.967800223585322e-05,
+      "loss": 0.2151,
+      "step": 1673
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9677472668982643e-05,
+      "loss": 0.2787,
+      "step": 1674
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9676942674136822e-05,
+      "loss": 0.2564,
+      "step": 1675
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9676412251339205e-05,
+      "loss": 0.2381,
+      "step": 1676
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9675881400613245e-05,
+      "loss": 0.2312,
+      "step": 1677
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9675350121982415e-05,
+      "loss": 0.2631,
+      "step": 1678
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.967481841547022e-05,
+      "loss": 0.241,
+      "step": 1679
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.967428628110016e-05,
+      "loss": 0.2576,
+      "step": 1680
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.967375371889578e-05,
+      "loss": 0.2549,
+      "step": 1681
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9673220728880628e-05,
+      "loss": 0.2464,
+      "step": 1682
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.967268731107827e-05,
+      "loss": 0.2465,
+      "step": 1683
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9672153465512305e-05,
+      "loss": 0.2303,
+      "step": 1684
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.967161919220634e-05,
+      "loss": 0.2762,
+      "step": 1685
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9671084491183993e-05,
+      "loss": 0.2651,
+      "step": 1686
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9670549362468923e-05,
+      "loss": 0.226,
+      "step": 1687
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9670013806084785e-05,
+      "loss": 0.264,
+      "step": 1688
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9669477822055273e-05,
+      "loss": 0.2581,
+      "step": 1689
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.966894141040408e-05,
+      "loss": 0.2694,
+      "step": 1690
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9668404571154937e-05,
+      "loss": 0.2184,
+      "step": 1691
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.966786730433158e-05,
+      "loss": 0.2191,
+      "step": 1692
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9667329609957773e-05,
+      "loss": 0.2327,
+      "step": 1693
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9666791488057292e-05,
+      "loss": 0.2464,
+      "step": 1694
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9666252938653937e-05,
+      "loss": 0.2549,
+      "step": 1695
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.966571396177152e-05,
+      "loss": 0.2586,
+      "step": 1696
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9665174557433882e-05,
+      "loss": 0.2761,
+      "step": 1697
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9664634725664874e-05,
+      "loss": 0.2722,
+      "step": 1698
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.966409446648837e-05,
+      "loss": 0.2588,
+      "step": 1699
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9663553779928267e-05,
+      "loss": 0.2804,
+      "step": 1700
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9663012666008468e-05,
+      "loss": 0.2266,
+      "step": 1701
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.966247112475291e-05,
+      "loss": 0.2825,
+      "step": 1702
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.966192915618554e-05,
+      "loss": 0.2674,
+      "step": 1703
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9661386760330322e-05,
+      "loss": 0.2173,
+      "step": 1704
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.966084393721125e-05,
+      "loss": 0.251,
+      "step": 1705
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9660300686852328e-05,
+      "loss": 0.2826,
+      "step": 1706
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9659757009277576e-05,
+      "loss": 0.223,
+      "step": 1707
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.965921290451104e-05,
+      "loss": 0.2521,
+      "step": 1708
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9658668372576784e-05,
+      "loss": 0.2172,
+      "step": 1709
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9658123413498886e-05,
+      "loss": 0.2365,
+      "step": 1710
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.965757802730145e-05,
+      "loss": 0.2634,
+      "step": 1711
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9657032214008596e-05,
+      "loss": 0.2739,
+      "step": 1712
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9656485973644454e-05,
+      "loss": 0.2633,
+      "step": 1713
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.965593930623319e-05,
+      "loss": 0.2648,
+      "step": 1714
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9655392211798975e-05,
+      "loss": 0.2955,
+      "step": 1715
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9654844690366005e-05,
+      "loss": 0.2507,
+      "step": 1716
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9654296741958488e-05,
+      "loss": 0.2681,
+      "step": 1717
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.965374836660067e-05,
+      "loss": 0.2267,
+      "step": 1718
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.965319956431679e-05,
+      "loss": 0.2834,
+      "step": 1719
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.965265033513112e-05,
+      "loss": 0.2342,
+      "step": 1720
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.965210067906795e-05,
+      "loss": 0.2495,
+      "step": 1721
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9651550596151593e-05,
+      "loss": 0.2341,
+      "step": 1722
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.965100008640637e-05,
+      "loss": 0.2425,
+      "step": 1723
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9650449149856627e-05,
+      "loss": 0.2302,
+      "step": 1724
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9649897786526732e-05,
+      "loss": 0.231,
+      "step": 1725
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9649345996441063e-05,
+      "loss": 0.2591,
+      "step": 1726
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.964879377962403e-05,
+      "loss": 0.2414,
+      "step": 1727
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.964824113610005e-05,
+      "loss": 0.2219,
+      "step": 1728
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9647688065893558e-05,
+      "loss": 0.2695,
+      "step": 1729
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.964713456902902e-05,
+      "loss": 0.2697,
+      "step": 1730
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9646580645530912e-05,
+      "loss": 0.2347,
+      "step": 1731
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.964602629542373e-05,
+      "loss": 0.2354,
+      "step": 1732
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9645471518731987e-05,
+      "loss": 0.2639,
+      "step": 1733
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.964491631548022e-05,
+      "loss": 0.2278,
+      "step": 1734
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9644360685692986e-05,
+      "loss": 0.2143,
+      "step": 1735
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.964380462939485e-05,
+      "loss": 0.2404,
+      "step": 1736
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.964324814661041e-05,
+      "loss": 0.2599,
+      "step": 1737
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9642691237364266e-05,
+      "loss": 0.239,
+      "step": 1738
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9642133901681056e-05,
+      "loss": 0.2796,
+      "step": 1739
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9641576139585425e-05,
+      "loss": 0.3044,
+      "step": 1740
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9641017951102038e-05,
+      "loss": 0.2847,
+      "step": 1741
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.964045933625558e-05,
+      "loss": 0.2499,
+      "step": 1742
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.963990029507076e-05,
+      "loss": 0.2679,
+      "step": 1743
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9639340827572294e-05,
+      "loss": 0.2407,
+      "step": 1744
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9638780933784924e-05,
+      "loss": 0.2827,
+      "step": 1745
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9638220613733415e-05,
+      "loss": 0.3004,
+      "step": 1746
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.963765986744255e-05,
+      "loss": 0.2584,
+      "step": 1747
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9637098694937116e-05,
+      "loss": 0.2211,
+      "step": 1748
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9636537096241942e-05,
+      "loss": 0.2046,
+      "step": 1749
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9635975071381853e-05,
+      "loss": 0.2039,
+      "step": 1750
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9635412620381714e-05,
+      "loss": 0.291,
+      "step": 1751
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9634849743266395e-05,
+      "loss": 0.2408,
+      "step": 1752
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.963428644006078e-05,
+      "loss": 0.2555,
+      "step": 1753
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9633722710789797e-05,
+      "loss": 0.2444,
+      "step": 1754
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9633158555478364e-05,
+      "loss": 0.2472,
+      "step": 1755
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9632593974151434e-05,
+      "loss": 0.2647,
+      "step": 1756
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9632028966833975e-05,
+      "loss": 0.2669,
+      "step": 1757
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9631463533550975e-05,
+      "loss": 0.2432,
+      "step": 1758
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9630897674327433e-05,
+      "loss": 0.2554,
+      "step": 1759
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9630331389188382e-05,
+      "loss": 0.3113,
+      "step": 1760
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9629764678158865e-05,
+      "loss": 0.2718,
+      "step": 1761
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9629197541263937e-05,
+      "loss": 0.246,
+      "step": 1762
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9628629978528683e-05,
+      "loss": 0.2924,
+      "step": 1763
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9628061989978206e-05,
+      "loss": 0.247,
+      "step": 1764
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.962749357563762e-05,
+      "loss": 0.2313,
+      "step": 1765
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9626924735532063e-05,
+      "loss": 0.2496,
+      "step": 1766
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9626355469686694e-05,
+      "loss": 0.2544,
+      "step": 1767
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9625785778126685e-05,
+      "loss": 0.2472,
+      "step": 1768
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9625215660877234e-05,
+      "loss": 0.2169,
+      "step": 1769
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.962464511796355e-05,
+      "loss": 0.2311,
+      "step": 1770
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9624074149410862e-05,
+      "loss": 0.2493,
+      "step": 1771
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9623502755244427e-05,
+      "loss": 0.265,
+      "step": 1772
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9622930935489513e-05,
+      "loss": 0.2349,
+      "step": 1773
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.962235869017141e-05,
+      "loss": 0.2395,
+      "step": 1774
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9621786019315415e-05,
+      "loss": 0.2341,
+      "step": 1775
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9621212922946863e-05,
+      "loss": 0.2231,
+      "step": 1776
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9620639401091097e-05,
+      "loss": 0.2642,
+      "step": 1777
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9620065453773473e-05,
+      "loss": 0.3007,
+      "step": 1778
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9619491081019385e-05,
+      "loss": 0.2417,
+      "step": 1779
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9618916282854225e-05,
+      "loss": 0.2563,
+      "step": 1780
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.961834105930342e-05,
+      "loss": 0.2495,
+      "step": 1781
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.96177654103924e-05,
+      "loss": 0.235,
+      "step": 1782
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9617189336146634e-05,
+      "loss": 0.2326,
+      "step": 1783
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9616612836591586e-05,
+      "loss": 0.247,
+      "step": 1784
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.961603591175276e-05,
+      "loss": 0.2684,
+      "step": 1785
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9615458561655662e-05,
+      "loss": 0.2331,
+      "step": 1786
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.961488078632583e-05,
+      "loss": 0.2504,
+      "step": 1787
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9614302585788815e-05,
+      "loss": 0.2806,
+      "step": 1788
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9613723960070186e-05,
+      "loss": 0.2415,
+      "step": 1789
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9613144909195534e-05,
+      "loss": 0.2406,
+      "step": 1790
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9612565433190464e-05,
+      "loss": 0.2487,
+      "step": 1791
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9611985532080605e-05,
+      "loss": 0.2273,
+      "step": 1792
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9611405205891602e-05,
+      "loss": 0.2322,
+      "step": 1793
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.961082445464912e-05,
+      "loss": 0.2785,
+      "step": 1794
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9610243278378837e-05,
+      "loss": 0.2788,
+      "step": 1795
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9609661677106466e-05,
+      "loss": 0.2539,
+      "step": 1796
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9609079650857715e-05,
+      "loss": 0.2197,
+      "step": 1797
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.960849719965833e-05,
+      "loss": 0.2442,
+      "step": 1798
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9607914323534067e-05,
+      "loss": 0.2311,
+      "step": 1799
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9607331022510706e-05,
+      "loss": 0.2688,
+      "step": 1800
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.960674729661404e-05,
+      "loss": 0.2309,
+      "step": 1801
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9606163145869886e-05,
+      "loss": 0.2433,
+      "step": 1802
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9605578570304072e-05,
+      "loss": 0.2739,
+      "step": 1803
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.960499356994246e-05,
+      "loss": 0.2443,
+      "step": 1804
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9604408144810914e-05,
+      "loss": 0.2569,
+      "step": 1805
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.960382229493532e-05,
+      "loss": 0.2319,
+      "step": 1806
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.96032360203416e-05,
+      "loss": 0.2353,
+      "step": 1807
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9602649321055667e-05,
+      "loss": 0.2516,
+      "step": 1808
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9602062197103472e-05,
+      "loss": 0.2349,
+      "step": 1809
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9601474648510983e-05,
+      "loss": 0.2604,
+      "step": 1810
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9600886675304182e-05,
+      "loss": 0.222,
+      "step": 1811
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.960029827750907e-05,
+      "loss": 0.2499,
+      "step": 1812
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.959970945515167e-05,
+      "loss": 0.2495,
+      "step": 1813
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9599120208258024e-05,
+      "loss": 0.2616,
+      "step": 1814
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9598530536854182e-05,
+      "loss": 0.2381,
+      "step": 1815
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9597940440966234e-05,
+      "loss": 0.2529,
+      "step": 1816
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9597349920620266e-05,
+      "loss": 0.2767,
+      "step": 1817
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9596758975842396e-05,
+      "loss": 0.2584,
+      "step": 1818
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.959616760665876e-05,
+      "loss": 0.2397,
+      "step": 1819
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9595575813095513e-05,
+      "loss": 0.2491,
+      "step": 1820
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9594983595178818e-05,
+      "loss": 0.2305,
+      "step": 1821
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.959439095293487e-05,
+      "loss": 0.2182,
+      "step": 1822
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.959379788638988e-05,
+      "loss": 0.2444,
+      "step": 1823
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9593204395570073e-05,
+      "loss": 0.2604,
+      "step": 1824
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9592610480501696e-05,
+      "loss": 0.2865,
+      "step": 1825
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9592016141211016e-05,
+      "loss": 0.2282,
+      "step": 1826
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9591421377724315e-05,
+      "loss": 0.2122,
+      "step": 1827
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.959082619006789e-05,
+      "loss": 0.2765,
+      "step": 1828
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9590230578268078e-05,
+      "loss": 0.271,
+      "step": 1829
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9589634542351204e-05,
+      "loss": 0.2432,
+      "step": 1830
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9589038082343636e-05,
+      "loss": 0.2826,
+      "step": 1831
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9588441198271744e-05,
+      "loss": 0.2489,
+      "step": 1832
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9587843890161934e-05,
+      "loss": 0.318,
+      "step": 1833
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.958724615804061e-05,
+      "loss": 0.2243,
+      "step": 1834
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9586648001934218e-05,
+      "loss": 0.2729,
+      "step": 1835
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9586049421869204e-05,
+      "loss": 0.275,
+      "step": 1836
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.958545041787204e-05,
+      "loss": 0.273,
+      "step": 1837
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9584850989969218e-05,
+      "loss": 0.2774,
+      "step": 1838
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9584251138187246e-05,
+      "loss": 0.2354,
+      "step": 1839
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.958365086255265e-05,
+      "loss": 0.2469,
+      "step": 1840
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9583050163091978e-05,
+      "loss": 0.2583,
+      "step": 1841
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9582449039831797e-05,
+      "loss": 0.2712,
+      "step": 1842
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.958184749279869e-05,
+      "loss": 0.2746,
+      "step": 1843
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9581245522019258e-05,
+      "loss": 0.2343,
+      "step": 1844
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9580643127520124e-05,
+      "loss": 0.2689,
+      "step": 1845
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9580040309327926e-05,
+      "loss": 0.2542,
+      "step": 1846
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9579437067469327e-05,
+      "loss": 0.2182,
+      "step": 1847
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9578833401971005e-05,
+      "loss": 0.2521,
+      "step": 1848
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.957822931285965e-05,
+      "loss": 0.2489,
+      "step": 1849
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9577624800161982e-05,
+      "loss": 0.2409,
+      "step": 1850
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9577019863904737e-05,
+      "loss": 0.2537,
+      "step": 1851
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9576414504114663e-05,
+      "loss": 0.2449,
+      "step": 1852
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.957580872081853e-05,
+      "loss": 0.2191,
+      "step": 1853
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9575202514043134e-05,
+      "loss": 0.2818,
+      "step": 1854
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9574595883815282e-05,
+      "loss": 0.2151,
+      "step": 1855
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9573988830161796e-05,
+      "loss": 0.2582,
+      "step": 1856
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.957338135310953e-05,
+      "loss": 0.2578,
+      "step": 1857
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9572773452685344e-05,
+      "loss": 0.2556,
+      "step": 1858
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9572165128916124e-05,
+      "loss": 0.27,
+      "step": 1859
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9571556381828772e-05,
+      "loss": 0.2405,
+      "step": 1860
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.957094721145021e-05,
+      "loss": 0.2662,
+      "step": 1861
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9570337617807372e-05,
+      "loss": 0.2168,
+      "step": 1862
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9569727600927227e-05,
+      "loss": 0.2525,
+      "step": 1863
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9569117160836746e-05,
+      "loss": 0.2299,
+      "step": 1864
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9568506297562922e-05,
+      "loss": 0.2064,
+      "step": 1865
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9567895011132777e-05,
+      "loss": 0.2416,
+      "step": 1866
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.956728330157334e-05,
+      "loss": 0.2579,
+      "step": 1867
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9566671168911666e-05,
+      "loss": 0.2288,
+      "step": 1868
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9566058613174822e-05,
+      "loss": 0.2418,
+      "step": 1869
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9565445634389897e-05,
+      "loss": 0.2825,
+      "step": 1870
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9564832232584002e-05,
+      "loss": 0.206,
+      "step": 1871
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9564218407784267e-05,
+      "loss": 0.2729,
+      "step": 1872
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9563604160017833e-05,
+      "loss": 0.2479,
+      "step": 1873
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9562989489311864e-05,
+      "loss": 0.2641,
+      "step": 1874
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9562374395693552e-05,
+      "loss": 0.2209,
+      "step": 1875
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9561758879190084e-05,
+      "loss": 0.2716,
+      "step": 1876
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9561142939828693e-05,
+      "loss": 0.2581,
+      "step": 1877
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.956052657763661e-05,
+      "loss": 0.248,
+      "step": 1878
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.95599097926411e-05,
+      "loss": 0.2193,
+      "step": 1879
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9559292584869432e-05,
+      "loss": 0.2385,
+      "step": 1880
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.955867495434891e-05,
+      "loss": 0.2548,
+      "step": 1881
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.955805690110684e-05,
+      "loss": 0.2885,
+      "step": 1882
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.955743842517056e-05,
+      "loss": 0.2763,
+      "step": 1883
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9556819526567418e-05,
+      "loss": 0.2503,
+      "step": 1884
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9556200205324785e-05,
+      "loss": 0.2491,
+      "step": 1885
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9555580461470053e-05,
+      "loss": 0.2582,
+      "step": 1886
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9554960295030627e-05,
+      "loss": 0.2461,
+      "step": 1887
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9554339706033932e-05,
+      "loss": 0.2454,
+      "step": 1888
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9553718694507414e-05,
+      "loss": 0.2534,
+      "step": 1889
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9553097260478538e-05,
+      "loss": 0.2564,
+      "step": 1890
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.955247540397478e-05,
+      "loss": 0.2369,
+      "step": 1891
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.955185312502365e-05,
+      "loss": 0.2297,
+      "step": 1892
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9551230423652662e-05,
+      "loss": 0.2348,
+      "step": 1893
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9550607299889353e-05,
+      "loss": 0.2465,
+      "step": 1894
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9549983753761285e-05,
+      "loss": 0.2835,
+      "step": 1895
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9549359785296027e-05,
+      "loss": 0.2471,
+      "step": 1896
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9548735394521178e-05,
+      "loss": 0.2604,
+      "step": 1897
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9548110581464355e-05,
+      "loss": 0.2366,
+      "step": 1898
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.954748534615318e-05,
+      "loss": 0.253,
+      "step": 1899
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.954685968861531e-05,
+      "loss": 0.2304,
+      "step": 1900
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9546233608878412e-05,
+      "loss": 0.2555,
+      "step": 1901
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9545607106970175e-05,
+      "loss": 0.2725,
+      "step": 1902
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9544980182918305e-05,
+      "loss": 0.269,
+      "step": 1903
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.954435283675052e-05,
+      "loss": 0.2419,
+      "step": 1904
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9543725068494576e-05,
+      "loss": 0.2687,
+      "step": 1905
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.954309687817823e-05,
+      "loss": 0.2555,
+      "step": 1906
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9542468265829256e-05,
+      "loss": 0.2623,
+      "step": 1907
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9541839231475466e-05,
+      "loss": 0.2447,
+      "step": 1908
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9541209775144668e-05,
+      "loss": 0.2574,
+      "step": 1909
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9540579896864705e-05,
+      "loss": 0.2342,
+      "step": 1910
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9539949596663428e-05,
+      "loss": 0.2846,
+      "step": 1911
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9539318874568718e-05,
+      "loss": 0.2559,
+      "step": 1912
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9538687730608463e-05,
+      "loss": 0.2615,
+      "step": 1913
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9538056164810573e-05,
+      "loss": 0.2381,
+      "step": 1914
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9537424177202985e-05,
+      "loss": 0.2586,
+      "step": 1915
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.953679176781364e-05,
+      "loss": 0.2413,
+      "step": 1916
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9536158936670513e-05,
+      "loss": 0.2614,
+      "step": 1917
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9535525683801585e-05,
+      "loss": 0.2457,
+      "step": 1918
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.953489200923486e-05,
+      "loss": 0.2291,
+      "step": 1919
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9534257912998368e-05,
+      "loss": 0.2175,
+      "step": 1920
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.953362339512015e-05,
+      "loss": 0.2518,
+      "step": 1921
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.953298845562826e-05,
+      "loss": 0.2597,
+      "step": 1922
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.953235309455078e-05,
+      "loss": 0.2272,
+      "step": 1923
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.953171731191581e-05,
+      "loss": 0.2755,
+      "step": 1924
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.953108110775147e-05,
+      "loss": 0.2356,
+      "step": 1925
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.953044448208589e-05,
+      "loss": 0.2805,
+      "step": 1926
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9529807434947227e-05,
+      "loss": 0.2169,
+      "step": 1927
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.952916996636365e-05,
+      "loss": 0.2271,
+      "step": 1928
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.952853207636336e-05,
+      "loss": 0.2538,
+      "step": 1929
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9527893764974554e-05,
+      "loss": 0.2571,
+      "step": 1930
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9527255032225465e-05,
+      "loss": 0.2733,
+      "step": 1931
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9526615878144347e-05,
+      "loss": 0.3154,
+      "step": 1932
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.952597630275946e-05,
+      "loss": 0.2283,
+      "step": 1933
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9525336306099082e-05,
+      "loss": 0.2377,
+      "step": 1934
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.952469588819153e-05,
+      "loss": 0.3002,
+      "step": 1935
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9524055049065117e-05,
+      "loss": 0.2499,
+      "step": 1936
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9523413788748186e-05,
+      "loss": 0.2203,
+      "step": 1937
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.95227721072691e-05,
+      "loss": 0.2279,
+      "step": 1938
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9522130004656224e-05,
+      "loss": 0.235,
+      "step": 1939
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9521487480937966e-05,
+      "loss": 0.249,
+      "step": 1940
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9520844536142736e-05,
+      "loss": 0.2369,
+      "step": 1941
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9520201170298972e-05,
+      "loss": 0.2576,
+      "step": 1942
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.951955738343512e-05,
+      "loss": 0.2431,
+      "step": 1943
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9518913175579653e-05,
+      "loss": 0.2114,
+      "step": 1944
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.951826854676106e-05,
+      "loss": 0.2281,
+      "step": 1945
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9517623497007855e-05,
+      "loss": 0.2361,
+      "step": 1946
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9516978026348555e-05,
+      "loss": 0.2638,
+      "step": 1947
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9516332134811713e-05,
+      "loss": 0.262,
+      "step": 1948
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9515685822425884e-05,
+      "loss": 0.2241,
+      "step": 1949
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.951503908921966e-05,
+      "loss": 0.2273,
+      "step": 1950
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.951439193522164e-05,
+      "loss": 0.224,
+      "step": 1951
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9513744360460437e-05,
+      "loss": 0.2407,
+      "step": 1952
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9513096364964698e-05,
+      "loss": 0.2372,
+      "step": 1953
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9512447948763073e-05,
+      "loss": 0.2294,
+      "step": 1954
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9511799111884242e-05,
+      "loss": 0.2277,
+      "step": 1955
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9511149854356897e-05,
+      "loss": 0.2357,
+      "step": 1956
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.951050017620975e-05,
+      "loss": 0.2152,
+      "step": 1957
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9509850077471535e-05,
+      "loss": 0.2308,
+      "step": 1958
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9509199558170995e-05,
+      "loss": 0.2426,
+      "step": 1959
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.950854861833691e-05,
+      "loss": 0.2557,
+      "step": 1960
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.950789725799806e-05,
+      "loss": 0.2643,
+      "step": 1961
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9507245477183246e-05,
+      "loss": 0.269,
+      "step": 1962
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9506593275921302e-05,
+      "loss": 0.2706,
+      "step": 1963
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9505940654241067e-05,
+      "loss": 0.2896,
+      "step": 1964
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.95052876121714e-05,
+      "loss": 0.2479,
+      "step": 1965
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9504634149741184e-05,
+      "loss": 0.244,
+      "step": 1966
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.950398026697932e-05,
+      "loss": 0.2741,
+      "step": 1967
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9503325963914717e-05,
+      "loss": 0.2439,
+      "step": 1968
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9502671240576316e-05,
+      "loss": 0.2474,
+      "step": 1969
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9502016096993073e-05,
+      "loss": 0.2381,
+      "step": 1970
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.950136053319396e-05,
+      "loss": 0.2205,
+      "step": 1971
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9500704549207962e-05,
+      "loss": 0.2317,
+      "step": 1972
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.95000481450641e-05,
+      "loss": 0.2499,
+      "step": 1973
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9499391320791395e-05,
+      "loss": 0.2683,
+      "step": 1974
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.94987340764189e-05,
+      "loss": 0.2296,
+      "step": 1975
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9498076411975673e-05,
+      "loss": 0.2284,
+      "step": 1976
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9497418327490805e-05,
+      "loss": 0.2783,
+      "step": 1977
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9496759822993397e-05,
+      "loss": 0.2472,
+      "step": 1978
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9496100898512572e-05,
+      "loss": 0.2333,
+      "step": 1979
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.949544155407747e-05,
+      "loss": 0.2472,
+      "step": 1980
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9494781789717246e-05,
+      "loss": 0.2327,
+      "step": 1981
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9494121605461083e-05,
+      "loss": 0.2904,
+      "step": 1982
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9493461001338172e-05,
+      "loss": 0.24,
+      "step": 1983
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9492799977377726e-05,
+      "loss": 0.2368,
+      "step": 1984
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9492138533608988e-05,
+      "loss": 0.2438,
+      "step": 1985
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.94914766700612e-05,
+      "loss": 0.2517,
+      "step": 1986
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9490814386763634e-05,
+      "loss": 0.1964,
+      "step": 1987
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9490151683745582e-05,
+      "loss": 0.2779,
+      "step": 1988
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.948948856103635e-05,
+      "loss": 0.2408,
+      "step": 1989
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.948882501866526e-05,
+      "loss": 0.2303,
+      "step": 1990
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9488161056661663e-05,
+      "loss": 0.2442,
+      "step": 1991
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9487496675054917e-05,
+      "loss": 0.2527,
+      "step": 1992
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9486831873874404e-05,
+      "loss": 0.2496,
+      "step": 1993
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9486166653149525e-05,
+      "loss": 0.2444,
+      "step": 1994
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.94855010129097e-05,
+      "loss": 0.2141,
+      "step": 1995
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9484834953184366e-05,
+      "loss": 0.2754,
+      "step": 1996
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9484168474002976e-05,
+      "loss": 0.2631,
+      "step": 1997
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9483501575395004e-05,
+      "loss": 0.2362,
+      "step": 1998
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.948283425738995e-05,
+      "loss": 0.2509,
+      "step": 1999
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9482166520017313e-05,
+      "loss": 0.2484,
+      "step": 2000
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9481498363306635e-05,
+      "loss": 0.2166,
+      "step": 2001
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9480829787287456e-05,
+      "loss": 0.2434,
+      "step": 2002
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9480160791989352e-05,
+      "loss": 0.2611,
+      "step": 2003
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.94794913774419e-05,
+      "loss": 0.2293,
+      "step": 2004
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9478821543674705e-05,
+      "loss": 0.2072,
+      "step": 2005
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9478151290717396e-05,
+      "loss": 0.1914,
+      "step": 2006
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9477480618599606e-05,
+      "loss": 0.2581,
+      "step": 2007
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9476809527351e-05,
+      "loss": 0.2397,
+      "step": 2008
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9476138017001254e-05,
+      "loss": 0.2539,
+      "step": 2009
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.947546608758007e-05,
+      "loss": 0.2259,
+      "step": 2010
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9474793739117155e-05,
+      "loss": 0.2838,
+      "step": 2011
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.947412097164225e-05,
+      "loss": 0.2792,
+      "step": 2012
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9473447785185103e-05,
+      "loss": 0.2429,
+      "step": 2013
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9472774179775486e-05,
+      "loss": 0.1895,
+      "step": 2014
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9472100155443186e-05,
+      "loss": 0.2296,
+      "step": 2015
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9471425712218018e-05,
+      "loss": 0.2326,
+      "step": 2016
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.94707508501298e-05,
+      "loss": 0.2598,
+      "step": 2017
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9470075569208384e-05,
+      "loss": 0.275,
+      "step": 2018
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.946939986948363e-05,
+      "loss": 0.2567,
+      "step": 2019
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.946872375098542e-05,
+      "loss": 0.2449,
+      "step": 2020
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9468047213743654e-05,
+      "loss": 0.2099,
+      "step": 2021
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9467370257788257e-05,
+      "loss": 0.2646,
+      "step": 2022
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9466692883149157e-05,
+      "loss": 0.239,
+      "step": 2023
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.946601508985632e-05,
+      "loss": 0.2553,
+      "step": 2024
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9465336877939712e-05,
+      "loss": 0.233,
+      "step": 2025
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.946465824742933e-05,
+      "loss": 0.2495,
+      "step": 2026
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9463979198355185e-05,
+      "loss": 0.242,
+      "step": 2027
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.946329973074731e-05,
+      "loss": 0.2204,
+      "step": 2028
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.946261984463575e-05,
+      "loss": 0.2658,
+      "step": 2029
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9461939540050573e-05,
+      "loss": 0.2555,
+      "step": 2030
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9461258817021866e-05,
+      "loss": 0.2686,
+      "step": 2031
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.946057767557973e-05,
+      "loss": 0.2168,
+      "step": 2032
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9459896115754295e-05,
+      "loss": 0.2455,
+      "step": 2033
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9459214137575694e-05,
+      "loss": 0.2682,
+      "step": 2034
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.945853174107409e-05,
+      "loss": 0.2717,
+      "step": 2035
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9457848926279666e-05,
+      "loss": 0.2715,
+      "step": 2036
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9457165693222608e-05,
+      "loss": 0.242,
+      "step": 2037
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9456482041933142e-05,
+      "loss": 0.2391,
+      "step": 2038
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9455797972441494e-05,
+      "loss": 0.2632,
+      "step": 2039
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.945511348477792e-05,
+      "loss": 0.237,
+      "step": 2040
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.945442857897269e-05,
+      "loss": 0.267,
+      "step": 2041
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.945374325505609e-05,
+      "loss": 0.2549,
+      "step": 2042
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9453057513058435e-05,
+      "loss": 0.2645,
+      "step": 2043
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9452371353010045e-05,
+      "loss": 0.2628,
+      "step": 2044
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9451684774941263e-05,
+      "loss": 0.2427,
+      "step": 2045
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9450997778882458e-05,
+      "loss": 0.2377,
+      "step": 2046
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9450310364864012e-05,
+      "loss": 0.2283,
+      "step": 2047
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9449622532916323e-05,
+      "loss": 0.2597,
+      "step": 2048
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.94489342830698e-05,
+      "loss": 0.2358,
+      "step": 2049
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.94482456153549e-05,
+      "loss": 0.2477,
+      "step": 2050
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9447556529802062e-05,
+      "loss": 0.2493,
+      "step": 2051
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9446867026441768e-05,
+      "loss": 0.2383,
+      "step": 2052
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9446177105304505e-05,
+      "loss": 0.2082,
+      "step": 2053
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9445486766420788e-05,
+      "loss": 0.2136,
+      "step": 2054
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9444796009821147e-05,
+      "loss": 0.2363,
+      "step": 2055
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.944410483553613e-05,
+      "loss": 0.2229,
+      "step": 2056
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9443413243596298e-05,
+      "loss": 0.2527,
+      "step": 2057
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.944272123403224e-05,
+      "loss": 0.2535,
+      "step": 2058
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.944202880687456e-05,
+      "loss": 0.2265,
+      "step": 2059
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9441335962153878e-05,
+      "loss": 0.261,
+      "step": 2060
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9440642699900837e-05,
+      "loss": 0.271,
+      "step": 2061
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9439949020146093e-05,
+      "loss": 0.2759,
+      "step": 2062
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9439254922920323e-05,
+      "loss": 0.2415,
+      "step": 2063
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9438560408254225e-05,
+      "loss": 0.2436,
+      "step": 2064
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.943786547617851e-05,
+      "loss": 0.2405,
+      "step": 2065
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9437170126723914e-05,
+      "loss": 0.2623,
+      "step": 2066
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9436474359921187e-05,
+      "loss": 0.2389,
+      "step": 2067
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9435778175801096e-05,
+      "loss": 0.2288,
+      "step": 2068
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9435081574394432e-05,
+      "loss": 0.2426,
+      "step": 2069
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9434384555732002e-05,
+      "loss": 0.2408,
+      "step": 2070
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9433687119844626e-05,
+      "loss": 0.2498,
+      "step": 2071
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9432989266763152e-05,
+      "loss": 0.307,
+      "step": 2072
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.943229099651844e-05,
+      "loss": 0.2413,
+      "step": 2073
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.943159230914137e-05,
+      "loss": 0.2548,
+      "step": 2074
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9430893204662843e-05,
+      "loss": 0.2603,
+      "step": 2075
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9430193683113773e-05,
+      "loss": 0.2434,
+      "step": 2076
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9429493744525096e-05,
+      "loss": 0.2478,
+      "step": 2077
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9428793388927768e-05,
+      "loss": 0.2458,
+      "step": 2078
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.942809261635276e-05,
+      "loss": 0.2019,
+      "step": 2079
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9427391426831058e-05,
+      "loss": 0.2475,
+      "step": 2080
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.942668982039368e-05,
+      "loss": 0.2338,
+      "step": 2081
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9425987797071655e-05,
+      "loss": 0.2194,
+      "step": 2082
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.942528535689602e-05,
+      "loss": 0.2441,
+      "step": 2083
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9424582499897844e-05,
+      "loss": 0.2977,
+      "step": 2084
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.942387922610821e-05,
+      "loss": 0.2121,
+      "step": 2085
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.942317553555822e-05,
+      "loss": 0.2703,
+      "step": 2086
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.942247142827899e-05,
+      "loss": 0.2835,
+      "step": 2087
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9421766904301663e-05,
+      "loss": 0.2223,
+      "step": 2088
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9421061963657396e-05,
+      "loss": 0.2418,
+      "step": 2089
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9420356606377362e-05,
+      "loss": 0.242,
+      "step": 2090
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.941965083249276e-05,
+      "loss": 0.2258,
+      "step": 2091
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.941894464203479e-05,
+      "loss": 0.2029,
+      "step": 2092
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9418238035034696e-05,
+      "loss": 0.2436,
+      "step": 2093
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9417531011523717e-05,
+      "loss": 0.2621,
+      "step": 2094
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9416823571533123e-05,
+      "loss": 0.2729,
+      "step": 2095
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9416115715094203e-05,
+      "loss": 0.2782,
+      "step": 2096
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9415407442238256e-05,
+      "loss": 0.238,
+      "step": 2097
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.941469875299661e-05,
+      "loss": 0.266,
+      "step": 2098
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9413989647400604e-05,
+      "loss": 0.2539,
+      "step": 2099
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9413280125481597e-05,
+      "loss": 0.2746,
+      "step": 2100
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9412570187270964e-05,
+      "loss": 0.264,
+      "step": 2101
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9411859832800108e-05,
+      "loss": 0.2769,
+      "step": 2102
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9411149062100433e-05,
+      "loss": 0.23,
+      "step": 2103
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9410437875203384e-05,
+      "loss": 0.2372,
+      "step": 2104
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9409726272140406e-05,
+      "loss": 0.2538,
+      "step": 2105
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.940901425294297e-05,
+      "loss": 0.2366,
+      "step": 2106
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9408301817642562e-05,
+      "loss": 0.2577,
+      "step": 2107
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.940758896627069e-05,
+      "loss": 0.2937,
+      "step": 2108
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.940687569885888e-05,
+      "loss": 0.2692,
+      "step": 2109
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.940616201543868e-05,
+      "loss": 0.2319,
+      "step": 2110
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.940544791604164e-05,
+      "loss": 0.2593,
+      "step": 2111
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9404733400699352e-05,
+      "loss": 0.3074,
+      "step": 2112
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9404018469443405e-05,
+      "loss": 0.3138,
+      "step": 2113
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9403303122305428e-05,
+      "loss": 0.2263,
+      "step": 2114
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9402587359317042e-05,
+      "loss": 0.2254,
+      "step": 2115
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9401871180509913e-05,
+      "loss": 0.2226,
+      "step": 2116
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9401154585915706e-05,
+      "loss": 0.2429,
+      "step": 2117
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9400437575566117e-05,
+      "loss": 0.2504,
+      "step": 2118
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9399720149492846e-05,
+      "loss": 0.2248,
+      "step": 2119
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.939900230772763e-05,
+      "loss": 0.2254,
+      "step": 2120
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.939828405030221e-05,
+      "loss": 0.2243,
+      "step": 2121
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9397565377248352e-05,
+      "loss": 0.2644,
+      "step": 2122
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9396846288597836e-05,
+      "loss": 0.2347,
+      "step": 2123
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9396126784382464e-05,
+      "loss": 0.257,
+      "step": 2124
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.939540686463406e-05,
+      "loss": 0.2793,
+      "step": 2125
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9394686529384453e-05,
+      "loss": 0.2304,
+      "step": 2126
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.93939657786655e-05,
+      "loss": 0.2276,
+      "step": 2127
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9393244612509087e-05,
+      "loss": 0.2437,
+      "step": 2128
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.939252303094709e-05,
+      "loss": 0.23,
+      "step": 2129
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9391801034011438e-05,
+      "loss": 0.2545,
+      "step": 2130
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9391078621734043e-05,
+      "loss": 0.2424,
+      "step": 2131
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9390355794146866e-05,
+      "loss": 0.234,
+      "step": 2132
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9389632551281865e-05,
+      "loss": 0.2152,
+      "step": 2133
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.938890889317103e-05,
+      "loss": 0.2634,
+      "step": 2134
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.938818481984636e-05,
+      "loss": 0.2447,
+      "step": 2135
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.938746033133988e-05,
+      "loss": 0.2321,
+      "step": 2136
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9386735427683626e-05,
+      "loss": 0.2159,
+      "step": 2137
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9386010108909654e-05,
+      "loss": 0.2589,
+      "step": 2138
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.938528437505005e-05,
+      "loss": 0.2608,
+      "step": 2139
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.93845582261369e-05,
+      "loss": 0.2112,
+      "step": 2140
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9383831662202322e-05,
+      "loss": 0.2528,
+      "step": 2141
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9383104683278444e-05,
+      "loss": 0.2519,
+      "step": 2142
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9382377289397415e-05,
+      "loss": 0.2253,
+      "step": 2143
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9381649480591412e-05,
+      "loss": 0.2591,
+      "step": 2144
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9380921256892608e-05,
+      "loss": 0.2269,
+      "step": 2145
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.938019261833322e-05,
+      "loss": 0.252,
+      "step": 2146
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.937946356494546e-05,
+      "loss": 0.2334,
+      "step": 2147
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9378734096761582e-05,
+      "loss": 0.2402,
+      "step": 2148
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9378004213813837e-05,
+      "loss": 0.2455,
+      "step": 2149
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9377273916134506e-05,
+      "loss": 0.1902,
+      "step": 2150
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9376543203755884e-05,
+      "loss": 0.2732,
+      "step": 2151
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9375812076710286e-05,
+      "loss": 0.2751,
+      "step": 2152
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.937508053503005e-05,
+      "loss": 0.2212,
+      "step": 2153
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9374348578747523e-05,
+      "loss": 0.2577,
+      "step": 2154
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9373616207895076e-05,
+      "loss": 0.261,
+      "step": 2155
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9372883422505098e-05,
+      "loss": 0.2302,
+      "step": 2156
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9372150222609997e-05,
+      "loss": 0.2626,
+      "step": 2157
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9371416608242192e-05,
+      "loss": 0.2559,
+      "step": 2158
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.937068257943413e-05,
+      "loss": 0.3068,
+      "step": 2159
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9369948136218274e-05,
+      "loss": 0.2395,
+      "step": 2160
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9369213278627103e-05,
+      "loss": 0.2522,
+      "step": 2161
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.936847800669311e-05,
+      "loss": 0.2413,
+      "step": 2162
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9367742320448823e-05,
+      "loss": 0.2572,
+      "step": 2163
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.936700621992677e-05,
+      "loss": 0.2458,
+      "step": 2164
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9366269705159502e-05,
+      "loss": 0.2805,
+      "step": 2165
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9365532776179594e-05,
+      "loss": 0.242,
+      "step": 2166
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9364795433019635e-05,
+      "loss": 0.2159,
+      "step": 2167
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9364057675712235e-05,
+      "loss": 0.2029,
+      "step": 2168
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9363319504290013e-05,
+      "loss": 0.2886,
+      "step": 2169
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9362580918785622e-05,
+      "loss": 0.2895,
+      "step": 2170
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9361841919231725e-05,
+      "loss": 0.2296,
+      "step": 2171
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9361102505661e-05,
+      "loss": 0.2339,
+      "step": 2172
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9360362678106142e-05,
+      "loss": 0.2472,
+      "step": 2173
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9359622436599884e-05,
+      "loss": 0.2661,
+      "step": 2174
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9358881781174946e-05,
+      "loss": 0.2644,
+      "step": 2175
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.935814071186409e-05,
+      "loss": 0.2368,
+      "step": 2176
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9357399228700094e-05,
+      "loss": 0.2511,
+      "step": 2177
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9356657331715737e-05,
+      "loss": 0.2497,
+      "step": 2178
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.935591502094384e-05,
+      "loss": 0.2287,
+      "step": 2179
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9355172296417226e-05,
+      "loss": 0.2533,
+      "step": 2180
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9354429158168737e-05,
+      "loss": 0.2434,
+      "step": 2181
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9353685606231245e-05,
+      "loss": 0.2168,
+      "step": 2182
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.935294164063763e-05,
+      "loss": 0.2066,
+      "step": 2183
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.935219726142079e-05,
+      "loss": 0.2433,
+      "step": 2184
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.935145246861365e-05,
+      "loss": 0.257,
+      "step": 2185
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9350707262249146e-05,
+      "loss": 0.2063,
+      "step": 2186
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9349961642360225e-05,
+      "loss": 0.2463,
+      "step": 2187
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9349215608979876e-05,
+      "loss": 0.2672,
+      "step": 2188
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.934846916214108e-05,
+      "loss": 0.2248,
+      "step": 2189
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.934772230187686e-05,
+      "loss": 0.2411,
+      "step": 2190
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.934697502822023e-05,
+      "loss": 0.2351,
+      "step": 2191
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.934622734120424e-05,
+      "loss": 0.2656,
+      "step": 2192
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9345479240861963e-05,
+      "loss": 0.2388,
+      "step": 2193
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9344730727226483e-05,
+      "loss": 0.2686,
+      "step": 2194
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.93439818003309e-05,
+      "loss": 0.2588,
+      "step": 2195
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9343232460208327e-05,
+      "loss": 0.253,
+      "step": 2196
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9342482706891913e-05,
+      "loss": 0.2412,
+      "step": 2197
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9341732540414812e-05,
+      "loss": 0.2394,
+      "step": 2198
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9340981960810195e-05,
+      "loss": 0.227,
+      "step": 2199
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.934023096811126e-05,
+      "loss": 0.2898,
+      "step": 2200
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9339479562351217e-05,
+      "loss": 0.2669,
+      "step": 2201
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9338727743563298e-05,
+      "loss": 0.2894,
+      "step": 2202
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.933797551178075e-05,
+      "loss": 0.2373,
+      "step": 2203
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.933722286703684e-05,
+      "loss": 0.2819,
+      "step": 2204
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9336469809364852e-05,
+      "loss": 0.2531,
+      "step": 2205
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.933571633879809e-05,
+      "loss": 0.2552,
+      "step": 2206
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9334962455369876e-05,
+      "loss": 0.2297,
+      "step": 2207
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9334208159113544e-05,
+      "loss": 0.2182,
+      "step": 2208
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.933345345006246e-05,
+      "loss": 0.2145,
+      "step": 2209
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.933269832825e-05,
+      "loss": 0.2366,
+      "step": 2210
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9331942793709552e-05,
+      "loss": 0.23,
+      "step": 2211
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.933118684647453e-05,
+      "loss": 0.2084,
+      "step": 2212
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9330430486578367e-05,
+      "loss": 0.2356,
+      "step": 2213
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9329673714054516e-05,
+      "loss": 0.2408,
+      "step": 2214
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9328916528936433e-05,
+      "loss": 0.2279,
+      "step": 2215
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.932815893125762e-05,
+      "loss": 0.2244,
+      "step": 2216
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9327400921051566e-05,
+      "loss": 0.2347,
+      "step": 2217
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9326642498351798e-05,
+      "loss": 0.2494,
+      "step": 2218
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9325883663191857e-05,
+      "loss": 0.225,
+      "step": 2219
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9325124415605306e-05,
+      "loss": 0.2613,
+      "step": 2220
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9324364755625714e-05,
+      "loss": 0.2666,
+      "step": 2221
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.932360468328668e-05,
+      "loss": 0.2583,
+      "step": 2222
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.932284419862182e-05,
+      "loss": 0.2798,
+      "step": 2223
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9322083301664763e-05,
+      "loss": 0.2757,
+      "step": 2224
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9321321992449156e-05,
+      "loss": 0.2187,
+      "step": 2225
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.932056027100867e-05,
+      "loss": 0.2703,
+      "step": 2226
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9319798137376993e-05,
+      "loss": 0.2257,
+      "step": 2227
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9319035591587826e-05,
+      "loss": 0.2653,
+      "step": 2228
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.931827263367489e-05,
+      "loss": 0.2522,
+      "step": 2229
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9317509263671935e-05,
+      "loss": 0.2342,
+      "step": 2230
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.931674548161271e-05,
+      "loss": 0.2428,
+      "step": 2231
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9315981287530996e-05,
+      "loss": 0.2209,
+      "step": 2232
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9315216681460593e-05,
+      "loss": 0.2693,
+      "step": 2233
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.931445166343531e-05,
+      "loss": 0.2767,
+      "step": 2234
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.931368623348898e-05,
+      "loss": 0.2399,
+      "step": 2235
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.931292039165545e-05,
+      "loss": 0.2629,
+      "step": 2236
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9312154137968594e-05,
+      "loss": 0.2402,
+      "step": 2237
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9311387472462297e-05,
+      "loss": 0.2698,
+      "step": 2238
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9310620395170464e-05,
+      "loss": 0.218,
+      "step": 2239
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.930985290612702e-05,
+      "loss": 0.2508,
+      "step": 2240
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9309085005365896e-05,
+      "loss": 0.2164,
+      "step": 2241
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.930831669292107e-05,
+      "loss": 0.2143,
+      "step": 2242
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.93075479688265e-05,
+      "loss": 0.2175,
+      "step": 2243
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9306778833116192e-05,
+      "loss": 0.2395,
+      "step": 2244
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9306009285824166e-05,
+      "loss": 0.2669,
+      "step": 2245
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9305239326984445e-05,
+      "loss": 0.2247,
+      "step": 2246
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9304468956631076e-05,
+      "loss": 0.2446,
+      "step": 2247
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.930369817479814e-05,
+      "loss": 0.2506,
+      "step": 2248
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9302926981519718e-05,
+      "loss": 0.2402,
+      "step": 2249
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9302155376829915e-05,
+      "loss": 0.2368,
+      "step": 2250
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9301383360762853e-05,
+      "loss": 0.2609,
+      "step": 2251
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9300610933352676e-05,
+      "loss": 0.2658,
+      "step": 2252
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.929983809463354e-05,
+      "loss": 0.2635,
+      "step": 2253
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9299064844639626e-05,
+      "loss": 0.2425,
+      "step": 2254
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9298291183405134e-05,
+      "loss": 0.2426,
+      "step": 2255
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9297517110964273e-05,
+      "loss": 0.2762,
+      "step": 2256
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9296742627351276e-05,
+      "loss": 0.2642,
+      "step": 2257
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.929596773260039e-05,
+      "loss": 0.216,
+      "step": 2258
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.929519242674589e-05,
+      "loss": 0.2494,
+      "step": 2259
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9294416709822063e-05,
+      "loss": 0.2234,
+      "step": 2260
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9293640581863212e-05,
+      "loss": 0.2829,
+      "step": 2261
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.929286404290366e-05,
+      "loss": 0.2251,
+      "step": 2262
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.929208709297775e-05,
+      "loss": 0.2322,
+      "step": 2263
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9291309732119837e-05,
+      "loss": 0.2398,
+      "step": 2264
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9290531960364305e-05,
+      "loss": 0.1936,
+      "step": 2265
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9289753777745548e-05,
+      "loss": 0.2388,
+      "step": 2266
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.928897518429798e-05,
+      "loss": 0.2405,
+      "step": 2267
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9288196180056035e-05,
+      "loss": 0.211,
+      "step": 2268
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9287416765054163e-05,
+      "loss": 0.2589,
+      "step": 2269
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9286636939326827e-05,
+      "loss": 0.2094,
+      "step": 2270
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9285856702908523e-05,
+      "loss": 0.2383,
+      "step": 2271
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.928507605583375e-05,
+      "loss": 0.2101,
+      "step": 2272
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9284294998137033e-05,
+      "loss": 0.2192,
+      "step": 2273
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.928351352985291e-05,
+      "loss": 0.2567,
+      "step": 2274
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.928273165101595e-05,
+      "loss": 0.2403,
+      "step": 2275
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.928194936166072e-05,
+      "loss": 0.2501,
+      "step": 2276
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9281166661821825e-05,
+      "loss": 0.2154,
+      "step": 2277
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.928038355153387e-05,
+      "loss": 0.2564,
+      "step": 2278
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9279600030831495e-05,
+      "loss": 0.2452,
+      "step": 2279
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9278816099749344e-05,
+      "loss": 0.2369,
+      "step": 2280
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.927803175832209e-05,
+      "loss": 0.2446,
+      "step": 2281
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.927724700658442e-05,
+      "loss": 0.2212,
+      "step": 2282
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9276461844571033e-05,
+      "loss": 0.2568,
+      "step": 2283
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9275676272316658e-05,
+      "loss": 0.2288,
+      "step": 2284
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9274890289856033e-05,
+      "loss": 0.2423,
+      "step": 2285
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.927410389722392e-05,
+      "loss": 0.207,
+      "step": 2286
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9273317094455094e-05,
+      "loss": 0.2214,
+      "step": 2287
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9272529881584344e-05,
+      "loss": 0.2536,
+      "step": 2288
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9271742258646496e-05,
+      "loss": 0.2411,
+      "step": 2289
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.927095422567638e-05,
+      "loss": 0.2339,
+      "step": 2290
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9270165782708833e-05,
+      "loss": 0.2499,
+      "step": 2291
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9269376929778742e-05,
+      "loss": 0.2551,
+      "step": 2292
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9268587666920974e-05,
+      "loss": 0.2338,
+      "step": 2293
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9267797994170445e-05,
+      "loss": 0.2332,
+      "step": 2294
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9267007911562078e-05,
+      "loss": 0.257,
+      "step": 2295
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9266217419130807e-05,
+      "loss": 0.2697,
+      "step": 2296
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9265426516911595e-05,
+      "loss": 0.2677,
+      "step": 2297
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9264635204939416e-05,
+      "loss": 0.2646,
+      "step": 2298
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9263843483249265e-05,
+      "loss": 0.2208,
+      "step": 2299
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.926305135187616e-05,
+      "loss": 0.262,
+      "step": 2300
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.926225881085513e-05,
+      "loss": 0.2891,
+      "step": 2301
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.926146586022122e-05,
+      "loss": 0.2284,
+      "step": 2302
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9260672500009495e-05,
+      "loss": 0.2732,
+      "step": 2303
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9259878730255054e-05,
+      "loss": 0.2378,
+      "step": 2304
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9259084550992987e-05,
+      "loss": 0.2385,
+      "step": 2305
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9258289962258422e-05,
+      "loss": 0.2207,
+      "step": 2306
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9257494964086497e-05,
+      "loss": 0.2135,
+      "step": 2307
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.925669955651237e-05,
+      "loss": 0.2526,
+      "step": 2308
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.925590373957122e-05,
+      "loss": 0.2436,
+      "step": 2309
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.925510751329824e-05,
+      "loss": 0.2573,
+      "step": 2310
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9254310877728636e-05,
+      "loss": 0.2641,
+      "step": 2311
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9253513832897645e-05,
+      "loss": 0.2101,
+      "step": 2312
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9252716378840514e-05,
+      "loss": 0.2379,
+      "step": 2313
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9251918515592513e-05,
+      "loss": 0.2347,
+      "step": 2314
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9251120243188913e-05,
+      "loss": 0.2272,
+      "step": 2315
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9250321561665032e-05,
+      "loss": 0.2156,
+      "step": 2316
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9249522471056187e-05,
+      "loss": 0.2475,
+      "step": 2317
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9248722971397714e-05,
+      "loss": 0.299,
+      "step": 2318
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9247923062724967e-05,
+      "loss": 0.2627,
+      "step": 2319
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.924712274507333e-05,
+      "loss": 0.2442,
+      "step": 2320
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9246322018478186e-05,
+      "loss": 0.2182,
+      "step": 2321
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9245520882974953e-05,
+      "loss": 0.2289,
+      "step": 2322
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.924471933859906e-05,
+      "loss": 0.2285,
+      "step": 2323
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9243917385385954e-05,
+      "loss": 0.256,
+      "step": 2324
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9243115023371096e-05,
+      "loss": 0.2895,
+      "step": 2325
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.924231225258997e-05,
+      "loss": 0.2729,
+      "step": 2326
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.924150907307809e-05,
+      "loss": 0.2131,
+      "step": 2327
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.924070548487096e-05,
+      "loss": 0.2167,
+      "step": 2328
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9239901488004126e-05,
+      "loss": 0.2493,
+      "step": 2329
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9239097082513144e-05,
+      "loss": 0.2253,
+      "step": 2330
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.923829226843358e-05,
+      "loss": 0.2888,
+      "step": 2331
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9237487045801034e-05,
+      "loss": 0.2276,
+      "step": 2332
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9236681414651118e-05,
+      "loss": 0.2023,
+      "step": 2333
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.923587537501945e-05,
+      "loss": 0.2496,
+      "step": 2334
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.923506892694169e-05,
+      "loss": 0.2345,
+      "step": 2335
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.923426207045349e-05,
+      "loss": 0.2486,
+      "step": 2336
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9233454805590534e-05,
+      "loss": 0.2572,
+      "step": 2337
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.923264713238853e-05,
+      "loss": 0.2263,
+      "step": 2338
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.923183905088319e-05,
+      "loss": 0.2735,
+      "step": 2339
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9231030561110252e-05,
+      "loss": 0.2272,
+      "step": 2340
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9230221663105467e-05,
+      "loss": 0.2274,
+      "step": 2341
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.922941235690462e-05,
+      "loss": 0.2882,
+      "step": 2342
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9228602642543483e-05,
+      "loss": 0.2538,
+      "step": 2343
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.922779252005788e-05,
+      "loss": 0.2734,
+      "step": 2344
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9226981989483633e-05,
+      "loss": 0.2367,
+      "step": 2345
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9226171050856583e-05,
+      "loss": 0.2309,
+      "step": 2346
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.92253597042126e-05,
+      "loss": 0.2523,
+      "step": 2347
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9224547949587557e-05,
+      "loss": 0.211,
+      "step": 2348
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9223735787017355e-05,
+      "loss": 0.2522,
+      "step": 2349
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.922292321653792e-05,
+      "loss": 0.2499,
+      "step": 2350
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9222110238185174e-05,
+      "loss": 0.2222,
+      "step": 2351
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9221296851995074e-05,
+      "loss": 0.2582,
+      "step": 2352
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9220483058003597e-05,
+      "loss": 0.2372,
+      "step": 2353
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.921966885624673e-05,
+      "loss": 0.2288,
+      "step": 2354
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9218854246760475e-05,
+      "loss": 0.2476,
+      "step": 2355
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9218039229580858e-05,
+      "loss": 0.2212,
+      "step": 2356
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.921722380474393e-05,
+      "loss": 0.2387,
+      "step": 2357
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.921640797228574e-05,
+      "loss": 0.2122,
+      "step": 2358
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.921559173224238e-05,
+      "loss": 0.2816,
+      "step": 2359
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9214775084649938e-05,
+      "loss": 0.2389,
+      "step": 2360
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9213958029544534e-05,
+      "loss": 0.2609,
+      "step": 2361
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.92131405669623e-05,
+      "loss": 0.2593,
+      "step": 2362
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9212322696939386e-05,
+      "loss": 0.2663,
+      "step": 2363
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9211504419511964e-05,
+      "loss": 0.2356,
+      "step": 2364
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.921068573471622e-05,
+      "loss": 0.245,
+      "step": 2365
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9209866642588356e-05,
+      "loss": 0.2236,
+      "step": 2366
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9209047143164605e-05,
+      "loss": 0.2346,
+      "step": 2367
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9208227236481198e-05,
+      "loss": 0.2376,
+      "step": 2368
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.92074069225744e-05,
+      "loss": 0.2335,
+      "step": 2369
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9206586201480487e-05,
+      "loss": 0.2586,
+      "step": 2370
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9205765073235755e-05,
+      "loss": 0.2497,
+      "step": 2371
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9204943537876517e-05,
+      "loss": 0.227,
+      "step": 2372
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9204121595439105e-05,
+      "loss": 0.2367,
+      "step": 2373
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.920329924595987e-05,
+      "loss": 0.2078,
+      "step": 2374
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9202476489475172e-05,
+      "loss": 0.2408,
+      "step": 2375
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9201653326021407e-05,
+      "loss": 0.2606,
+      "step": 2376
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9200829755634972e-05,
+      "loss": 0.2534,
+      "step": 2377
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.920000577835229e-05,
+      "loss": 0.2162,
+      "step": 2378
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.91991813942098e-05,
+      "loss": 0.242,
+      "step": 2379
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9198356603243958e-05,
+      "loss": 0.2441,
+      "step": 2380
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9197531405491245e-05,
+      "loss": 0.2456,
+      "step": 2381
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.919670580098815e-05,
+      "loss": 0.2814,
+      "step": 2382
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9195879789771182e-05,
+      "loss": 0.2511,
+      "step": 2383
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9195053371876876e-05,
+      "loss": 0.1948,
+      "step": 2384
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9194226547341775e-05,
+      "loss": 0.2162,
+      "step": 2385
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.919339931620245e-05,
+      "loss": 0.2468,
+      "step": 2386
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.919257167849548e-05,
+      "loss": 0.2297,
+      "step": 2387
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9191743634257463e-05,
+      "loss": 0.2926,
+      "step": 2388
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9190915183525027e-05,
+      "loss": 0.227,
+      "step": 2389
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.91900863263348e-05,
+      "loss": 0.2169,
+      "step": 2390
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9189257062723445e-05,
+      "loss": 0.2365,
+      "step": 2391
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9188427392727634e-05,
+      "loss": 0.226,
+      "step": 2392
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9187597316384052e-05,
+      "loss": 0.2553,
+      "step": 2393
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9186766833729415e-05,
+      "loss": 0.2695,
+      "step": 2394
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.918593594480045e-05,
+      "loss": 0.2322,
+      "step": 2395
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9185104649633895e-05,
+      "loss": 0.227,
+      "step": 2396
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.918427294826652e-05,
+      "loss": 0.253,
+      "step": 2397
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9183440840735104e-05,
+      "loss": 0.2205,
+      "step": 2398
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.918260832707645e-05,
+      "loss": 0.2583,
+      "step": 2399
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9181775407327365e-05,
+      "loss": 0.2697,
+      "step": 2400
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.918094208152469e-05,
+      "loss": 0.234,
+      "step": 2401
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9180108349705282e-05,
+      "loss": 0.2307,
+      "step": 2402
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9179274211906003e-05,
+      "loss": 0.2438,
+      "step": 2403
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.917843966816375e-05,
+      "loss": 0.2474,
+      "step": 2404
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9177604718515425e-05,
+      "loss": 0.233,
+      "step": 2405
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9176769362997953e-05,
+      "loss": 0.2549,
+      "step": 2406
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.917593360164828e-05,
+      "loss": 0.2476,
+      "step": 2407
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.917509743450336e-05,
+      "loss": 0.2128,
+      "step": 2408
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9174260861600177e-05,
+      "loss": 0.2576,
+      "step": 2409
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9173423882975728e-05,
+      "loss": 0.2219,
+      "step": 2410
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9172586498667024e-05,
+      "loss": 0.2325,
+      "step": 2411
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9171748708711098e-05,
+      "loss": 0.2352,
+      "step": 2412
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9170910513145004e-05,
+      "loss": 0.2277,
+      "step": 2413
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9170071912005804e-05,
+      "loss": 0.2633,
+      "step": 2414
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9169232905330593e-05,
+      "loss": 0.2488,
+      "step": 2415
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9168393493156465e-05,
+      "loss": 0.2348,
+      "step": 2416
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.916755367552055e-05,
+      "loss": 0.245,
+      "step": 2417
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9166713452459982e-05,
+      "loss": 0.248,
+      "step": 2418
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9165872824011922e-05,
+      "loss": 0.2198,
+      "step": 2419
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9165031790213545e-05,
+      "loss": 0.2708,
+      "step": 2420
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.916419035110205e-05,
+      "loss": 0.263,
+      "step": 2421
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.916334850671464e-05,
+      "loss": 0.2196,
+      "step": 2422
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9162506257088548e-05,
+      "loss": 0.2588,
+      "step": 2423
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9161663602261024e-05,
+      "loss": 0.2477,
+      "step": 2424
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.916082054226933e-05,
+      "loss": 0.2734,
+      "step": 2425
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9159977077150752e-05,
+      "loss": 0.2522,
+      "step": 2426
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.915913320694259e-05,
+      "loss": 0.2249,
+      "step": 2427
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9158288931682164e-05,
+      "loss": 0.2161,
+      "step": 2428
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.915744425140681e-05,
+      "loss": 0.2359,
+      "step": 2429
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9156599166153884e-05,
+      "loss": 0.2248,
+      "step": 2430
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.915575367596076e-05,
+      "loss": 0.2561,
+      "step": 2431
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9154907780864825e-05,
+      "loss": 0.2207,
+      "step": 2432
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.915406148090349e-05,
+      "loss": 0.279,
+      "step": 2433
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9153214776114185e-05,
+      "loss": 0.2388,
+      "step": 2434
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.915236766653435e-05,
+      "loss": 0.2181,
+      "step": 2435
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9151520152201445e-05,
+      "loss": 0.2539,
+      "step": 2436
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9150672233152957e-05,
+      "loss": 0.2386,
+      "step": 2437
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.914982390942638e-05,
+      "loss": 0.2481,
+      "step": 2438
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9148975181059236e-05,
+      "loss": 0.247,
+      "step": 2439
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9148126048089054e-05,
+      "loss": 0.2129,
+      "step": 2440
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9147276510553385e-05,
+      "loss": 0.2133,
+      "step": 2441
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9146426568489796e-05,
+      "loss": 0.2331,
+      "step": 2442
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9145576221935883e-05,
+      "loss": 0.2253,
+      "step": 2443
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.914472547092925e-05,
+      "loss": 0.2624,
+      "step": 2444
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9143874315507515e-05,
+      "loss": 0.2073,
+      "step": 2445
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.914302275570832e-05,
+      "loss": 0.2204,
+      "step": 2446
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.914217079156933e-05,
+      "loss": 0.2745,
+      "step": 2447
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9141318423128223e-05,
+      "loss": 0.2143,
+      "step": 2448
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9140465650422687e-05,
+      "loss": 0.219,
+      "step": 2449
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9139612473490433e-05,
+      "loss": 0.2349,
+      "step": 2450
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9138758892369202e-05,
+      "loss": 0.2429,
+      "step": 2451
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9137904907096737e-05,
+      "loss": 0.2471,
+      "step": 2452
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9137050517710803e-05,
+      "loss": 0.2069,
+      "step": 2453
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.913619572424919e-05,
+      "loss": 0.23,
+      "step": 2454
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9135340526749694e-05,
+      "loss": 0.2364,
+      "step": 2455
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9134484925250135e-05,
+      "loss": 0.2382,
+      "step": 2456
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9133628919788357e-05,
+      "loss": 0.2283,
+      "step": 2457
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9132772510402214e-05,
+      "loss": 0.2428,
+      "step": 2458
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9131915697129576e-05,
+      "loss": 0.256,
+      "step": 2459
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.913105848000834e-05,
+      "loss": 0.2624,
+      "step": 2460
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9130200859076406e-05,
+      "loss": 0.2197,
+      "step": 2461
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9129342834371714e-05,
+      "loss": 0.2268,
+      "step": 2462
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.91284844059322e-05,
+      "loss": 0.2347,
+      "step": 2463
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.912762557379583e-05,
+      "loss": 0.2505,
+      "step": 2464
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9126766338000588e-05,
+      "loss": 0.2142,
+      "step": 2465
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9125906698584466e-05,
+      "loss": 0.2625,
+      "step": 2466
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9125046655585483e-05,
+      "loss": 0.2319,
+      "step": 2467
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9124186209041674e-05,
+      "loss": 0.241,
+      "step": 2468
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9123325358991095e-05,
+      "loss": 0.1903,
+      "step": 2469
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.912246410547181e-05,
+      "loss": 0.2367,
+      "step": 2470
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.912160244852191e-05,
+      "loss": 0.2335,
+      "step": 2471
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9120740388179503e-05,
+      "loss": 0.229,
+      "step": 2472
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.911987792448271e-05,
+      "loss": 0.2008,
+      "step": 2473
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9119015057469673e-05,
+      "loss": 0.2605,
+      "step": 2474
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9118151787178552e-05,
+      "loss": 0.2263,
+      "step": 2475
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.911728811364752e-05,
+      "loss": 0.2339,
+      "step": 2476
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9116424036914778e-05,
+      "loss": 0.2523,
+      "step": 2477
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9115559557018534e-05,
+      "loss": 0.2257,
+      "step": 2478
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.911469467399702e-05,
+      "loss": 0.2119,
+      "step": 2479
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9113829387888488e-05,
+      "loss": 0.2625,
+      "step": 2480
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.91129636987312e-05,
+      "loss": 0.2275,
+      "step": 2481
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.911209760656344e-05,
+      "loss": 0.2424,
+      "step": 2482
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9111231111423514e-05,
+      "loss": 0.2537,
+      "step": 2483
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9110364213349738e-05,
+      "loss": 0.2146,
+      "step": 2484
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.910949691238045e-05,
+      "loss": 0.2728,
+      "step": 2485
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9108629208554007e-05,
+      "loss": 0.2277,
+      "step": 2486
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.910776110190878e-05,
+      "loss": 0.2388,
+      "step": 2487
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9106892592483165e-05,
+      "loss": 0.2141,
+      "step": 2488
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9106023680315565e-05,
+      "loss": 0.2265,
+      "step": 2489
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.910515436544441e-05,
+      "loss": 0.2309,
+      "step": 2490
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.910428464790814e-05,
+      "loss": 0.2635,
+      "step": 2491
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9103414527745225e-05,
+      "loss": 0.2397,
+      "step": 2492
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.910254400499414e-05,
+      "loss": 0.2147,
+      "step": 2493
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9101673079693383e-05,
+      "loss": 0.2479,
+      "step": 2494
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9100801751881472e-05,
+      "loss": 0.2298,
+      "step": 2495
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9099930021596937e-05,
+      "loss": 0.2397,
+      "step": 2496
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.909905788887833e-05,
+      "loss": 0.2716,
+      "step": 2497
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.909818535376422e-05,
+      "loss": 0.2425,
+      "step": 2498
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9097312416293197e-05,
+      "loss": 0.2616,
+      "step": 2499
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9096439076503866e-05,
+      "loss": 0.2197,
+      "step": 2500
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9095565334434843e-05,
+      "loss": 0.2987,
+      "step": 2501
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9094691190124775e-05,
+      "loss": 0.2605,
+      "step": 2502
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9093816643612317e-05,
+      "loss": 0.2259,
+      "step": 2503
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9092941694936147e-05,
+      "loss": 0.2736,
+      "step": 2504
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9092066344134955e-05,
+      "loss": 0.2257,
+      "step": 2505
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9091190591247456e-05,
+      "loss": 0.274,
+      "step": 2506
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9090314436312374e-05,
+      "loss": 0.224,
+      "step": 2507
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9089437879368466e-05,
+      "loss": 0.2513,
+      "step": 2508
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9088560920454484e-05,
+      "loss": 0.2177,
+      "step": 2509
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9087683559609217e-05,
+      "loss": 0.2463,
+      "step": 2510
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9086805796871465e-05,
+      "loss": 0.215,
+      "step": 2511
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.908592763228005e-05,
+      "loss": 0.2271,
+      "step": 2512
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9085049065873798e-05,
+      "loss": 0.254,
+      "step": 2513
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9084170097691573e-05,
+      "loss": 0.2051,
+      "step": 2514
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.908329072777224e-05,
+      "loss": 0.2535,
+      "step": 2515
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9082410956154694e-05,
+      "loss": 0.2352,
+      "step": 2516
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9081530782877834e-05,
+      "loss": 0.2705,
+      "step": 2517
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9080650207980587e-05,
+      "loss": 0.2072,
+      "step": 2518
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.90797692315019e-05,
+      "loss": 0.2397,
+      "step": 2519
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9078887853480727e-05,
+      "loss": 0.2255,
+      "step": 2520
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9078006073956055e-05,
+      "loss": 0.2659,
+      "step": 2521
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9077123892966867e-05,
+      "loss": 0.2302,
+      "step": 2522
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9076241310552186e-05,
+      "loss": 0.2382,
+      "step": 2523
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9075358326751042e-05,
+      "loss": 0.2385,
+      "step": 2524
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9074474941602482e-05,
+      "loss": 0.2222,
+      "step": 2525
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9073591155145572e-05,
+      "loss": 0.2358,
+      "step": 2526
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.90727069674194e-05,
+      "loss": 0.2421,
+      "step": 2527
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.907182237846306e-05,
+      "loss": 0.2304,
+      "step": 2528
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9070937388315684e-05,
+      "loss": 0.2384,
+      "step": 2529
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9070051997016395e-05,
+      "loss": 0.2142,
+      "step": 2530
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9069166204604367e-05,
+      "loss": 0.2289,
+      "step": 2531
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9068280011118754e-05,
+      "loss": 0.2473,
+      "step": 2532
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9067393416598763e-05,
+      "loss": 0.2628,
+      "step": 2533
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9066506421083593e-05,
+      "loss": 0.2299,
+      "step": 2534
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.906561902461247e-05,
+      "loss": 0.2426,
+      "step": 2535
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9064731227224647e-05,
+      "loss": 0.2348,
+      "step": 2536
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9063843028959375e-05,
+      "loss": 0.2532,
+      "step": 2537
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.906295442985594e-05,
+      "loss": 0.3001,
+      "step": 2538
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.906206542995364e-05,
+      "loss": 0.2567,
+      "step": 2539
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9061176029291782e-05,
+      "loss": 0.2567,
+      "step": 2540
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.906028622790971e-05,
+      "loss": 0.3036,
+      "step": 2541
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9059396025846765e-05,
+      "loss": 0.245,
+      "step": 2542
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.905850542314232e-05,
+      "loss": 0.2417,
+      "step": 2543
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.905761441983576e-05,
+      "loss": 0.223,
+      "step": 2544
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9056723015966486e-05,
+      "loss": 0.2211,
+      "step": 2545
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.905583121157393e-05,
+      "loss": 0.244,
+      "step": 2546
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9054939006697516e-05,
+      "loss": 0.2276,
+      "step": 2547
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.905404640137671e-05,
+      "loss": 0.2198,
+      "step": 2548
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9053153395650984e-05,
+      "loss": 0.2269,
+      "step": 2549
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9052259989559825e-05,
+      "loss": 0.2331,
+      "step": 2550
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9051366183142756e-05,
+      "loss": 0.2692,
+      "step": 2551
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.905047197643929e-05,
+      "loss": 0.2114,
+      "step": 2552
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9049577369488984e-05,
+      "loss": 0.2307,
+      "step": 2553
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9048682362331394e-05,
+      "loss": 0.2366,
+      "step": 2554
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9047786955006103e-05,
+      "loss": 0.2677,
+      "step": 2555
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.904689114755271e-05,
+      "loss": 0.2292,
+      "step": 2556
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.904599494001083e-05,
+      "loss": 0.1999,
+      "step": 2557
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9045098332420096e-05,
+      "loss": 0.2353,
+      "step": 2558
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.904420132482016e-05,
+      "loss": 0.2622,
+      "step": 2559
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.904330391725069e-05,
+      "loss": 0.2365,
+      "step": 2560
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9042406109751376e-05,
+      "loss": 0.2571,
+      "step": 2561
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9041507902361922e-05,
+      "loss": 0.2749,
+      "step": 2562
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9040609295122047e-05,
+      "loss": 0.2332,
+      "step": 2563
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9039710288071495e-05,
+      "loss": 0.2448,
+      "step": 2564
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9038810881250018e-05,
+      "loss": 0.2652,
+      "step": 2565
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9037911074697396e-05,
+      "loss": 0.2319,
+      "step": 2566
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9037010868453423e-05,
+      "loss": 0.231,
+      "step": 2567
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9036110262557906e-05,
+      "loss": 0.2332,
+      "step": 2568
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9035209257050672e-05,
+      "loss": 0.246,
+      "step": 2569
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.903430785197157e-05,
+      "loss": 0.227,
+      "step": 2570
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9033406047360464e-05,
+      "loss": 0.2538,
+      "step": 2571
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9032503843257236e-05,
+      "loss": 0.2542,
+      "step": 2572
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.903160123970178e-05,
+      "loss": 0.2289,
+      "step": 2573
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.903069823673402e-05,
+      "loss": 0.2548,
+      "step": 2574
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9029794834393883e-05,
+      "loss": 0.2519,
+      "step": 2575
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9028891032721325e-05,
+      "loss": 0.2386,
+      "step": 2576
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9027986831756315e-05,
+      "loss": 0.2263,
+      "step": 2577
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.902708223153884e-05,
+      "loss": 0.2792,
+      "step": 2578
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9026177232108906e-05,
+      "loss": 0.2481,
+      "step": 2579
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9025271833506532e-05,
+      "loss": 0.2215,
+      "step": 2580
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.902436603577176e-05,
+      "loss": 0.2191,
+      "step": 2581
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.902345983894465e-05,
+      "loss": 0.2525,
+      "step": 2582
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9022553243065277e-05,
+      "loss": 0.2253,
+      "step": 2583
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9021646248173735e-05,
+      "loss": 0.2175,
+      "step": 2584
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.902073885431013e-05,
+      "loss": 0.2569,
+      "step": 2585
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9019831061514594e-05,
+      "loss": 0.2349,
+      "step": 2586
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9018922869827273e-05,
+      "loss": 0.2193,
+      "step": 2587
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.901801427928833e-05,
+      "loss": 0.2308,
+      "step": 2588
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9017105289937946e-05,
+      "loss": 0.2169,
+      "step": 2589
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9016195901816323e-05,
+      "loss": 0.2171,
+      "step": 2590
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9015286114963675e-05,
+      "loss": 0.2406,
+      "step": 2591
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9014375929420237e-05,
+      "loss": 0.2396,
+      "step": 2592
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.901346534522626e-05,
+      "loss": 0.2262,
+      "step": 2593
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9012554362422012e-05,
+      "loss": 0.2313,
+      "step": 2594
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9011642981047784e-05,
+      "loss": 0.2071,
+      "step": 2595
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9010731201143883e-05,
+      "loss": 0.2363,
+      "step": 2596
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.900981902275062e-05,
+      "loss": 0.2385,
+      "step": 2597
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.900890644590835e-05,
+      "loss": 0.2221,
+      "step": 2598
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9007993470657417e-05,
+      "loss": 0.2776,
+      "step": 2599
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9007080097038207e-05,
+      "loss": 0.2403,
+      "step": 2600
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9006166325091108e-05,
+      "loss": 0.2432,
+      "step": 2601
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.900525215485653e-05,
+      "loss": 0.2195,
+      "step": 2602
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9004337586374904e-05,
+      "loss": 0.2604,
+      "step": 2603
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.900342261968667e-05,
+      "loss": 0.2248,
+      "step": 2604
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9002507254832297e-05,
+      "loss": 0.2444,
+      "step": 2605
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.900159149185227e-05,
+      "loss": 0.2153,
+      "step": 2606
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9000675330787075e-05,
+      "loss": 0.2541,
+      "step": 2607
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.899975877167724e-05,
+      "loss": 0.2521,
+      "step": 2608
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8998841814563288e-05,
+      "loss": 0.2587,
+      "step": 2609
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.899792445948578e-05,
+      "loss": 0.2509,
+      "step": 2610
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8997006706485283e-05,
+      "loss": 0.2355,
+      "step": 2611
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8996088555602378e-05,
+      "loss": 0.1963,
+      "step": 2612
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8995170006877676e-05,
+      "loss": 0.2289,
+      "step": 2613
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8994251060351793e-05,
+      "loss": 0.235,
+      "step": 2614
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.899333171606537e-05,
+      "loss": 0.2409,
+      "step": 2615
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.899241197405907e-05,
+      "loss": 0.2247,
+      "step": 2616
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.899149183437356e-05,
+      "loss": 0.2215,
+      "step": 2617
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8990571297049536e-05,
+      "loss": 0.2115,
+      "step": 2618
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.898965036212771e-05,
+      "loss": 0.2321,
+      "step": 2619
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.89887290296488e-05,
+      "loss": 0.2239,
+      "step": 2620
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8987807299653555e-05,
+      "loss": 0.2515,
+      "step": 2621
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8986885172182742e-05,
+      "loss": 0.2033,
+      "step": 2622
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8985962647277137e-05,
+      "loss": 0.2507,
+      "step": 2623
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8985039724977545e-05,
+      "loss": 0.2488,
+      "step": 2624
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8984116405324768e-05,
+      "loss": 0.2421,
+      "step": 2625
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8983192688359647e-05,
+      "loss": 0.2239,
+      "step": 2626
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8982268574123033e-05,
+      "loss": 0.2612,
+      "step": 2627
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.898134406265579e-05,
+      "loss": 0.2261,
+      "step": 2628
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8980419153998807e-05,
+      "loss": 0.2293,
+      "step": 2629
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8979493848192984e-05,
+      "loss": 0.2513,
+      "step": 2630
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8978568145279246e-05,
+      "loss": 0.2478,
+      "step": 2631
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8977642045298527e-05,
+      "loss": 0.2065,
+      "step": 2632
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.897671554829178e-05,
+      "loss": 0.2437,
+      "step": 2633
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8975788654299988e-05,
+      "loss": 0.2408,
+      "step": 2634
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.897486136336413e-05,
+      "loss": 0.2266,
+      "step": 2635
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.897393367552523e-05,
+      "loss": 0.2312,
+      "step": 2636
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8973005590824295e-05,
+      "loss": 0.2031,
+      "step": 2637
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8972077109302383e-05,
+      "loss": 0.2527,
+      "step": 2638
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8971148231000545e-05,
+      "loss": 0.2415,
+      "step": 2639
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8970218955959872e-05,
+      "loss": 0.2166,
+      "step": 2640
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.896928928422145e-05,
+      "loss": 0.2257,
+      "step": 2641
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.896835921582639e-05,
+      "loss": 0.2217,
+      "step": 2642
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8967428750815836e-05,
+      "loss": 0.2245,
+      "step": 2643
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8966497889230923e-05,
+      "loss": 0.234,
+      "step": 2644
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8965566631112828e-05,
+      "loss": 0.2601,
+      "step": 2645
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.896463497650273e-05,
+      "loss": 0.233,
+      "step": 2646
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.896370292544183e-05,
+      "loss": 0.2305,
+      "step": 2647
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8962770477971346e-05,
+      "loss": 0.2249,
+      "step": 2648
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8961837634132517e-05,
+      "loss": 0.2012,
+      "step": 2649
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.89609043939666e-05,
+      "loss": 0.2328,
+      "step": 2650
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8959970757514856e-05,
+      "loss": 0.2215,
+      "step": 2651
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8959036724818586e-05,
+      "loss": 0.2295,
+      "step": 2652
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8958102295919088e-05,
+      "loss": 0.1926,
+      "step": 2653
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8957167470857692e-05,
+      "loss": 0.2164,
+      "step": 2654
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8956232249675735e-05,
+      "loss": 0.2293,
+      "step": 2655
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.895529663241458e-05,
+      "loss": 0.2718,
+      "step": 2656
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.89543606191156e-05,
+      "loss": 0.2526,
+      "step": 2657
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8953424209820184e-05,
+      "loss": 0.233,
+      "step": 2658
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.895248740456976e-05,
+      "loss": 0.2277,
+      "step": 2659
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8951550203405743e-05,
+      "loss": 0.2255,
+      "step": 2660
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8950612606369586e-05,
+      "loss": 0.2289,
+      "step": 2661
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.894967461350275e-05,
+      "loss": 0.2812,
+      "step": 2662
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8948736224846714e-05,
+      "loss": 0.2087,
+      "step": 2663
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8947797440442987e-05,
+      "loss": 0.2511,
+      "step": 2664
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8946858260333078e-05,
+      "loss": 0.2641,
+      "step": 2665
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8945918684558523e-05,
+      "loss": 0.2385,
+      "step": 2666
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8944978713160873e-05,
+      "loss": 0.2278,
+      "step": 2667
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.89440383461817e-05,
+      "loss": 0.2972,
+      "step": 2668
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8943097583662584e-05,
+      "loss": 0.2481,
+      "step": 2669
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8942156425645138e-05,
+      "loss": 0.2563,
+      "step": 2670
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8941214872170977e-05,
+      "loss": 0.2638,
+      "step": 2671
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.894027292328174e-05,
+      "loss": 0.2314,
+      "step": 2672
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8939330579019087e-05,
+      "loss": 0.2284,
+      "step": 2673
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8938387839424693e-05,
+      "loss": 0.2434,
+      "step": 2674
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.893744470454025e-05,
+      "loss": 0.2134,
+      "step": 2675
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8936501174407462e-05,
+      "loss": 0.2525,
+      "step": 2676
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8935557249068057e-05,
+      "loss": 0.2324,
+      "step": 2677
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.893461292856378e-05,
+      "loss": 0.2518,
+      "step": 2678
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8933668212936396e-05,
+      "loss": 0.2627,
+      "step": 2679
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8932723102227677e-05,
+      "loss": 0.2512,
+      "step": 2680
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8931777596479427e-05,
+      "loss": 0.2765,
+      "step": 2681
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8930831695733452e-05,
+      "loss": 0.2004,
+      "step": 2682
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8929885400031593e-05,
+      "loss": 0.2134,
+      "step": 2683
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.892893870941569e-05,
+      "loss": 0.2168,
+      "step": 2684
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8927991623927614e-05,
+      "loss": 0.2473,
+      "step": 2685
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8927044143609246e-05,
+      "loss": 0.2081,
+      "step": 2686
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8926096268502486e-05,
+      "loss": 0.2299,
+      "step": 2687
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.892514799864926e-05,
+      "loss": 0.2734,
+      "step": 2688
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.89241993340915e-05,
+      "loss": 0.253,
+      "step": 2689
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8923250274871155e-05,
+      "loss": 0.2213,
+      "step": 2690
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8922300821030205e-05,
+      "loss": 0.2426,
+      "step": 2691
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.892135097261063e-05,
+      "loss": 0.2172,
+      "step": 2692
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.892040072965444e-05,
+      "loss": 0.243,
+      "step": 2693
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.891945009220366e-05,
+      "loss": 0.259,
+      "step": 2694
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.891849906030033e-05,
+      "loss": 0.2304,
+      "step": 2695
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8917547633986504e-05,
+      "loss": 0.2817,
+      "step": 2696
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8916595813304262e-05,
+      "loss": 0.2484,
+      "step": 2697
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8915643598295698e-05,
+      "loss": 0.2209,
+      "step": 2698
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.891469098900292e-05,
+      "loss": 0.2241,
+      "step": 2699
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8913737985468054e-05,
+      "loss": 0.215,
+      "step": 2700
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8912784587733254e-05,
+      "loss": 0.244,
+      "step": 2701
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8911830795840674e-05,
+      "loss": 0.2502,
+      "step": 2702
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8910876609832497e-05,
+      "loss": 0.2369,
+      "step": 2703
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8909922029750925e-05,
+      "loss": 0.2339,
+      "step": 2704
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8908967055638167e-05,
+      "loss": 0.2122,
+      "step": 2705
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.890801168753646e-05,
+      "loss": 0.2204,
+      "step": 2706
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.890705592548805e-05,
+      "loss": 0.2359,
+      "step": 2707
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.890609976953521e-05,
+      "loss": 0.2272,
+      "step": 2708
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.890514321972022e-05,
+      "loss": 0.2047,
+      "step": 2709
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8904186276085384e-05,
+      "loss": 0.2358,
+      "step": 2710
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8903228938673024e-05,
+      "loss": 0.2136,
+      "step": 2711
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8902271207525472e-05,
+      "loss": 0.2726,
+      "step": 2712
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.890131308268509e-05,
+      "loss": 0.2275,
+      "step": 2713
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.890035456419424e-05,
+      "loss": 0.2625,
+      "step": 2714
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.889939565209532e-05,
+      "loss": 0.2207,
+      "step": 2715
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8898436346430733e-05,
+      "loss": 0.2164,
+      "step": 2716
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8897476647242907e-05,
+      "loss": 0.244,
+      "step": 2717
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8896516554574278e-05,
+      "loss": 0.2235,
+      "step": 2718
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.889555606846731e-05,
+      "loss": 0.2155,
+      "step": 2719
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8894595188964472e-05,
+      "loss": 0.2382,
+      "step": 2720
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8893633916108266e-05,
+      "loss": 0.234,
+      "step": 2721
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.88926722499412e-05,
+      "loss": 0.2086,
+      "step": 2722
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8891710190505806e-05,
+      "loss": 0.2497,
+      "step": 2723
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8890747737844624e-05,
+      "loss": 0.2439,
+      "step": 2724
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.888978489200022e-05,
+      "loss": 0.2262,
+      "step": 2725
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8888821653015174e-05,
+      "loss": 0.24,
+      "step": 2726
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8887858020932082e-05,
+      "loss": 0.2074,
+      "step": 2727
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8886893995793567e-05,
+      "loss": 0.2191,
+      "step": 2728
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8885929577642258e-05,
+      "loss": 0.2258,
+      "step": 2729
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8884964766520803e-05,
+      "loss": 0.2335,
+      "step": 2730
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.888399956247187e-05,
+      "loss": 0.2383,
+      "step": 2731
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8883033965538148e-05,
+      "loss": 0.1935,
+      "step": 2732
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8882067975762334e-05,
+      "loss": 0.2249,
+      "step": 2733
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8881101593187155e-05,
+      "loss": 0.2261,
+      "step": 2734
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.888013481785534e-05,
+      "loss": 0.2471,
+      "step": 2735
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.887916764980965e-05,
+      "loss": 0.238,
+      "step": 2736
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.887820008909285e-05,
+      "loss": 0.2269,
+      "step": 2737
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.887723213574774e-05,
+      "loss": 0.2518,
+      "step": 2738
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8876263789817115e-05,
+      "loss": 0.2521,
+      "step": 2739
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8875295051343805e-05,
+      "loss": 0.2487,
+      "step": 2740
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8874325920370654e-05,
+      "loss": 0.2861,
+      "step": 2741
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8873356396940514e-05,
+      "loss": 0.235,
+      "step": 2742
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8872386481096266e-05,
+      "loss": 0.2278,
+      "step": 2743
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.88714161728808e-05,
+      "loss": 0.206,
+      "step": 2744
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8870445472337028e-05,
+      "loss": 0.2719,
+      "step": 2745
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.886947437950788e-05,
+      "loss": 0.2339,
+      "step": 2746
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.88685028944363e-05,
+      "loss": 0.2314,
+      "step": 2747
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.886753101716525e-05,
+      "loss": 0.2722,
+      "step": 2748
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8866558747737716e-05,
+      "loss": 0.2317,
+      "step": 2749
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8865586086196684e-05,
+      "loss": 0.2204,
+      "step": 2750
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8864613032585182e-05,
+      "loss": 0.2432,
+      "step": 2751
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8863639586946233e-05,
+      "loss": 0.2215,
+      "step": 2752
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8862665749322885e-05,
+      "loss": 0.2667,
+      "step": 2753
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8861691519758212e-05,
+      "loss": 0.2331,
+      "step": 2754
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8860716898295297e-05,
+      "loss": 0.2811,
+      "step": 2755
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.885974188497724e-05,
+      "loss": 0.2382,
+      "step": 2756
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8858766479847156e-05,
+      "loss": 0.2489,
+      "step": 2757
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8857790682948188e-05,
+      "loss": 0.256,
+      "step": 2758
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8856814494323485e-05,
+      "loss": 0.2455,
+      "step": 2759
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8855837914016222e-05,
+      "loss": 0.2364,
+      "step": 2760
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8854860942069582e-05,
+      "loss": 0.2413,
+      "step": 2761
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8853883578526772e-05,
+      "loss": 0.2155,
+      "step": 2762
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8852905823431015e-05,
+      "loss": 0.2343,
+      "step": 2763
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8851927676825555e-05,
+      "loss": 0.2446,
+      "step": 2764
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8850949138753646e-05,
+      "loss": 0.1963,
+      "step": 2765
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.884997020925856e-05,
+      "loss": 0.2014,
+      "step": 2766
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8848990888383593e-05,
+      "loss": 0.2322,
+      "step": 2767
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8848011176172055e-05,
+      "loss": 0.259,
+      "step": 2768
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8847031072667267e-05,
+      "loss": 0.2264,
+      "step": 2769
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.884605057791258e-05,
+      "loss": 0.2372,
+      "step": 2770
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.884506969195135e-05,
+      "loss": 0.2381,
+      "step": 2771
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.884408841482696e-05,
+      "loss": 0.2212,
+      "step": 2772
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.88431067465828e-05,
+      "loss": 0.2539,
+      "step": 2773
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8842124687262292e-05,
+      "loss": 0.2472,
+      "step": 2774
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.884114223690886e-05,
+      "loss": 0.2552,
+      "step": 2775
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8840159395565953e-05,
+      "loss": 0.2576,
+      "step": 2776
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8839176163277034e-05,
+      "loss": 0.211,
+      "step": 2777
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.883819254008559e-05,
+      "loss": 0.2486,
+      "step": 2778
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8837208526035113e-05,
+      "loss": 0.2226,
+      "step": 2779
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.883622412116913e-05,
+      "loss": 0.2662,
+      "step": 2780
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.883523932553117e-05,
+      "loss": 0.2278,
+      "step": 2781
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.883425413916478e-05,
+      "loss": 0.2084,
+      "step": 2782
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8833268562113538e-05,
+      "loss": 0.271,
+      "step": 2783
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8832282594421024e-05,
+      "loss": 0.2502,
+      "step": 2784
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.883129623613084e-05,
+      "loss": 0.2189,
+      "step": 2785
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8830309487286612e-05,
+      "loss": 0.2302,
+      "step": 2786
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8829322347931973e-05,
+      "loss": 0.221,
+      "step": 2787
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8828334818110584e-05,
+      "loss": 0.2572,
+      "step": 2788
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.882734689786611e-05,
+      "loss": 0.2168,
+      "step": 2789
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8826358587242245e-05,
+      "loss": 0.2976,
+      "step": 2790
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8825369886282695e-05,
+      "loss": 0.2407,
+      "step": 2791
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8824380795031187e-05,
+      "loss": 0.2183,
+      "step": 2792
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8823391313531453e-05,
+      "loss": 0.2411,
+      "step": 2793
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8822401441827263e-05,
+      "loss": 0.1859,
+      "step": 2794
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.882141117996239e-05,
+      "loss": 0.2486,
+      "step": 2795
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8820420527980622e-05,
+      "loss": 0.2495,
+      "step": 2796
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8819429485925774e-05,
+      "loss": 0.2609,
+      "step": 2797
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8818438053841673e-05,
+      "loss": 0.2395,
+      "step": 2798
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8817446231772165e-05,
+      "loss": 0.2182,
+      "step": 2799
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.881645401976111e-05,
+      "loss": 0.2803,
+      "step": 2800
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.881546141785239e-05,
+      "loss": 0.2296,
+      "step": 2801
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.88144684260899e-05,
+      "loss": 0.2399,
+      "step": 2802
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.881347504451755e-05,
+      "loss": 0.2127,
+      "step": 2803
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8812481273179277e-05,
+      "loss": 0.2434,
+      "step": 2804
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.881148711211903e-05,
+      "loss": 0.2235,
+      "step": 2805
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.881049256138077e-05,
+      "loss": 0.252,
+      "step": 2806
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8809497621008482e-05,
+      "loss": 0.2152,
+      "step": 2807
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8808502291046165e-05,
+      "loss": 0.2393,
+      "step": 2808
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8807506571537838e-05,
+      "loss": 0.2774,
+      "step": 2809
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.880651046252754e-05,
+      "loss": 0.2219,
+      "step": 2810
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.880551396405931e-05,
+      "loss": 0.241,
+      "step": 2811
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.880451707617723e-05,
+      "loss": 0.2337,
+      "step": 2812
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.880351979892538e-05,
+      "loss": 0.2043,
+      "step": 2813
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8802522132347863e-05,
+      "loss": 0.243,
+      "step": 2814
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8801524076488805e-05,
+      "loss": 0.2082,
+      "step": 2815
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8800525631392337e-05,
+      "loss": 0.2279,
+      "step": 2816
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8799526797102618e-05,
+      "loss": 0.1986,
+      "step": 2817
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8798527573663824e-05,
+      "loss": 0.2454,
+      "step": 2818
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.879752796112013e-05,
+      "loss": 0.228,
+      "step": 2819
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.879652795951576e-05,
+      "loss": 0.243,
+      "step": 2820
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8795527568894934e-05,
+      "loss": 0.2188,
+      "step": 2821
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8794526789301884e-05,
+      "loss": 0.2227,
+      "step": 2822
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8793525620780876e-05,
+      "loss": 0.2459,
+      "step": 2823
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.879252406337618e-05,
+      "loss": 0.2087,
+      "step": 2824
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8791522117132098e-05,
+      "loss": 0.2222,
+      "step": 2825
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8790519782092935e-05,
+      "loss": 0.2457,
+      "step": 2826
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8789517058303013e-05,
+      "loss": 0.2432,
+      "step": 2827
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8788513945806683e-05,
+      "loss": 0.2503,
+      "step": 2828
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8787510444648302e-05,
+      "loss": 0.2428,
+      "step": 2829
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8786506554872256e-05,
+      "loss": 0.2392,
+      "step": 2830
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8785502276522932e-05,
+      "loss": 0.2263,
+      "step": 2831
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8784497609644745e-05,
+      "loss": 0.2569,
+      "step": 2832
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8783492554282134e-05,
+      "loss": 0.2261,
+      "step": 2833
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8782487110479534e-05,
+      "loss": 0.2338,
+      "step": 2834
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.878148127828142e-05,
+      "loss": 0.1884,
+      "step": 2835
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8780475057732267e-05,
+      "loss": 0.209,
+      "step": 2836
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8779468448876573e-05,
+      "loss": 0.2223,
+      "step": 2837
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.877846145175886e-05,
+      "loss": 0.2275,
+      "step": 2838
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8777454066423658e-05,
+      "loss": 0.215,
+      "step": 2839
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.877644629291552e-05,
+      "loss": 0.2345,
+      "step": 2840
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8775438131279008e-05,
+      "loss": 0.2621,
+      "step": 2841
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8774429581558714e-05,
+      "loss": 0.2239,
+      "step": 2842
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8773420643799233e-05,
+      "loss": 0.2287,
+      "step": 2843
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.877241131804519e-05,
+      "loss": 0.2084,
+      "step": 2844
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.877140160434122e-05,
+      "loss": 0.275,
+      "step": 2845
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8770391502731975e-05,
+      "loss": 0.3078,
+      "step": 2846
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8769381013262125e-05,
+      "loss": 0.2133,
+      "step": 2847
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8768370135976357e-05,
+      "loss": 0.2347,
+      "step": 2848
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.876735887091938e-05,
+      "loss": 0.2174,
+      "step": 2849
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.876634721813591e-05,
+      "loss": 0.2254,
+      "step": 2850
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.876533517767069e-05,
+      "loss": 0.2144,
+      "step": 2851
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.876432274956848e-05,
+      "loss": 0.2459,
+      "step": 2852
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8763309933874046e-05,
+      "loss": 0.2505,
+      "step": 2853
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8762296730632186e-05,
+      "loss": 0.2058,
+      "step": 2854
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8761283139887704e-05,
+      "loss": 0.2774,
+      "step": 2855
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.876026916168542e-05,
+      "loss": 0.2303,
+      "step": 2856
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8759254796070183e-05,
+      "loss": 0.2656,
+      "step": 2857
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.875824004308685e-05,
+      "loss": 0.2279,
+      "step": 2858
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8757224902780298e-05,
+      "loss": 0.2455,
+      "step": 2859
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8756209375195422e-05,
+      "loss": 0.2104,
+      "step": 2860
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8755193460377125e-05,
+      "loss": 0.2769,
+      "step": 2861
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8754177158370344e-05,
+      "loss": 0.1989,
+      "step": 2862
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.875316046922002e-05,
+      "loss": 0.2553,
+      "step": 2863
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.875214339297111e-05,
+      "loss": 0.2501,
+      "step": 2864
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8751125929668603e-05,
+      "loss": 0.2471,
+      "step": 2865
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.875010807935749e-05,
+      "loss": 0.2101,
+      "step": 2866
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.874908984208278e-05,
+      "loss": 0.2576,
+      "step": 2867
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8748071217889514e-05,
+      "loss": 0.2281,
+      "step": 2868
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8747052206822726e-05,
+      "loss": 0.2265,
+      "step": 2869
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8746032808927494e-05,
+      "loss": 0.1911,
+      "step": 2870
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.874501302424889e-05,
+      "loss": 0.2376,
+      "step": 2871
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8743992852832016e-05,
+      "loss": 0.2518,
+      "step": 2872
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.874297229472199e-05,
+      "loss": 0.2492,
+      "step": 2873
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.874195134996394e-05,
+      "loss": 0.245,
+      "step": 2874
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.874093001860302e-05,
+      "loss": 0.2079,
+      "step": 2875
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.87399083006844e-05,
+      "loss": 0.2462,
+      "step": 2876
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8738886196253258e-05,
+      "loss": 0.2374,
+      "step": 2877
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.87378637053548e-05,
+      "loss": 0.2635,
+      "step": 2878
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.873684082803424e-05,
+      "loss": 0.2258,
+      "step": 2879
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.873581756433682e-05,
+      "loss": 0.2579,
+      "step": 2880
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8734793914307787e-05,
+      "loss": 0.2179,
+      "step": 2881
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.873376987799241e-05,
+      "loss": 0.2258,
+      "step": 2882
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8732745455435983e-05,
+      "loss": 0.2673,
+      "step": 2883
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8731720646683804e-05,
+      "loss": 0.2525,
+      "step": 2884
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8730695451781197e-05,
+      "loss": 0.2112,
+      "step": 2885
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.87296698707735e-05,
+      "loss": 0.2803,
+      "step": 2886
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.872864390370606e-05,
+      "loss": 0.2346,
+      "step": 2887
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.872761755062426e-05,
+      "loss": 0.235,
+      "step": 2888
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8726590811573486e-05,
+      "loss": 0.2068,
+      "step": 2889
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8725563686599145e-05,
+      "loss": 0.2626,
+      "step": 2890
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8724536175746653e-05,
+      "loss": 0.2192,
+      "step": 2891
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8723508279061465e-05,
+      "loss": 0.2188,
+      "step": 2892
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8722479996589023e-05,
+      "loss": 0.215,
+      "step": 2893
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8721451328374812e-05,
+      "loss": 0.2376,
+      "step": 2894
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.872042227446432e-05,
+      "loss": 0.2639,
+      "step": 2895
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.871939283490306e-05,
+      "loss": 0.271,
+      "step": 2896
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.871836300973655e-05,
+      "loss": 0.2925,
+      "step": 2897
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.871733279901034e-05,
+      "loss": 0.211,
+      "step": 2898
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8716302202769982e-05,
+      "loss": 0.2597,
+      "step": 2899
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8715271221061058e-05,
+      "loss": 0.2308,
+      "step": 2900
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8714239853929162e-05,
+      "loss": 0.199,
+      "step": 2901
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8713208101419908e-05,
+      "loss": 0.2376,
+      "step": 2902
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.871217596357892e-05,
+      "loss": 0.2393,
+      "step": 2903
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8711143440451846e-05,
+      "loss": 0.2351,
+      "step": 2904
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.871011053208434e-05,
+      "loss": 0.2493,
+      "step": 2905
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8709077238522093e-05,
+      "loss": 0.2271,
+      "step": 2906
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8708043559810792e-05,
+      "loss": 0.251,
+      "step": 2907
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8707009495996155e-05,
+      "loss": 0.2405,
+      "step": 2908
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.870597504712391e-05,
+      "loss": 0.2401,
+      "step": 2909
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.870494021323981e-05,
+      "loss": 0.1862,
+      "step": 2910
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.870390499438961e-05,
+      "loss": 0.2382,
+      "step": 2911
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8702869390619095e-05,
+      "loss": 0.2301,
+      "step": 2912
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.870183340197407e-05,
+      "loss": 0.2602,
+      "step": 2913
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8700797028500342e-05,
+      "loss": 0.2118,
+      "step": 2914
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8699760270243745e-05,
+      "loss": 0.2556,
+      "step": 2915
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.869872312725013e-05,
+      "loss": 0.2159,
+      "step": 2916
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8697685599565368e-05,
+      "loss": 0.224,
+      "step": 2917
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8696647687235335e-05,
+      "loss": 0.2732,
+      "step": 2918
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8695609390305934e-05,
+      "loss": 0.2514,
+      "step": 2919
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8694570708823083e-05,
+      "loss": 0.2426,
+      "step": 2920
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8693531642832716e-05,
+      "loss": 0.2474,
+      "step": 2921
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8692492192380785e-05,
+      "loss": 0.2715,
+      "step": 2922
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.869145235751326e-05,
+      "loss": 0.2318,
+      "step": 2923
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8690412138276122e-05,
+      "loss": 0.2224,
+      "step": 2924
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8689371534715377e-05,
+      "loss": 0.2647,
+      "step": 2925
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.868833054687705e-05,
+      "loss": 0.2257,
+      "step": 2926
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8687289174807163e-05,
+      "loss": 0.2313,
+      "step": 2927
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.868624741855178e-05,
+      "loss": 0.224,
+      "step": 2928
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8685205278156972e-05,
+      "loss": 0.2305,
+      "step": 2929
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.868416275366882e-05,
+      "loss": 0.2399,
+      "step": 2930
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8683119845133438e-05,
+      "loss": 0.2246,
+      "step": 2931
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.868207655259694e-05,
+      "loss": 0.233,
+      "step": 2932
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8681032876105463e-05,
+      "loss": 0.2505,
+      "step": 2933
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8679988815705167e-05,
+      "loss": 0.2041,
+      "step": 2934
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8678944371442225e-05,
+      "loss": 0.2311,
+      "step": 2935
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8677899543362823e-05,
+      "loss": 0.2356,
+      "step": 2936
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.867685433151317e-05,
+      "loss": 0.2162,
+      "step": 2937
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8675808735939484e-05,
+      "loss": 0.2469,
+      "step": 2938
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8674762756688012e-05,
+      "loss": 0.2286,
+      "step": 2939
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8673716393805012e-05,
+      "loss": 0.2365,
+      "step": 2940
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8672669647336748e-05,
+      "loss": 0.2558,
+      "step": 2941
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8671622517329522e-05,
+      "loss": 0.2144,
+      "step": 2942
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8670575003829638e-05,
+      "loss": 0.2285,
+      "step": 2943
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.866952710688342e-05,
+      "loss": 0.2052,
+      "step": 2944
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8668478826537212e-05,
+      "loss": 0.2189,
+      "step": 2945
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8667430162837374e-05,
+      "loss": 0.2255,
+      "step": 2946
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8666381115830274e-05,
+      "loss": 0.245,
+      "step": 2947
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.866533168556232e-05,
+      "loss": 0.216,
+      "step": 2948
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8664281872079906e-05,
+      "loss": 0.2256,
+      "step": 2949
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.866323167542947e-05,
+      "loss": 0.2348,
+      "step": 2950
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8662181095657447e-05,
+      "loss": 0.2296,
+      "step": 2951
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8661130132810306e-05,
+      "loss": 0.2368,
+      "step": 2952
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8660078786934517e-05,
+      "loss": 0.2455,
+      "step": 2953
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8659027058076583e-05,
+      "loss": 0.2623,
+      "step": 2954
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.865797494628301e-05,
+      "loss": 0.2323,
+      "step": 2955
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8656922451600326e-05,
+      "loss": 0.2248,
+      "step": 2956
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8655869574075076e-05,
+      "loss": 0.2083,
+      "step": 2957
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8654816313753828e-05,
+      "loss": 0.2276,
+      "step": 2958
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8653762670683154e-05,
+      "loss": 0.2365,
+      "step": 2959
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.865270864490966e-05,
+      "loss": 0.2563,
+      "step": 2960
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8651654236479944e-05,
+      "loss": 0.2541,
+      "step": 2961
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.865059944544065e-05,
+      "loss": 0.2245,
+      "step": 2962
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8649544271838418e-05,
+      "loss": 0.2308,
+      "step": 2963
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8648488715719913e-05,
+      "loss": 0.2358,
+      "step": 2964
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8647432777131815e-05,
+      "loss": 0.2142,
+      "step": 2965
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8646376456120824e-05,
+      "loss": 0.2821,
+      "step": 2966
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8645319752733655e-05,
+      "loss": 0.2818,
+      "step": 2967
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8644262667017036e-05,
+      "loss": 0.2431,
+      "step": 2968
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8643205199017718e-05,
+      "loss": 0.2167,
+      "step": 2969
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8642147348782466e-05,
+      "loss": 0.2293,
+      "step": 2970
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.864108911635806e-05,
+      "loss": 0.2091,
+      "step": 2971
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8640030501791306e-05,
+      "loss": 0.2489,
+      "step": 2972
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8638971505129012e-05,
+      "loss": 0.2237,
+      "step": 2973
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8637912126418013e-05,
+      "loss": 0.2202,
+      "step": 2974
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.863685236570516e-05,
+      "loss": 0.2242,
+      "step": 2975
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.863579222303732e-05,
+      "loss": 0.2398,
+      "step": 2976
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8634731698461377e-05,
+      "loss": 0.2232,
+      "step": 2977
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.863367079202423e-05,
+      "loss": 0.2256,
+      "step": 2978
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8632609503772798e-05,
+      "loss": 0.2758,
+      "step": 2979
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8631547833754013e-05,
+      "loss": 0.218,
+      "step": 2980
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.863048578201483e-05,
+      "loss": 0.2245,
+      "step": 2981
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8629423348602208e-05,
+      "loss": 0.2279,
+      "step": 2982
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8628360533563147e-05,
+      "loss": 0.2298,
+      "step": 2983
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8627297336944633e-05,
+      "loss": 0.2539,
+      "step": 2984
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8626233758793698e-05,
+      "loss": 0.2516,
+      "step": 2985
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8625169799157365e-05,
+      "loss": 0.2298,
+      "step": 2986
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8624105458082696e-05,
+      "loss": 0.209,
+      "step": 2987
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8623040735616758e-05,
+      "loss": 0.2417,
+      "step": 2988
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8621975631806636e-05,
+      "loss": 0.1966,
+      "step": 2989
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8620910146699432e-05,
+      "loss": 0.2407,
+      "step": 2990
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.861984428034227e-05,
+      "loss": 0.251,
+      "step": 2991
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8618778032782278e-05,
+      "loss": 0.236,
+      "step": 2992
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8617711404066618e-05,
+      "loss": 0.2284,
+      "step": 2993
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8616644394242457e-05,
+      "loss": 0.2048,
+      "step": 2994
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8615577003356987e-05,
+      "loss": 0.2259,
+      "step": 2995
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8614509231457405e-05,
+      "loss": 0.2298,
+      "step": 2996
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8613441078590938e-05,
+      "loss": 0.2312,
+      "step": 2997
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8612372544804817e-05,
+      "loss": 0.2331,
+      "step": 2998
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8611303630146303e-05,
+      "loss": 0.2654,
+      "step": 2999
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8610234334662666e-05,
+      "loss": 0.243,
+      "step": 3000
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8609164658401193e-05,
+      "loss": 0.2389,
+      "step": 3001
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.860809460140919e-05,
+      "loss": 0.2757,
+      "step": 3002
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.860702416373398e-05,
+      "loss": 0.2422,
+      "step": 3003
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.86059533454229e-05,
+      "loss": 0.2381,
+      "step": 3004
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8604882146523306e-05,
+      "loss": 0.2625,
+      "step": 3005
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8603810567082572e-05,
+      "loss": 0.2347,
+      "step": 3006
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8602738607148088e-05,
+      "loss": 0.2708,
+      "step": 3007
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8601666266767256e-05,
+      "loss": 0.2351,
+      "step": 3008
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.86005935459875e-05,
+      "loss": 0.1978,
+      "step": 3009
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8599520444856268e-05,
+      "loss": 0.2534,
+      "step": 3010
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8598446963421004e-05,
+      "loss": 0.2142,
+      "step": 3011
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.859737310172919e-05,
+      "loss": 0.2153,
+      "step": 3012
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8596298859828316e-05,
+      "loss": 0.2066,
+      "step": 3013
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.859522423776589e-05,
+      "loss": 0.2215,
+      "step": 3014
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8594149235589427e-05,
+      "loss": 0.2393,
+      "step": 3015
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8593073853346476e-05,
+      "loss": 0.252,
+      "step": 3016
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8591998091084592e-05,
+      "loss": 0.2502,
+      "step": 3017
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8590921948851352e-05,
+      "loss": 0.2321,
+      "step": 3018
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8589845426694344e-05,
+      "loss": 0.2502,
+      "step": 3019
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.858876852466118e-05,
+      "loss": 0.2364,
+      "step": 3020
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8587691242799474e-05,
+      "loss": 0.2315,
+      "step": 3021
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.858661358115688e-05,
+      "loss": 0.2524,
+      "step": 3022
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8585535539781055e-05,
+      "loss": 0.2254,
+      "step": 3023
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8584457118719664e-05,
+      "loss": 0.2135,
+      "step": 3024
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8583378318020413e-05,
+      "loss": 0.224,
+      "step": 3025
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8582299137731e-05,
+      "loss": 0.2728,
+      "step": 3026
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.858121957789915e-05,
+      "loss": 0.2248,
+      "step": 3027
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8580139638572618e-05,
+      "loss": 0.2137,
+      "step": 3028
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.857905931979915e-05,
+      "loss": 0.2482,
+      "step": 3029
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8577978621626527e-05,
+      "loss": 0.2484,
+      "step": 3030
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.857689754410254e-05,
+      "loss": 0.1888,
+      "step": 3031
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8575816087274998e-05,
+      "loss": 0.2409,
+      "step": 3032
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.857473425119173e-05,
+      "loss": 0.2124,
+      "step": 3033
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.857365203590058e-05,
+      "loss": 0.2445,
+      "step": 3034
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8572569441449405e-05,
+      "loss": 0.2501,
+      "step": 3035
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8571486467886077e-05,
+      "loss": 0.2105,
+      "step": 3036
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.85704031152585e-05,
+      "loss": 0.2207,
+      "step": 3037
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8569319383614575e-05,
+      "loss": 0.2066,
+      "step": 3038
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8568235273002234e-05,
+      "loss": 0.2287,
+      "step": 3039
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8567150783469414e-05,
+      "loss": 0.2146,
+      "step": 3040
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8566065915064085e-05,
+      "loss": 0.2473,
+      "step": 3041
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8564980667834218e-05,
+      "loss": 0.2678,
+      "step": 3042
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8563895041827805e-05,
+      "loss": 0.2221,
+      "step": 3043
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8562809037092863e-05,
+      "loss": 0.2464,
+      "step": 3044
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.856172265367741e-05,
+      "loss": 0.2129,
+      "step": 3045
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8560635891629505e-05,
+      "loss": 0.2671,
+      "step": 3046
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.855954875099719e-05,
+      "loss": 0.2637,
+      "step": 3047
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8558461231828556e-05,
+      "loss": 0.2591,
+      "step": 3048
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8557373334171695e-05,
+      "loss": 0.2561,
+      "step": 3049
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8556285058074713e-05,
+      "loss": 0.2347,
+      "step": 3050
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8555196403585745e-05,
+      "loss": 0.2369,
+      "step": 3051
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.855410737075293e-05,
+      "loss": 0.2599,
+      "step": 3052
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8553017959624432e-05,
+      "loss": 0.263,
+      "step": 3053
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.855192817024843e-05,
+      "loss": 0.2021,
+      "step": 3054
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.855083800267311e-05,
+      "loss": 0.2504,
+      "step": 3055
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8549747456946694e-05,
+      "loss": 0.2556,
+      "step": 3056
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8548656533117408e-05,
+      "loss": 0.2525,
+      "step": 3057
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.854756523123349e-05,
+      "loss": 0.2262,
+      "step": 3058
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.854647355134321e-05,
+      "loss": 0.2599,
+      "step": 3059
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8545381493494844e-05,
+      "loss": 0.2102,
+      "step": 3060
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.854428905773668e-05,
+      "loss": 0.2248,
+      "step": 3061
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8543196244117045e-05,
+      "loss": 0.2526,
+      "step": 3062
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.854210305268425e-05,
+      "loss": 0.2188,
+      "step": 3063
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8541009483486652e-05,
+      "loss": 0.2948,
+      "step": 3064
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.853991553657261e-05,
+      "loss": 0.2335,
+      "step": 3065
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.85388212119905e-05,
+      "loss": 0.2514,
+      "step": 3066
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.853772650978872e-05,
+      "loss": 0.2525,
+      "step": 3067
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8536631430015682e-05,
+      "loss": 0.244,
+      "step": 3068
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.853553597271981e-05,
+      "loss": 0.2053,
+      "step": 3069
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8534440137949557e-05,
+      "loss": 0.2193,
+      "step": 3070
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.853334392575338e-05,
+      "loss": 0.2192,
+      "step": 3071
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8532247336179754e-05,
+      "loss": 0.2272,
+      "step": 3072
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8531150369277186e-05,
+      "loss": 0.2603,
+      "step": 3073
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8530053025094177e-05,
+      "loss": 0.2327,
+      "step": 3074
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.852895530367926e-05,
+      "loss": 0.2499,
+      "step": 3075
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8527857205080983e-05,
+      "loss": 0.2049,
+      "step": 3076
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8526758729347906e-05,
+      "loss": 0.2391,
+      "step": 3077
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8525659876528605e-05,
+      "loss": 0.2818,
+      "step": 3078
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.852456064667168e-05,
+      "loss": 0.2077,
+      "step": 3079
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.852346103982574e-05,
+      "loss": 0.2156,
+      "step": 3080
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8522361056039416e-05,
+      "loss": 0.2264,
+      "step": 3081
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8521260695361353e-05,
+      "loss": 0.2186,
+      "step": 3082
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8520159957840212e-05,
+      "loss": 0.2507,
+      "step": 3083
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8519058843524674e-05,
+      "loss": 0.2467,
+      "step": 3084
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.851795735246343e-05,
+      "loss": 0.2296,
+      "step": 3085
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.85168554847052e-05,
+      "loss": 0.2422,
+      "step": 3086
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8515753240298705e-05,
+      "loss": 0.2665,
+      "step": 3087
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8514650619292697e-05,
+      "loss": 0.2338,
+      "step": 3088
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8513547621735932e-05,
+      "loss": 0.2301,
+      "step": 3089
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8512444247677194e-05,
+      "loss": 0.2087,
+      "step": 3090
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8511340497165276e-05,
+      "loss": 0.2388,
+      "step": 3091
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8510236370248987e-05,
+      "loss": 0.2341,
+      "step": 3092
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8509131866977165e-05,
+      "loss": 0.2501,
+      "step": 3093
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8508026987398644e-05,
+      "loss": 0.2037,
+      "step": 3094
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8506921731562293e-05,
+      "loss": 0.2148,
+      "step": 3095
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.850581609951699e-05,
+      "loss": 0.2245,
+      "step": 3096
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8504710091311632e-05,
+      "loss": 0.2153,
+      "step": 3097
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8503603706995123e-05,
+      "loss": 0.2371,
+      "step": 3098
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8502496946616402e-05,
+      "loss": 0.2057,
+      "step": 3099
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8501389810224408e-05,
+      "loss": 0.2186,
+      "step": 3100
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8500282297868105e-05,
+      "loss": 0.246,
+      "step": 3101
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.849917440959647e-05,
+      "loss": 0.2672,
+      "step": 3102
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8498066145458497e-05,
+      "loss": 0.1894,
+      "step": 3103
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.84969575055032e-05,
+      "loss": 0.2374,
+      "step": 3104
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8495848489779612e-05,
+      "loss": 0.2237,
+      "step": 3105
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8494739098336765e-05,
+      "loss": 0.2275,
+      "step": 3106
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8493629331223734e-05,
+      "loss": 0.2522,
+      "step": 3107
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.849251918848959e-05,
+      "loss": 0.2067,
+      "step": 3108
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8491408670183428e-05,
+      "loss": 0.2187,
+      "step": 3109
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8490297776354363e-05,
+      "loss": 0.251,
+      "step": 3110
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.848918650705152e-05,
+      "loss": 0.218,
+      "step": 3111
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8488074862324042e-05,
+      "loss": 0.2389,
+      "step": 3112
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8486962842221096e-05,
+      "loss": 0.2386,
+      "step": 3113
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8485850446791857e-05,
+      "loss": 0.2141,
+      "step": 3114
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8484737676085514e-05,
+      "loss": 0.274,
+      "step": 3115
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8483624530151287e-05,
+      "loss": 0.2031,
+      "step": 3116
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.84825110090384e-05,
+      "loss": 0.217,
+      "step": 3117
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8481397112796092e-05,
+      "loss": 0.222,
+      "step": 3118
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8480282841473634e-05,
+      "loss": 0.2726,
+      "step": 3119
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8479168195120297e-05,
+      "loss": 0.2376,
+      "step": 3120
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8478053173785374e-05,
+      "loss": 0.2591,
+      "step": 3121
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8476937777518177e-05,
+      "loss": 0.2274,
+      "step": 3122
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8475822006368038e-05,
+      "loss": 0.2519,
+      "step": 3123
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8474705860384292e-05,
+      "loss": 0.2392,
+      "step": 3124
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8473589339616304e-05,
+      "loss": 0.2181,
+      "step": 3125
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8472472444113455e-05,
+      "loss": 0.2327,
+      "step": 3126
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.847135517392513e-05,
+      "loss": 0.2257,
+      "step": 3127
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.847023752910074e-05,
+      "loss": 0.234,
+      "step": 3128
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8469119509689717e-05,
+      "loss": 0.2108,
+      "step": 3129
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8468001115741502e-05,
+      "loss": 0.2334,
+      "step": 3130
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8466882347305553e-05,
+      "loss": 0.2391,
+      "step": 3131
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8465763204431352e-05,
+      "loss": 0.2828,
+      "step": 3132
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8464643687168384e-05,
+      "loss": 0.2408,
+      "step": 3133
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.846352379556616e-05,
+      "loss": 0.214,
+      "step": 3134
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8462403529674215e-05,
+      "loss": 0.2222,
+      "step": 3135
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8461282889542075e-05,
+      "loss": 0.2094,
+      "step": 3136
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8460161875219317e-05,
+      "loss": 0.21,
+      "step": 3137
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.84590404867555e-05,
+      "loss": 0.225,
+      "step": 3138
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8457918724200228e-05,
+      "loss": 0.2308,
+      "step": 3139
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8456796587603105e-05,
+      "loss": 0.253,
+      "step": 3140
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8455674077013756e-05,
+      "loss": 0.2398,
+      "step": 3141
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8454551192481824e-05,
+      "loss": 0.2555,
+      "step": 3142
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.845342793405697e-05,
+      "loss": 0.2408,
+      "step": 3143
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8452304301788858e-05,
+      "loss": 0.214,
+      "step": 3144
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8451180295727197e-05,
+      "loss": 0.2392,
+      "step": 3145
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8450055915921678e-05,
+      "loss": 0.2662,
+      "step": 3146
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8448931162422034e-05,
+      "loss": 0.2214,
+      "step": 3147
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8447806035278004e-05,
+      "loss": 0.2302,
+      "step": 3148
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8446680534539348e-05,
+      "loss": 0.2317,
+      "step": 3149
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8445554660255836e-05,
+      "loss": 0.224,
+      "step": 3150
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.844442841247726e-05,
+      "loss": 0.2453,
+      "step": 3151
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.844330179125343e-05,
+      "loss": 0.2108,
+      "step": 3152
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.844217479663416e-05,
+      "loss": 0.2622,
+      "step": 3153
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8441047428669304e-05,
+      "loss": 0.2164,
+      "step": 3154
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8439919687408712e-05,
+      "loss": 0.2267,
+      "step": 3155
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8438791572902253e-05,
+      "loss": 0.2477,
+      "step": 3156
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8437663085199822e-05,
+      "loss": 0.2357,
+      "step": 3157
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.843653422435132e-05,
+      "loss": 0.2705,
+      "step": 3158
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8435404990406674e-05,
+      "loss": 0.2635,
+      "step": 3159
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8434275383415825e-05,
+      "loss": 0.2241,
+      "step": 3160
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8433145403428723e-05,
+      "loss": 0.2384,
+      "step": 3161
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8432015050495344e-05,
+      "loss": 0.2398,
+      "step": 3162
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8430884324665675e-05,
+      "loss": 0.2032,
+      "step": 3163
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.842975322598972e-05,
+      "loss": 0.2664,
+      "step": 3164
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8428621754517504e-05,
+      "loss": 0.2695,
+      "step": 3165
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.842748991029906e-05,
+      "loss": 0.2356,
+      "step": 3166
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8426357693384446e-05,
+      "loss": 0.2512,
+      "step": 3167
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8425225103823735e-05,
+      "loss": 0.1982,
+      "step": 3168
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.842409214166701e-05,
+      "loss": 0.2082,
+      "step": 3169
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8422958806964377e-05,
+      "loss": 0.2196,
+      "step": 3170
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8421825099765955e-05,
+      "loss": 0.224,
+      "step": 3171
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8420691020121884e-05,
+      "loss": 0.274,
+      "step": 3172
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8419556568082316e-05,
+      "loss": 0.2,
+      "step": 3173
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8418421743697415e-05,
+      "loss": 0.2221,
+      "step": 3174
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.841728654701738e-05,
+      "loss": 0.254,
+      "step": 3175
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8416150978092405e-05,
+      "loss": 0.2454,
+      "step": 3176
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8415015036972707e-05,
+      "loss": 0.2305,
+      "step": 3177
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8413878723708528e-05,
+      "loss": 0.2809,
+      "step": 3178
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8412742038350117e-05,
+      "loss": 0.2147,
+      "step": 3179
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8411604980947744e-05,
+      "loss": 0.2085,
+      "step": 3180
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8410467551551693e-05,
+      "loss": 0.2425,
+      "step": 3181
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8409329750212266e-05,
+      "loss": 0.2292,
+      "step": 3182
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8408191576979782e-05,
+      "loss": 0.258,
+      "step": 3183
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.840705303190457e-05,
+      "loss": 0.2552,
+      "step": 3184
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8405914115036992e-05,
+      "loss": 0.2513,
+      "step": 3185
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.84047748264274e-05,
+      "loss": 0.2715,
+      "step": 3186
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8403635166126194e-05,
+      "loss": 0.2685,
+      "step": 3187
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.840249513418376e-05,
+      "loss": 0.2438,
+      "step": 3188
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8401354730650526e-05,
+      "loss": 0.2312,
+      "step": 3189
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.840021395557692e-05,
+      "loss": 0.2153,
+      "step": 3190
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8399072809013384e-05,
+      "loss": 0.228,
+      "step": 3191
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.83979312910104e-05,
+      "loss": 0.2075,
+      "step": 3192
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.839678940161844e-05,
+      "loss": 0.2012,
+      "step": 3193
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8395647140888e-05,
+      "loss": 0.2011,
+      "step": 3194
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.83945045088696e-05,
+      "loss": 0.2627,
+      "step": 3195
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8393361505613776e-05,
+      "loss": 0.2425,
+      "step": 3196
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.839221813117107e-05,
+      "loss": 0.2377,
+      "step": 3197
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8391074385592047e-05,
+      "loss": 0.2404,
+      "step": 3198
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.838993026892729e-05,
+      "loss": 0.226,
+      "step": 3199
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8388785781227393e-05,
+      "loss": 0.2492,
+      "step": 3200
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.838764092254297e-05,
+      "loss": 0.2324,
+      "step": 3201
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8386495692924654e-05,
+      "loss": 0.2437,
+      "step": 3202
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8385350092423094e-05,
+      "loss": 0.2091,
+      "step": 3203
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8384204121088944e-05,
+      "loss": 0.2482,
+      "step": 3204
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.838305777897289e-05,
+      "loss": 0.2251,
+      "step": 3205
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8381911066125627e-05,
+      "loss": 0.2353,
+      "step": 3206
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8380763982597864e-05,
+      "loss": 0.2252,
+      "step": 3207
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8379616528440332e-05,
+      "loss": 0.2322,
+      "step": 3208
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8378468703703774e-05,
+      "loss": 0.2013,
+      "step": 3209
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8377320508438956e-05,
+      "loss": 0.2441,
+      "step": 3210
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.837617194269665e-05,
+      "loss": 0.2373,
+      "step": 3211
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.837502300652765e-05,
+      "loss": 0.2302,
+      "step": 3212
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8373873699982772e-05,
+      "loss": 0.2149,
+      "step": 3213
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.837272402311284e-05,
+      "loss": 0.2245,
+      "step": 3214
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8371573975968694e-05,
+      "loss": 0.2097,
+      "step": 3215
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.83704235586012e-05,
+      "loss": 0.2062,
+      "step": 3216
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8369272771061224e-05,
+      "loss": 0.2191,
+      "step": 3217
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.836812161339967e-05,
+      "loss": 0.2132,
+      "step": 3218
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8366970085667435e-05,
+      "loss": 0.2528,
+      "step": 3219
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8365818187915458e-05,
+      "loss": 0.1979,
+      "step": 3220
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8364665920194664e-05,
+      "loss": 0.255,
+      "step": 3221
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8363513282556022e-05,
+      "loss": 0.2505,
+      "step": 3222
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8362360275050503e-05,
+      "loss": 0.2253,
+      "step": 3223
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.83612068977291e-05,
+      "loss": 0.2169,
+      "step": 3224
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8360053150642813e-05,
+      "loss": 0.2458,
+      "step": 3225
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8358899033842675e-05,
+      "loss": 0.2342,
+      "step": 3226
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8357744547379713e-05,
+      "loss": 0.2329,
+      "step": 3227
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8356589691304994e-05,
+      "loss": 0.2539,
+      "step": 3228
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8355434465669587e-05,
+      "loss": 0.2219,
+      "step": 3229
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8354278870524577e-05,
+      "loss": 0.2033,
+      "step": 3230
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8353122905921072e-05,
+      "loss": 0.2349,
+      "step": 3231
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8351966571910195e-05,
+      "loss": 0.2175,
+      "step": 3232
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8350809868543076e-05,
+      "loss": 0.2477,
+      "step": 3233
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.834965279587088e-05,
+      "loss": 0.2278,
+      "step": 3234
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.834849535394477e-05,
+      "loss": 0.2048,
+      "step": 3235
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8347337542815932e-05,
+      "loss": 0.2369,
+      "step": 3236
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8346179362535573e-05,
+      "loss": 0.2091,
+      "step": 3237
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.834502081315491e-05,
+      "loss": 0.219,
+      "step": 3238
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8343861894725177e-05,
+      "loss": 0.2445,
+      "step": 3239
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8342702607297624e-05,
+      "loss": 0.2731,
+      "step": 3240
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.834154295092353e-05,
+      "loss": 0.1963,
+      "step": 3241
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8340382925654165e-05,
+      "loss": 0.2256,
+      "step": 3242
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.833922253154084e-05,
+      "loss": 0.2285,
+      "step": 3243
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.833806176863487e-05,
+      "loss": 0.228,
+      "step": 3244
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8336900636987588e-05,
+      "loss": 0.2164,
+      "step": 3245
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.833573913665034e-05,
+      "loss": 0.1994,
+      "step": 3246
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8334577267674496e-05,
+      "loss": 0.2484,
+      "step": 3247
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.833341503011144e-05,
+      "loss": 0.1795,
+      "step": 3248
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8332252424012567e-05,
+      "loss": 0.2256,
+      "step": 3249
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.833108944942929e-05,
+      "loss": 0.2171,
+      "step": 3250
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8329926106413044e-05,
+      "loss": 0.2408,
+      "step": 3251
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8328762395015278e-05,
+      "loss": 0.187,
+      "step": 3252
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8327598315287454e-05,
+      "loss": 0.2235,
+      "step": 3253
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.832643386728105e-05,
+      "loss": 0.2403,
+      "step": 3254
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8325269051047566e-05,
+      "loss": 0.2333,
+      "step": 3255
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8324103866638512e-05,
+      "loss": 0.2462,
+      "step": 3256
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8322938314105415e-05,
+      "loss": 0.2337,
+      "step": 3257
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8321772393499826e-05,
+      "loss": 0.2159,
+      "step": 3258
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8320606104873303e-05,
+      "loss": 0.2204,
+      "step": 3259
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.831943944827742e-05,
+      "loss": 0.2606,
+      "step": 3260
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8318272423763784e-05,
+      "loss": 0.1969,
+      "step": 3261
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8317105031383992e-05,
+      "loss": 0.2549,
+      "step": 3262
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8315937271189677e-05,
+      "loss": 0.237,
+      "step": 3263
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8314769143232477e-05,
+      "loss": 0.2262,
+      "step": 3264
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8313600647564054e-05,
+      "loss": 0.2323,
+      "step": 3265
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8312431784236085e-05,
+      "loss": 0.2036,
+      "step": 3266
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8311262553300263e-05,
+      "loss": 0.2377,
+      "step": 3267
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8310092954808292e-05,
+      "loss": 0.2573,
+      "step": 3268
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8308922988811894e-05,
+      "loss": 0.2401,
+      "step": 3269
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8307752655362815e-05,
+      "loss": 0.2446,
+      "step": 3270
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8306581954512812e-05,
+      "loss": 0.2076,
+      "step": 3271
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8305410886313653e-05,
+      "loss": 0.1947,
+      "step": 3272
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8304239450817128e-05,
+      "loss": 0.247,
+      "step": 3273
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.830306764807505e-05,
+      "loss": 0.241,
+      "step": 3274
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8301895478139228e-05,
+      "loss": 0.2066,
+      "step": 3275
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8300722941061505e-05,
+      "loss": 0.2505,
+      "step": 3276
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8299550036893743e-05,
+      "loss": 0.2081,
+      "step": 3277
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8298376765687805e-05,
+      "loss": 0.2251,
+      "step": 3278
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8297203127495575e-05,
+      "loss": 0.1879,
+      "step": 3279
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.829602912236896e-05,
+      "loss": 0.2511,
+      "step": 3280
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8294854750359882e-05,
+      "loss": 0.2446,
+      "step": 3281
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.829368001152027e-05,
+      "loss": 0.2378,
+      "step": 3282
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.829250490590208e-05,
+      "loss": 0.2169,
+      "step": 3283
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8291329433557275e-05,
+      "loss": 0.27,
+      "step": 3284
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.829015359453784e-05,
+      "loss": 0.2308,
+      "step": 3285
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.828897738889578e-05,
+      "loss": 0.2286,
+      "step": 3286
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.828780081668311e-05,
+      "loss": 0.2566,
+      "step": 3287
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.828662387795186e-05,
+      "loss": 0.2588,
+      "step": 3288
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.828544657275408e-05,
+      "loss": 0.2514,
+      "step": 3289
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.828426890114183e-05,
+      "loss": 0.2587,
+      "step": 3290
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8283090863167204e-05,
+      "loss": 0.2538,
+      "step": 3291
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8281912458882286e-05,
+      "loss": 0.2013,
+      "step": 3292
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.82807336883392e-05,
+      "loss": 0.2419,
+      "step": 3293
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8279554551590065e-05,
+      "loss": 0.2571,
+      "step": 3294
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8278375048687033e-05,
+      "loss": 0.2256,
+      "step": 3295
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8277195179682266e-05,
+      "loss": 0.2203,
+      "step": 3296
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8276014944627945e-05,
+      "loss": 0.251,
+      "step": 3297
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8274834343576258e-05,
+      "loss": 0.2171,
+      "step": 3298
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8273653376579423e-05,
+      "loss": 0.249,
+      "step": 3299
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.827247204368966e-05,
+      "loss": 0.231,
+      "step": 3300
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8271290344959214e-05,
+      "loss": 0.2423,
+      "step": 3301
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8270108280440352e-05,
+      "loss": 0.256,
+      "step": 3302
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.826892585018534e-05,
+      "loss": 0.2479,
+      "step": 3303
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8267743054246472e-05,
+      "loss": 0.2042,
+      "step": 3304
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8266559892676056e-05,
+      "loss": 0.2201,
+      "step": 3305
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.826537636552642e-05,
+      "loss": 0.2151,
+      "step": 3306
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8264192472849897e-05,
+      "loss": 0.2515,
+      "step": 3307
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.826300821469885e-05,
+      "loss": 0.2097,
+      "step": 3308
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8261823591125647e-05,
+      "loss": 0.261,
+      "step": 3309
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8260638602182678e-05,
+      "loss": 0.2852,
+      "step": 3310
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.825945324792235e-05,
+      "loss": 0.2374,
+      "step": 3311
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.825826752839708e-05,
+      "loss": 0.2113,
+      "step": 3312
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8257081443659307e-05,
+      "loss": 0.2117,
+      "step": 3313
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8255894993761486e-05,
+      "loss": 0.2221,
+      "step": 3314
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8254708178756083e-05,
+      "loss": 0.2068,
+      "step": 3315
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8253520998695588e-05,
+      "loss": 0.245,
+      "step": 3316
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8252333453632495e-05,
+      "loss": 0.2109,
+      "step": 3317
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.825114554361933e-05,
+      "loss": 0.2734,
+      "step": 3318
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.824995726870862e-05,
+      "loss": 0.2056,
+      "step": 3319
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8248768628952924e-05,
+      "loss": 0.222,
+      "step": 3320
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.82475796244048e-05,
+      "loss": 0.2696,
+      "step": 3321
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8246390255116838e-05,
+      "loss": 0.2174,
+      "step": 3322
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8245200521141625e-05,
+      "loss": 0.2271,
+      "step": 3323
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.824401042253179e-05,
+      "loss": 0.2305,
+      "step": 3324
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.824281995933995e-05,
+      "loss": 0.2444,
+      "step": 3325
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8241629131618763e-05,
+      "loss": 0.2896,
+      "step": 3326
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8240437939420884e-05,
+      "loss": 0.2314,
+      "step": 3327
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8239246382798996e-05,
+      "loss": 0.2341,
+      "step": 3328
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8238054461805793e-05,
+      "loss": 0.2153,
+      "step": 3329
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.823686217649399e-05,
+      "loss": 0.2616,
+      "step": 3330
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8235669526916307e-05,
+      "loss": 0.2297,
+      "step": 3331
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8234476513125496e-05,
+      "loss": 0.2575,
+      "step": 3332
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.823328313517431e-05,
+      "loss": 0.2886,
+      "step": 3333
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8232089393115527e-05,
+      "loss": 0.2337,
+      "step": 3334
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.823089528700194e-05,
+      "loss": 0.1992,
+      "step": 3335
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8229700816886356e-05,
+      "loss": 0.2396,
+      "step": 3336
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8228505982821602e-05,
+      "loss": 0.2555,
+      "step": 3337
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8227310784860514e-05,
+      "loss": 0.269,
+      "step": 3338
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.822611522305595e-05,
+      "loss": 0.2255,
+      "step": 3339
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8224919297460784e-05,
+      "loss": 0.2165,
+      "step": 3340
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.82237230081279e-05,
+      "loss": 0.2796,
+      "step": 3341
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8222526355110206e-05,
+      "loss": 0.2263,
+      "step": 3342
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8221329338460624e-05,
+      "loss": 0.2371,
+      "step": 3343
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8220131958232087e-05,
+      "loss": 0.2413,
+      "step": 3344
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8218934214477544e-05,
+      "loss": 0.2597,
+      "step": 3345
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8217736107249975e-05,
+      "loss": 0.233,
+      "step": 3346
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.821653763660236e-05,
+      "loss": 0.2742,
+      "step": 3347
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8215338802587698e-05,
+      "loss": 0.2535,
+      "step": 3348
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8214139605259006e-05,
+      "loss": 0.2454,
+      "step": 3349
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8212940044669315e-05,
+      "loss": 0.2165,
+      "step": 3350
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8211740120871683e-05,
+      "loss": 0.2212,
+      "step": 3351
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8210539833919164e-05,
+      "loss": 0.2313,
+      "step": 3352
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.820933918386485e-05,
+      "loss": 0.2578,
+      "step": 3353
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8208138170761832e-05,
+      "loss": 0.185,
+      "step": 3354
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.820693679466323e-05,
+      "loss": 0.2179,
+      "step": 3355
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.820573505562216e-05,
+      "loss": 0.2507,
+      "step": 3356
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.820453295369178e-05,
+      "loss": 0.2239,
+      "step": 3357
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8203330488925247e-05,
+      "loss": 0.2218,
+      "step": 3358
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8202127661375743e-05,
+      "loss": 0.2496,
+      "step": 3359
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.820092447109645e-05,
+      "loss": 0.2341,
+      "step": 3360
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.819972091814059e-05,
+      "loss": 0.2348,
+      "step": 3361
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8198517002561385e-05,
+      "loss": 0.2348,
+      "step": 3362
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8197312724412076e-05,
+      "loss": 0.2119,
+      "step": 3363
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8196108083745923e-05,
+      "loss": 0.2098,
+      "step": 3364
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8194903080616194e-05,
+      "loss": 0.257,
+      "step": 3365
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8193697715076185e-05,
+      "loss": 0.2287,
+      "step": 3366
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8192491987179198e-05,
+      "loss": 0.2161,
+      "step": 3367
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8191285896978555e-05,
+      "loss": 0.2423,
+      "step": 3368
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8190079444527603e-05,
+      "loss": 0.2501,
+      "step": 3369
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.818887262987968e-05,
+      "loss": 0.2419,
+      "step": 3370
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.818766545308817e-05,
+      "loss": 0.2218,
+      "step": 3371
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8186457914206453e-05,
+      "loss": 0.2027,
+      "step": 3372
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.818525001328793e-05,
+      "loss": 0.2311,
+      "step": 3373
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.818404175038602e-05,
+      "loss": 0.2133,
+      "step": 3374
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.818283312555416e-05,
+      "loss": 0.2798,
+      "step": 3375
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8181624138845798e-05,
+      "loss": 0.2402,
+      "step": 3376
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.81804147903144e-05,
+      "loss": 0.223,
+      "step": 3377
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8179205080013447e-05,
+      "loss": 0.2224,
+      "step": 3378
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8177995007996436e-05,
+      "loss": 0.2408,
+      "step": 3379
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8176784574316885e-05,
+      "loss": 0.2606,
+      "step": 3380
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8175573779028323e-05,
+      "loss": 0.2208,
+      "step": 3381
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8174362622184297e-05,
+      "loss": 0.2553,
+      "step": 3382
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8173151103838362e-05,
+      "loss": 0.2132,
+      "step": 3383
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8171939224044105e-05,
+      "loss": 0.2362,
+      "step": 3384
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8170726982855114e-05,
+      "loss": 0.2204,
+      "step": 3385
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8169514380325e-05,
+      "loss": 0.2305,
+      "step": 3386
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8168301416507398e-05,
+      "loss": 0.2319,
+      "step": 3387
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8167088091455934e-05,
+      "loss": 0.2239,
+      "step": 3388
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8165874405224276e-05,
+      "loss": 0.1819,
+      "step": 3389
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8164660357866097e-05,
+      "loss": 0.2301,
+      "step": 3390
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8163445949435086e-05,
+      "loss": 0.2534,
+      "step": 3391
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.816223117998495e-05,
+      "loss": 0.2355,
+      "step": 3392
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8161016049569406e-05,
+      "loss": 0.221,
+      "step": 3393
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.81598005582422e-05,
+      "loss": 0.2444,
+      "step": 3394
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8158584706057075e-05,
+      "loss": 0.2288,
+      "step": 3395
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.815736849306781e-05,
+      "loss": 0.2526,
+      "step": 3396
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8156151919328185e-05,
+      "loss": 0.2177,
+      "step": 3397
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.815493498489201e-05,
+      "loss": 0.2294,
+      "step": 3398
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8153717689813087e-05,
+      "loss": 0.2572,
+      "step": 3399
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8152500034145266e-05,
+      "loss": 0.2402,
+      "step": 3400
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8151282017942388e-05,
+      "loss": 0.232,
+      "step": 3401
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8150063641258317e-05,
+      "loss": 0.2212,
+      "step": 3402
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.814884490414694e-05,
+      "loss": 0.1907,
+      "step": 3403
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.814762580666215e-05,
+      "loss": 0.2661,
+      "step": 3404
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8146406348857864e-05,
+      "loss": 0.254,
+      "step": 3405
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8145186530788004e-05,
+      "loss": 0.2667,
+      "step": 3406
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8143966352506525e-05,
+      "loss": 0.2664,
+      "step": 3407
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8142745814067382e-05,
+      "loss": 0.1945,
+      "step": 3408
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8141524915524548e-05,
+      "loss": 0.2373,
+      "step": 3409
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8140303656932023e-05,
+      "loss": 0.2345,
+      "step": 3410
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8139082038343817e-05,
+      "loss": 0.2321,
+      "step": 3411
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8137860059813945e-05,
+      "loss": 0.2031,
+      "step": 3412
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8136637721396456e-05,
+      "loss": 0.2362,
+      "step": 3413
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8135415023145406e-05,
+      "loss": 0.2306,
+      "step": 3414
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8134191965114865e-05,
+      "loss": 0.2468,
+      "step": 3415
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8132968547358918e-05,
+      "loss": 0.2464,
+      "step": 3416
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.813174476993168e-05,
+      "loss": 0.2302,
+      "step": 3417
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.813052063288726e-05,
+      "loss": 0.1817,
+      "step": 3418
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8129296136279797e-05,
+      "loss": 0.189,
+      "step": 3419
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8128071280163446e-05,
+      "loss": 0.2452,
+      "step": 3420
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.812684606459237e-05,
+      "loss": 0.2218,
+      "step": 3421
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.812562048962076e-05,
+      "loss": 0.2427,
+      "step": 3422
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8124394555302812e-05,
+      "loss": 0.2332,
+      "step": 3423
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8123168261692742e-05,
+      "loss": 0.2154,
+      "step": 3424
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8121941608844778e-05,
+      "loss": 0.2815,
+      "step": 3425
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8120714596813173e-05,
+      "loss": 0.2533,
+      "step": 3426
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8119487225652185e-05,
+      "loss": 0.2725,
+      "step": 3427
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8118259495416093e-05,
+      "loss": 0.2479,
+      "step": 3428
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8117031406159197e-05,
+      "loss": 0.2476,
+      "step": 3429
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8115802957935807e-05,
+      "loss": 0.2217,
+      "step": 3430
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8114574150800247e-05,
+      "loss": 0.2848,
+      "step": 3431
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8113344984806856e-05,
+      "loss": 0.2372,
+      "step": 3432
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8112115460010004e-05,
+      "loss": 0.2409,
+      "step": 3433
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.811088557646405e-05,
+      "loss": 0.2017,
+      "step": 3434
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8109655334223396e-05,
+      "loss": 0.2511,
+      "step": 3435
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.810842473334245e-05,
+      "loss": 0.2717,
+      "step": 3436
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8107193773875622e-05,
+      "loss": 0.2164,
+      "step": 3437
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.810596245587736e-05,
+      "loss": 0.2195,
+      "step": 3438
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8104730779402108e-05,
+      "loss": 0.2369,
+      "step": 3439
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8103498744504347e-05,
+      "loss": 0.2321,
+      "step": 3440
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8102266351238554e-05,
+      "loss": 0.2276,
+      "step": 3441
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8101033599659237e-05,
+      "loss": 0.2465,
+      "step": 3442
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8099800489820904e-05,
+      "loss": 0.2454,
+      "step": 3443
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8098567021778098e-05,
+      "loss": 0.2409,
+      "step": 3444
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.809733319558536e-05,
+      "loss": 0.2258,
+      "step": 3445
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8096099011297256e-05,
+      "loss": 0.2275,
+      "step": 3446
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8094864468968367e-05,
+      "loss": 0.2204,
+      "step": 3447
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.809362956865329e-05,
+      "loss": 0.2397,
+      "step": 3448
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.809239431040664e-05,
+      "loss": 0.2362,
+      "step": 3449
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.809115869428304e-05,
+      "loss": 0.2287,
+      "step": 3450
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8089922720337135e-05,
+      "loss": 0.204,
+      "step": 3451
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8088686388623585e-05,
+      "loss": 0.2015,
+      "step": 3452
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8087449699197065e-05,
+      "loss": 0.2485,
+      "step": 3453
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8086212652112264e-05,
+      "loss": 0.2223,
+      "step": 3454
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8084975247423896e-05,
+      "loss": 0.2247,
+      "step": 3455
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8083737485186678e-05,
+      "loss": 0.2478,
+      "step": 3456
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.808249936545535e-05,
+      "loss": 0.2565,
+      "step": 3457
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8081260888284668e-05,
+      "loss": 0.2184,
+      "step": 3458
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8080022053729397e-05,
+      "loss": 0.1854,
+      "step": 3459
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.807878286184433e-05,
+      "loss": 0.2055,
+      "step": 3460
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.807754331268426e-05,
+      "loss": 0.222,
+      "step": 3461
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.807630340630402e-05,
+      "loss": 0.2902,
+      "step": 3462
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8075063142758426e-05,
+      "loss": 0.2093,
+      "step": 3463
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8073822522102337e-05,
+      "loss": 0.2145,
+      "step": 3464
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8072581544390618e-05,
+      "loss": 0.2438,
+      "step": 3465
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8071340209678144e-05,
+      "loss": 0.2407,
+      "step": 3466
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8070098518019817e-05,
+      "loss": 0.2458,
+      "step": 3467
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8068856469470547e-05,
+      "loss": 0.2224,
+      "step": 3468
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8067614064085264e-05,
+      "loss": 0.2346,
+      "step": 3469
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.806637130191891e-05,
+      "loss": 0.2332,
+      "step": 3470
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.806512818302645e-05,
+      "loss": 0.2293,
+      "step": 3471
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.806388470746285e-05,
+      "loss": 0.2297,
+      "step": 3472
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.806264087528311e-05,
+      "loss": 0.189,
+      "step": 3473
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8061396686542232e-05,
+      "loss": 0.2512,
+      "step": 3474
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.806015214129524e-05,
+      "loss": 0.2309,
+      "step": 3475
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8058907239597174e-05,
+      "loss": 0.2067,
+      "step": 3476
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.805766198150309e-05,
+      "loss": 0.2355,
+      "step": 3477
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.805641636706805e-05,
+      "loss": 0.2715,
+      "step": 3478
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8055170396347147e-05,
+      "loss": 0.2297,
+      "step": 3479
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8053924069395484e-05,
+      "loss": 0.2256,
+      "step": 3480
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.805267738626817e-05,
+      "loss": 0.2105,
+      "step": 3481
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.805143034702035e-05,
+      "loss": 0.224,
+      "step": 3482
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8050182951707164e-05,
+      "loss": 0.1988,
+      "step": 3483
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.804893520038378e-05,
+      "loss": 0.1951,
+      "step": 3484
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.804768709310537e-05,
+      "loss": 0.2886,
+      "step": 3485
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.804643862992715e-05,
+      "loss": 0.2252,
+      "step": 3486
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.804518981090431e-05,
+      "loss": 0.257,
+      "step": 3487
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.804394063609209e-05,
+      "loss": 0.2507,
+      "step": 3488
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8042691105545734e-05,
+      "loss": 0.2651,
+      "step": 3489
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.804144121932049e-05,
+      "loss": 0.2241,
+      "step": 3490
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8040190977471647e-05,
+      "loss": 0.2031,
+      "step": 3491
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8038940380054485e-05,
+      "loss": 0.2168,
+      "step": 3492
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8037689427124317e-05,
+      "loss": 0.2178,
+      "step": 3493
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8036438118736458e-05,
+      "loss": 0.2098,
+      "step": 3494
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8035186454946253e-05,
+      "loss": 0.2257,
+      "step": 3495
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8033934435809053e-05,
+      "loss": 0.2615,
+      "step": 3496
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8032682061380224e-05,
+      "loss": 0.2058,
+      "step": 3497
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8031429331715155e-05,
+      "loss": 0.2396,
+      "step": 3498
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8030176246869243e-05,
+      "loss": 0.2915,
+      "step": 3499
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8028922806897907e-05,
+      "loss": 0.2215,
+      "step": 3500
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.802766901185658e-05,
+      "loss": 0.1959,
+      "step": 3501
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.80264148618007e-05,
+      "loss": 0.2228,
+      "step": 3502
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8025160356785746e-05,
+      "loss": 0.1934,
+      "step": 3503
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8023905496867185e-05,
+      "loss": 0.2351,
+      "step": 3504
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8022650282100518e-05,
+      "loss": 0.2454,
+      "step": 3505
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8021394712541253e-05,
+      "loss": 0.211,
+      "step": 3506
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.802013878824491e-05,
+      "loss": 0.1984,
+      "step": 3507
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8018882509267045e-05,
+      "loss": 0.2104,
+      "step": 3508
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8017625875663206e-05,
+      "loss": 0.2284,
+      "step": 3509
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8016368887488967e-05,
+      "loss": 0.2515,
+      "step": 3510
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8015111544799916e-05,
+      "loss": 0.2324,
+      "step": 3511
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8013853847651663e-05,
+      "loss": 0.2055,
+      "step": 3512
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.801259579609982e-05,
+      "loss": 0.2172,
+      "step": 3513
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.801133739020003e-05,
+      "loss": 0.2158,
+      "step": 3514
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.801007863000794e-05,
+      "loss": 0.2004,
+      "step": 3515
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8008819515579222e-05,
+      "loss": 0.3128,
+      "step": 3516
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8007560046969557e-05,
+      "loss": 0.2279,
+      "step": 3517
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.800630022423464e-05,
+      "loss": 0.2629,
+      "step": 3518
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8005040047430187e-05,
+      "loss": 0.2501,
+      "step": 3519
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.800377951661193e-05,
+      "loss": 0.2389,
+      "step": 3520
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8002518631835616e-05,
+      "loss": 0.2418,
+      "step": 3521
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8001257393157e-05,
+      "loss": 0.2393,
+      "step": 3522
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7999995800631862e-05,
+      "loss": 0.2566,
+      "step": 3523
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7998733854316e-05,
+      "loss": 0.2421,
+      "step": 3524
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.799747155426521e-05,
+      "loss": 0.2315,
+      "step": 3525
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7996208900535327e-05,
+      "loss": 0.2289,
+      "step": 3526
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7994945893182185e-05,
+      "loss": 0.231,
+      "step": 3527
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.799368253226164e-05,
+      "loss": 0.2083,
+      "step": 3528
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7992418817829565e-05,
+      "loss": 0.2474,
+      "step": 3529
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7991154749941842e-05,
+      "loss": 0.2008,
+      "step": 3530
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.798989032865438e-05,
+      "loss": 0.2496,
+      "step": 3531
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7988625554023086e-05,
+      "loss": 0.2209,
+      "step": 3532
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7987360426103902e-05,
+      "loss": 0.2297,
+      "step": 3533
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7986094944952772e-05,
+      "loss": 0.2286,
+      "step": 3534
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7984829110625664e-05,
+      "loss": 0.2525,
+      "step": 3535
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7983562923178557e-05,
+      "loss": 0.2246,
+      "step": 3536
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7982296382667447e-05,
+      "loss": 0.2455,
+      "step": 3537
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7981029489148343e-05,
+      "loss": 0.2089,
+      "step": 3538
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7979762242677273e-05,
+      "loss": 0.2336,
+      "step": 3539
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7978494643310283e-05,
+      "loss": 0.2205,
+      "step": 3540
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7977226691103424e-05,
+      "loss": 0.2453,
+      "step": 3541
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7975958386112776e-05,
+      "loss": 0.2389,
+      "step": 3542
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7974689728394425e-05,
+      "loss": 0.2427,
+      "step": 3543
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.797342071800448e-05,
+      "loss": 0.2178,
+      "step": 3544
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7972151354999057e-05,
+      "loss": 0.2621,
+      "step": 3545
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7970881639434292e-05,
+      "loss": 0.2271,
+      "step": 3546
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.796961157136634e-05,
+      "loss": 0.2372,
+      "step": 3547
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7968341150851368e-05,
+      "loss": 0.2465,
+      "step": 3548
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7967070377945556e-05,
+      "loss": 0.2687,
+      "step": 3549
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.796579925270511e-05,
+      "loss": 0.222,
+      "step": 3550
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.796452777518623e-05,
+      "loss": 0.2187,
+      "step": 3551
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.796325594544516e-05,
+      "loss": 0.2068,
+      "step": 3552
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7961983763538137e-05,
+      "loss": 0.2289,
+      "step": 3553
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7960711229521425e-05,
+      "loss": 0.2206,
+      "step": 3554
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7959438343451298e-05,
+      "loss": 0.2188,
+      "step": 3555
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.795816510538405e-05,
+      "loss": 0.2246,
+      "step": 3556
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7956891515375987e-05,
+      "loss": 0.1936,
+      "step": 3557
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7955617573483435e-05,
+      "loss": 0.2341,
+      "step": 3558
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7954343279762727e-05,
+      "loss": 0.2548,
+      "step": 3559
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7953068634270226e-05,
+      "loss": 0.2302,
+      "step": 3560
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7951793637062293e-05,
+      "loss": 0.2297,
+      "step": 3561
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7950518288195316e-05,
+      "loss": 0.244,
+      "step": 3562
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.79492425877257e-05,
+      "loss": 0.2264,
+      "step": 3563
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.794796653570986e-05,
+      "loss": 0.2126,
+      "step": 3564
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7946690132204216e-05,
+      "loss": 0.2143,
+      "step": 3565
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7945413377265234e-05,
+      "loss": 0.2336,
+      "step": 3566
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7944136270949365e-05,
+      "loss": 0.2179,
+      "step": 3567
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7942858813313093e-05,
+      "loss": 0.217,
+      "step": 3568
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7941581004412912e-05,
+      "loss": 0.265,
+      "step": 3569
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7940302844305325e-05,
+      "loss": 0.2278,
+      "step": 3570
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7939024333046864e-05,
+      "loss": 0.2188,
+      "step": 3571
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7937745470694073e-05,
+      "loss": 0.2175,
+      "step": 3572
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7936466257303497e-05,
+      "loss": 0.2149,
+      "step": 3573
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7935186692931714e-05,
+      "loss": 0.2372,
+      "step": 3574
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7933906777635312e-05,
+      "loss": 0.2093,
+      "step": 3575
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7932626511470897e-05,
+      "loss": 0.2244,
+      "step": 3576
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.793134589449508e-05,
+      "loss": 0.2407,
+      "step": 3577
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7930064926764495e-05,
+      "loss": 0.1878,
+      "step": 3578
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7928783608335797e-05,
+      "loss": 0.2548,
+      "step": 3579
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.792750193926565e-05,
+      "loss": 0.2194,
+      "step": 3580
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.792621991961073e-05,
+      "loss": 0.2146,
+      "step": 3581
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7924937549427738e-05,
+      "loss": 0.2228,
+      "step": 3582
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.792365482877338e-05,
+      "loss": 0.2335,
+      "step": 3583
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7922371757704385e-05,
+      "loss": 0.2156,
+      "step": 3584
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.79210883362775e-05,
+      "loss": 0.2447,
+      "step": 3585
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7919804564549474e-05,
+      "loss": 0.2047,
+      "step": 3586
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.791852044257709e-05,
+      "loss": 0.2232,
+      "step": 3587
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7917235970417127e-05,
+      "loss": 0.2422,
+      "step": 3588
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.79159511481264e-05,
+      "loss": 0.2169,
+      "step": 3589
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.791466597576172e-05,
+      "loss": 0.1948,
+      "step": 3590
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7913380453379927e-05,
+      "loss": 0.2056,
+      "step": 3591
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7912094581037867e-05,
+      "loss": 0.2544,
+      "step": 3592
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7910808358792416e-05,
+      "loss": 0.205,
+      "step": 3593
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7909521786700443e-05,
+      "loss": 0.2151,
+      "step": 3594
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7908234864818852e-05,
+      "loss": 0.2385,
+      "step": 3595
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7906947593204557e-05,
+      "loss": 0.2398,
+      "step": 3596
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7905659971914486e-05,
+      "loss": 0.2401,
+      "step": 3597
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7904372001005578e-05,
+      "loss": 0.2346,
+      "step": 3598
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7903083680534793e-05,
+      "loss": 0.2286,
+      "step": 3599
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.790179501055911e-05,
+      "loss": 0.2553,
+      "step": 3600
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7900505991135518e-05,
+      "loss": 0.2672,
+      "step": 3601
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.789921662232102e-05,
+      "loss": 0.2218,
+      "step": 3602
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7897926904172632e-05,
+      "loss": 0.213,
+      "step": 3603
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.78966368367474e-05,
+      "loss": 0.213,
+      "step": 3604
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7895346420102372e-05,
+      "loss": 0.2608,
+      "step": 3605
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7894055654294614e-05,
+      "loss": 0.2693,
+      "step": 3606
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7892764539381213e-05,
+      "loss": 0.2333,
+      "step": 3607
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.789147307541926e-05,
+      "loss": 0.2431,
+      "step": 3608
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7890181262465876e-05,
+      "loss": 0.2499,
+      "step": 3609
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.788888910057818e-05,
+      "loss": 0.2316,
+      "step": 3610
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7887596589813324e-05,
+      "loss": 0.2507,
+      "step": 3611
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7886303730228473e-05,
+      "loss": 0.2209,
+      "step": 3612
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.788501052188079e-05,
+      "loss": 0.257,
+      "step": 3613
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.788371696482747e-05,
+      "loss": 0.2159,
+      "step": 3614
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7882423059125725e-05,
+      "loss": 0.2345,
+      "step": 3615
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7881128804832767e-05,
+      "loss": 0.248,
+      "step": 3616
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7879834202005842e-05,
+      "loss": 0.2418,
+      "step": 3617
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7878539250702194e-05,
+      "loss": 0.2097,
+      "step": 3618
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.78772439509791e-05,
+      "loss": 0.2059,
+      "step": 3619
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.787594830289383e-05,
+      "loss": 0.2436,
+      "step": 3620
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7874652306503695e-05,
+      "loss": 0.222,
+      "step": 3621
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7873355961866008e-05,
+      "loss": 0.1921,
+      "step": 3622
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.787205926903809e-05,
+      "loss": 0.2166,
+      "step": 3623
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7870762228077297e-05,
+      "loss": 0.2462,
+      "step": 3624
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.786946483904098e-05,
+      "loss": 0.2367,
+      "step": 3625
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7868167101986513e-05,
+      "loss": 0.2407,
+      "step": 3626
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7866869016971296e-05,
+      "loss": 0.201,
+      "step": 3627
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7865570584052726e-05,
+      "loss": 0.2372,
+      "step": 3628
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7864271803288234e-05,
+      "loss": 0.2182,
+      "step": 3629
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.786297267473525e-05,
+      "loss": 0.2352,
+      "step": 3630
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7861673198451232e-05,
+      "loss": 0.2644,
+      "step": 3631
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.786037337449364e-05,
+      "loss": 0.2087,
+      "step": 3632
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.785907320291997e-05,
+      "loss": 0.2559,
+      "step": 3633
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7857772683787703e-05,
+      "loss": 0.2455,
+      "step": 3634
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.785647181715437e-05,
+      "loss": 0.2562,
+      "step": 3635
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.785517060307749e-05,
+      "loss": 0.1967,
+      "step": 3636
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7853869041614614e-05,
+      "loss": 0.2069,
+      "step": 3637
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7852567132823293e-05,
+      "loss": 0.2743,
+      "step": 3638
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.785126487676111e-05,
+      "loss": 0.2432,
+      "step": 3639
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.784996227348566e-05,
+      "loss": 0.2594,
+      "step": 3640
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.784865932305454e-05,
+      "loss": 0.2532,
+      "step": 3641
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7847356025525373e-05,
+      "loss": 0.2283,
+      "step": 3642
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7846052380955796e-05,
+      "loss": 0.2525,
+      "step": 3643
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7844748389403465e-05,
+      "loss": 0.2309,
+      "step": 3644
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7843444050926043e-05,
+      "loss": 0.2062,
+      "step": 3645
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7842139365581222e-05,
+      "loss": 0.2313,
+      "step": 3646
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.784083433342669e-05,
+      "loss": 0.2218,
+      "step": 3647
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.783952895452016e-05,
+      "loss": 0.2245,
+      "step": 3648
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.783822322891937e-05,
+      "loss": 0.23,
+      "step": 3649
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7836917156682056e-05,
+      "loss": 0.2055,
+      "step": 3650
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.783561073786598e-05,
+      "loss": 0.2057,
+      "step": 3651
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.783430397252892e-05,
+      "loss": 0.2158,
+      "step": 3652
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7832996860728663e-05,
+      "loss": 0.245,
+      "step": 3653
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7831689402523017e-05,
+      "loss": 0.2392,
+      "step": 3654
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.78303815979698e-05,
+      "loss": 0.214,
+      "step": 3655
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.782907344712685e-05,
+      "loss": 0.2052,
+      "step": 3656
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7827764950052015e-05,
+      "loss": 0.211,
+      "step": 3657
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7826456106803166e-05,
+      "loss": 0.1976,
+      "step": 3658
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7825146917438186e-05,
+      "loss": 0.2367,
+      "step": 3659
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7823837382014968e-05,
+      "loss": 0.2209,
+      "step": 3660
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.782252750059143e-05,
+      "loss": 0.2558,
+      "step": 3661
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.782121727322549e-05,
+      "loss": 0.2349,
+      "step": 3662
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7819906699975104e-05,
+      "loss": 0.1998,
+      "step": 3663
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7818595780898226e-05,
+      "loss": 0.2086,
+      "step": 3664
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7817284516052827e-05,
+      "loss": 0.214,
+      "step": 3665
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7815972905496896e-05,
+      "loss": 0.214,
+      "step": 3666
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.781466094928844e-05,
+      "loss": 0.233,
+      "step": 3667
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.781334864748548e-05,
+      "loss": 0.316,
+      "step": 3668
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.781203600014605e-05,
+      "loss": 0.2156,
+      "step": 3669
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7810723007328195e-05,
+      "loss": 0.2448,
+      "step": 3670
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.780940966908999e-05,
+      "loss": 0.2577,
+      "step": 3671
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7808095985489508e-05,
+      "loss": 0.2386,
+      "step": 3672
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.780678195658485e-05,
+      "loss": 0.1952,
+      "step": 3673
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7805467582434123e-05,
+      "loss": 0.2429,
+      "step": 3674
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.780415286309546e-05,
+      "loss": 0.2308,
+      "step": 3675
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7802837798626996e-05,
+      "loss": 0.1948,
+      "step": 3676
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7801522389086897e-05,
+      "loss": 0.2193,
+      "step": 3677
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7800206634533325e-05,
+      "loss": 0.2368,
+      "step": 3678
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7798890535024474e-05,
+      "loss": 0.2718,
+      "step": 3679
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.779757409061855e-05,
+      "loss": 0.2263,
+      "step": 3680
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.779625730137376e-05,
+      "loss": 0.2612,
+      "step": 3681
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7794940167348348e-05,
+      "loss": 0.2348,
+      "step": 3682
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.779362268860056e-05,
+      "loss": 0.2262,
+      "step": 3683
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7792304865188658e-05,
+      "loss": 0.2482,
+      "step": 3684
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7790986697170926e-05,
+      "loss": 0.2347,
+      "step": 3685
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7789668184605653e-05,
+      "loss": 0.2093,
+      "step": 3686
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.778834932755115e-05,
+      "loss": 0.2421,
+      "step": 3687
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7787030126065743e-05,
+      "loss": 0.2266,
+      "step": 3688
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7785710580207773e-05,
+      "loss": 0.2507,
+      "step": 3689
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7784390690035592e-05,
+      "loss": 0.2347,
+      "step": 3690
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7783070455607578e-05,
+      "loss": 0.2387,
+      "step": 3691
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7781749876982108e-05,
+      "loss": 0.2299,
+      "step": 3692
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7780428954217587e-05,
+      "loss": 0.2147,
+      "step": 3693
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7779107687372434e-05,
+      "loss": 0.2334,
+      "step": 3694
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7777786076505076e-05,
+      "loss": 0.2728,
+      "step": 3695
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.777646412167396e-05,
+      "loss": 0.2397,
+      "step": 3696
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.777514182293755e-05,
+      "loss": 0.2684,
+      "step": 3697
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7773819180354322e-05,
+      "loss": 0.2254,
+      "step": 3698
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7772496193982775e-05,
+      "loss": 0.2176,
+      "step": 3699
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7771172863881405e-05,
+      "loss": 0.2506,
+      "step": 3700
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7769849190108743e-05,
+      "loss": 0.245,
+      "step": 3701
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7768525172723323e-05,
+      "loss": 0.2003,
+      "step": 3702
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7767200811783696e-05,
+      "loss": 0.2248,
+      "step": 3703
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7765876107348443e-05,
+      "loss": 0.2228,
+      "step": 3704
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7764551059476128e-05,
+      "loss": 0.2344,
+      "step": 3705
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7763225668225367e-05,
+      "loss": 0.2418,
+      "step": 3706
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7761899933654766e-05,
+      "loss": 0.2499,
+      "step": 3707
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7760573855822954e-05,
+      "loss": 0.2584,
+      "step": 3708
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.775924743478858e-05,
+      "loss": 0.2261,
+      "step": 3709
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7757920670610292e-05,
+      "loss": 0.2277,
+      "step": 3710
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.775659356334678e-05,
+      "loss": 0.2115,
+      "step": 3711
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7755266113056726e-05,
+      "loss": 0.2235,
+      "step": 3712
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.775393831979883e-05,
+      "loss": 0.2379,
+      "step": 3713
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.775261018363182e-05,
+      "loss": 0.2516,
+      "step": 3714
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7751281704614432e-05,
+      "loss": 0.2103,
+      "step": 3715
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7749952882805415e-05,
+      "loss": 0.2335,
+      "step": 3716
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7748623718263527e-05,
+      "loss": 0.2288,
+      "step": 3717
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7747294211047558e-05,
+      "loss": 0.2636,
+      "step": 3718
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.77459643612163e-05,
+      "loss": 0.2261,
+      "step": 3719
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7744634168828563e-05,
+      "loss": 0.237,
+      "step": 3720
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7743303633943176e-05,
+      "loss": 0.2225,
+      "step": 3721
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.774197275661898e-05,
+      "loss": 0.2331,
+      "step": 3722
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7740641536914832e-05,
+      "loss": 0.2342,
+      "step": 3723
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.77393099748896e-05,
+      "loss": 0.2224,
+      "step": 3724
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7737978070602177e-05,
+      "loss": 0.2208,
+      "step": 3725
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.773664582411146e-05,
+      "loss": 0.2023,
+      "step": 3726
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7735313235476367e-05,
+      "loss": 0.2072,
+      "step": 3727
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.773398030475583e-05,
+      "loss": 0.2285,
+      "step": 3728
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7732647032008798e-05,
+      "loss": 0.1928,
+      "step": 3729
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.773131341729423e-05,
+      "loss": 0.2328,
+      "step": 3730
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.772997946067111e-05,
+      "loss": 0.2497,
+      "step": 3731
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7728645162198425e-05,
+      "loss": 0.241,
+      "step": 3732
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7727310521935186e-05,
+      "loss": 0.2116,
+      "step": 3733
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7725975539940415e-05,
+      "loss": 0.213,
+      "step": 3734
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7724640216273145e-05,
+      "loss": 0.2297,
+      "step": 3735
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.772330455099244e-05,
+      "loss": 0.2122,
+      "step": 3736
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7721968544157357e-05,
+      "loss": 0.2199,
+      "step": 3737
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.772063219582699e-05,
+      "loss": 0.2238,
+      "step": 3738
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.771929550606043e-05,
+      "loss": 0.223,
+      "step": 3739
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.771795847491679e-05,
+      "loss": 0.2346,
+      "step": 3740
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7716621102455204e-05,
+      "loss": 0.2509,
+      "step": 3741
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.771528338873481e-05,
+      "loss": 0.2247,
+      "step": 3742
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7713945333814776e-05,
+      "loss": 0.2514,
+      "step": 3743
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7712606937754266e-05,
+      "loss": 0.2351,
+      "step": 3744
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7711268200612473e-05,
+      "loss": 0.2103,
+      "step": 3745
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7709929122448598e-05,
+      "loss": 0.2008,
+      "step": 3746
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7708589703321866e-05,
+      "loss": 0.241,
+      "step": 3747
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7707249943291512e-05,
+      "loss": 0.2518,
+      "step": 3748
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.770590984241678e-05,
+      "loss": 0.2232,
+      "step": 3749
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7704569400756937e-05,
+      "loss": 0.2399,
+      "step": 3750
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7703228618371257e-05,
+      "loss": 0.2298,
+      "step": 3751
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7701887495319045e-05,
+      "loss": 0.248,
+      "step": 3752
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.77005460316596e-05,
+      "loss": 0.1998,
+      "step": 3753
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.769920422745226e-05,
+      "loss": 0.2153,
+      "step": 3754
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.769786208275635e-05,
+      "loss": 0.2136,
+      "step": 3755
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7696519597631232e-05,
+      "loss": 0.2514,
+      "step": 3756
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7695176772136275e-05,
+      "loss": 0.2749,
+      "step": 3757
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7693833606330866e-05,
+      "loss": 0.2159,
+      "step": 3758
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7692490100274403e-05,
+      "loss": 0.229,
+      "step": 3759
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7691146254026302e-05,
+      "loss": 0.2265,
+      "step": 3760
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7689802067645993e-05,
+      "loss": 0.223,
+      "step": 3761
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.768845754119292e-05,
+      "loss": 0.2284,
+      "step": 3762
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7687112674726542e-05,
+      "loss": 0.2433,
+      "step": 3763
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7685767468306335e-05,
+      "loss": 0.2377,
+      "step": 3764
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7684421921991792e-05,
+      "loss": 0.2365,
+      "step": 3765
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7683076035842414e-05,
+      "loss": 0.1925,
+      "step": 3766
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.768172980991773e-05,
+      "loss": 0.2214,
+      "step": 3767
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.768038324427726e-05,
+      "loss": 0.2405,
+      "step": 3768
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.767903633898057e-05,
+      "loss": 0.1922,
+      "step": 3769
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7677689094087212e-05,
+      "loss": 0.205,
+      "step": 3770
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.767634150965678e-05,
+      "loss": 0.2453,
+      "step": 3771
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.767499358574886e-05,
+      "loss": 0.2388,
+      "step": 3772
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.767364532242306e-05,
+      "loss": 0.2141,
+      "step": 3773
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7672296719739016e-05,
+      "loss": 0.2171,
+      "step": 3774
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.767094777775636e-05,
+      "loss": 0.229,
+      "step": 3775
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.766959849653475e-05,
+      "loss": 0.198,
+      "step": 3776
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7668248876133858e-05,
+      "loss": 0.2247,
+      "step": 3777
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.766689891661337e-05,
+      "loss": 0.2881,
+      "step": 3778
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.766554861803298e-05,
+      "loss": 0.2384,
+      "step": 3779
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7664197980452408e-05,
+      "loss": 0.2339,
+      "step": 3780
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7662847003931386e-05,
+      "loss": 0.2167,
+      "step": 3781
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7661495688529657e-05,
+      "loss": 0.2189,
+      "step": 3782
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.766014403430698e-05,
+      "loss": 0.2201,
+      "step": 3783
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7658792041323135e-05,
+      "loss": 0.2825,
+      "step": 3784
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.765743970963791e-05,
+      "loss": 0.2333,
+      "step": 3785
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7656087039311107e-05,
+      "loss": 0.2175,
+      "step": 3786
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7654734030402552e-05,
+      "loss": 0.2148,
+      "step": 3787
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7653380682972076e-05,
+      "loss": 0.2262,
+      "step": 3788
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7652026997079532e-05,
+      "loss": 0.2176,
+      "step": 3789
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.765067297278478e-05,
+      "loss": 0.2144,
+      "step": 3790
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7649318610147708e-05,
+      "loss": 0.2196,
+      "step": 3791
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7647963909228208e-05,
+      "loss": 0.2343,
+      "step": 3792
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.764660887008619e-05,
+      "loss": 0.2033,
+      "step": 3793
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7645253492781576e-05,
+      "loss": 0.2093,
+      "step": 3794
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.764389777737431e-05,
+      "loss": 0.2017,
+      "step": 3795
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7642541723924347e-05,
+      "loss": 0.248,
+      "step": 3796
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7641185332491652e-05,
+      "loss": 0.2127,
+      "step": 3797
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7639828603136217e-05,
+      "loss": 0.201,
+      "step": 3798
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7638471535918035e-05,
+      "loss": 0.211,
+      "step": 3799
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7637114130897127e-05,
+      "loss": 0.1997,
+      "step": 3800
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7635756388133517e-05,
+      "loss": 0.1907,
+      "step": 3801
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7634398307687252e-05,
+      "loss": 0.2271,
+      "step": 3802
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.763303988961839e-05,
+      "loss": 0.2659,
+      "step": 3803
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7631681133987016e-05,
+      "loss": 0.2425,
+      "step": 3804
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7630322040853202e-05,
+      "loss": 0.1988,
+      "step": 3805
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7628962610277063e-05,
+      "loss": 0.2526,
+      "step": 3806
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.762760284231872e-05,
+      "loss": 0.2237,
+      "step": 3807
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.76262427370383e-05,
+      "loss": 0.2349,
+      "step": 3808
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7624882294495956e-05,
+      "loss": 0.2753,
+      "step": 3809
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7623521514751855e-05,
+      "loss": 0.2687,
+      "step": 3810
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7622160397866167e-05,
+      "loss": 0.1917,
+      "step": 3811
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7620798943899097e-05,
+      "loss": 0.2172,
+      "step": 3812
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7619437152910847e-05,
+      "loss": 0.2301,
+      "step": 3813
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.761807502496164e-05,
+      "loss": 0.2041,
+      "step": 3814
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7616712560111717e-05,
+      "loss": 0.26,
+      "step": 3815
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7615349758421332e-05,
+      "loss": 0.1956,
+      "step": 3816
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.761398661995075e-05,
+      "loss": 0.2216,
+      "step": 3817
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7612623144760256e-05,
+      "loss": 0.2532,
+      "step": 3818
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7611259332910152e-05,
+      "loss": 0.1952,
+      "step": 3819
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7609895184460745e-05,
+      "loss": 0.205,
+      "step": 3820
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.760853069947237e-05,
+      "loss": 0.1937,
+      "step": 3821
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7607165878005358e-05,
+      "loss": 0.2213,
+      "step": 3822
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.760580072012008e-05,
+      "loss": 0.215,
+      "step": 3823
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.76044352258769e-05,
+      "loss": 0.2164,
+      "step": 3824
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7603069395336212e-05,
+      "loss": 0.2303,
+      "step": 3825
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.760170322855841e-05,
+      "loss": 0.2419,
+      "step": 3826
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7600336725603917e-05,
+      "loss": 0.2127,
+      "step": 3827
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7598969886533167e-05,
+      "loss": 0.2252,
+      "step": 3828
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.75976027114066e-05,
+      "loss": 0.1974,
+      "step": 3829
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.759623520028468e-05,
+      "loss": 0.2152,
+      "step": 3830
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7594867353227884e-05,
+      "loss": 0.2432,
+      "step": 3831
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.759349917029671e-05,
+      "loss": 0.2538,
+      "step": 3832
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7592130651551654e-05,
+      "loss": 0.2321,
+      "step": 3833
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7590761797053244e-05,
+      "loss": 0.2035,
+      "step": 3834
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7589392606862013e-05,
+      "loss": 0.2441,
+      "step": 3835
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7588023081038513e-05,
+      "loss": 0.2453,
+      "step": 3836
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.758665321964331e-05,
+      "loss": 0.2579,
+      "step": 3837
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7585283022736983e-05,
+      "loss": 0.2698,
+      "step": 3838
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7583912490380132e-05,
+      "loss": 0.2073,
+      "step": 3839
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.758254162263336e-05,
+      "loss": 0.2005,
+      "step": 3840
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7581170419557298e-05,
+      "loss": 0.2131,
+      "step": 3841
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7579798881212582e-05,
+      "loss": 0.2487,
+      "step": 3842
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.757842700765987e-05,
+      "loss": 0.2156,
+      "step": 3843
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.757705479895983e-05,
+      "loss": 0.2827,
+      "step": 3844
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7575682255173144e-05,
+      "loss": 0.2175,
+      "step": 3845
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7574309376360513e-05,
+      "loss": 0.2448,
+      "step": 3846
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7572936162582652e-05,
+      "loss": 0.2167,
+      "step": 3847
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7571562613900294e-05,
+      "loss": 0.2884,
+      "step": 3848
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7570188730374173e-05,
+      "loss": 0.2339,
+      "step": 3849
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7568814512065056e-05,
+      "loss": 0.2369,
+      "step": 3850
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7567439959033712e-05,
+      "loss": 0.1973,
+      "step": 3851
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.756606507134093e-05,
+      "loss": 0.1965,
+      "step": 3852
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7564689849047508e-05,
+      "loss": 0.2172,
+      "step": 3853
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7563314292214275e-05,
+      "loss": 0.2177,
+      "step": 3854
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7561938400902052e-05,
+      "loss": 0.2339,
+      "step": 3855
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7560562175171694e-05,
+      "loss": 0.2353,
+      "step": 3856
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7559185615084056e-05,
+      "loss": 0.2229,
+      "step": 3857
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.755780872070002e-05,
+      "loss": 0.2261,
+      "step": 3858
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.755643149208048e-05,
+      "loss": 0.2278,
+      "step": 3859
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7555053929286333e-05,
+      "loss": 0.2141,
+      "step": 3860
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.755367603237851e-05,
+      "loss": 0.2216,
+      "step": 3861
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7552297801417943e-05,
+      "loss": 0.2543,
+      "step": 3862
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.755091923646558e-05,
+      "loss": 0.2533,
+      "step": 3863
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.754954033758239e-05,
+      "loss": 0.2058,
+      "step": 3864
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7548161104829352e-05,
+      "loss": 0.2317,
+      "step": 3865
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7546781538267463e-05,
+      "loss": 0.2266,
+      "step": 3866
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.754540163795773e-05,
+      "loss": 0.2384,
+      "step": 3867
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7544021403961177e-05,
+      "loss": 0.2269,
+      "step": 3868
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7542640836338848e-05,
+      "loss": 0.215,
+      "step": 3869
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.754125993515179e-05,
+      "loss": 0.2296,
+      "step": 3870
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7539878700461076e-05,
+      "loss": 0.2497,
+      "step": 3871
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.753849713232779e-05,
+      "loss": 0.2522,
+      "step": 3872
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.753711523081303e-05,
+      "loss": 0.2203,
+      "step": 3873
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.753573299597791e-05,
+      "loss": 0.2235,
+      "step": 3874
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7534350427883555e-05,
+      "loss": 0.2421,
+      "step": 3875
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7532967526591106e-05,
+      "loss": 0.2066,
+      "step": 3876
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7531584292161724e-05,
+      "loss": 0.2185,
+      "step": 3877
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7530200724656582e-05,
+      "loss": 0.2215,
+      "step": 3878
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7528816824136862e-05,
+      "loss": 0.2164,
+      "step": 3879
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7527432590663767e-05,
+      "loss": 0.2172,
+      "step": 3880
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7526048024298518e-05,
+      "loss": 0.2759,
+      "step": 3881
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.752466312510234e-05,
+      "loss": 0.229,
+      "step": 3882
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7523277893136482e-05,
+      "loss": 0.221,
+      "step": 3883
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7521892328462203e-05,
+      "loss": 0.1979,
+      "step": 3884
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.752050643114078e-05,
+      "loss": 0.206,
+      "step": 3885
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7519120201233494e-05,
+      "loss": 0.2005,
+      "step": 3886
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.751773363880166e-05,
+      "loss": 0.2371,
+      "step": 3887
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7516346743906597e-05,
+      "loss": 0.2452,
+      "step": 3888
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.751495951660963e-05,
+      "loss": 0.2447,
+      "step": 3889
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7513571956972113e-05,
+      "loss": 0.2627,
+      "step": 3890
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.751218406505541e-05,
+      "loss": 0.2276,
+      "step": 3891
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7510795840920898e-05,
+      "loss": 0.2247,
+      "step": 3892
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.750940728462997e-05,
+      "loss": 0.2937,
+      "step": 3893
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.750801839624403e-05,
+      "loss": 0.2264,
+      "step": 3894
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.750662917582451e-05,
+      "loss": 0.2061,
+      "step": 3895
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7505239623432834e-05,
+      "loss": 0.2153,
+      "step": 3896
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7503849739130458e-05,
+      "loss": 0.2427,
+      "step": 3897
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.750245952297885e-05,
+      "loss": 0.2274,
+      "step": 3898
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7501068975039494e-05,
+      "loss": 0.2119,
+      "step": 3899
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7499678095373875e-05,
+      "loss": 0.2141,
+      "step": 3900
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7498286884043513e-05,
+      "loss": 0.2344,
+      "step": 3901
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7496895341109927e-05,
+      "loss": 0.2489,
+      "step": 3902
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.749550346663466e-05,
+      "loss": 0.2453,
+      "step": 3903
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7494111260679264e-05,
+      "loss": 0.1823,
+      "step": 3904
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.749271872330531e-05,
+      "loss": 0.2103,
+      "step": 3905
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7491325854574378e-05,
+      "loss": 0.2229,
+      "step": 3906
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7489932654548062e-05,
+      "loss": 0.2322,
+      "step": 3907
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7488539123287987e-05,
+      "loss": 0.2123,
+      "step": 3908
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.748714526085577e-05,
+      "loss": 0.2103,
+      "step": 3909
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.748575106731306e-05,
+      "loss": 0.2205,
+      "step": 3910
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.748435654272151e-05,
+      "loss": 0.2175,
+      "step": 3911
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.748296168714279e-05,
+      "loss": 0.1983,
+      "step": 3912
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7481566500638585e-05,
+      "loss": 0.2432,
+      "step": 3913
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7480170983270604e-05,
+      "loss": 0.2181,
+      "step": 3914
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7478775135100553e-05,
+      "loss": 0.226,
+      "step": 3915
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7477378956190164e-05,
+      "loss": 0.24,
+      "step": 3916
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7475982446601182e-05,
+      "loss": 0.2429,
+      "step": 3917
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7474585606395365e-05,
+      "loss": 0.2011,
+      "step": 3918
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7473188435634494e-05,
+      "loss": 0.2199,
+      "step": 3919
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7471790934380347e-05,
+      "loss": 0.2283,
+      "step": 3920
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.747039310269473e-05,
+      "loss": 0.2315,
+      "step": 3921
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7468994940639465e-05,
+      "loss": 0.2457,
+      "step": 3922
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.746759644827638e-05,
+      "loss": 0.1872,
+      "step": 3923
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.746619762566732e-05,
+      "loss": 0.2142,
+      "step": 3924
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7464798472874155e-05,
+      "loss": 0.1799,
+      "step": 3925
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.746339898995875e-05,
+      "loss": 0.2333,
+      "step": 3926
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7461999176983e-05,
+      "loss": 0.2111,
+      "step": 3927
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.746059903400881e-05,
+      "loss": 0.2637,
+      "step": 3928
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7459198561098106e-05,
+      "loss": 0.1968,
+      "step": 3929
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7457797758312815e-05,
+      "loss": 0.2448,
+      "step": 3930
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7456396625714883e-05,
+      "loss": 0.2139,
+      "step": 3931
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.745499516336628e-05,
+      "loss": 0.2108,
+      "step": 3932
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7453593371328984e-05,
+      "loss": 0.2208,
+      "step": 3933
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.745219124966498e-05,
+      "loss": 0.2376,
+      "step": 3934
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7450788798436288e-05,
+      "loss": 0.207,
+      "step": 3935
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.744938601770492e-05,
+      "loss": 0.2374,
+      "step": 3936
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7447982907532916e-05,
+      "loss": 0.2606,
+      "step": 3937
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7446579467982328e-05,
+      "loss": 0.2258,
+      "step": 3938
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7445175699115216e-05,
+      "loss": 0.2242,
+      "step": 3939
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7443771600993665e-05,
+      "loss": 0.258,
+      "step": 3940
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.744236717367977e-05,
+      "loss": 0.2098,
+      "step": 3941
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7440962417235637e-05,
+      "loss": 0.2496,
+      "step": 3942
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7439557331723395e-05,
+      "loss": 0.2383,
+      "step": 3943
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7438151917205176e-05,
+      "loss": 0.2382,
+      "step": 3944
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.743674617374313e-05,
+      "loss": 0.2111,
+      "step": 3945
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7435340101399437e-05,
+      "loss": 0.237,
+      "step": 3946
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7433933700236272e-05,
+      "loss": 0.256,
+      "step": 3947
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.743252697031583e-05,
+      "loss": 0.2235,
+      "step": 3948
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7431119911700323e-05,
+      "loss": 0.2187,
+      "step": 3949
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7429712524451975e-05,
+      "loss": 0.2201,
+      "step": 3950
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7428304808633032e-05,
+      "loss": 0.2496,
+      "step": 3951
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7426896764305743e-05,
+      "loss": 0.2285,
+      "step": 3952
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.742548839153238e-05,
+      "loss": 0.2212,
+      "step": 3953
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7424079690375224e-05,
+      "loss": 0.1898,
+      "step": 3954
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7422670660896576e-05,
+      "loss": 0.2378,
+      "step": 3955
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7421261303158745e-05,
+      "loss": 0.2161,
+      "step": 3956
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7419851617224064e-05,
+      "loss": 0.2185,
+      "step": 3957
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.741844160315487e-05,
+      "loss": 0.2325,
+      "step": 3958
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.741703126101352e-05,
+      "loss": 0.2475,
+      "step": 3959
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.741562059086239e-05,
+      "loss": 0.2401,
+      "step": 3960
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7414209592763856e-05,
+      "loss": 0.2018,
+      "step": 3961
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7412798266780323e-05,
+      "loss": 0.2027,
+      "step": 3962
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7411386612974206e-05,
+      "loss": 0.1881,
+      "step": 3963
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7409974631407934e-05,
+      "loss": 0.2131,
+      "step": 3964
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7408562322143947e-05,
+      "loss": 0.2154,
+      "step": 3965
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7407149685244706e-05,
+      "loss": 0.2256,
+      "step": 3966
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.740573672077268e-05,
+      "loss": 0.2442,
+      "step": 3967
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7404323428790363e-05,
+      "loss": 0.2609,
+      "step": 3968
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7402909809360246e-05,
+      "loss": 0.2107,
+      "step": 3969
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7401495862544854e-05,
+      "loss": 0.2526,
+      "step": 3970
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7400081588406708e-05,
+      "loss": 0.2367,
+      "step": 3971
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.739866698700836e-05,
+      "loss": 0.1969,
+      "step": 3972
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7397252058412367e-05,
+      "loss": 0.224,
+      "step": 3973
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.73958368026813e-05,
+      "loss": 0.1999,
+      "step": 3974
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7394421219877754e-05,
+      "loss": 0.2436,
+      "step": 3975
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7393005310064323e-05,
+      "loss": 0.2025,
+      "step": 3976
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7391589073303632e-05,
+      "loss": 0.2482,
+      "step": 3977
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7390172509658306e-05,
+      "loss": 0.2435,
+      "step": 3978
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7388755619190997e-05,
+      "loss": 0.2185,
+      "step": 3979
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7387338401964357e-05,
+      "loss": 0.2455,
+      "step": 3980
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7385920858041066e-05,
+      "loss": 0.237,
+      "step": 3981
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7384502987483812e-05,
+      "loss": 0.2353,
+      "step": 3982
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7383084790355305e-05,
+      "loss": 0.2459,
+      "step": 3983
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7381666266718254e-05,
+      "loss": 0.2248,
+      "step": 3984
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7380247416635393e-05,
+      "loss": 0.2136,
+      "step": 3985
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7378828240169474e-05,
+      "loss": 0.2425,
+      "step": 3986
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7377408737383254e-05,
+      "loss": 0.2175,
+      "step": 3987
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7375988908339513e-05,
+      "loss": 0.2116,
+      "step": 3988
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.737456875310104e-05,
+      "loss": 0.2136,
+      "step": 3989
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7373148271730635e-05,
+      "loss": 0.2189,
+      "step": 3990
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7371727464291127e-05,
+      "loss": 0.2761,
+      "step": 3991
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7370306330845338e-05,
+      "loss": 0.2536,
+      "step": 3992
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7368884871456124e-05,
+      "loss": 0.2327,
+      "step": 3993
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7367463086186346e-05,
+      "loss": 0.2516,
+      "step": 3994
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7366040975098876e-05,
+      "loss": 0.2466,
+      "step": 3995
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7364618538256613e-05,
+      "loss": 0.2408,
+      "step": 3996
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7363195775722456e-05,
+      "loss": 0.2036,
+      "step": 3997
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7361772687559327e-05,
+      "loss": 0.2248,
+      "step": 3998
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7360349273830164e-05,
+      "loss": 0.2428,
+      "step": 3999
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.735892553459791e-05,
+      "loss": 0.2285,
+      "step": 4000
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7357501469925535e-05,
+      "loss": 0.23,
+      "step": 4001
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7356077079876012e-05,
+      "loss": 0.2305,
+      "step": 4002
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7354652364512335e-05,
+      "loss": 0.214,
+      "step": 4003
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7353227323897507e-05,
+      "loss": 0.197,
+      "step": 4004
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.735180195809455e-05,
+      "loss": 0.204,
+      "step": 4005
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7350376267166504e-05,
+      "loss": 0.2467,
+      "step": 4006
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7348950251176415e-05,
+      "loss": 0.2262,
+      "step": 4007
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7347523910187343e-05,
+      "loss": 0.2649,
+      "step": 4008
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7346097244262376e-05,
+      "loss": 0.2428,
+      "step": 4009
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7344670253464596e-05,
+      "loss": 0.2425,
+      "step": 4010
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7343242937857117e-05,
+      "loss": 0.2371,
+      "step": 4011
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.734181529750306e-05,
+      "loss": 0.2432,
+      "step": 4012
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7340387332465557e-05,
+      "loss": 0.2063,
+      "step": 4013
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.733895904280776e-05,
+      "loss": 0.2208,
+      "step": 4014
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7337530428592834e-05,
+      "loss": 0.2235,
+      "step": 4015
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7336101489883962e-05,
+      "loss": 0.2081,
+      "step": 4016
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7334672226744325e-05,
+      "loss": 0.2057,
+      "step": 4017
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7333242639237148e-05,
+      "loss": 0.2119,
+      "step": 4018
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7331812727425636e-05,
+      "loss": 0.213,
+      "step": 4019
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7330382491373035e-05,
+      "loss": 0.2277,
+      "step": 4020
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.732895193114259e-05,
+      "loss": 0.2319,
+      "step": 4021
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.732752104679757e-05,
+      "loss": 0.2447,
+      "step": 4022
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7326089838401258e-05,
+      "loss": 0.2498,
+      "step": 4023
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7324658306016936e-05,
+      "loss": 0.2442,
+      "step": 4024
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7323226449707923e-05,
+      "loss": 0.2619,
+      "step": 4025
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7321794269537537e-05,
+      "loss": 0.213,
+      "step": 4026
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7320361765569114e-05,
+      "loss": 0.2506,
+      "step": 4027
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7318928937866e-05,
+      "loss": 0.2438,
+      "step": 4028
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7317495786491573e-05,
+      "loss": 0.196,
+      "step": 4029
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7316062311509202e-05,
+      "loss": 0.2294,
+      "step": 4030
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.731462851298228e-05,
+      "loss": 0.224,
+      "step": 4031
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7313194390974226e-05,
+      "loss": 0.2454,
+      "step": 4032
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.731175994554845e-05,
+      "loss": 0.2009,
+      "step": 4033
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7310325176768398e-05,
+      "loss": 0.2331,
+      "step": 4034
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.730889008469751e-05,
+      "loss": 0.2225,
+      "step": 4035
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7307454669399266e-05,
+      "loss": 0.2345,
+      "step": 4036
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7306018930937133e-05,
+      "loss": 0.2314,
+      "step": 4037
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7304582869374615e-05,
+      "loss": 0.2055,
+      "step": 4038
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.730314648477521e-05,
+      "loss": 0.2159,
+      "step": 4039
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7301709777202445e-05,
+      "loss": 0.23,
+      "step": 4040
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7300272746719863e-05,
+      "loss": 0.2226,
+      "step": 4041
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7298835393391e-05,
+      "loss": 0.2591,
+      "step": 4042
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.729739771727944e-05,
+      "loss": 0.2358,
+      "step": 4043
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7295959718448748e-05,
+      "loss": 0.2313,
+      "step": 4044
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7294521396962525e-05,
+      "loss": 0.2611,
+      "step": 4045
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7293082752884378e-05,
+      "loss": 0.2435,
+      "step": 4046
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7291643786277927e-05,
+      "loss": 0.2227,
+      "step": 4047
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7290204497206812e-05,
+      "loss": 0.2481,
+      "step": 4048
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.728876488573468e-05,
+      "loss": 0.2139,
+      "step": 4049
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.72873249519252e-05,
+      "loss": 0.2613,
+      "step": 4050
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7285884695842048e-05,
+      "loss": 0.2325,
+      "step": 4051
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7284444117548922e-05,
+      "loss": 0.2563,
+      "step": 4052
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7283003217109525e-05,
+      "loss": 0.2282,
+      "step": 4053
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7281561994587585e-05,
+      "loss": 0.207,
+      "step": 4054
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7280120450046832e-05,
+      "loss": 0.2326,
+      "step": 4055
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.727867858355102e-05,
+      "loss": 0.189,
+      "step": 4056
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7277236395163916e-05,
+      "loss": 0.2301,
+      "step": 4057
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7275793884949295e-05,
+      "loss": 0.2452,
+      "step": 4058
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.727435105297095e-05,
+      "loss": 0.2153,
+      "step": 4059
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7272907899292696e-05,
+      "loss": 0.2067,
+      "step": 4060
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7271464423978344e-05,
+      "loss": 0.2364,
+      "step": 4061
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.727002062709174e-05,
+      "loss": 0.2654,
+      "step": 4062
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.726857650869673e-05,
+      "loss": 0.224,
+      "step": 4063
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7267132068857172e-05,
+      "loss": 0.1915,
+      "step": 4064
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7265687307636956e-05,
+      "loss": 0.2311,
+      "step": 4065
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7264242225099973e-05,
+      "loss": 0.2291,
+      "step": 4066
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.726279682131012e-05,
+      "loss": 0.3077,
+      "step": 4067
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7261351096331332e-05,
+      "loss": 0.223,
+      "step": 4068
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7259905050227537e-05,
+      "loss": 0.1989,
+      "step": 4069
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7258458683062686e-05,
+      "loss": 0.2,
+      "step": 4070
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.725701199490074e-05,
+      "loss": 0.2235,
+      "step": 4071
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.725556498580568e-05,
+      "loss": 0.2482,
+      "step": 4072
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7254117655841502e-05,
+      "loss": 0.2243,
+      "step": 4073
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7252670005072202e-05,
+      "loss": 0.1978,
+      "step": 4074
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7251222033561816e-05,
+      "loss": 0.2659,
+      "step": 4075
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7249773741374364e-05,
+      "loss": 0.2038,
+      "step": 4076
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7248325128573907e-05,
+      "loss": 0.1799,
+      "step": 4077
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.72468761952245e-05,
+      "loss": 0.2274,
+      "step": 4078
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.724542694139022e-05,
+      "loss": 0.217,
+      "step": 4079
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7243977367135162e-05,
+      "loss": 0.2225,
+      "step": 4080
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7242527472523435e-05,
+      "loss": 0.2377,
+      "step": 4081
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7241077257619156e-05,
+      "loss": 0.2175,
+      "step": 4082
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7239626722486457e-05,
+      "loss": 0.2058,
+      "step": 4083
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7238175867189487e-05,
+      "loss": 0.1979,
+      "step": 4084
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.723672469179241e-05,
+      "loss": 0.2031,
+      "step": 4085
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7235273196359402e-05,
+      "loss": 0.2062,
+      "step": 4086
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7233821380954655e-05,
+      "loss": 0.2079,
+      "step": 4087
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.723236924564237e-05,
+      "loss": 0.2118,
+      "step": 4088
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.723091679048677e-05,
+      "loss": 0.2466,
+      "step": 4089
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7229464015552087e-05,
+      "loss": 0.2035,
+      "step": 4090
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7228010920902564e-05,
+      "loss": 0.2158,
+      "step": 4091
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7226557506602472e-05,
+      "loss": 0.2182,
+      "step": 4092
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.722510377271608e-05,
+      "loss": 0.202,
+      "step": 4093
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7223649719307676e-05,
+      "loss": 0.2063,
+      "step": 4094
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.722219534644157e-05,
+      "loss": 0.1903,
+      "step": 4095
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7220740654182077e-05,
+      "loss": 0.2203,
+      "step": 4096
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7219285642593524e-05,
+      "loss": 0.2284,
+      "step": 4097
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7217830311740266e-05,
+      "loss": 0.1943,
+      "step": 4098
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.721637466168666e-05,
+      "loss": 0.1997,
+      "step": 4099
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7214918692497074e-05,
+      "loss": 0.2298,
+      "step": 4100
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.721346240423591e-05,
+      "loss": 0.2129,
+      "step": 4101
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7212005796967558e-05,
+      "loss": 0.2051,
+      "step": 4102
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.721054887075644e-05,
+      "loss": 0.2425,
+      "step": 4103
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7209091625666988e-05,
+      "loss": 0.2105,
+      "step": 4104
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7207634061763648e-05,
+      "loss": 0.2554,
+      "step": 4105
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7206176179110876e-05,
+      "loss": 0.2321,
+      "step": 4106
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7204717977773145e-05,
+      "loss": 0.2189,
+      "step": 4107
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7203259457814945e-05,
+      "loss": 0.2235,
+      "step": 4108
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7201800619300772e-05,
+      "loss": 0.2059,
+      "step": 4109
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.720034146229515e-05,
+      "loss": 0.2306,
+      "step": 4110
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.71988819868626e-05,
+      "loss": 0.2275,
+      "step": 4111
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.719742219306767e-05,
+      "loss": 0.2382,
+      "step": 4112
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.719596208097492e-05,
+      "loss": 0.2381,
+      "step": 4113
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.719450165064892e-05,
+      "loss": 0.2326,
+      "step": 4114
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7193040902154252e-05,
+      "loss": 0.2074,
+      "step": 4115
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7191579835555518e-05,
+      "loss": 0.2371,
+      "step": 4116
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.719011845091733e-05,
+      "loss": 0.2102,
+      "step": 4117
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7188656748304325e-05,
+      "loss": 0.2419,
+      "step": 4118
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7187194727781138e-05,
+      "loss": 0.2129,
+      "step": 4119
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7185732389412425e-05,
+      "loss": 0.2553,
+      "step": 4120
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.718426973326286e-05,
+      "loss": 0.2515,
+      "step": 4121
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7182806759397123e-05,
+      "loss": 0.2314,
+      "step": 4122
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7181343467879914e-05,
+      "loss": 0.2154,
+      "step": 4123
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7179879858775945e-05,
+      "loss": 0.225,
+      "step": 4124
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7178415932149945e-05,
+      "loss": 0.2002,
+      "step": 4125
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.717695168806665e-05,
+      "loss": 0.209,
+      "step": 4126
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.717548712659082e-05,
+      "loss": 0.1843,
+      "step": 4127
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.717402224778722e-05,
+      "loss": 0.1971,
+      "step": 4128
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.717255705172063e-05,
+      "loss": 0.2303,
+      "step": 4129
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.717109153845585e-05,
+      "loss": 0.2375,
+      "step": 4130
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7169625708057694e-05,
+      "loss": 0.1831,
+      "step": 4131
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7168159560590983e-05,
+      "loss": 0.2438,
+      "step": 4132
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7166693096120554e-05,
+      "loss": 0.2118,
+      "step": 4133
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7165226314711264e-05,
+      "loss": 0.2691,
+      "step": 4134
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7163759216427975e-05,
+      "loss": 0.2466,
+      "step": 4135
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.716229180133557e-05,
+      "loss": 0.225,
+      "step": 4136
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7160824069498944e-05,
+      "loss": 0.1944,
+      "step": 4137
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.715935602098301e-05,
+      "loss": 0.221,
+      "step": 4138
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.715788765585268e-05,
+      "loss": 0.258,
+      "step": 4139
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.71564189741729e-05,
+      "loss": 0.2166,
+      "step": 4140
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7154949976008623e-05,
+      "loss": 0.2233,
+      "step": 4141
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7153480661424802e-05,
+      "loss": 0.2174,
+      "step": 4142
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7152011030486425e-05,
+      "loss": 0.2021,
+      "step": 4143
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7150541083258484e-05,
+      "loss": 0.2429,
+      "step": 4144
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.714907081980598e-05,
+      "loss": 0.2165,
+      "step": 4145
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.714760024019394e-05,
+      "loss": 0.2087,
+      "step": 4146
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7146129344487396e-05,
+      "loss": 0.2358,
+      "step": 4147
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.71446581327514e-05,
+      "loss": 0.1931,
+      "step": 4148
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7143186605051007e-05,
+      "loss": 0.2343,
+      "step": 4149
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.71417147614513e-05,
+      "loss": 0.218,
+      "step": 4150
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.714024260201737e-05,
+      "loss": 0.2503,
+      "step": 4151
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.713877012681432e-05,
+      "loss": 0.207,
+      "step": 4152
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7137297335907268e-05,
+      "loss": 0.2459,
+      "step": 4153
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7135824229361345e-05,
+      "loss": 0.206,
+      "step": 4154
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.71343508072417e-05,
+      "loss": 0.2075,
+      "step": 4155
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7132877069613495e-05,
+      "loss": 0.2431,
+      "step": 4156
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7131403016541898e-05,
+      "loss": 0.2152,
+      "step": 4157
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7129928648092105e-05,
+      "loss": 0.2622,
+      "step": 4158
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7128453964329313e-05,
+      "loss": 0.217,
+      "step": 4159
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7126978965318737e-05,
+      "loss": 0.2543,
+      "step": 4160
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7125503651125613e-05,
+      "loss": 0.252,
+      "step": 4161
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7124028021815182e-05,
+      "loss": 0.2233,
+      "step": 4162
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7122552077452703e-05,
+      "loss": 0.2446,
+      "step": 4163
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7121075818103444e-05,
+      "loss": 0.2446,
+      "step": 4164
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.71195992438327e-05,
+      "loss": 0.2229,
+      "step": 4165
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7118122354705758e-05,
+      "loss": 0.2062,
+      "step": 4166
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7116645150787942e-05,
+      "loss": 0.2084,
+      "step": 4167
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7115167632144576e-05,
+      "loss": 0.2064,
+      "step": 4168
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7113689798841e-05,
+      "loss": 0.2402,
+      "step": 4169
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7112211650942572e-05,
+      "loss": 0.2456,
+      "step": 4170
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7110733188514662e-05,
+      "loss": 0.2007,
+      "step": 4171
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.710925441162265e-05,
+      "loss": 0.213,
+      "step": 4172
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7107775320331937e-05,
+      "loss": 0.2362,
+      "step": 4173
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7106295914707934e-05,
+      "loss": 0.2222,
+      "step": 4174
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7104816194816063e-05,
+      "loss": 0.2293,
+      "step": 4175
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7103336160721765e-05,
+      "loss": 0.2203,
+      "step": 4176
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.710185581249049e-05,
+      "loss": 0.2342,
+      "step": 4177
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7100375150187708e-05,
+      "loss": 0.2333,
+      "step": 4178
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7098894173878896e-05,
+      "loss": 0.215,
+      "step": 4179
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7097412883629554e-05,
+      "loss": 0.2188,
+      "step": 4180
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7095931279505187e-05,
+      "loss": 0.2142,
+      "step": 4181
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.709444936157132e-05,
+      "loss": 0.2288,
+      "step": 4182
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7092967129893482e-05,
+      "loss": 0.2061,
+      "step": 4183
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.709148458453723e-05,
+      "loss": 0.2167,
+      "step": 4184
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7090001725568125e-05,
+      "loss": 0.2152,
+      "step": 4185
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7088518553051745e-05,
+      "loss": 0.2363,
+      "step": 4186
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7087035067053683e-05,
+      "loss": 0.27,
+      "step": 4187
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7085551267639545e-05,
+      "loss": 0.245,
+      "step": 4188
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7084067154874947e-05,
+      "loss": 0.2166,
+      "step": 4189
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7082582728825524e-05,
+      "loss": 0.2075,
+      "step": 4190
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7081097989556923e-05,
+      "loss": 0.237,
+      "step": 4191
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7079612937134808e-05,
+      "loss": 0.2049,
+      "step": 4192
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.707812757162485e-05,
+      "loss": 0.2136,
+      "step": 4193
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7076641893092733e-05,
+      "loss": 0.2211,
+      "step": 4194
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.707515590160417e-05,
+      "loss": 0.2286,
+      "step": 4195
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.707366959722487e-05,
+      "loss": 0.2177,
+      "step": 4196
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7072182980020565e-05,
+      "loss": 0.2383,
+      "step": 4197
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7070696050057e-05,
+      "loss": 0.2181,
+      "step": 4198
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7069208807399935e-05,
+      "loss": 0.2148,
+      "step": 4199
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7067721252115132e-05,
+      "loss": 0.2555,
+      "step": 4200
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.706623338426839e-05,
+      "loss": 0.2267,
+      "step": 4201
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7064745203925495e-05,
+      "loss": 0.2235,
+      "step": 4202
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.706325671115227e-05,
+      "loss": 0.2191,
+      "step": 4203
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7061767906014538e-05,
+      "loss": 0.206,
+      "step": 4204
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7060278788578143e-05,
+      "loss": 0.2261,
+      "step": 4205
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.705878935890893e-05,
+      "loss": 0.2371,
+      "step": 4206
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7057299617072776e-05,
+      "loss": 0.1921,
+      "step": 4207
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7055809563135565e-05,
+      "loss": 0.2226,
+      "step": 4208
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7054319197163187e-05,
+      "loss": 0.2311,
+      "step": 4209
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7052828519221554e-05,
+      "loss": 0.236,
+      "step": 4210
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.705133752937659e-05,
+      "loss": 0.2039,
+      "step": 4211
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.704984622769423e-05,
+      "loss": 0.227,
+      "step": 4212
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7048354614240424e-05,
+      "loss": 0.2179,
+      "step": 4213
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7046862689081144e-05,
+      "loss": 0.2034,
+      "step": 4214
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7045370452282362e-05,
+      "loss": 0.2448,
+      "step": 4215
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7043877903910076e-05,
+      "loss": 0.2183,
+      "step": 4216
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7042385044030284e-05,
+      "loss": 0.2155,
+      "step": 4217
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7040891872709013e-05,
+      "loss": 0.2319,
+      "step": 4218
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7039398390012295e-05,
+      "loss": 0.2047,
+      "step": 4219
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7037904596006178e-05,
+      "loss": 0.2363,
+      "step": 4220
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7036410490756723e-05,
+      "loss": 0.2596,
+      "step": 4221
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7034916074330006e-05,
+      "loss": 0.2271,
+      "step": 4222
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7033421346792113e-05,
+      "loss": 0.2627,
+      "step": 4223
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7031926308209146e-05,
+      "loss": 0.2264,
+      "step": 4224
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7030430958647232e-05,
+      "loss": 0.2434,
+      "step": 4225
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7028935298172485e-05,
+      "loss": 0.2053,
+      "step": 4226
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.702743932685106e-05,
+      "loss": 0.2225,
+      "step": 4227
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.702594304474911e-05,
+      "loss": 0.1969,
+      "step": 4228
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.702444645193281e-05,
+      "loss": 0.2393,
+      "step": 4229
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.702294954846834e-05,
+      "loss": 0.2444,
+      "step": 4230
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7021452334421904e-05,
+      "loss": 0.2221,
+      "step": 4231
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7019954809859714e-05,
+      "loss": 0.2296,
+      "step": 4232
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7018456974847996e-05,
+      "loss": 0.2548,
+      "step": 4233
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7016958829452983e-05,
+      "loss": 0.2349,
+      "step": 4234
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7015460373740935e-05,
+      "loss": 0.2119,
+      "step": 4235
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.701396160777812e-05,
+      "loss": 0.2331,
+      "step": 4236
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.701246253163082e-05,
+      "loss": 0.2086,
+      "step": 4237
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.701096314536533e-05,
+      "loss": 0.2366,
+      "step": 4238
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.700946344904795e-05,
+      "loss": 0.2119,
+      "step": 4239
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.700796344274502e-05,
+      "loss": 0.2039,
+      "step": 4240
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7006463126522856e-05,
+      "loss": 0.2125,
+      "step": 4241
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7004962500447818e-05,
+      "loss": 0.2342,
+      "step": 4242
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.700346156458627e-05,
+      "loss": 0.1985,
+      "step": 4243
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.700196031900459e-05,
+      "loss": 0.2287,
+      "step": 4244
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7000458763769167e-05,
+      "loss": 0.2197,
+      "step": 4245
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6998956898946405e-05,
+      "loss": 0.243,
+      "step": 4246
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.699745472460272e-05,
+      "loss": 0.2234,
+      "step": 4247
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6995952240804548e-05,
+      "loss": 0.2311,
+      "step": 4248
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.699444944761833e-05,
+      "loss": 0.1996,
+      "step": 4249
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6992946345110534e-05,
+      "loss": 0.2103,
+      "step": 4250
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.699144293334762e-05,
+      "loss": 0.2184,
+      "step": 4251
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6989939212396088e-05,
+      "loss": 0.2093,
+      "step": 4252
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.698843518232243e-05,
+      "loss": 0.2171,
+      "step": 4253
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6986930843193164e-05,
+      "loss": 0.2548,
+      "step": 4254
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6985426195074814e-05,
+      "loss": 0.2819,
+      "step": 4255
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6983921238033923e-05,
+      "loss": 0.2042,
+      "step": 4256
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6982415972137053e-05,
+      "loss": 0.2504,
+      "step": 4257
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.698091039745076e-05,
+      "loss": 0.2137,
+      "step": 4258
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6979404514041632e-05,
+      "loss": 0.2153,
+      "step": 4259
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6977898321976268e-05,
+      "loss": 0.1758,
+      "step": 4260
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6976391821321273e-05,
+      "loss": 0.2196,
+      "step": 4261
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6974885012143276e-05,
+      "loss": 0.2463,
+      "step": 4262
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6973377894508906e-05,
+      "loss": 0.2461,
+      "step": 4263
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.697187046848482e-05,
+      "loss": 0.248,
+      "step": 4264
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6970362734137684e-05,
+      "loss": 0.2049,
+      "step": 4265
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6968854691534168e-05,
+      "loss": 0.2313,
+      "step": 4266
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6967346340740967e-05,
+      "loss": 0.2454,
+      "step": 4267
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6965837681824786e-05,
+      "loss": 0.2449,
+      "step": 4268
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6964328714852344e-05,
+      "loss": 0.2403,
+      "step": 4269
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6962819439890378e-05,
+      "loss": 0.258,
+      "step": 4270
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6961309857005626e-05,
+      "loss": 0.2174,
+      "step": 4271
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6959799966264853e-05,
+      "loss": 0.2015,
+      "step": 4272
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6958289767734827e-05,
+      "loss": 0.2305,
+      "step": 4273
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.695677926148234e-05,
+      "loss": 0.2307,
+      "step": 4274
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6955268447574194e-05,
+      "loss": 0.2218,
+      "step": 4275
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6953757326077198e-05,
+      "loss": 0.2565,
+      "step": 4276
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6952245897058178e-05,
+      "loss": 0.2312,
+      "step": 4277
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.695073416058398e-05,
+      "loss": 0.2023,
+      "step": 4278
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.694922211672146e-05,
+      "loss": 0.2406,
+      "step": 4279
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.694770976553748e-05,
+      "loss": 0.2153,
+      "step": 4280
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6946197107098926e-05,
+      "loss": 0.223,
+      "step": 4281
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6944684141472697e-05,
+      "loss": 0.2437,
+      "step": 4282
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6943170868725697e-05,
+      "loss": 0.2306,
+      "step": 4283
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.694165728892485e-05,
+      "loss": 0.1846,
+      "step": 4284
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.694014340213709e-05,
+      "loss": 0.2358,
+      "step": 4285
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.693862920842937e-05,
+      "loss": 0.2034,
+      "step": 4286
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6937114707868655e-05,
+      "loss": 0.1853,
+      "step": 4287
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.693559990052192e-05,
+      "loss": 0.2214,
+      "step": 4288
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6934084786456157e-05,
+      "loss": 0.1935,
+      "step": 4289
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6932569365738367e-05,
+      "loss": 0.2457,
+      "step": 4290
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6931053638435572e-05,
+      "loss": 0.205,
+      "step": 4291
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.69295376046148e-05,
+      "loss": 0.2436,
+      "step": 4292
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6928021264343095e-05,
+      "loss": 0.2144,
+      "step": 4293
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6926504617687517e-05,
+      "loss": 0.2111,
+      "step": 4294
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6924987664715146e-05,
+      "loss": 0.2383,
+      "step": 4295
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6923470405493052e-05,
+      "loss": 0.2002,
+      "step": 4296
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.692195284008834e-05,
+      "loss": 0.2352,
+      "step": 4297
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6920434968568133e-05,
+      "loss": 0.2341,
+      "step": 4298
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.691891679099954e-05,
+      "loss": 0.2426,
+      "step": 4299
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6917398307449717e-05,
+      "loss": 0.181,
+      "step": 4300
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.691587951798581e-05,
+      "loss": 0.2173,
+      "step": 4301
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.691436042267498e-05,
+      "loss": 0.1867,
+      "step": 4302
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6912841021584413e-05,
+      "loss": 0.2606,
+      "step": 4303
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6911321314781302e-05,
+      "loss": 0.1958,
+      "step": 4304
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6909801302332857e-05,
+      "loss": 0.2327,
+      "step": 4305
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6908280984306292e-05,
+      "loss": 0.2083,
+      "step": 4306
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.690676036076885e-05,
+      "loss": 0.2584,
+      "step": 4307
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6905239431787776e-05,
+      "loss": 0.2235,
+      "step": 4308
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.690371819743033e-05,
+      "loss": 0.2254,
+      "step": 4309
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6902196657763785e-05,
+      "loss": 0.2292,
+      "step": 4310
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.690067481285543e-05,
+      "loss": 0.2178,
+      "step": 4311
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6899152662772572e-05,
+      "loss": 0.2424,
+      "step": 4312
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6897630207582518e-05,
+      "loss": 0.2434,
+      "step": 4313
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6896107447352602e-05,
+      "loss": 0.2377,
+      "step": 4314
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6894584382150168e-05,
+      "loss": 0.2112,
+      "step": 4315
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6893061012042568e-05,
+      "loss": 0.2356,
+      "step": 4316
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6891537337097175e-05,
+      "loss": 0.1986,
+      "step": 4317
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6890013357381366e-05,
+      "loss": 0.2496,
+      "step": 4318
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6888489072962543e-05,
+      "loss": 0.2049,
+      "step": 4319
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6886964483908114e-05,
+      "loss": 0.2489,
+      "step": 4320
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.68854395902855e-05,
+      "loss": 0.241,
+      "step": 4321
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6883914392162144e-05,
+      "loss": 0.2169,
+      "step": 4322
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6882388889605487e-05,
+      "loss": 0.2096,
+      "step": 4323
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6880863082683e-05,
+      "loss": 0.2111,
+      "step": 4324
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6879336971462156e-05,
+      "loss": 0.2349,
+      "step": 4325
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.687781055601045e-05,
+      "loss": 0.2235,
+      "step": 4326
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.687628383639538e-05,
+      "loss": 0.2009,
+      "step": 4327
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6874756812684463e-05,
+      "loss": 0.239,
+      "step": 4328
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.687322948494524e-05,
+      "loss": 0.2088,
+      "step": 4329
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6871701853245245e-05,
+      "loss": 0.2396,
+      "step": 4330
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.687017391765204e-05,
+      "loss": 0.209,
+      "step": 4331
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6868645678233195e-05,
+      "loss": 0.2156,
+      "step": 4332
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6867117135056296e-05,
+      "loss": 0.2303,
+      "step": 4333
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6865588288188937e-05,
+      "loss": 0.2141,
+      "step": 4334
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6864059137698736e-05,
+      "loss": 0.2397,
+      "step": 4335
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.686252968365332e-05,
+      "loss": 0.2041,
+      "step": 4336
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6860999926120313e-05,
+      "loss": 0.213,
+      "step": 4337
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.685946986516738e-05,
+      "loss": 0.2002,
+      "step": 4338
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6857939500862183e-05,
+      "loss": 0.2159,
+      "step": 4339
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6856408833272398e-05,
+      "loss": 0.209,
+      "step": 4340
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.685487786246572e-05,
+      "loss": 0.2504,
+      "step": 4341
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.685334658850985e-05,
+      "loss": 0.23,
+      "step": 4342
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6851815011472514e-05,
+      "loss": 0.2065,
+      "step": 4343
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6850283131421438e-05,
+      "loss": 0.2262,
+      "step": 4344
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.684875094842437e-05,
+      "loss": 0.2162,
+      "step": 4345
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6847218462549073e-05,
+      "loss": 0.2372,
+      "step": 4346
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.684568567386331e-05,
+      "loss": 0.2493,
+      "step": 4347
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6844152582434877e-05,
+      "loss": 0.2112,
+      "step": 4348
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6842619188331564e-05,
+      "loss": 0.2266,
+      "step": 4349
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.684108549162119e-05,
+      "loss": 0.2275,
+      "step": 4350
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6839551492371582e-05,
+      "loss": 0.2463,
+      "step": 4351
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6838017190650572e-05,
+      "loss": 0.2075,
+      "step": 4352
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6836482586526024e-05,
+      "loss": 0.2226,
+      "step": 4353
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.683494768006579e-05,
+      "loss": 0.217,
+      "step": 4354
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.683341247133776e-05,
+      "loss": 0.2761,
+      "step": 4355
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6831876960409827e-05,
+      "loss": 0.2543,
+      "step": 4356
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.683034114734989e-05,
+      "loss": 0.2464,
+      "step": 4357
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6828805032225874e-05,
+      "loss": 0.2241,
+      "step": 4358
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.682726861510571e-05,
+      "loss": 0.202,
+      "step": 4359
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6825731896057346e-05,
+      "loss": 0.2294,
+      "step": 4360
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.682419487514874e-05,
+      "loss": 0.2116,
+      "step": 4361
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6822657552447865e-05,
+      "loss": 0.2425,
+      "step": 4362
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6821119928022706e-05,
+      "loss": 0.2092,
+      "step": 4363
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.681958200194127e-05,
+      "loss": 0.2691,
+      "step": 4364
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6818043774271558e-05,
+      "loss": 0.2198,
+      "step": 4365
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.681650524508161e-05,
+      "loss": 0.2161,
+      "step": 4366
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6814966414439453e-05,
+      "loss": 0.2232,
+      "step": 4367
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6813427282413146e-05,
+      "loss": 0.2442,
+      "step": 4368
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6811887849070755e-05,
+      "loss": 0.2123,
+      "step": 4369
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6810348114480364e-05,
+      "loss": 0.2092,
+      "step": 4370
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6808808078710058e-05,
+      "loss": 0.2139,
+      "step": 4371
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.680726774182795e-05,
+      "loss": 0.2301,
+      "step": 4372
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6805727103902155e-05,
+      "loss": 0.2368,
+      "step": 4373
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6804186165000808e-05,
+      "loss": 0.2287,
+      "step": 4374
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6802644925192053e-05,
+      "loss": 0.2631,
+      "step": 4375
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6801103384544056e-05,
+      "loss": 0.1728,
+      "step": 4376
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6799561543124984e-05,
+      "loss": 0.1795,
+      "step": 4377
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.679801940100302e-05,
+      "loss": 0.2155,
+      "step": 4378
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6796476958246373e-05,
+      "loss": 0.227,
+      "step": 4379
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.679493421492325e-05,
+      "loss": 0.224,
+      "step": 4380
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6793391171101875e-05,
+      "loss": 0.2103,
+      "step": 4381
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6791847826850493e-05,
+      "loss": 0.2172,
+      "step": 4382
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6790304182237358e-05,
+      "loss": 0.2364,
+      "step": 4383
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6788760237330725e-05,
+      "loss": 0.2248,
+      "step": 4384
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6787215992198885e-05,
+      "loss": 0.2415,
+      "step": 4385
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6785671446910123e-05,
+      "loss": 0.2111,
+      "step": 4386
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6784126601532745e-05,
+      "loss": 0.2297,
+      "step": 4387
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6782581456135078e-05,
+      "loss": 0.227,
+      "step": 4388
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6781036010785446e-05,
+      "loss": 0.2129,
+      "step": 4389
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.67794902655522e-05,
+      "loss": 0.2148,
+      "step": 4390
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.677794422050369e-05,
+      "loss": 0.2038,
+      "step": 4391
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6776397875708298e-05,
+      "loss": 0.2145,
+      "step": 4392
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6774851231234404e-05,
+      "loss": 0.2678,
+      "step": 4393
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6773304287150412e-05,
+      "loss": 0.2283,
+      "step": 4394
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6771757043524727e-05,
+      "loss": 0.2119,
+      "step": 4395
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6770209500425776e-05,
+      "loss": 0.2399,
+      "step": 4396
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6768661657922e-05,
+      "loss": 0.2422,
+      "step": 4397
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.676711351608185e-05,
+      "loss": 0.2229,
+      "step": 4398
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6765565074973785e-05,
+      "loss": 0.249,
+      "step": 4399
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6764016334666297e-05,
+      "loss": 0.2227,
+      "step": 4400
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6762467295227863e-05,
+      "loss": 0.2646,
+      "step": 4401
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.676091795672699e-05,
+      "loss": 0.2093,
+      "step": 4402
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.67593683192322e-05,
+      "loss": 0.2639,
+      "step": 4403
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6757818382812024e-05,
+      "loss": 0.2217,
+      "step": 4404
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6756268147535003e-05,
+      "loss": 0.2087,
+      "step": 4405
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6754717613469694e-05,
+      "loss": 0.2508,
+      "step": 4406
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6753166780684675e-05,
+      "loss": 0.2152,
+      "step": 4407
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.675161564924852e-05,
+      "loss": 0.2086,
+      "step": 4408
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6750064219229832e-05,
+      "loss": 0.2372,
+      "step": 4409
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6748512490697218e-05,
+      "loss": 0.2444,
+      "step": 4410
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6746960463719305e-05,
+      "loss": 0.2079,
+      "step": 4411
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.674540813836473e-05,
+      "loss": 0.1959,
+      "step": 4412
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6743855514702133e-05,
+      "loss": 0.2329,
+      "step": 4413
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6742302592800192e-05,
+      "loss": 0.2535,
+      "step": 4414
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6740749372727573e-05,
+      "loss": 0.2557,
+      "step": 4415
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6739195854552967e-05,
+      "loss": 0.2022,
+      "step": 4416
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.673764203834508e-05,
+      "loss": 0.2506,
+      "step": 4417
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6736087924172622e-05,
+      "loss": 0.2181,
+      "step": 4418
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6734533512104327e-05,
+      "loss": 0.2219,
+      "step": 4419
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6732978802208933e-05,
+      "loss": 0.2205,
+      "step": 4420
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6731423794555204e-05,
+      "loss": 0.2256,
+      "step": 4421
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6729868489211898e-05,
+      "loss": 0.2201,
+      "step": 4422
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.67283128862478e-05,
+      "loss": 0.2149,
+      "step": 4423
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.672675698573171e-05,
+      "loss": 0.2238,
+      "step": 4424
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.672520078773243e-05,
+      "loss": 0.2246,
+      "step": 4425
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6723644292318775e-05,
+      "loss": 0.2075,
+      "step": 4426
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6722087499559597e-05,
+      "loss": 0.2156,
+      "step": 4427
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.672053040952373e-05,
+      "loss": 0.2272,
+      "step": 4428
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6718973022280036e-05,
+      "loss": 0.2213,
+      "step": 4429
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.671741533789739e-05,
+      "loss": 0.216,
+      "step": 4430
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6715857356444684e-05,
+      "loss": 0.2209,
+      "step": 4431
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6714299077990807e-05,
+      "loss": 0.2179,
+      "step": 4432
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.671274050260468e-05,
+      "loss": 0.2036,
+      "step": 4433
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.671118163035523e-05,
+      "loss": 0.2246,
+      "step": 4434
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6709622461311396e-05,
+      "loss": 0.2145,
+      "step": 4435
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6708062995542126e-05,
+      "loss": 0.205,
+      "step": 4436
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.670650323311638e-05,
+      "loss": 0.2197,
+      "step": 4437
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6704943174103158e-05,
+      "loss": 0.2123,
+      "step": 4438
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.670338281857143e-05,
+      "loss": 0.2309,
+      "step": 4439
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6701822166590212e-05,
+      "loss": 0.1929,
+      "step": 4440
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.670026121822852e-05,
+      "loss": 0.2308,
+      "step": 4441
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6698699973555384e-05,
+      "loss": 0.2465,
+      "step": 4442
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6697138432639852e-05,
+      "loss": 0.2316,
+      "step": 4443
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6695576595550974e-05,
+      "loss": 0.2007,
+      "step": 4444
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6694014462357828e-05,
+      "loss": 0.221,
+      "step": 4445
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6692452033129493e-05,
+      "loss": 0.1961,
+      "step": 4446
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6690889307935067e-05,
+      "loss": 0.2201,
+      "step": 4447
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.668932628684366e-05,
+      "loss": 0.2267,
+      "step": 4448
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6687762969924396e-05,
+      "loss": 0.2203,
+      "step": 4449
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.668619935724641e-05,
+      "loss": 0.2419,
+      "step": 4450
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6684635448878848e-05,
+      "loss": 0.2174,
+      "step": 4451
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6683071244890876e-05,
+      "loss": 0.2295,
+      "step": 4452
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.668150674535167e-05,
+      "loss": 0.2201,
+      "step": 4453
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.667994195033041e-05,
+      "loss": 0.2231,
+      "step": 4454
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6678376859896305e-05,
+      "loss": 0.2142,
+      "step": 4455
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.667681147411857e-05,
+      "loss": 0.2304,
+      "step": 4456
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.667524579306643e-05,
+      "loss": 0.2077,
+      "step": 4457
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6673679816809126e-05,
+      "loss": 0.2005,
+      "step": 4458
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.667211354541591e-05,
+      "loss": 0.2157,
+      "step": 4459
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6670546978956047e-05,
+      "loss": 0.2046,
+      "step": 4460
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6668980117498823e-05,
+      "loss": 0.238,
+      "step": 4461
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.666741296111353e-05,
+      "loss": 0.1976,
+      "step": 4462
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6665845509869462e-05,
+      "loss": 0.1907,
+      "step": 4463
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6664277763835954e-05,
+      "loss": 0.2004,
+      "step": 4464
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6662709723082327e-05,
+      "loss": 0.2043,
+      "step": 4465
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6661141387677928e-05,
+      "loss": 0.2294,
+      "step": 4466
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.665957275769212e-05,
+      "loss": 0.2236,
+      "step": 4467
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6658003833194266e-05,
+      "loss": 0.2267,
+      "step": 4468
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.665643461425375e-05,
+      "loss": 0.1947,
+      "step": 4469
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6654865100939977e-05,
+      "loss": 0.1798,
+      "step": 4470
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6653295293322355e-05,
+      "loss": 0.2175,
+      "step": 4471
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.66517251914703e-05,
+      "loss": 0.2253,
+      "step": 4472
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.665015479545325e-05,
+      "loss": 0.237,
+      "step": 4473
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6648584105340664e-05,
+      "loss": 0.2015,
+      "step": 4474
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6647013121201992e-05,
+      "loss": 0.2437,
+      "step": 4475
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6645441843106716e-05,
+      "loss": 0.2416,
+      "step": 4476
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.664387027112432e-05,
+      "loss": 0.243,
+      "step": 4477
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6642298405324303e-05,
+      "loss": 0.1984,
+      "step": 4478
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6640726245776187e-05,
+      "loss": 0.2212,
+      "step": 4479
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.663915379254949e-05,
+      "loss": 0.212,
+      "step": 4480
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.663758104571376e-05,
+      "loss": 0.2232,
+      "step": 4481
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6636008005338547e-05,
+      "loss": 0.1932,
+      "step": 4482
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6634434671493417e-05,
+      "loss": 0.2156,
+      "step": 4483
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6632861044247942e-05,
+      "loss": 0.2159,
+      "step": 4484
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6631287123671722e-05,
+      "loss": 0.2366,
+      "step": 4485
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6629712909834363e-05,
+      "loss": 0.2034,
+      "step": 4486
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6628138402805476e-05,
+      "loss": 0.2275,
+      "step": 4487
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.66265636026547e-05,
+      "loss": 0.2339,
+      "step": 4488
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6624988509451667e-05,
+      "loss": 0.2385,
+      "step": 4489
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6623413123266045e-05,
+      "loss": 0.2136,
+      "step": 4490
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6621837444167498e-05,
+      "loss": 0.2381,
+      "step": 4491
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6620261472225714e-05,
+      "loss": 0.2193,
+      "step": 4492
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.661868520751038e-05,
+      "loss": 0.228,
+      "step": 4493
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6617108650091213e-05,
+      "loss": 0.1921,
+      "step": 4494
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6615531800037928e-05,
+      "loss": 0.2082,
+      "step": 4495
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6613954657420263e-05,
+      "loss": 0.2213,
+      "step": 4496
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6612377222307965e-05,
+      "loss": 0.2332,
+      "step": 4497
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6610799494770794e-05,
+      "loss": 0.2137,
+      "step": 4498
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.660922147487852e-05,
+      "loss": 0.2336,
+      "step": 4499
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6607643162700932e-05,
+      "loss": 0.2252,
+      "step": 4500
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.660606455830783e-05,
+      "loss": 0.212,
+      "step": 4501
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6604485661769026e-05,
+      "loss": 0.2195,
+      "step": 4502
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6602906473154345e-05,
+      "loss": 0.2032,
+      "step": 4503
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6601326992533625e-05,
+      "loss": 0.2348,
+      "step": 4504
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.659974721997671e-05,
+      "loss": 0.2113,
+      "step": 4505
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6598167155553473e-05,
+      "loss": 0.2255,
+      "step": 4506
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.659658679933379e-05,
+      "loss": 0.2082,
+      "step": 4507
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.659500615138754e-05,
+      "loss": 0.2155,
+      "step": 4508
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6593425211784638e-05,
+      "loss": 0.2273,
+      "step": 4509
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.659184398059499e-05,
+      "loss": 0.2501,
+      "step": 4510
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6590262457888527e-05,
+      "loss": 0.2366,
+      "step": 4511
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6588680643735195e-05,
+      "loss": 0.2306,
+      "step": 4512
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6587098538204945e-05,
+      "loss": 0.2058,
+      "step": 4513
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6585516141367742e-05,
+      "loss": 0.2262,
+      "step": 4514
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.658393345329356e-05,
+      "loss": 0.2409,
+      "step": 4515
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6582350474052406e-05,
+      "loss": 0.2015,
+      "step": 4516
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6580767203714275e-05,
+      "loss": 0.2156,
+      "step": 4517
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6579183642349186e-05,
+      "loss": 0.2332,
+      "step": 4518
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6577599790027172e-05,
+      "loss": 0.2287,
+      "step": 4519
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6576015646818278e-05,
+      "loss": 0.2428,
+      "step": 4520
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6574431212792562e-05,
+      "loss": 0.2569,
+      "step": 4521
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.657284648802009e-05,
+      "loss": 0.2399,
+      "step": 4522
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.657126147257095e-05,
+      "loss": 0.2205,
+      "step": 4523
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6569676166515228e-05,
+      "loss": 0.2384,
+      "step": 4524
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.656809056992304e-05,
+      "loss": 0.2391,
+      "step": 4525
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6566504682864507e-05,
+      "loss": 0.2004,
+      "step": 4526
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6564918505409764e-05,
+      "loss": 0.2443,
+      "step": 4527
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6563332037628953e-05,
+      "loss": 0.2408,
+      "step": 4528
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6561745279592234e-05,
+      "loss": 0.228,
+      "step": 4529
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6560158231369787e-05,
+      "loss": 0.2051,
+      "step": 4530
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6558570893031785e-05,
+      "loss": 0.2286,
+      "step": 4531
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.655698326464844e-05,
+      "loss": 0.2557,
+      "step": 4532
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6555395346289956e-05,
+      "loss": 0.2232,
+      "step": 4533
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6553807138026553e-05,
+      "loss": 0.2254,
+      "step": 4534
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6552218639928475e-05,
+      "loss": 0.2351,
+      "step": 4535
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.655062985206597e-05,
+      "loss": 0.2226,
+      "step": 4536
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6549040774509295e-05,
+      "loss": 0.2442,
+      "step": 4537
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6547451407328732e-05,
+      "loss": 0.226,
+      "step": 4538
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6545861750594564e-05,
+      "loss": 0.2287,
+      "step": 4539
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6544271804377097e-05,
+      "loss": 0.1901,
+      "step": 4540
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.654268156874664e-05,
+      "loss": 0.2394,
+      "step": 4541
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.654109104377352e-05,
+      "loss": 0.2336,
+      "step": 4542
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.653950022952808e-05,
+      "loss": 0.2672,
+      "step": 4543
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6537909126080665e-05,
+      "loss": 0.2545,
+      "step": 4544
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6536317733501648e-05,
+      "loss": 0.2316,
+      "step": 4545
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.65347260518614e-05,
+      "loss": 0.2435,
+      "step": 4546
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.653313408123031e-05,
+      "loss": 0.2343,
+      "step": 4547
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.653154182167879e-05,
+      "loss": 0.2606,
+      "step": 4548
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6529949273277247e-05,
+      "loss": 0.2383,
+      "step": 4549
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6528356436096114e-05,
+      "loss": 0.2094,
+      "step": 4550
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.652676331020583e-05,
+      "loss": 0.2199,
+      "step": 4551
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6525169895676854e-05,
+      "loss": 0.2502,
+      "step": 4552
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.652357619257965e-05,
+      "loss": 0.2107,
+      "step": 4553
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6521982200984692e-05,
+      "loss": 0.2388,
+      "step": 4554
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6520387920962484e-05,
+      "loss": 0.2202,
+      "step": 4555
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6518793352583518e-05,
+      "loss": 0.2262,
+      "step": 4556
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6517198495918325e-05,
+      "loss": 0.2235,
+      "step": 4557
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.651560335103743e-05,
+      "loss": 0.2394,
+      "step": 4558
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6514007918011375e-05,
+      "loss": 0.2492,
+      "step": 4559
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6512412196910716e-05,
+      "loss": 0.2101,
+      "step": 4560
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6510816187806024e-05,
+      "loss": 0.2348,
+      "step": 4561
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.650921989076788e-05,
+      "loss": 0.1993,
+      "step": 4562
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6507623305866878e-05,
+      "loss": 0.2232,
+      "step": 4563
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6506026433173627e-05,
+      "loss": 0.1967,
+      "step": 4564
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6504429272758744e-05,
+      "loss": 0.226,
+      "step": 4565
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6502831824692865e-05,
+      "loss": 0.2311,
+      "step": 4566
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6501234089046633e-05,
+      "loss": 0.2319,
+      "step": 4567
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6499636065890705e-05,
+      "loss": 0.2504,
+      "step": 4568
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6498037755295754e-05,
+      "loss": 0.2082,
+      "step": 4569
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6496439157332464e-05,
+      "loss": 0.2298,
+      "step": 4570
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.649484027207153e-05,
+      "loss": 0.2354,
+      "step": 4571
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6493241099583657e-05,
+      "loss": 0.2558,
+      "step": 4572
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6491641639939574e-05,
+      "loss": 0.2165,
+      "step": 4573
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.649004189321001e-05,
+      "loss": 0.2457,
+      "step": 4574
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6488441859465716e-05,
+      "loss": 0.2277,
+      "step": 4575
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6486841538777446e-05,
+      "loss": 0.2276,
+      "step": 4576
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6485240931215976e-05,
+      "loss": 0.2405,
+      "step": 4577
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6483640036852095e-05,
+      "loss": 0.2578,
+      "step": 4578
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6482038855756593e-05,
+      "loss": 0.2227,
+      "step": 4579
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6480437388000284e-05,
+      "loss": 0.221,
+      "step": 4580
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6478835633653995e-05,
+      "loss": 0.1933,
+      "step": 4581
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6477233592788553e-05,
+      "loss": 0.2366,
+      "step": 4582
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6475631265474813e-05,
+      "loss": 0.2341,
+      "step": 4583
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6474028651783635e-05,
+      "loss": 0.232,
+      "step": 4584
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.647242575178589e-05,
+      "loss": 0.2342,
+      "step": 4585
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6470822565552466e-05,
+      "loss": 0.2157,
+      "step": 4586
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6469219093154265e-05,
+      "loss": 0.2103,
+      "step": 4587
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6467615334662198e-05,
+      "loss": 0.1994,
+      "step": 4588
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6466011290147187e-05,
+      "loss": 0.2318,
+      "step": 4589
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6464406959680165e-05,
+      "loss": 0.2483,
+      "step": 4590
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.646280234333209e-05,
+      "loss": 0.2384,
+      "step": 4591
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6461197441173918e-05,
+      "loss": 0.245,
+      "step": 4592
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6459592253276633e-05,
+      "loss": 0.2144,
+      "step": 4593
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.645798677971121e-05,
+      "loss": 0.2157,
+      "step": 4594
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.645638102054866e-05,
+      "loss": 0.2429,
+      "step": 4595
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.645477497585999e-05,
+      "loss": 0.2224,
+      "step": 4596
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6453168645716227e-05,
+      "loss": 0.2125,
+      "step": 4597
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.645156203018841e-05,
+      "loss": 0.2172,
+      "step": 4598
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6449955129347586e-05,
+      "loss": 0.2164,
+      "step": 4599
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6448347943264823e-05,
+      "loss": 0.2167,
+      "step": 4600
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.64467404720112e-05,
+      "loss": 0.2114,
+      "step": 4601
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6445132715657792e-05,
+      "loss": 0.1999,
+      "step": 4602
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6443524674275714e-05,
+      "loss": 0.2384,
+      "step": 4603
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6441916347936072e-05,
+      "loss": 0.2372,
+      "step": 4604
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.644030773671e-05,
+      "loss": 0.244,
+      "step": 4605
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.643869884066863e-05,
+      "loss": 0.2421,
+      "step": 4606
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6437089659883114e-05,
+      "loss": 0.2337,
+      "step": 4607
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.643548019442462e-05,
+      "loss": 0.2109,
+      "step": 4608
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6433870444364323e-05,
+      "loss": 0.2142,
+      "step": 4609
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6432260409773414e-05,
+      "loss": 0.2529,
+      "step": 4610
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.643065009072309e-05,
+      "loss": 0.2377,
+      "step": 4611
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6429039487284568e-05,
+      "loss": 0.1971,
+      "step": 4612
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6427428599529078e-05,
+      "loss": 0.2129,
+      "step": 4613
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6425817427527853e-05,
+      "loss": 0.2505,
+      "step": 4614
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6424205971352157e-05,
+      "loss": 0.2062,
+      "step": 4615
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6422594231073246e-05,
+      "loss": 0.2125,
+      "step": 4616
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6420982206762394e-05,
+      "loss": 0.2368,
+      "step": 4617
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.64193698984909e-05,
+      "loss": 0.2028,
+      "step": 4618
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.641775730633006e-05,
+      "loss": 0.199,
+      "step": 4619
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6416144430351197e-05,
+      "loss": 0.2052,
+      "step": 4620
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6414531270625626e-05,
+      "loss": 0.2045,
+      "step": 4621
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6412917827224695e-05,
+      "loss": 0.2068,
+      "step": 4622
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.641130410021976e-05,
+      "loss": 0.2271,
+      "step": 4623
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.640969008968218e-05,
+      "loss": 0.239,
+      "step": 4624
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6408075795683335e-05,
+      "loss": 0.2182,
+      "step": 4625
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6406461218294617e-05,
+      "loss": 0.2104,
+      "step": 4626
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6404846357587422e-05,
+      "loss": 0.2106,
+      "step": 4627
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6403231213633175e-05,
+      "loss": 0.2126,
+      "step": 4628
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.64016157865033e-05,
+      "loss": 0.2015,
+      "step": 4629
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6400000076269234e-05,
+      "loss": 0.1959,
+      "step": 4630
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.639838408300243e-05,
+      "loss": 0.2162,
+      "step": 4631
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6396767806774364e-05,
+      "loss": 0.2191,
+      "step": 4632
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6395151247656504e-05,
+      "loss": 0.2225,
+      "step": 4633
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6393534405720336e-05,
+      "loss": 0.226,
+      "step": 4634
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6391917281037378e-05,
+      "loss": 0.197,
+      "step": 4635
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6390299873679132e-05,
+      "loss": 0.2261,
+      "step": 4636
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6388682183717132e-05,
+      "loss": 0.2448,
+      "step": 4637
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6387064211222918e-05,
+      "loss": 0.2049,
+      "step": 4638
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6385445956268045e-05,
+      "loss": 0.2074,
+      "step": 4639
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6383827418924076e-05,
+      "loss": 0.235,
+      "step": 4640
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6382208599262585e-05,
+      "loss": 0.2282,
+      "step": 4641
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6380589497355172e-05,
+      "loss": 0.247,
+      "step": 4642
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6378970113273433e-05,
+      "loss": 0.2703,
+      "step": 4643
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6377350447088984e-05,
+      "loss": 0.2134,
+      "step": 4644
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6375730498873453e-05,
+      "loss": 0.1993,
+      "step": 4645
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6374110268698482e-05,
+      "loss": 0.2307,
+      "step": 4646
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6372489756635724e-05,
+      "loss": 0.2537,
+      "step": 4647
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.637086896275684e-05,
+      "loss": 0.2087,
+      "step": 4648
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6369247887133517e-05,
+      "loss": 0.237,
+      "step": 4649
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.636762652983743e-05,
+      "loss": 0.2394,
+      "step": 4650
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6366004890940303e-05,
+      "loss": 0.2163,
+      "step": 4651
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6364382970513833e-05,
+      "loss": 0.2014,
+      "step": 4652
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6362760768629754e-05,
+      "loss": 0.2205,
+      "step": 4653
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6361138285359805e-05,
+      "loss": 0.1838,
+      "step": 4654
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.635951552077574e-05,
+      "loss": 0.2547,
+      "step": 4655
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6357892474949327e-05,
+      "loss": 0.2339,
+      "step": 4656
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6356269147952332e-05,
+      "loss": 0.2575,
+      "step": 4657
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6354645539856558e-05,
+      "loss": 0.2211,
+      "step": 4658
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6353021650733802e-05,
+      "loss": 0.2693,
+      "step": 4659
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.635139748065588e-05,
+      "loss": 0.2173,
+      "step": 4660
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.634977302969461e-05,
+      "loss": 0.2073,
+      "step": 4661
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6348148297921845e-05,
+      "loss": 0.241,
+      "step": 4662
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6346523285409425e-05,
+      "loss": 0.2532,
+      "step": 4663
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6344897992229226e-05,
+      "loss": 0.2102,
+      "step": 4664
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6343272418453117e-05,
+      "loss": 0.2419,
+      "step": 4665
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6341646564152988e-05,
+      "loss": 0.2186,
+      "step": 4666
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6340020429400745e-05,
+      "loss": 0.2156,
+      "step": 4667
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6338394014268293e-05,
+      "loss": 0.2097,
+      "step": 4668
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6336767318827568e-05,
+      "loss": 0.2384,
+      "step": 4669
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6335140343150504e-05,
+      "loss": 0.2066,
+      "step": 4670
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.633351308730905e-05,
+      "loss": 0.225,
+      "step": 4671
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6331885551375176e-05,
+      "loss": 0.225,
+      "step": 4672
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.633025773542085e-05,
+      "loss": 0.2479,
+      "step": 4673
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.632862963951807e-05,
+      "loss": 0.2354,
+      "step": 4674
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.632700126373883e-05,
+      "loss": 0.2238,
+      "step": 4675
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6325372608155143e-05,
+      "loss": 0.2205,
+      "step": 4676
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6323743672839035e-05,
+      "loss": 0.2285,
+      "step": 4677
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6322114457862544e-05,
+      "loss": 0.2327,
+      "step": 4678
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6320484963297723e-05,
+      "loss": 0.2191,
+      "step": 4679
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.631885518921663e-05,
+      "loss": 0.2167,
+      "step": 4680
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6317225135691346e-05,
+      "loss": 0.2097,
+      "step": 4681
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.631559480279395e-05,
+      "loss": 0.2456,
+      "step": 4682
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6313964190596553e-05,
+      "loss": 0.2353,
+      "step": 4683
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.631233329917125e-05,
+      "loss": 0.2174,
+      "step": 4684
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6310702128590184e-05,
+      "loss": 0.2173,
+      "step": 4685
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6309070678925477e-05,
+      "loss": 0.2494,
+      "step": 4686
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6307438950249292e-05,
+      "loss": 0.2038,
+      "step": 4687
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6305806942633777e-05,
+      "loss": 0.2199,
+      "step": 4688
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.630417465615111e-05,
+      "loss": 0.2124,
+      "step": 4689
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6302542090873483e-05,
+      "loss": 0.2111,
+      "step": 4690
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.630090924687308e-05,
+      "loss": 0.2231,
+      "step": 4691
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.629927612422213e-05,
+      "loss": 0.2051,
+      "step": 4692
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6297642722992846e-05,
+      "loss": 0.2219,
+      "step": 4693
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6296009043257464e-05,
+      "loss": 0.2449,
+      "step": 4694
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.629437508508823e-05,
+      "loss": 0.1904,
+      "step": 4695
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6292740848557407e-05,
+      "loss": 0.2281,
+      "step": 4696
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6291106333737267e-05,
+      "loss": 0.2203,
+      "step": 4697
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6289471540700095e-05,
+      "loss": 0.2038,
+      "step": 4698
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6287836469518184e-05,
+      "loss": 0.2137,
+      "step": 4699
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6286201120263848e-05,
+      "loss": 0.2389,
+      "step": 4700
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6284565493009405e-05,
+      "loss": 0.2296,
+      "step": 4701
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6282929587827193e-05,
+      "loss": 0.2343,
+      "step": 4702
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6281293404789548e-05,
+      "loss": 0.2038,
+      "step": 4703
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6279656943968843e-05,
+      "loss": 0.1993,
+      "step": 4704
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6278020205437437e-05,
+      "loss": 0.2065,
+      "step": 4705
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.627638318926772e-05,
+      "loss": 0.2064,
+      "step": 4706
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6274745895532084e-05,
+      "loss": 0.2277,
+      "step": 4707
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6273108324302932e-05,
+      "loss": 0.2147,
+      "step": 4708
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6271470475652693e-05,
+      "loss": 0.2315,
+      "step": 4709
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6269832349653793e-05,
+      "loss": 0.2574,
+      "step": 4710
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.626819394637868e-05,
+      "loss": 0.2215,
+      "step": 4711
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6266555265899805e-05,
+      "loss": 0.2166,
+      "step": 4712
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.626491630828964e-05,
+      "loss": 0.2217,
+      "step": 4713
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6263277073620666e-05,
+      "loss": 0.2513,
+      "step": 4714
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.626163756196538e-05,
+      "loss": 0.2229,
+      "step": 4715
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.625999777339628e-05,
+      "loss": 0.207,
+      "step": 4716
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6258357707985888e-05,
+      "loss": 0.2086,
+      "step": 4717
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6256717365806736e-05,
+      "loss": 0.2331,
+      "step": 4718
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.625507674693136e-05,
+      "loss": 0.209,
+      "step": 4719
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.625343585143232e-05,
+      "loss": 0.2164,
+      "step": 4720
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6251794679382182e-05,
+      "loss": 0.2152,
+      "step": 4721
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.625015323085352e-05,
+      "loss": 0.2344,
+      "step": 4722
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6248511505918938e-05,
+      "loss": 0.2417,
+      "step": 4723
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6246869504651026e-05,
+      "loss": 0.2297,
+      "step": 4724
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6245227227122404e-05,
+      "loss": 0.2064,
+      "step": 4725
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.62435846734057e-05,
+      "loss": 0.228,
+      "step": 4726
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6241941843573553e-05,
+      "loss": 0.1998,
+      "step": 4727
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.624029873769862e-05,
+      "loss": 0.2174,
+      "step": 4728
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6238655355853562e-05,
+      "loss": 0.2228,
+      "step": 4729
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6237011698111054e-05,
+      "loss": 0.247,
+      "step": 4730
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6235367764543787e-05,
+      "loss": 0.2211,
+      "step": 4731
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6233723555224462e-05,
+      "loss": 0.2254,
+      "step": 4732
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6232079070225793e-05,
+      "loss": 0.2279,
+      "step": 4733
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6230434309620503e-05,
+      "loss": 0.2003,
+      "step": 4734
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.622878927348133e-05,
+      "loss": 0.2458,
+      "step": 4735
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.622714396188103e-05,
+      "loss": 0.2186,
+      "step": 4736
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6225498374892358e-05,
+      "loss": 0.2444,
+      "step": 4737
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6223852512588088e-05,
+      "loss": 0.2273,
+      "step": 4738
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6222206375041007e-05,
+      "loss": 0.2214,
+      "step": 4739
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.622055996232392e-05,
+      "loss": 0.194,
+      "step": 4740
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.621891327450963e-05,
+      "loss": 0.2246,
+      "step": 4741
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.621726631167097e-05,
+      "loss": 0.2122,
+      "step": 4742
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6215619073880758e-05,
+      "loss": 0.2548,
+      "step": 4743
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6213971561211855e-05,
+      "loss": 0.2282,
+      "step": 4744
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.621232377373712e-05,
+      "loss": 0.2172,
+      "step": 4745
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6210675711529416e-05,
+      "loss": 0.1961,
+      "step": 4746
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6209027374661633e-05,
+      "loss": 0.2266,
+      "step": 4747
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.620737876320667e-05,
+      "loss": 0.236,
+      "step": 4748
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6205729877237426e-05,
+      "loss": 0.2216,
+      "step": 4749
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6204080716826824e-05,
+      "loss": 0.214,
+      "step": 4750
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6202431282047803e-05,
+      "loss": 0.2246,
+      "step": 4751
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.62007815729733e-05,
+      "loss": 0.2399,
+      "step": 4752
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6199131589676274e-05,
+      "loss": 0.2356,
+      "step": 4753
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6197481332229693e-05,
+      "loss": 0.219,
+      "step": 4754
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6195830800706537e-05,
+      "loss": 0.2577,
+      "step": 4755
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6194179995179803e-05,
+      "loss": 0.2237,
+      "step": 4756
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6192528915722496e-05,
+      "loss": 0.2101,
+      "step": 4757
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6190877562407622e-05,
+      "loss": 0.2218,
+      "step": 4758
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6189225935308223e-05,
+      "loss": 0.2294,
+      "step": 4759
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6187574034497334e-05,
+      "loss": 0.2144,
+      "step": 4760
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6185921860048013e-05,
+      "loss": 0.1968,
+      "step": 4761
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.618426941203332e-05,
+      "loss": 0.2292,
+      "step": 4762
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6182616690526338e-05,
+      "loss": 0.2594,
+      "step": 4763
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6180963695600155e-05,
+      "loss": 0.2117,
+      "step": 4764
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6179310427327868e-05,
+      "loss": 0.2158,
+      "step": 4765
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6177656885782596e-05,
+      "loss": 0.2118,
+      "step": 4766
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6176003071037466e-05,
+      "loss": 0.2032,
+      "step": 4767
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6174348983165613e-05,
+      "loss": 0.2147,
+      "step": 4768
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6172694622240187e-05,
+      "loss": 0.2061,
+      "step": 4769
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.617103998833435e-05,
+      "loss": 0.2312,
+      "step": 4770
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6169385081521284e-05,
+      "loss": 0.205,
+      "step": 4771
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6167729901874164e-05,
+      "loss": 0.2347,
+      "step": 4772
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6166074449466195e-05,
+      "loss": 0.2758,
+      "step": 4773
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6164418724370582e-05,
+      "loss": 0.2378,
+      "step": 4774
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6162762726660556e-05,
+      "loss": 0.24,
+      "step": 4775
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6161106456409345e-05,
+      "loss": 0.2346,
+      "step": 4776
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.61594499136902e-05,
+      "loss": 0.2388,
+      "step": 4777
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6157793098576374e-05,
+      "loss": 0.2097,
+      "step": 4778
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6156136011141144e-05,
+      "loss": 0.1998,
+      "step": 4779
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.615447865145779e-05,
+      "loss": 0.2277,
+      "step": 4780
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6152821019599603e-05,
+      "loss": 0.1983,
+      "step": 4781
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.61511631156399e-05,
+      "loss": 0.2393,
+      "step": 4782
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.614950493965199e-05,
+      "loss": 0.2216,
+      "step": 4783
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.614784649170921e-05,
+      "loss": 0.2227,
+      "step": 4784
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6146187771884898e-05,
+      "loss": 0.2204,
+      "step": 4785
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6144528780252413e-05,
+      "loss": 0.2305,
+      "step": 4786
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.614286951688512e-05,
+      "loss": 0.2129,
+      "step": 4787
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.61412099818564e-05,
+      "loss": 0.2483,
+      "step": 4788
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6139550175239645e-05,
+      "loss": 0.2184,
+      "step": 4789
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.613789009710825e-05,
+      "loss": 0.2281,
+      "step": 4790
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.613622974753564e-05,
+      "loss": 0.2204,
+      "step": 4791
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.613456912659524e-05,
+      "loss": 0.2053,
+      "step": 4792
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.613290823436048e-05,
+      "loss": 0.2341,
+      "step": 4793
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6131247070904824e-05,
+      "loss": 0.2354,
+      "step": 4794
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6129585636301726e-05,
+      "loss": 0.2192,
+      "step": 4795
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.612792393062467e-05,
+      "loss": 0.2016,
+      "step": 4796
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6126261953947128e-05,
+      "loss": 0.2342,
+      "step": 4797
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6124599706342613e-05,
+      "loss": 0.2731,
+      "step": 4798
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6122937187884632e-05,
+      "loss": 0.1939,
+      "step": 4799
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.612127439864671e-05,
+      "loss": 0.2137,
+      "step": 4800
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6119611338702378e-05,
+      "loss": 0.2459,
+      "step": 4801
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.611794800812518e-05,
+      "loss": 0.2254,
+      "step": 4802
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6116284406988682e-05,
+      "loss": 0.1991,
+      "step": 4803
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.611462053536645e-05,
+      "loss": 0.2004,
+      "step": 4804
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6112956393332074e-05,
+      "loss": 0.2261,
+      "step": 4805
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.611129198095914e-05,
+      "loss": 0.1713,
+      "step": 4806
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6109627298321256e-05,
+      "loss": 0.2355,
+      "step": 4807
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.610796234549205e-05,
+      "loss": 0.2218,
+      "step": 4808
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.610629712254514e-05,
+      "loss": 0.2199,
+      "step": 4809
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6104631629554175e-05,
+      "loss": 0.2314,
+      "step": 4810
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.610296586659281e-05,
+      "loss": 0.1909,
+      "step": 4811
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.610129983373471e-05,
+      "loss": 0.2676,
+      "step": 4812
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.609963353105356e-05,
+      "loss": 0.2092,
+      "step": 4813
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6097966958623034e-05,
+      "loss": 0.2003,
+      "step": 4814
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6096300116516846e-05,
+      "loss": 0.2241,
+      "step": 4815
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6094633004808715e-05,
+      "loss": 0.2463,
+      "step": 4816
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6092965623572355e-05,
+      "loss": 0.2272,
+      "step": 4817
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6091297972881516e-05,
+      "loss": 0.2347,
+      "step": 4818
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6089630052809934e-05,
+      "loss": 0.2085,
+      "step": 4819
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6087961863431382e-05,
+      "loss": 0.24,
+      "step": 4820
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6086293404819634e-05,
+      "loss": 0.2486,
+      "step": 4821
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6084624677048465e-05,
+      "loss": 0.2247,
+      "step": 4822
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6082955680191684e-05,
+      "loss": 0.1986,
+      "step": 4823
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6081286414323093e-05,
+      "loss": 0.2408,
+      "step": 4824
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6079616879516523e-05,
+      "loss": 0.2078,
+      "step": 4825
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6077947075845796e-05,
+      "loss": 0.2278,
+      "step": 4826
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.607627700338476e-05,
+      "loss": 0.2375,
+      "step": 4827
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.607460666220728e-05,
+      "loss": 0.2097,
+      "step": 4828
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6072936052387214e-05,
+      "loss": 0.2142,
+      "step": 4829
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.607126517399845e-05,
+      "loss": 0.227,
+      "step": 4830
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6069594027114876e-05,
+      "loss": 0.2117,
+      "step": 4831
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.60679226118104e-05,
+      "loss": 0.2006,
+      "step": 4832
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.606625092815894e-05,
+      "loss": 0.2339,
+      "step": 4833
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.606457897623442e-05,
+      "loss": 0.2469,
+      "step": 4834
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6062906756110782e-05,
+      "loss": 0.2129,
+      "step": 4835
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6061234267861977e-05,
+      "loss": 0.2591,
+      "step": 4836
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6059561511561975e-05,
+      "loss": 0.2079,
+      "step": 4837
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6057888487284737e-05,
+      "loss": 0.2461,
+      "step": 4838
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.605621519510427e-05,
+      "loss": 0.2053,
+      "step": 4839
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.605454163509456e-05,
+      "loss": 0.2241,
+      "step": 4840
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6052867807329623e-05,
+      "loss": 0.2072,
+      "step": 4841
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6051193711883478e-05,
+      "loss": 0.2045,
+      "step": 4842
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.604951934883016e-05,
+      "loss": 0.2194,
+      "step": 4843
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6047844718243732e-05,
+      "loss": 0.2135,
+      "step": 4844
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.604616982019823e-05,
+      "loss": 0.1829,
+      "step": 4845
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6044494654767736e-05,
+      "loss": 0.2248,
+      "step": 4846
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.604281922202633e-05,
+      "loss": 0.211,
+      "step": 4847
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6041143522048105e-05,
+      "loss": 0.2102,
+      "step": 4848
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6039467554907168e-05,
+      "loss": 0.213,
+      "step": 4849
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.603779132067764e-05,
+      "loss": 0.2131,
+      "step": 4850
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6036114819433648e-05,
+      "loss": 0.2263,
+      "step": 4851
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.603443805124933e-05,
+      "loss": 0.2051,
+      "step": 4852
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6032761016198844e-05,
+      "loss": 0.2098,
+      "step": 4853
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6031083714356352e-05,
+      "loss": 0.2436,
+      "step": 4854
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6029406145796033e-05,
+      "loss": 0.2529,
+      "step": 4855
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6027728310592073e-05,
+      "loss": 0.2065,
+      "step": 4856
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6026050208818676e-05,
+      "loss": 0.2481,
+      "step": 4857
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.602437184055005e-05,
+      "loss": 0.2156,
+      "step": 4858
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.602269320586042e-05,
+      "loss": 0.2235,
+      "step": 4859
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6021014304824026e-05,
+      "loss": 0.2141,
+      "step": 4860
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.601933513751511e-05,
+      "loss": 0.2946,
+      "step": 4861
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6017655704007935e-05,
+      "loss": 0.2228,
+      "step": 4862
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.601597600437677e-05,
+      "loss": 0.2125,
+      "step": 4863
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6014296038695896e-05,
+      "loss": 0.2215,
+      "step": 4864
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.601261580703961e-05,
+      "loss": 0.2014,
+      "step": 4865
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6010935309482222e-05,
+      "loss": 0.2108,
+      "step": 4866
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6009254546098042e-05,
+      "loss": 0.2094,
+      "step": 4867
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6007573516961408e-05,
+      "loss": 0.243,
+      "step": 4868
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6005892222146656e-05,
+      "loss": 0.2407,
+      "step": 4869
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6004210661728138e-05,
+      "loss": 0.2099,
+      "step": 4870
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6002528835780228e-05,
+      "loss": 0.1731,
+      "step": 4871
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.600084674437729e-05,
+      "loss": 0.2071,
+      "step": 4872
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5999164387593724e-05,
+      "loss": 0.2364,
+      "step": 4873
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5997481765503927e-05,
+      "loss": 0.2037,
+      "step": 4874
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5995798878182308e-05,
+      "loss": 0.2203,
+      "step": 4875
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.599411572570329e-05,
+      "loss": 0.215,
+      "step": 4876
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5992432308141313e-05,
+      "loss": 0.2019,
+      "step": 4877
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5990748625570827e-05,
+      "loss": 0.2591,
+      "step": 4878
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.598906467806628e-05,
+      "loss": 0.2043,
+      "step": 4879
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5987380465702154e-05,
+      "loss": 0.2511,
+      "step": 4880
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5985695988552927e-05,
+      "loss": 0.2127,
+      "step": 4881
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.598401124669309e-05,
+      "loss": 0.204,
+      "step": 4882
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.598232624019715e-05,
+      "loss": 0.2172,
+      "step": 4883
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5980640969139628e-05,
+      "loss": 0.2346,
+      "step": 4884
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.597895543359505e-05,
+      "loss": 0.2558,
+      "step": 4885
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5977269633637957e-05,
+      "loss": 0.2178,
+      "step": 4886
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5975583569342903e-05,
+      "loss": 0.2373,
+      "step": 4887
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.597389724078445e-05,
+      "loss": 0.2285,
+      "step": 4888
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.597221064803718e-05,
+      "loss": 0.2623,
+      "step": 4889
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5970523791175677e-05,
+      "loss": 0.2397,
+      "step": 4890
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5968836670274536e-05,
+      "loss": 0.2118,
+      "step": 4891
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5967149285408374e-05,
+      "loss": 0.2075,
+      "step": 4892
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5965461636651808e-05,
+      "loss": 0.204,
+      "step": 4893
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.596377372407948e-05,
+      "loss": 0.2207,
+      "step": 4894
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5962085547766032e-05,
+      "loss": 0.1951,
+      "step": 4895
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5960397107786117e-05,
+      "loss": 0.2052,
+      "step": 4896
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.595870840421441e-05,
+      "loss": 0.2129,
+      "step": 4897
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5957019437125595e-05,
+      "loss": 0.201,
+      "step": 4898
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5955330206594356e-05,
+      "loss": 0.2346,
+      "step": 4899
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5953640712695404e-05,
+      "loss": 0.2184,
+      "step": 4900
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5951950955503453e-05,
+      "loss": 0.224,
+      "step": 4901
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.595026093509323e-05,
+      "loss": 0.242,
+      "step": 4902
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5948570651539474e-05,
+      "loss": 0.2129,
+      "step": 4903
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5946880104916936e-05,
+      "loss": 0.2721,
+      "step": 4904
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.594518929530038e-05,
+      "loss": 0.1989,
+      "step": 4905
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5943498222764578e-05,
+      "loss": 0.203,
+      "step": 4906
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5941806887384313e-05,
+      "loss": 0.2083,
+      "step": 4907
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5940115289234393e-05,
+      "loss": 0.2083,
+      "step": 4908
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5938423428389616e-05,
+      "loss": 0.2417,
+      "step": 4909
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.593673130492481e-05,
+      "loss": 0.1992,
+      "step": 4910
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5935038918914806e-05,
+      "loss": 0.2087,
+      "step": 4911
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.593334627043444e-05,
+      "loss": 0.2284,
+      "step": 4912
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5931653359558578e-05,
+      "loss": 0.2583,
+      "step": 4913
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5929960186362077e-05,
+      "loss": 0.1987,
+      "step": 4914
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5928266750919826e-05,
+      "loss": 0.2176,
+      "step": 4915
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5926573053306708e-05,
+      "loss": 0.2301,
+      "step": 4916
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.592487909359763e-05,
+      "loss": 0.2148,
+      "step": 4917
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.59231848718675e-05,
+      "loss": 0.2159,
+      "step": 4918
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5921490388191245e-05,
+      "loss": 0.2234,
+      "step": 4919
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5919795642643805e-05,
+      "loss": 0.223,
+      "step": 4920
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.591810063530012e-05,
+      "loss": 0.2021,
+      "step": 4921
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5916405366235165e-05,
+      "loss": 0.1958,
+      "step": 4922
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.591470983552389e-05,
+      "loss": 0.2131,
+      "step": 4923
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.59130140432413e-05,
+      "loss": 0.2121,
+      "step": 4924
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5911317989462375e-05,
+      "loss": 0.215,
+      "step": 4925
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5909621674262124e-05,
+      "loss": 0.2307,
+      "step": 4926
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5907925097715566e-05,
+      "loss": 0.2184,
+      "step": 4927
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5906228259897728e-05,
+      "loss": 0.2191,
+      "step": 4928
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5904531160883657e-05,
+      "loss": 0.2352,
+      "step": 4929
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5902833800748397e-05,
+      "loss": 0.1972,
+      "step": 4930
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5901136179567015e-05,
+      "loss": 0.2114,
+      "step": 4931
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5899438297414592e-05,
+      "loss": 0.2141,
+      "step": 4932
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.589774015436621e-05,
+      "loss": 0.2254,
+      "step": 4933
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5896041750496962e-05,
+      "loss": 0.2416,
+      "step": 4934
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5894343085881967e-05,
+      "loss": 0.2363,
+      "step": 4935
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.589264416059634e-05,
+      "loss": 0.23,
+      "step": 4936
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.589094497471522e-05,
+      "loss": 0.2312,
+      "step": 4937
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5889245528313748e-05,
+      "loss": 0.2141,
+      "step": 4938
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.588754582146708e-05,
+      "loss": 0.2108,
+      "step": 4939
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5885845854250382e-05,
+      "loss": 0.2526,
+      "step": 4940
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.588414562673884e-05,
+      "loss": 0.2244,
+      "step": 4941
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.588244513900764e-05,
+      "loss": 0.215,
+      "step": 4942
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.588074439113198e-05,
+      "loss": 0.2382,
+      "step": 4943
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.587904338318708e-05,
+      "loss": 0.2137,
+      "step": 4944
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5877342115248163e-05,
+      "loss": 0.2334,
+      "step": 4945
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5875640587390468e-05,
+      "loss": 0.2286,
+      "step": 4946
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5873938799689235e-05,
+      "loss": 0.2278,
+      "step": 4947
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.587223675221973e-05,
+      "loss": 0.1925,
+      "step": 4948
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5870534445057226e-05,
+      "loss": 0.2108,
+      "step": 4949
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5868831878277002e-05,
+      "loss": 0.2418,
+      "step": 4950
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5867129051954357e-05,
+      "loss": 0.2594,
+      "step": 4951
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5865425966164588e-05,
+      "loss": 0.2285,
+      "step": 4952
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5863722620983017e-05,
+      "loss": 0.2256,
+      "step": 4953
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5862019016484972e-05,
+      "loss": 0.226,
+      "step": 4954
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5860315152745793e-05,
+      "loss": 0.2344,
+      "step": 4955
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5858611029840836e-05,
+      "loss": 0.2019,
+      "step": 4956
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5856906647845452e-05,
+      "loss": 0.1769,
+      "step": 4957
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.585520200683503e-05,
+      "loss": 0.2333,
+      "step": 4958
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5853497106884948e-05,
+      "loss": 0.1865,
+      "step": 4959
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.58517919480706e-05,
+      "loss": 0.2236,
+      "step": 4960
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.58500865304674e-05,
+      "loss": 0.2153,
+      "step": 4961
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.584838085415077e-05,
+      "loss": 0.2704,
+      "step": 4962
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5846674919196137e-05,
+      "loss": 0.2294,
+      "step": 4963
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.584496872567894e-05,
+      "loss": 0.214,
+      "step": 4964
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5843262273674645e-05,
+      "loss": 0.2289,
+      "step": 4965
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.584155556325871e-05,
+      "loss": 0.2409,
+      "step": 4966
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5839848594506617e-05,
+      "loss": 0.2096,
+      "step": 4967
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5838141367493845e-05,
+      "loss": 0.2005,
+      "step": 4968
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5836433882295905e-05,
+      "loss": 0.2361,
+      "step": 4969
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.583472613898831e-05,
+      "loss": 0.2337,
+      "step": 4970
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.583301813764657e-05,
+      "loss": 0.2064,
+      "step": 4971
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5831309878346227e-05,
+      "loss": 0.2156,
+      "step": 4972
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5829601361162832e-05,
+      "loss": 0.2679,
+      "step": 4973
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5827892586171936e-05,
+      "loss": 0.1951,
+      "step": 4974
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5826183553449106e-05,
+      "loss": 0.2296,
+      "step": 4975
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5824474263069927e-05,
+      "loss": 0.214,
+      "step": 4976
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5822764715109987e-05,
+      "loss": 0.2228,
+      "step": 4977
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5821054909644892e-05,
+      "loss": 0.2218,
+      "step": 4978
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.581934484675025e-05,
+      "loss": 0.2457,
+      "step": 4979
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5817634526501696e-05,
+      "loss": 0.2477,
+      "step": 4980
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5815923948974862e-05,
+      "loss": 0.2072,
+      "step": 4981
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5814213114245392e-05,
+      "loss": 0.2201,
+      "step": 4982
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5812502022388955e-05,
+      "loss": 0.2038,
+      "step": 4983
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5810790673481213e-05,
+      "loss": 0.2584,
+      "step": 4984
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.580907906759785e-05,
+      "loss": 0.2272,
+      "step": 4985
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.580736720481457e-05,
+      "loss": 0.2374,
+      "step": 4986
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5805655085207063e-05,
+      "loss": 0.1888,
+      "step": 4987
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5803942708851056e-05,
+      "loss": 0.2294,
+      "step": 4988
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5802230075822274e-05,
+      "loss": 0.2051,
+      "step": 4989
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.580051718619646e-05,
+      "loss": 0.2239,
+      "step": 4990
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5798804040049354e-05,
+      "loss": 0.211,
+      "step": 4991
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5797090637456726e-05,
+      "loss": 0.2004,
+      "step": 4992
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.579537697849435e-05,
+      "loss": 0.2426,
+      "step": 4993
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.579366306323801e-05,
+      "loss": 0.2158,
+      "step": 4994
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.57919488917635e-05,
+      "loss": 0.222,
+      "step": 4995
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5790234464146624e-05,
+      "loss": 0.236,
+      "step": 4996
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.578851978046321e-05,
+      "loss": 0.1986,
+      "step": 4997
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.578680484078908e-05,
+      "loss": 0.1957,
+      "step": 4998
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5785089645200075e-05,
+      "loss": 0.1939,
+      "step": 4999
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5783374193772056e-05,
+      "loss": 0.2183,
+      "step": 5000
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.578165848658088e-05,
+      "loss": 0.2144,
+      "step": 5001
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5779942523702415e-05,
+      "loss": 0.224,
+      "step": 5002
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5778226305212566e-05,
+      "loss": 0.2174,
+      "step": 5003
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5776509831187216e-05,
+      "loss": 0.2588,
+      "step": 5004
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5774793101702278e-05,
+      "loss": 0.1892,
+      "step": 5005
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5773076116833674e-05,
+      "loss": 0.2121,
+      "step": 5006
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5771358876657336e-05,
+      "loss": 0.2305,
+      "step": 5007
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.576964138124921e-05,
+      "loss": 0.2204,
+      "step": 5008
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5767923630685237e-05,
+      "loss": 0.2225,
+      "step": 5009
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.57662056250414e-05,
+      "loss": 0.2219,
+      "step": 5010
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.576448736439366e-05,
+      "loss": 0.2204,
+      "step": 5011
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.576276884881802e-05,
+      "loss": 0.2206,
+      "step": 5012
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5761050078390467e-05,
+      "loss": 0.219,
+      "step": 5013
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.575933105318702e-05,
+      "loss": 0.231,
+      "step": 5014
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5757611773283696e-05,
+      "loss": 0.2361,
+      "step": 5015
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5755892238756532e-05,
+      "loss": 0.2189,
+      "step": 5016
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5754172449681568e-05,
+      "loss": 0.2721,
+      "step": 5017
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5752452406134864e-05,
+      "loss": 0.2524,
+      "step": 5018
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5750732108192483e-05,
+      "loss": 0.2154,
+      "step": 5019
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5749011555930506e-05,
+      "loss": 0.2078,
+      "step": 5020
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.574729074942502e-05,
+      "loss": 0.2049,
+      "step": 5021
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.574556968875213e-05,
+      "loss": 0.2432,
+      "step": 5022
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5743848373987948e-05,
+      "loss": 0.2254,
+      "step": 5023
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5742126805208592e-05,
+      "loss": 0.2229,
+      "step": 5024
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.57404049824902e-05,
+      "loss": 0.2213,
+      "step": 5025
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5738682905908913e-05,
+      "loss": 0.1992,
+      "step": 5026
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5736960575540892e-05,
+      "loss": 0.208,
+      "step": 5027
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.573523799146231e-05,
+      "loss": 0.2131,
+      "step": 5028
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5733515153749337e-05,
+      "loss": 0.2214,
+      "step": 5029
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.573179206247817e-05,
+      "loss": 0.237,
+      "step": 5030
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5730068717725005e-05,
+      "loss": 0.2517,
+      "step": 5031
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.572834511956606e-05,
+      "loss": 0.2215,
+      "step": 5032
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5726621268077556e-05,
+      "loss": 0.2444,
+      "step": 5033
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5724897163335732e-05,
+      "loss": 0.2222,
+      "step": 5034
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.572317280541683e-05,
+      "loss": 0.2021,
+      "step": 5035
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5721448194397114e-05,
+      "loss": 0.1825,
+      "step": 5036
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.571972333035285e-05,
+      "loss": 0.2198,
+      "step": 5037
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5717998213360314e-05,
+      "loss": 0.219,
+      "step": 5038
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.57162728434958e-05,
+      "loss": 0.2051,
+      "step": 5039
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5714547220835613e-05,
+      "loss": 0.2067,
+      "step": 5040
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5712821345456062e-05,
+      "loss": 0.2152,
+      "step": 5041
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.571109521743348e-05,
+      "loss": 0.1788,
+      "step": 5042
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5709368836844193e-05,
+      "loss": 0.2125,
+      "step": 5043
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5707642203764553e-05,
+      "loss": 0.2431,
+      "step": 5044
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.570591531827092e-05,
+      "loss": 0.2401,
+      "step": 5045
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5704188180439666e-05,
+      "loss": 0.2257,
+      "step": 5046
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.570246079034716e-05,
+      "loss": 0.245,
+      "step": 5047
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.570073314806981e-05,
+      "loss": 0.219,
+      "step": 5048
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5699005253684e-05,
+      "loss": 0.2032,
+      "step": 5049
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5697277107266162e-05,
+      "loss": 0.2195,
+      "step": 5050
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5695548708892713e-05,
+      "loss": 0.222,
+      "step": 5051
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.569382005864009e-05,
+      "loss": 0.2217,
+      "step": 5052
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5692091156584744e-05,
+      "loss": 0.246,
+      "step": 5053
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5690362002803124e-05,
+      "loss": 0.2465,
+      "step": 5054
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5688632597371712e-05,
+      "loss": 0.2683,
+      "step": 5055
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5686902940366985e-05,
+      "loss": 0.2038,
+      "step": 5056
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.568517303186543e-05,
+      "loss": 0.2219,
+      "step": 5057
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5683442871943554e-05,
+      "loss": 0.2015,
+      "step": 5058
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.568171246067787e-05,
+      "loss": 0.199,
+      "step": 5059
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5679981798144906e-05,
+      "loss": 0.1993,
+      "step": 5060
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.56782508844212e-05,
+      "loss": 0.2174,
+      "step": 5061
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5676519719583295e-05,
+      "loss": 0.2416,
+      "step": 5062
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5674788303707754e-05,
+      "loss": 0.2236,
+      "step": 5063
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5673056636871144e-05,
+      "loss": 0.2309,
+      "step": 5064
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5671324719150047e-05,
+      "loss": 0.2159,
+      "step": 5065
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5669592550621057e-05,
+      "loss": 0.2026,
+      "step": 5066
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5667860131360766e-05,
+      "loss": 0.2061,
+      "step": 5067
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5666127461445806e-05,
+      "loss": 0.2424,
+      "step": 5068
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5664394540952793e-05,
+      "loss": 0.2257,
+      "step": 5069
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5662661369958362e-05,
+      "loss": 0.2127,
+      "step": 5070
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5660927948539165e-05,
+      "loss": 0.2259,
+      "step": 5071
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5659194276771855e-05,
+      "loss": 0.2133,
+      "step": 5072
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5657460354733107e-05,
+      "loss": 0.2633,
+      "step": 5073
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.56557261824996e-05,
+      "loss": 0.2323,
+      "step": 5074
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5653991760148023e-05,
+      "loss": 0.2147,
+      "step": 5075
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5652257087755085e-05,
+      "loss": 0.2047,
+      "step": 5076
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5650522165397493e-05,
+      "loss": 0.1781,
+      "step": 5077
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5648786993151975e-05,
+      "loss": 0.1978,
+      "step": 5078
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5647051571095272e-05,
+      "loss": 0.2184,
+      "step": 5079
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5645315899304125e-05,
+      "loss": 0.2175,
+      "step": 5080
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5643579977855288e-05,
+      "loss": 0.2374,
+      "step": 5081
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.564184380682554e-05,
+      "loss": 0.2174,
+      "step": 5082
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5640107386291658e-05,
+      "loss": 0.2044,
+      "step": 5083
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5638370716330428e-05,
+      "loss": 0.2048,
+      "step": 5084
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5636633797018656e-05,
+      "loss": 0.2397,
+      "step": 5085
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5634896628433162e-05,
+      "loss": 0.239,
+      "step": 5086
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5633159210650757e-05,
+      "loss": 0.2346,
+      "step": 5087
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5631421543748286e-05,
+      "loss": 0.2404,
+      "step": 5088
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5629683627802593e-05,
+      "loss": 0.2131,
+      "step": 5089
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5627945462890536e-05,
+      "loss": 0.2219,
+      "step": 5090
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5626207049088977e-05,
+      "loss": 0.229,
+      "step": 5091
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5624468386474806e-05,
+      "loss": 0.2163,
+      "step": 5092
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5622729475124905e-05,
+      "loss": 0.2096,
+      "step": 5093
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.562099031511618e-05,
+      "loss": 0.2053,
+      "step": 5094
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.561925090652554e-05,
+      "loss": 0.1773,
+      "step": 5095
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5617511249429913e-05,
+      "loss": 0.2234,
+      "step": 5096
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.561577134390623e-05,
+      "loss": 0.2168,
+      "step": 5097
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5614031190031433e-05,
+      "loss": 0.214,
+      "step": 5098
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5612290787882487e-05,
+      "loss": 0.227,
+      "step": 5099
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5610550137536354e-05,
+      "loss": 0.1925,
+      "step": 5100
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5608809239070008e-05,
+      "loss": 0.204,
+      "step": 5101
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5607068092560448e-05,
+      "loss": 0.2354,
+      "step": 5102
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.560532669808467e-05,
+      "loss": 0.2134,
+      "step": 5103
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.560358505571968e-05,
+      "loss": 0.2137,
+      "step": 5104
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5601843165542504e-05,
+      "loss": 0.2045,
+      "step": 5105
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.560010102763018e-05,
+      "loss": 0.2277,
+      "step": 5106
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5598358642059744e-05,
+      "loss": 0.2166,
+      "step": 5107
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.559661600890826e-05,
+      "loss": 0.2355,
+      "step": 5108
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5594873128252784e-05,
+      "loss": 0.2393,
+      "step": 5109
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5593130000170403e-05,
+      "loss": 0.173,
+      "step": 5110
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5591386624738193e-05,
+      "loss": 0.2236,
+      "step": 5111
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5589643002033264e-05,
+      "loss": 0.2501,
+      "step": 5112
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.558789913213272e-05,
+      "loss": 0.2068,
+      "step": 5113
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5586155015113684e-05,
+      "loss": 0.2154,
+      "step": 5114
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.558441065105328e-05,
+      "loss": 0.2184,
+      "step": 5115
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5582666040028662e-05,
+      "loss": 0.2167,
+      "step": 5116
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.558092118211698e-05,
+      "loss": 0.2089,
+      "step": 5117
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5579176077395392e-05,
+      "loss": 0.21,
+      "step": 5118
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.557743072594108e-05,
+      "loss": 0.2193,
+      "step": 5119
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.557568512783123e-05,
+      "loss": 0.2124,
+      "step": 5120
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5573939283143036e-05,
+      "loss": 0.202,
+      "step": 5121
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5572193191953706e-05,
+      "loss": 0.2278,
+      "step": 5122
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5570446854340465e-05,
+      "loss": 0.2324,
+      "step": 5123
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5568700270380532e-05,
+      "loss": 0.2171,
+      "step": 5124
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5566953440151157e-05,
+      "loss": 0.213,
+      "step": 5125
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5565206363729582e-05,
+      "loss": 0.2037,
+      "step": 5126
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5563459041193085e-05,
+      "loss": 0.1896,
+      "step": 5127
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5561711472618923e-05,
+      "loss": 0.1744,
+      "step": 5128
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5559963658084392e-05,
+      "loss": 0.2187,
+      "step": 5129
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.555821559766678e-05,
+      "loss": 0.2551,
+      "step": 5130
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5556467291443396e-05,
+      "loss": 0.2136,
+      "step": 5131
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.555471873949155e-05,
+      "loss": 0.2241,
+      "step": 5132
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5552969941888586e-05,
+      "loss": 0.2037,
+      "step": 5133
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.555122089871183e-05,
+      "loss": 0.199,
+      "step": 5134
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5549471610038632e-05,
+      "loss": 0.1889,
+      "step": 5135
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5547722075946353e-05,
+      "loss": 0.2068,
+      "step": 5136
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5545972296512367e-05,
+      "loss": 0.1691,
+      "step": 5137
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5544222271814055e-05,
+      "loss": 0.2012,
+      "step": 5138
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5542472001928808e-05,
+      "loss": 0.1923,
+      "step": 5139
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5540721486934033e-05,
+      "loss": 0.211,
+      "step": 5140
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.553897072690714e-05,
+      "loss": 0.2184,
+      "step": 5141
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5537219721925557e-05,
+      "loss": 0.1912,
+      "step": 5142
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5535468472066725e-05,
+      "loss": 0.2159,
+      "step": 5143
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5533716977408083e-05,
+      "loss": 0.1936,
+      "step": 5144
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5531965238027087e-05,
+      "loss": 0.2018,
+      "step": 5145
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5530213254001215e-05,
+      "loss": 0.207,
+      "step": 5146
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5528461025407945e-05,
+      "loss": 0.2144,
+      "step": 5147
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.552670855232476e-05,
+      "loss": 0.2122,
+      "step": 5148
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5524955834829166e-05,
+      "loss": 0.2122,
+      "step": 5149
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.552320287299868e-05,
+      "loss": 0.1932,
+      "step": 5150
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5521449666910816e-05,
+      "loss": 0.2184,
+      "step": 5151
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.551969621664311e-05,
+      "loss": 0.1757,
+      "step": 5152
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5517942522273106e-05,
+      "loss": 0.2288,
+      "step": 5153
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5516188583878363e-05,
+      "loss": 0.2022,
+      "step": 5154
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.551443440153644e-05,
+      "loss": 0.172,
+      "step": 5155
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5512679975324924e-05,
+      "loss": 0.2228,
+      "step": 5156
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5510925305321394e-05,
+      "loss": 0.1932,
+      "step": 5157
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5509170391603452e-05,
+      "loss": 0.1883,
+      "step": 5158
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5507415234248704e-05,
+      "loss": 0.161,
+      "step": 5159
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5505659833334772e-05,
+      "loss": 0.2238,
+      "step": 5160
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5503904188939288e-05,
+      "loss": 0.1944,
+      "step": 5161
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.550214830113989e-05,
+      "loss": 0.1673,
+      "step": 5162
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.550039217001423e-05,
+      "loss": 0.1774,
+      "step": 5163
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.549863579563998e-05,
+      "loss": 0.2194,
+      "step": 5164
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5496879178094803e-05,
+      "loss": 0.1906,
+      "step": 5165
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5495122317456382e-05,
+      "loss": 0.185,
+      "step": 5166
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.549336521380242e-05,
+      "loss": 0.2083,
+      "step": 5167
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5491607867210622e-05,
+      "loss": 0.1869,
+      "step": 5168
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.54898502777587e-05,
+      "loss": 0.215,
+      "step": 5169
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5488092445524385e-05,
+      "loss": 0.2143,
+      "step": 5170
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5486334370585416e-05,
+      "loss": 0.2187,
+      "step": 5171
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5484576053019534e-05,
+      "loss": 0.1895,
+      "step": 5172
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5482817492904508e-05,
+      "loss": 0.2177,
+      "step": 5173
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.548105869031811e-05,
+      "loss": 0.1935,
+      "step": 5174
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.547929964533811e-05,
+      "loss": 0.1933,
+      "step": 5175
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5477540358042306e-05,
+      "loss": 0.2235,
+      "step": 5176
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5475780828508503e-05,
+      "loss": 0.2195,
+      "step": 5177
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5474021056814513e-05,
+      "loss": 0.222,
+      "step": 5178
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.547226104303815e-05,
+      "loss": 0.2014,
+      "step": 5179
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5470500787257267e-05,
+      "loss": 0.2081,
+      "step": 5180
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5468740289549694e-05,
+      "loss": 0.169,
+      "step": 5181
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.546697954999329e-05,
+      "loss": 0.1674,
+      "step": 5182
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5465218568665928e-05,
+      "loss": 0.1953,
+      "step": 5183
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.546345734564548e-05,
+      "loss": 0.1777,
+      "step": 5184
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.546169588100984e-05,
+      "loss": 0.1877,
+      "step": 5185
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5459934174836894e-05,
+      "loss": 0.1682,
+      "step": 5186
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5458172227204564e-05,
+      "loss": 0.202,
+      "step": 5187
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5456410038190765e-05,
+      "loss": 0.1852,
+      "step": 5188
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.545464760787343e-05,
+      "loss": 0.2099,
+      "step": 5189
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5452884936330496e-05,
+      "loss": 0.2006,
+      "step": 5190
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5451122023639923e-05,
+      "loss": 0.2295,
+      "step": 5191
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5449358869879665e-05,
+      "loss": 0.2086,
+      "step": 5192
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5447595475127703e-05,
+      "loss": 0.2198,
+      "step": 5193
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5445831839462013e-05,
+      "loss": 0.205,
+      "step": 5194
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5444067962960597e-05,
+      "loss": 0.1986,
+      "step": 5195
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.544230384570146e-05,
+      "loss": 0.2223,
+      "step": 5196
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.544053948776261e-05,
+      "loss": 0.1768,
+      "step": 5197
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5438774889222087e-05,
+      "loss": 0.2223,
+      "step": 5198
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5437010050157915e-05,
+      "loss": 0.1998,
+      "step": 5199
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.543524497064815e-05,
+      "loss": 0.1926,
+      "step": 5200
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.543347965077085e-05,
+      "loss": 0.1976,
+      "step": 5201
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5431714090604084e-05,
+      "loss": 0.1923,
+      "step": 5202
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5429948290225926e-05,
+      "loss": 0.2038,
+      "step": 5203
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5428182249714474e-05,
+      "loss": 0.1989,
+      "step": 5204
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5426415969147827e-05,
+      "loss": 0.2002,
+      "step": 5205
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5424649448604097e-05,
+      "loss": 0.2007,
+      "step": 5206
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5422882688161406e-05,
+      "loss": 0.209,
+      "step": 5207
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5421115687897884e-05,
+      "loss": 0.1592,
+      "step": 5208
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.541934844789168e-05,
+      "loss": 0.206,
+      "step": 5209
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.541758096822094e-05,
+      "loss": 0.1991,
+      "step": 5210
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5415813248963842e-05,
+      "loss": 0.1904,
+      "step": 5211
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5414045290198548e-05,
+      "loss": 0.2293,
+      "step": 5212
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5412277092003254e-05,
+      "loss": 0.2034,
+      "step": 5213
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5410508654456147e-05,
+      "loss": 0.2092,
+      "step": 5214
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5408739977635443e-05,
+      "loss": 0.1927,
+      "step": 5215
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.540697106161935e-05,
+      "loss": 0.1787,
+      "step": 5216
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.540520190648611e-05,
+      "loss": 0.2122,
+      "step": 5217
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.540343251231395e-05,
+      "loss": 0.2377,
+      "step": 5218
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5401662879181122e-05,
+      "loss": 0.2134,
+      "step": 5219
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.539989300716589e-05,
+      "loss": 0.176,
+      "step": 5220
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5398122896346522e-05,
+      "loss": 0.2133,
+      "step": 5221
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5396352546801302e-05,
+      "loss": 0.1943,
+      "step": 5222
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5394581958608516e-05,
+      "loss": 0.2198,
+      "step": 5223
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5392811131846465e-05,
+      "loss": 0.2022,
+      "step": 5224
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5391040066593474e-05,
+      "loss": 0.2053,
+      "step": 5225
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5389268762927855e-05,
+      "loss": 0.2049,
+      "step": 5226
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5387497220927944e-05,
+      "loss": 0.1788,
+      "step": 5227
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5385725440672085e-05,
+      "loss": 0.2019,
+      "step": 5228
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.538395342223864e-05,
+      "loss": 0.2404,
+      "step": 5229
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5382181165705965e-05,
+      "loss": 0.2136,
+      "step": 5230
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5380408671152442e-05,
+      "loss": 0.2156,
+      "step": 5231
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5378635938656455e-05,
+      "loss": 0.1733,
+      "step": 5232
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.53768629682964e-05,
+      "loss": 0.2187,
+      "step": 5233
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.537508976015069e-05,
+      "loss": 0.1923,
+      "step": 5234
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.537331631429774e-05,
+      "loss": 0.1959,
+      "step": 5235
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5371542630815975e-05,
+      "loss": 0.2087,
+      "step": 5236
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5369768709783837e-05,
+      "loss": 0.2052,
+      "step": 5237
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.536799455127978e-05,
+      "loss": 0.2453,
+      "step": 5238
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5366220155382258e-05,
+      "loss": 0.191,
+      "step": 5239
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5364445522169738e-05,
+      "loss": 0.2028,
+      "step": 5240
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5362670651720715e-05,
+      "loss": 0.2079,
+      "step": 5241
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.536089554411367e-05,
+      "loss": 0.1685,
+      "step": 5242
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.535912019942711e-05,
+      "loss": 0.1787,
+      "step": 5243
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5357344617739543e-05,
+      "loss": 0.214,
+      "step": 5244
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5355568799129495e-05,
+      "loss": 0.1991,
+      "step": 5245
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5353792743675497e-05,
+      "loss": 0.1747,
+      "step": 5246
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5352016451456094e-05,
+      "loss": 0.2011,
+      "step": 5247
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5350239922549842e-05,
+      "loss": 0.2359,
+      "step": 5248
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.534846315703531e-05,
+      "loss": 0.163,
+      "step": 5249
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5346686154991063e-05,
+      "loss": 0.2125,
+      "step": 5250
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5344908916495698e-05,
+      "loss": 0.1854,
+      "step": 5251
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.53431314416278e-05,
+      "loss": 0.1678,
+      "step": 5252
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5341353730465987e-05,
+      "loss": 0.2095,
+      "step": 5253
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.533957578308887e-05,
+      "loss": 0.1717,
+      "step": 5254
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5337797599575076e-05,
+      "loss": 0.197,
+      "step": 5255
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5336019180003248e-05,
+      "loss": 0.1759,
+      "step": 5256
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5334240524452032e-05,
+      "loss": 0.1886,
+      "step": 5257
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5332461633000084e-05,
+      "loss": 0.2088,
+      "step": 5258
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5330682505726072e-05,
+      "loss": 0.2081,
+      "step": 5259
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.532890314270869e-05,
+      "loss": 0.2661,
+      "step": 5260
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5327123544026613e-05,
+      "loss": 0.1802,
+      "step": 5261
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5325343709758547e-05,
+      "loss": 0.2205,
+      "step": 5262
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.53235636399832e-05,
+      "loss": 0.2022,
+      "step": 5263
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.53217833347793e-05,
+      "loss": 0.2151,
+      "step": 5264
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5320002794225577e-05,
+      "loss": 0.1938,
+      "step": 5265
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5318222018400773e-05,
+      "loss": 0.1924,
+      "step": 5266
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5316441007383637e-05,
+      "loss": 0.1976,
+      "step": 5267
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5314659761252933e-05,
+      "loss": 0.1997,
+      "step": 5268
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5312878280087436e-05,
+      "loss": 0.2074,
+      "step": 5269
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.531109656396594e-05,
+      "loss": 0.1869,
+      "step": 5270
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.530931461296722e-05,
+      "loss": 0.214,
+      "step": 5271
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5307532427170096e-05,
+      "loss": 0.2071,
+      "step": 5272
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5305750006653374e-05,
+      "loss": 0.2077,
+      "step": 5273
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.530396735149589e-05,
+      "loss": 0.2092,
+      "step": 5274
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5302184461776468e-05,
+      "loss": 0.192,
+      "step": 5275
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.530040133757396e-05,
+      "loss": 0.208,
+      "step": 5276
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5298617978967222e-05,
+      "loss": 0.2021,
+      "step": 5277
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5296834386035122e-05,
+      "loss": 0.2282,
+      "step": 5278
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5295050558856538e-05,
+      "loss": 0.1841,
+      "step": 5279
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5293266497510357e-05,
+      "loss": 0.1879,
+      "step": 5280
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5291482202075473e-05,
+      "loss": 0.1952,
+      "step": 5281
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5289697672630797e-05,
+      "loss": 0.189,
+      "step": 5282
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.528791290925525e-05,
+      "loss": 0.1771,
+      "step": 5283
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5286127912027764e-05,
+      "loss": 0.2028,
+      "step": 5284
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.528434268102727e-05,
+      "loss": 0.2301,
+      "step": 5285
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5282557216332724e-05,
+      "loss": 0.2017,
+      "step": 5286
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5280771518023082e-05,
+      "loss": 0.2085,
+      "step": 5287
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5278985586177316e-05,
+      "loss": 0.199,
+      "step": 5288
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5277199420874412e-05,
+      "loss": 0.199,
+      "step": 5289
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5275413022193356e-05,
+      "loss": 0.233,
+      "step": 5290
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5273626390213147e-05,
+      "loss": 0.1784,
+      "step": 5291
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.52718395250128e-05,
+      "loss": 0.1895,
+      "step": 5292
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.527005242667134e-05,
+      "loss": 0.1967,
+      "step": 5293
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5268265095267793e-05,
+      "loss": 0.1792,
+      "step": 5294
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.526647753088121e-05,
+      "loss": 0.2396,
+      "step": 5295
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.526468973359063e-05,
+      "loss": 0.1963,
+      "step": 5296
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5262901703475135e-05,
+      "loss": 0.2075,
+      "step": 5297
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5261113440613784e-05,
+      "loss": 0.1807,
+      "step": 5298
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5259324945085668e-05,
+      "loss": 0.2054,
+      "step": 5299
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5257536216969877e-05,
+      "loss": 0.1995,
+      "step": 5300
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5255747256345518e-05,
+      "loss": 0.2132,
+      "step": 5301
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5253958063291705e-05,
+      "loss": 0.2041,
+      "step": 5302
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5252168637887565e-05,
+      "loss": 0.1989,
+      "step": 5303
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5250378980212231e-05,
+      "loss": 0.2372,
+      "step": 5304
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5248589090344847e-05,
+      "loss": 0.1965,
+      "step": 5305
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5246798968364571e-05,
+      "loss": 0.1963,
+      "step": 5306
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5245008614350573e-05,
+      "loss": 0.1769,
+      "step": 5307
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5243218028382023e-05,
+      "loss": 0.2288,
+      "step": 5308
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5241427210538113e-05,
+      "loss": 0.2217,
+      "step": 5309
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.523963616089803e-05,
+      "loss": 0.2084,
+      "step": 5310
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5237844879540995e-05,
+      "loss": 0.1988,
+      "step": 5311
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5236053366546214e-05,
+      "loss": 0.2091,
+      "step": 5312
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5234261621992925e-05,
+      "loss": 0.1786,
+      "step": 5313
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5232469645960355e-05,
+      "loss": 0.2296,
+      "step": 5314
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.523067743852776e-05,
+      "loss": 0.1885,
+      "step": 5315
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5228884999774391e-05,
+      "loss": 0.2111,
+      "step": 5316
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5227092329779525e-05,
+      "loss": 0.2175,
+      "step": 5317
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5225299428622436e-05,
+      "loss": 0.2387,
+      "step": 5318
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5223506296382416e-05,
+      "loss": 0.2068,
+      "step": 5319
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5221712933138757e-05,
+      "loss": 0.2264,
+      "step": 5320
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.521991933897078e-05,
+      "loss": 0.2117,
+      "step": 5321
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5218125513957794e-05,
+      "loss": 0.1808,
+      "step": 5322
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.521633145817913e-05,
+      "loss": 0.2072,
+      "step": 5323
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5214537171714137e-05,
+      "loss": 0.2038,
+      "step": 5324
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5212742654642159e-05,
+      "loss": 0.1991,
+      "step": 5325
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5210947907042556e-05,
+      "loss": 0.1901,
+      "step": 5326
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5209152928994697e-05,
+      "loss": 0.1944,
+      "step": 5327
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5207357720577967e-05,
+      "loss": 0.2007,
+      "step": 5328
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5205562281871755e-05,
+      "loss": 0.1944,
+      "step": 5329
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5203766612955465e-05,
+      "loss": 0.1826,
+      "step": 5330
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5201970713908507e-05,
+      "loss": 0.19,
+      "step": 5331
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5200174584810296e-05,
+      "loss": 0.214,
+      "step": 5332
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5198378225740276e-05,
+      "loss": 0.2163,
+      "step": 5333
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5196581636777876e-05,
+      "loss": 0.1972,
+      "step": 5334
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5194784818002562e-05,
+      "loss": 0.2014,
+      "step": 5335
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5192987769493784e-05,
+      "loss": 0.2001,
+      "step": 5336
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5191190491331021e-05,
+      "loss": 0.2111,
+      "step": 5337
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5189392983593754e-05,
+      "loss": 0.1778,
+      "step": 5338
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5187595246361472e-05,
+      "loss": 0.2142,
+      "step": 5339
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5185797279713686e-05,
+      "loss": 0.2203,
+      "step": 5340
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5183999083729903e-05,
+      "loss": 0.1888,
+      "step": 5341
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5182200658489647e-05,
+      "loss": 0.2138,
+      "step": 5342
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5180402004072451e-05,
+      "loss": 0.1864,
+      "step": 5343
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5178603120557864e-05,
+      "loss": 0.1916,
+      "step": 5344
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5176804008025432e-05,
+      "loss": 0.2093,
+      "step": 5345
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5175004666554723e-05,
+      "loss": 0.2201,
+      "step": 5346
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5173205096225307e-05,
+      "loss": 0.2029,
+      "step": 5347
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5171405297116773e-05,
+      "loss": 0.1815,
+      "step": 5348
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5169605269308711e-05,
+      "loss": 0.2308,
+      "step": 5349
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5167805012880728e-05,
+      "loss": 0.2082,
+      "step": 5350
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5166004527912437e-05,
+      "loss": 0.213,
+      "step": 5351
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5164203814483463e-05,
+      "loss": 0.1763,
+      "step": 5352
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5162402872673439e-05,
+      "loss": 0.2092,
+      "step": 5353
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.516060170256201e-05,
+      "loss": 0.2132,
+      "step": 5354
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5158800304228832e-05,
+      "loss": 0.1711,
+      "step": 5355
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5156998677753569e-05,
+      "loss": 0.2045,
+      "step": 5356
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5155196823215896e-05,
+      "loss": 0.2149,
+      "step": 5357
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.51533947406955e-05,
+      "loss": 0.2161,
+      "step": 5358
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5151592430272074e-05,
+      "loss": 0.191,
+      "step": 5359
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5149789892025316e-05,
+      "loss": 0.1936,
+      "step": 5360
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5147987126034958e-05,
+      "loss": 0.2187,
+      "step": 5361
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.514618413238071e-05,
+      "loss": 0.2215,
+      "step": 5362
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5144380911142315e-05,
+      "loss": 0.2049,
+      "step": 5363
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5142577462399518e-05,
+      "loss": 0.1909,
+      "step": 5364
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.514077378623207e-05,
+      "loss": 0.1882,
+      "step": 5365
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.513896988271974e-05,
+      "loss": 0.2073,
+      "step": 5366
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5137165751942302e-05,
+      "loss": 0.207,
+      "step": 5367
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5135361393979544e-05,
+      "loss": 0.1947,
+      "step": 5368
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.513355680891126e-05,
+      "loss": 0.1975,
+      "step": 5369
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5131751996817255e-05,
+      "loss": 0.188,
+      "step": 5370
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5129946957777346e-05,
+      "loss": 0.177,
+      "step": 5371
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5128141691871364e-05,
+      "loss": 0.2023,
+      "step": 5372
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5126336199179133e-05,
+      "loss": 0.1889,
+      "step": 5373
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5124530479780506e-05,
+      "loss": 0.2078,
+      "step": 5374
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5122724533755339e-05,
+      "loss": 0.177,
+      "step": 5375
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.51209183611835e-05,
+      "loss": 0.1913,
+      "step": 5376
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.511911196214486e-05,
+      "loss": 0.1985,
+      "step": 5377
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5117305336719305e-05,
+      "loss": 0.189,
+      "step": 5378
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5115498484986734e-05,
+      "loss": 0.1811,
+      "step": 5379
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5113691407027054e-05,
+      "loss": 0.1753,
+      "step": 5380
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5111884102920176e-05,
+      "loss": 0.1843,
+      "step": 5381
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.511007657274603e-05,
+      "loss": 0.181,
+      "step": 5382
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.510826881658455e-05,
+      "loss": 0.2022,
+      "step": 5383
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5106460834515683e-05,
+      "loss": 0.2033,
+      "step": 5384
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5104652626619387e-05,
+      "loss": 0.2142,
+      "step": 5385
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5102844192975626e-05,
+      "loss": 0.1927,
+      "step": 5386
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5101035533664373e-05,
+      "loss": 0.2025,
+      "step": 5387
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5099226648765615e-05,
+      "loss": 0.1999,
+      "step": 5388
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5097417538359354e-05,
+      "loss": 0.1861,
+      "step": 5389
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.509560820252559e-05,
+      "loss": 0.2066,
+      "step": 5390
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5093798641344337e-05,
+      "loss": 0.1976,
+      "step": 5391
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5091988854895623e-05,
+      "loss": 0.1768,
+      "step": 5392
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5090178843259491e-05,
+      "loss": 0.2237,
+      "step": 5393
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5088368606515975e-05,
+      "loss": 0.1851,
+      "step": 5394
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5086558144745138e-05,
+      "loss": 0.1798,
+      "step": 5395
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5084747458027042e-05,
+      "loss": 0.1714,
+      "step": 5396
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5082936546441764e-05,
+      "loss": 0.2381,
+      "step": 5397
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5081125410069392e-05,
+      "loss": 0.2029,
+      "step": 5398
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5079314048990022e-05,
+      "loss": 0.1973,
+      "step": 5399
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.507750246328375e-05,
+      "loss": 0.2004,
+      "step": 5400
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.50756906530307e-05,
+      "loss": 0.1859,
+      "step": 5401
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5073878618310995e-05,
+      "loss": 0.2249,
+      "step": 5402
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5072066359204772e-05,
+      "loss": 0.1783,
+      "step": 5403
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5070253875792174e-05,
+      "loss": 0.2242,
+      "step": 5404
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5068441168153353e-05,
+      "loss": 0.1991,
+      "step": 5405
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5066628236368482e-05,
+      "loss": 0.1895,
+      "step": 5406
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5064815080517727e-05,
+      "loss": 0.1934,
+      "step": 5407
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5063001700681277e-05,
+      "loss": 0.191,
+      "step": 5408
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.506118809693933e-05,
+      "loss": 0.1828,
+      "step": 5409
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5059374269372085e-05,
+      "loss": 0.2348,
+      "step": 5410
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5057560218059759e-05,
+      "loss": 0.2093,
+      "step": 5411
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5055745943082576e-05,
+      "loss": 0.1676,
+      "step": 5412
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5053931444520773e-05,
+      "loss": 0.2238,
+      "step": 5413
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5052116722454585e-05,
+      "loss": 0.1828,
+      "step": 5414
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5050301776964276e-05,
+      "loss": 0.2385,
+      "step": 5415
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5048486608130107e-05,
+      "loss": 0.202,
+      "step": 5416
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5046671216032352e-05,
+      "loss": 0.1974,
+      "step": 5417
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.504485560075129e-05,
+      "loss": 0.1748,
+      "step": 5418
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.504303976236722e-05,
+      "loss": 0.2007,
+      "step": 5419
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5041223700960444e-05,
+      "loss": 0.2177,
+      "step": 5420
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5039407416611274e-05,
+      "loss": 0.209,
+      "step": 5421
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5037590909400035e-05,
+      "loss": 0.1971,
+      "step": 5422
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5035774179407062e-05,
+      "loss": 0.1806,
+      "step": 5423
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5033957226712688e-05,
+      "loss": 0.2244,
+      "step": 5424
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5032140051397276e-05,
+      "loss": 0.2009,
+      "step": 5425
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5030322653541185e-05,
+      "loss": 0.1761,
+      "step": 5426
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5028505033224787e-05,
+      "loss": 0.1979,
+      "step": 5427
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5026687190528466e-05,
+      "loss": 0.2205,
+      "step": 5428
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5024869125532608e-05,
+      "loss": 0.1913,
+      "step": 5429
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5023050838317624e-05,
+      "loss": 0.1616,
+      "step": 5430
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5021232328963919e-05,
+      "loss": 0.1779,
+      "step": 5431
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.501941359755192e-05,
+      "loss": 0.1932,
+      "step": 5432
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.501759464416205e-05,
+      "loss": 0.2118,
+      "step": 5433
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5015775468874755e-05,
+      "loss": 0.2014,
+      "step": 5434
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5013956071770491e-05,
+      "loss": 0.193,
+      "step": 5435
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5012136452929711e-05,
+      "loss": 0.1771,
+      "step": 5436
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5010316612432886e-05,
+      "loss": 0.183,
+      "step": 5437
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5008496550360502e-05,
+      "loss": 0.1919,
+      "step": 5438
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5006676266793045e-05,
+      "loss": 0.2116,
+      "step": 5439
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5004855761811019e-05,
+      "loss": 0.2096,
+      "step": 5440
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5003035035494927e-05,
+      "loss": 0.183,
+      "step": 5441
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5001214087925292e-05,
+      "loss": 0.2113,
+      "step": 5442
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4999392919182643e-05,
+      "loss": 0.1863,
+      "step": 5443
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4997571529347524e-05,
+      "loss": 0.2182,
+      "step": 5444
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4995749918500475e-05,
+      "loss": 0.1831,
+      "step": 5445
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4993928086722063e-05,
+      "loss": 0.1977,
+      "step": 5446
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.499210603409285e-05,
+      "loss": 0.2082,
+      "step": 5447
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4990283760693417e-05,
+      "loss": 0.2012,
+      "step": 5448
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4988461266604353e-05,
+      "loss": 0.1883,
+      "step": 5449
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4986638551906254e-05,
+      "loss": 0.2462,
+      "step": 5450
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4984815616679727e-05,
+      "loss": 0.1789,
+      "step": 5451
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4982992461005393e-05,
+      "loss": 0.2051,
+      "step": 5452
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4981169084963874e-05,
+      "loss": 0.1752,
+      "step": 5453
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4979345488635813e-05,
+      "loss": 0.1871,
+      "step": 5454
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4977521672101849e-05,
+      "loss": 0.1956,
+      "step": 5455
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4975697635442638e-05,
+      "loss": 0.2049,
+      "step": 5456
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4973873378738856e-05,
+      "loss": 0.177,
+      "step": 5457
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4972048902071172e-05,
+      "loss": 0.1901,
+      "step": 5458
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.497022420552027e-05,
+      "loss": 0.2262,
+      "step": 5459
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4968399289166846e-05,
+      "loss": 0.1983,
+      "step": 5460
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4966574153091609e-05,
+      "loss": 0.1898,
+      "step": 5461
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4964748797375265e-05,
+      "loss": 0.2044,
+      "step": 5462
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4962923222098549e-05,
+      "loss": 0.2035,
+      "step": 5463
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4961097427342188e-05,
+      "loss": 0.2313,
+      "step": 5464
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4959271413186927e-05,
+      "loss": 0.1937,
+      "step": 5465
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4957445179713518e-05,
+      "loss": 0.1729,
+      "step": 5466
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4955618727002728e-05,
+      "loss": 0.1945,
+      "step": 5467
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4953792055135328e-05,
+      "loss": 0.1895,
+      "step": 5468
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.49519651641921e-05,
+      "loss": 0.2367,
+      "step": 5469
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4950138054253832e-05,
+      "loss": 0.1924,
+      "step": 5470
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4948310725401336e-05,
+      "loss": 0.1798,
+      "step": 5471
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4946483177715417e-05,
+      "loss": 0.188,
+      "step": 5472
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4944655411276895e-05,
+      "loss": 0.1929,
+      "step": 5473
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4942827426166602e-05,
+      "loss": 0.1947,
+      "step": 5474
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.494099922246538e-05,
+      "loss": 0.214,
+      "step": 5475
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.493917080025408e-05,
+      "loss": 0.2134,
+      "step": 5476
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4937342159613559e-05,
+      "loss": 0.2016,
+      "step": 5477
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4935513300624692e-05,
+      "loss": 0.2167,
+      "step": 5478
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4933684223368348e-05,
+      "loss": 0.2098,
+      "step": 5479
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4931854927925425e-05,
+      "loss": 0.206,
+      "step": 5480
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4930025414376822e-05,
+      "loss": 0.1984,
+      "step": 5481
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4928195682803439e-05,
+      "loss": 0.219,
+      "step": 5482
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4926365733286198e-05,
+      "loss": 0.1889,
+      "step": 5483
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4924535565906029e-05,
+      "loss": 0.1828,
+      "step": 5484
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4922705180743869e-05,
+      "loss": 0.1942,
+      "step": 5485
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4920874577880661e-05,
+      "loss": 0.2005,
+      "step": 5486
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4919043757397363e-05,
+      "loss": 0.1934,
+      "step": 5487
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4917212719374941e-05,
+      "loss": 0.1761,
+      "step": 5488
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4915381463894373e-05,
+      "loss": 0.2077,
+      "step": 5489
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4913549991036637e-05,
+      "loss": 0.193,
+      "step": 5490
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4911718300882738e-05,
+      "loss": 0.2066,
+      "step": 5491
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4909886393513675e-05,
+      "loss": 0.1945,
+      "step": 5492
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4908054269010457e-05,
+      "loss": 0.1534,
+      "step": 5493
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4906221927454117e-05,
+      "loss": 0.2153,
+      "step": 5494
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4904389368925682e-05,
+      "loss": 0.2114,
+      "step": 5495
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.49025565935062e-05,
+      "loss": 0.1679,
+      "step": 5496
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4900723601276716e-05,
+      "loss": 0.1645,
+      "step": 5497
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4898890392318299e-05,
+      "loss": 0.2029,
+      "step": 5498
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4897056966712016e-05,
+      "loss": 0.2295,
+      "step": 5499
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4895223324538951e-05,
+      "loss": 0.2278,
+      "step": 5500
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4893389465880195e-05,
+      "loss": 0.1679,
+      "step": 5501
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4891555390816845e-05,
+      "loss": 0.1877,
+      "step": 5502
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4889721099430013e-05,
+      "loss": 0.1997,
+      "step": 5503
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.488788659180082e-05,
+      "loss": 0.216,
+      "step": 5504
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4886051868010393e-05,
+      "loss": 0.1972,
+      "step": 5505
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.488421692813987e-05,
+      "loss": 0.2036,
+      "step": 5506
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4882381772270397e-05,
+      "loss": 0.1996,
+      "step": 5507
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.488054640048314e-05,
+      "loss": 0.1845,
+      "step": 5508
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.487871081285926e-05,
+      "loss": 0.1991,
+      "step": 5509
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4876875009479935e-05,
+      "loss": 0.1811,
+      "step": 5510
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.487503899042635e-05,
+      "loss": 0.1883,
+      "step": 5511
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4873202755779702e-05,
+      "loss": 0.1843,
+      "step": 5512
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4871366305621195e-05,
+      "loss": 0.1921,
+      "step": 5513
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4869529640032047e-05,
+      "loss": 0.2031,
+      "step": 5514
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4867692759093482e-05,
+      "loss": 0.1818,
+      "step": 5515
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.486585566288673e-05,
+      "loss": 0.1796,
+      "step": 5516
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.486401835149304e-05,
+      "loss": 0.1951,
+      "step": 5517
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4862180824993657e-05,
+      "loss": 0.1748,
+      "step": 5518
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4860343083469855e-05,
+      "loss": 0.19,
+      "step": 5519
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4858505127002896e-05,
+      "loss": 0.199,
+      "step": 5520
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4856666955674066e-05,
+      "loss": 0.2021,
+      "step": 5521
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4854828569564656e-05,
+      "loss": 0.1783,
+      "step": 5522
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4852989968755966e-05,
+      "loss": 0.2194,
+      "step": 5523
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4851151153329306e-05,
+      "loss": 0.2008,
+      "step": 5524
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4849312123365993e-05,
+      "loss": 0.1871,
+      "step": 5525
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4847472878947363e-05,
+      "loss": 0.207,
+      "step": 5526
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4845633420154746e-05,
+      "loss": 0.1875,
+      "step": 5527
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4843793747069494e-05,
+      "loss": 0.209,
+      "step": 5528
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4841953859772966e-05,
+      "loss": 0.2165,
+      "step": 5529
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4840113758346529e-05,
+      "loss": 0.2149,
+      "step": 5530
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4838273442871554e-05,
+      "loss": 0.1934,
+      "step": 5531
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4836432913429436e-05,
+      "loss": 0.1812,
+      "step": 5532
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.483459217010156e-05,
+      "loss": 0.1904,
+      "step": 5533
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4832751212969334e-05,
+      "loss": 0.1711,
+      "step": 5534
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4830910042114178e-05,
+      "loss": 0.1805,
+      "step": 5535
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4829068657617511e-05,
+      "loss": 0.2046,
+      "step": 5536
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4827227059560765e-05,
+      "loss": 0.1844,
+      "step": 5537
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4825385248025385e-05,
+      "loss": 0.1919,
+      "step": 5538
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4823543223092824e-05,
+      "loss": 0.1726,
+      "step": 5539
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4821700984844541e-05,
+      "loss": 0.1883,
+      "step": 5540
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4819858533362005e-05,
+      "loss": 0.1862,
+      "step": 5541
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.48180158687267e-05,
+      "loss": 0.2202,
+      "step": 5542
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4816172991020118e-05,
+      "loss": 0.1853,
+      "step": 5543
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4814329900323751e-05,
+      "loss": 0.1811,
+      "step": 5544
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4812486596719113e-05,
+      "loss": 0.1928,
+      "step": 5545
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4810643080287723e-05,
+      "loss": 0.2147,
+      "step": 5546
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4808799351111105e-05,
+      "loss": 0.1988,
+      "step": 5547
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4806955409270793e-05,
+      "loss": 0.1868,
+      "step": 5548
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4805111254848341e-05,
+      "loss": 0.181,
+      "step": 5549
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.48032668879253e-05,
+      "loss": 0.1867,
+      "step": 5550
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4801422308583236e-05,
+      "loss": 0.1963,
+      "step": 5551
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.479957751690372e-05,
+      "loss": 0.2077,
+      "step": 5552
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4797732512968343e-05,
+      "loss": 0.187,
+      "step": 5553
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4795887296858691e-05,
+      "loss": 0.1875,
+      "step": 5554
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.479404186865637e-05,
+      "loss": 0.2043,
+      "step": 5555
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4792196228442996e-05,
+      "loss": 0.1939,
+      "step": 5556
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.479035037630018e-05,
+      "loss": 0.189,
+      "step": 5557
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4788504312309562e-05,
+      "loss": 0.1837,
+      "step": 5558
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4786658036552779e-05,
+      "loss": 0.2299,
+      "step": 5559
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.478481154911148e-05,
+      "loss": 0.2172,
+      "step": 5560
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4782964850067325e-05,
+      "loss": 0.1832,
+      "step": 5561
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4781117939501977e-05,
+      "loss": 0.2008,
+      "step": 5562
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4779270817497122e-05,
+      "loss": 0.211,
+      "step": 5563
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.477742348413444e-05,
+      "loss": 0.1895,
+      "step": 5564
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4775575939495634e-05,
+      "loss": 0.1842,
+      "step": 5565
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4773728183662402e-05,
+      "loss": 0.178,
+      "step": 5566
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4771880216716463e-05,
+      "loss": 0.2036,
+      "step": 5567
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4770032038739542e-05,
+      "loss": 0.2193,
+      "step": 5568
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4768183649813367e-05,
+      "loss": 0.2019,
+      "step": 5569
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4766335050019691e-05,
+      "loss": 0.1815,
+      "step": 5570
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4764486239440255e-05,
+      "loss": 0.1742,
+      "step": 5571
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4762637218156828e-05,
+      "loss": 0.1919,
+      "step": 5572
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.476078798625118e-05,
+      "loss": 0.2013,
+      "step": 5573
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4758938543805089e-05,
+      "loss": 0.1951,
+      "step": 5574
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4757088890900347e-05,
+      "loss": 0.1967,
+      "step": 5575
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4755239027618747e-05,
+      "loss": 0.1862,
+      "step": 5576
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4753388954042103e-05,
+      "loss": 0.1994,
+      "step": 5577
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.475153867025223e-05,
+      "loss": 0.1855,
+      "step": 5578
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4749688176330956e-05,
+      "loss": 0.1856,
+      "step": 5579
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4747837472360118e-05,
+      "loss": 0.2283,
+      "step": 5580
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4745986558421559e-05,
+      "loss": 0.2012,
+      "step": 5581
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4744135434597134e-05,
+      "loss": 0.2002,
+      "step": 5582
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4742284100968706e-05,
+      "loss": 0.182,
+      "step": 5583
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.474043255761815e-05,
+      "loss": 0.1997,
+      "step": 5584
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4738580804627347e-05,
+      "loss": 0.186,
+      "step": 5585
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.473672884207819e-05,
+      "loss": 0.2135,
+      "step": 5586
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.473487667005258e-05,
+      "loss": 0.2266,
+      "step": 5587
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4733024288632426e-05,
+      "loss": 0.215,
+      "step": 5588
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4731171697899647e-05,
+      "loss": 0.2027,
+      "step": 5589
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4729318897936174e-05,
+      "loss": 0.1946,
+      "step": 5590
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4727465888823942e-05,
+      "loss": 0.2314,
+      "step": 5591
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.47256126706449e-05,
+      "loss": 0.1693,
+      "step": 5592
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4723759243481004e-05,
+      "loss": 0.2135,
+      "step": 5593
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4721905607414222e-05,
+      "loss": 0.2154,
+      "step": 5594
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4720051762526524e-05,
+      "loss": 0.1982,
+      "step": 5595
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.47181977088999e-05,
+      "loss": 0.1901,
+      "step": 5596
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4716343446616339e-05,
+      "loss": 0.2038,
+      "step": 5597
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4714488975757844e-05,
+      "loss": 0.1899,
+      "step": 5598
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4712634296406429e-05,
+      "loss": 0.1962,
+      "step": 5599
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4710779408644116e-05,
+      "loss": 0.2117,
+      "step": 5600
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4708924312552933e-05,
+      "loss": 0.1848,
+      "step": 5601
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4707069008214919e-05,
+      "loss": 0.1905,
+      "step": 5602
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.470521349571212e-05,
+      "loss": 0.204,
+      "step": 5603
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4703357775126604e-05,
+      "loss": 0.1861,
+      "step": 5604
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4701501846540429e-05,
+      "loss": 0.172,
+      "step": 5605
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4699645710035673e-05,
+      "loss": 0.2214,
+      "step": 5606
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4697789365694423e-05,
+      "loss": 0.2149,
+      "step": 5607
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4695932813598775e-05,
+      "loss": 0.1879,
+      "step": 5608
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.469407605383083e-05,
+      "loss": 0.221,
+      "step": 5609
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.46922190864727e-05,
+      "loss": 0.1695,
+      "step": 5610
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4690361911606517e-05,
+      "loss": 0.2027,
+      "step": 5611
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4688504529314398e-05,
+      "loss": 0.198,
+      "step": 5612
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.468664693967849e-05,
+      "loss": 0.2142,
+      "step": 5613
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4684789142780951e-05,
+      "loss": 0.2134,
+      "step": 5614
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4682931138703926e-05,
+      "loss": 0.1822,
+      "step": 5615
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4681072927529591e-05,
+      "loss": 0.191,
+      "step": 5616
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.467921450934012e-05,
+      "loss": 0.2094,
+      "step": 5617
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4677355884217705e-05,
+      "loss": 0.2112,
+      "step": 5618
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4675497052244537e-05,
+      "loss": 0.2079,
+      "step": 5619
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.467363801350282e-05,
+      "loss": 0.1932,
+      "step": 5620
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4671778768074771e-05,
+      "loss": 0.1877,
+      "step": 5621
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4669919316042611e-05,
+      "loss": 0.2029,
+      "step": 5622
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4668059657488572e-05,
+      "loss": 0.1878,
+      "step": 5623
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4666199792494897e-05,
+      "loss": 0.1918,
+      "step": 5624
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4664339721143837e-05,
+      "loss": 0.194,
+      "step": 5625
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4662479443517644e-05,
+      "loss": 0.1898,
+      "step": 5626
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4660618959698596e-05,
+      "loss": 0.2194,
+      "step": 5627
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4658758269768968e-05,
+      "loss": 0.214,
+      "step": 5628
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4656897373811046e-05,
+      "loss": 0.2074,
+      "step": 5629
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4655036271907127e-05,
+      "loss": 0.22,
+      "step": 5630
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4653174964139512e-05,
+      "loss": 0.1828,
+      "step": 5631
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.465131345059052e-05,
+      "loss": 0.1843,
+      "step": 5632
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4649451731342474e-05,
+      "loss": 0.1985,
+      "step": 5633
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4647589806477704e-05,
+      "loss": 0.2189,
+      "step": 5634
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4645727676078556e-05,
+      "loss": 0.2192,
+      "step": 5635
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4643865340227375e-05,
+      "loss": 0.1895,
+      "step": 5636
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4642002799006523e-05,
+      "loss": 0.2043,
+      "step": 5637
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.464014005249837e-05,
+      "loss": 0.21,
+      "step": 5638
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4638277100785291e-05,
+      "loss": 0.186,
+      "step": 5639
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4636413943949674e-05,
+      "loss": 0.208,
+      "step": 5640
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4634550582073918e-05,
+      "loss": 0.1994,
+      "step": 5641
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4632687015240427e-05,
+      "loss": 0.1606,
+      "step": 5642
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4630823243531612e-05,
+      "loss": 0.1779,
+      "step": 5643
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4628959267029896e-05,
+      "loss": 0.2243,
+      "step": 5644
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4627095085817718e-05,
+      "loss": 0.2187,
+      "step": 5645
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4625230699977511e-05,
+      "loss": 0.1958,
+      "step": 5646
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4623366109591728e-05,
+      "loss": 0.1692,
+      "step": 5647
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4621501314742832e-05,
+      "loss": 0.2053,
+      "step": 5648
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4619636315513286e-05,
+      "loss": 0.2294,
+      "step": 5649
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4617771111985572e-05,
+      "loss": 0.1895,
+      "step": 5650
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4615905704242175e-05,
+      "loss": 0.1994,
+      "step": 5651
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.461404009236559e-05,
+      "loss": 0.1968,
+      "step": 5652
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.461217427643832e-05,
+      "loss": 0.1885,
+      "step": 5653
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4610308256542883e-05,
+      "loss": 0.1861,
+      "step": 5654
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.46084420327618e-05,
+      "loss": 0.1994,
+      "step": 5655
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.46065756051776e-05,
+      "loss": 0.1993,
+      "step": 5656
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4604708973872825e-05,
+      "loss": 0.1974,
+      "step": 5657
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4602842138930022e-05,
+      "loss": 0.1836,
+      "step": 5658
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.460097510043176e-05,
+      "loss": 0.1921,
+      "step": 5659
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4599107858460592e-05,
+      "loss": 0.2087,
+      "step": 5660
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4597240413099108e-05,
+      "loss": 0.2079,
+      "step": 5661
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4595372764429882e-05,
+      "loss": 0.2058,
+      "step": 5662
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4593504912535519e-05,
+      "loss": 0.1773,
+      "step": 5663
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4591636857498615e-05,
+      "loss": 0.1839,
+      "step": 5664
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4589768599401786e-05,
+      "loss": 0.1824,
+      "step": 5665
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4587900138327653e-05,
+      "loss": 0.1924,
+      "step": 5666
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4586031474358846e-05,
+      "loss": 0.205,
+      "step": 5667
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4584162607578005e-05,
+      "loss": 0.1903,
+      "step": 5668
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4582293538067781e-05,
+      "loss": 0.2015,
+      "step": 5669
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.458042426591083e-05,
+      "loss": 0.1712,
+      "step": 5670
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4578554791189814e-05,
+      "loss": 0.2109,
+      "step": 5671
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4576685113987412e-05,
+      "loss": 0.2074,
+      "step": 5672
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4574815234386308e-05,
+      "loss": 0.2015,
+      "step": 5673
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4572945152469197e-05,
+      "loss": 0.2088,
+      "step": 5674
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4571074868318783e-05,
+      "loss": 0.1939,
+      "step": 5675
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.456920438201777e-05,
+      "loss": 0.2199,
+      "step": 5676
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4567333693648881e-05,
+      "loss": 0.2104,
+      "step": 5677
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.456546280329485e-05,
+      "loss": 0.1831,
+      "step": 5678
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4563591711038412e-05,
+      "loss": 0.1936,
+      "step": 5679
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.456172041696231e-05,
+      "loss": 0.1936,
+      "step": 5680
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4559848921149303e-05,
+      "loss": 0.1687,
+      "step": 5681
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4557977223682157e-05,
+      "loss": 0.1916,
+      "step": 5682
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4556105324643646e-05,
+      "loss": 0.1948,
+      "step": 5683
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.455423322411655e-05,
+      "loss": 0.236,
+      "step": 5684
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4552360922183662e-05,
+      "loss": 0.183,
+      "step": 5685
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4550488418927779e-05,
+      "loss": 0.2064,
+      "step": 5686
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4548615714431717e-05,
+      "loss": 0.1574,
+      "step": 5687
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4546742808778289e-05,
+      "loss": 0.2043,
+      "step": 5688
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4544869702050326e-05,
+      "loss": 0.1884,
+      "step": 5689
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.454299639433066e-05,
+      "loss": 0.1891,
+      "step": 5690
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4541122885702135e-05,
+      "loss": 0.2013,
+      "step": 5691
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4539249176247612e-05,
+      "loss": 0.2015,
+      "step": 5692
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.453737526604995e-05,
+      "loss": 0.1948,
+      "step": 5693
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4535501155192016e-05,
+      "loss": 0.2053,
+      "step": 5694
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4533626843756692e-05,
+      "loss": 0.1899,
+      "step": 5695
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4531752331826874e-05,
+      "loss": 0.2102,
+      "step": 5696
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4529877619485456e-05,
+      "loss": 0.1921,
+      "step": 5697
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4528002706815343e-05,
+      "loss": 0.2181,
+      "step": 5698
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.452612759389945e-05,
+      "loss": 0.2087,
+      "step": 5699
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4524252280820707e-05,
+      "loss": 0.2185,
+      "step": 5700
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4522376767662045e-05,
+      "loss": 0.2048,
+      "step": 5701
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4520501054506406e-05,
+      "loss": 0.2002,
+      "step": 5702
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4518625141436742e-05,
+      "loss": 0.1904,
+      "step": 5703
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.451674902853601e-05,
+      "loss": 0.1799,
+      "step": 5704
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4514872715887186e-05,
+      "loss": 0.1964,
+      "step": 5705
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4512996203573241e-05,
+      "loss": 0.2191,
+      "step": 5706
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4511119491677165e-05,
+      "loss": 0.1644,
+      "step": 5707
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.450924258028195e-05,
+      "loss": 0.1762,
+      "step": 5708
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4507365469470602e-05,
+      "loss": 0.194,
+      "step": 5709
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.450548815932614e-05,
+      "loss": 0.1606,
+      "step": 5710
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4503610649931579e-05,
+      "loss": 0.1983,
+      "step": 5711
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.450173294136995e-05,
+      "loss": 0.2119,
+      "step": 5712
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4499855033724295e-05,
+      "loss": 0.2153,
+      "step": 5713
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.449797692707766e-05,
+      "loss": 0.2054,
+      "step": 5714
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4496098621513106e-05,
+      "loss": 0.2092,
+      "step": 5715
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4494220117113695e-05,
+      "loss": 0.1834,
+      "step": 5716
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4492341413962503e-05,
+      "loss": 0.2209,
+      "step": 5717
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4490462512142614e-05,
+      "loss": 0.1979,
+      "step": 5718
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4488583411737117e-05,
+      "loss": 0.1861,
+      "step": 5719
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4486704112829123e-05,
+      "loss": 0.2158,
+      "step": 5720
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4484824615501732e-05,
+      "loss": 0.2154,
+      "step": 5721
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.448294491983806e-05,
+      "loss": 0.2029,
+      "step": 5722
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4481065025921246e-05,
+      "loss": 0.1891,
+      "step": 5723
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4479184933834419e-05,
+      "loss": 0.1991,
+      "step": 5724
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4477304643660725e-05,
+      "loss": 0.2029,
+      "step": 5725
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4475424155483317e-05,
+      "loss": 0.2272,
+      "step": 5726
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4473543469385358e-05,
+      "loss": 0.2181,
+      "step": 5727
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4471662585450022e-05,
+      "loss": 0.2051,
+      "step": 5728
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4469781503760483e-05,
+      "loss": 0.2209,
+      "step": 5729
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4467900224399935e-05,
+      "loss": 0.2047,
+      "step": 5730
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.446601874745157e-05,
+      "loss": 0.1761,
+      "step": 5731
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4464137072998603e-05,
+      "loss": 0.2002,
+      "step": 5732
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.446225520112424e-05,
+      "loss": 0.1845,
+      "step": 5733
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4460373131911713e-05,
+      "loss": 0.1988,
+      "step": 5734
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4458490865444245e-05,
+      "loss": 0.2071,
+      "step": 5735
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4456608401805081e-05,
+      "loss": 0.1861,
+      "step": 5736
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4454725741077475e-05,
+      "loss": 0.2087,
+      "step": 5737
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4452842883344682e-05,
+      "loss": 0.2377,
+      "step": 5738
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4450959828689968e-05,
+      "loss": 0.194,
+      "step": 5739
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.444907657719661e-05,
+      "loss": 0.2121,
+      "step": 5740
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4447193128947893e-05,
+      "loss": 0.2034,
+      "step": 5741
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.444530948402711e-05,
+      "loss": 0.1799,
+      "step": 5742
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4443425642517564e-05,
+      "loss": 0.192,
+      "step": 5743
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4441541604502567e-05,
+      "loss": 0.1835,
+      "step": 5744
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4439657370065433e-05,
+      "loss": 0.216,
+      "step": 5745
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4437772939289497e-05,
+      "loss": 0.1826,
+      "step": 5746
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.443588831225809e-05,
+      "loss": 0.2053,
+      "step": 5747
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4434003489054563e-05,
+      "loss": 0.1874,
+      "step": 5748
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4432118469762265e-05,
+      "loss": 0.213,
+      "step": 5749
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4430233254464557e-05,
+      "loss": 0.189,
+      "step": 5750
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4428347843244822e-05,
+      "loss": 0.1838,
+      "step": 5751
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4426462236186427e-05,
+      "loss": 0.2139,
+      "step": 5752
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.442457643337277e-05,
+      "loss": 0.2005,
+      "step": 5753
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4422690434887243e-05,
+      "loss": 0.211,
+      "step": 5754
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4420804240813256e-05,
+      "loss": 0.1761,
+      "step": 5755
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.441891785123422e-05,
+      "loss": 0.2335,
+      "step": 5756
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.441703126623356e-05,
+      "loss": 0.2229,
+      "step": 5757
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4415144485894711e-05,
+      "loss": 0.2231,
+      "step": 5758
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4413257510301107e-05,
+      "loss": 0.1938,
+      "step": 5759
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4411370339536203e-05,
+      "loss": 0.1941,
+      "step": 5760
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4409482973683459e-05,
+      "loss": 0.205,
+      "step": 5761
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4407595412826335e-05,
+      "loss": 0.2308,
+      "step": 5762
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4405707657048312e-05,
+      "loss": 0.2058,
+      "step": 5763
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4403819706432869e-05,
+      "loss": 0.1996,
+      "step": 5764
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4401931561063501e-05,
+      "loss": 0.2054,
+      "step": 5765
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4400043221023707e-05,
+      "loss": 0.2153,
+      "step": 5766
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4398154686397e-05,
+      "loss": 0.1937,
+      "step": 5767
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.43962659572669e-05,
+      "loss": 0.1826,
+      "step": 5768
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4394377033716927e-05,
+      "loss": 0.1919,
+      "step": 5769
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.439248791583062e-05,
+      "loss": 0.2106,
+      "step": 5770
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4390598603691525e-05,
+      "loss": 0.1902,
+      "step": 5771
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4388709097383191e-05,
+      "loss": 0.1943,
+      "step": 5772
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.438681939698918e-05,
+      "loss": 0.1909,
+      "step": 5773
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4384929502593065e-05,
+      "loss": 0.1923,
+      "step": 5774
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4383039414278425e-05,
+      "loss": 0.1974,
+      "step": 5775
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4381149132128842e-05,
+      "loss": 0.1658,
+      "step": 5776
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4379258656227915e-05,
+      "loss": 0.2053,
+      "step": 5777
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4377367986659246e-05,
+      "loss": 0.2019,
+      "step": 5778
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4375477123506447e-05,
+      "loss": 0.1951,
+      "step": 5779
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4373586066853145e-05,
+      "loss": 0.1992,
+      "step": 5780
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4371694816782963e-05,
+      "loss": 0.2048,
+      "step": 5781
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4369803373379546e-05,
+      "loss": 0.2162,
+      "step": 5782
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4367911736726534e-05,
+      "loss": 0.2123,
+      "step": 5783
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4366019906907587e-05,
+      "loss": 0.1907,
+      "step": 5784
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4364127884006372e-05,
+      "loss": 0.2143,
+      "step": 5785
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.436223566810655e-05,
+      "loss": 0.1818,
+      "step": 5786
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4360343259291813e-05,
+      "loss": 0.1978,
+      "step": 5787
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4358450657645848e-05,
+      "loss": 0.2153,
+      "step": 5788
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4356557863252352e-05,
+      "loss": 0.1804,
+      "step": 5789
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4354664876195031e-05,
+      "loss": 0.2019,
+      "step": 5790
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.43527716965576e-05,
+      "loss": 0.1756,
+      "step": 5791
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4350878324423787e-05,
+      "loss": 0.1788,
+      "step": 5792
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.434898475987732e-05,
+      "loss": 0.2222,
+      "step": 5793
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4347091003001941e-05,
+      "loss": 0.2166,
+      "step": 5794
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4345197053881398e-05,
+      "loss": 0.2176,
+      "step": 5795
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4343302912599451e-05,
+      "loss": 0.1861,
+      "step": 5796
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4341408579239864e-05,
+      "loss": 0.202,
+      "step": 5797
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4339514053886417e-05,
+      "loss": 0.2031,
+      "step": 5798
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4337619336622886e-05,
+      "loss": 0.1901,
+      "step": 5799
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4335724427533065e-05,
+      "loss": 0.1911,
+      "step": 5800
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4333829326700754e-05,
+      "loss": 0.1877,
+      "step": 5801
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4331934034209767e-05,
+      "loss": 0.1882,
+      "step": 5802
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4330038550143914e-05,
+      "loss": 0.1855,
+      "step": 5803
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4328142874587023e-05,
+      "loss": 0.2015,
+      "step": 5804
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.432624700762293e-05,
+      "loss": 0.1831,
+      "step": 5805
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4324350949335475e-05,
+      "loss": 0.2023,
+      "step": 5806
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4322454699808512e-05,
+      "loss": 0.1888,
+      "step": 5807
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4320558259125896e-05,
+      "loss": 0.1742,
+      "step": 5808
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4318661627371499e-05,
+      "loss": 0.2173,
+      "step": 5809
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4316764804629197e-05,
+      "loss": 0.2252,
+      "step": 5810
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4314867790982874e-05,
+      "loss": 0.1844,
+      "step": 5811
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4312970586516424e-05,
+      "loss": 0.2245,
+      "step": 5812
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4311073191313745e-05,
+      "loss": 0.1865,
+      "step": 5813
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.430917560545875e-05,
+      "loss": 0.1741,
+      "step": 5814
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.430727782903536e-05,
+      "loss": 0.1927,
+      "step": 5815
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4305379862127497e-05,
+      "loss": 0.1851,
+      "step": 5816
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.43034817048191e-05,
+      "loss": 0.2038,
+      "step": 5817
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4301583357194111e-05,
+      "loss": 0.1743,
+      "step": 5818
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4299684819336484e-05,
+      "loss": 0.2044,
+      "step": 5819
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.429778609133018e-05,
+      "loss": 0.1917,
+      "step": 5820
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4295887173259164e-05,
+      "loss": 0.1907,
+      "step": 5821
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.429398806520742e-05,
+      "loss": 0.2079,
+      "step": 5822
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4292088767258927e-05,
+      "loss": 0.165,
+      "step": 5823
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4290189279497683e-05,
+      "loss": 0.1943,
+      "step": 5824
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4288289602007693e-05,
+      "loss": 0.1965,
+      "step": 5825
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4286389734872965e-05,
+      "loss": 0.1894,
+      "step": 5826
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4284489678177517e-05,
+      "loss": 0.1907,
+      "step": 5827
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4282589432005376e-05,
+      "loss": 0.2422,
+      "step": 5828
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4280688996440587e-05,
+      "loss": 0.1598,
+      "step": 5829
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4278788371567184e-05,
+      "loss": 0.1997,
+      "step": 5830
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4276887557469225e-05,
+      "loss": 0.2061,
+      "step": 5831
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4274986554230772e-05,
+      "loss": 0.1947,
+      "step": 5832
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4273085361935892e-05,
+      "loss": 0.1804,
+      "step": 5833
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4271183980668668e-05,
+      "loss": 0.1734,
+      "step": 5834
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.426928241051318e-05,
+      "loss": 0.1858,
+      "step": 5835
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4267380651553527e-05,
+      "loss": 0.1914,
+      "step": 5836
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4265478703873809e-05,
+      "loss": 0.1847,
+      "step": 5837
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.426357656755814e-05,
+      "loss": 0.1532,
+      "step": 5838
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4261674242690642e-05,
+      "loss": 0.1802,
+      "step": 5839
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4259771729355438e-05,
+      "loss": 0.1962,
+      "step": 5840
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4257869027636669e-05,
+      "loss": 0.1787,
+      "step": 5841
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4255966137618474e-05,
+      "loss": 0.1989,
+      "step": 5842
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4254063059385012e-05,
+      "loss": 0.1714,
+      "step": 5843
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4252159793020441e-05,
+      "loss": 0.1915,
+      "step": 5844
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4250256338608935e-05,
+      "loss": 0.2104,
+      "step": 5845
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4248352696234666e-05,
+      "loss": 0.1969,
+      "step": 5846
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4246448865981826e-05,
+      "loss": 0.195,
+      "step": 5847
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4244544847934605e-05,
+      "loss": 0.2136,
+      "step": 5848
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.424264064217721e-05,
+      "loss": 0.2127,
+      "step": 5849
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4240736248793852e-05,
+      "loss": 0.1939,
+      "step": 5850
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4238831667868746e-05,
+      "loss": 0.2113,
+      "step": 5851
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4236926899486125e-05,
+      "loss": 0.2165,
+      "step": 5852
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4235021943730228e-05,
+      "loss": 0.1914,
+      "step": 5853
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.423311680068529e-05,
+      "loss": 0.2182,
+      "step": 5854
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4231211470435567e-05,
+      "loss": 0.2113,
+      "step": 5855
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4229305953065326e-05,
+      "loss": 0.2217,
+      "step": 5856
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.422740024865883e-05,
+      "loss": 0.2013,
+      "step": 5857
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.422549435730036e-05,
+      "loss": 0.2002,
+      "step": 5858
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4223588279074201e-05,
+      "loss": 0.1778,
+      "step": 5859
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4221682014064644e-05,
+      "loss": 0.1795,
+      "step": 5860
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4219775562355998e-05,
+      "loss": 0.2069,
+      "step": 5861
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4217868924032568e-05,
+      "loss": 0.2134,
+      "step": 5862
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4215962099178678e-05,
+      "loss": 0.2131,
+      "step": 5863
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4214055087878644e-05,
+      "loss": 0.1949,
+      "step": 5864
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4212147890216814e-05,
+      "loss": 0.2056,
+      "step": 5865
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4210240506277528e-05,
+      "loss": 0.1962,
+      "step": 5866
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.420833293614514e-05,
+      "loss": 0.1826,
+      "step": 5867
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4206425179904004e-05,
+      "loss": 0.1982,
+      "step": 5868
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4204517237638489e-05,
+      "loss": 0.2012,
+      "step": 5869
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4202609109432977e-05,
+      "loss": 0.2201,
+      "step": 5870
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4200700795371848e-05,
+      "loss": 0.1756,
+      "step": 5871
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4198792295539499e-05,
+      "loss": 0.2049,
+      "step": 5872
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4196883610020326e-05,
+      "loss": 0.1791,
+      "step": 5873
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4194974738898744e-05,
+      "loss": 0.1959,
+      "step": 5874
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4193065682259168e-05,
+      "loss": 0.1825,
+      "step": 5875
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4191156440186023e-05,
+      "loss": 0.1908,
+      "step": 5876
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4189247012763748e-05,
+      "loss": 0.1993,
+      "step": 5877
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4187337400076775e-05,
+      "loss": 0.1732,
+      "step": 5878
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4185427602209565e-05,
+      "loss": 0.2184,
+      "step": 5879
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4183517619246572e-05,
+      "loss": 0.1827,
+      "step": 5880
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4181607451272262e-05,
+      "loss": 0.1656,
+      "step": 5881
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4179697098371112e-05,
+      "loss": 0.2122,
+      "step": 5882
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4177786560627603e-05,
+      "loss": 0.1973,
+      "step": 5883
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4175875838126229e-05,
+      "loss": 0.2085,
+      "step": 5884
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.417396493095149e-05,
+      "loss": 0.1956,
+      "step": 5885
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.417205383918789e-05,
+      "loss": 0.1695,
+      "step": 5886
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4170142562919948e-05,
+      "loss": 0.1897,
+      "step": 5887
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4168231102232187e-05,
+      "loss": 0.2039,
+      "step": 5888
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.416631945720914e-05,
+      "loss": 0.1811,
+      "step": 5889
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4164407627935346e-05,
+      "loss": 0.1955,
+      "step": 5890
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4162495614495352e-05,
+      "loss": 0.198,
+      "step": 5891
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4160583416973716e-05,
+      "loss": 0.1964,
+      "step": 5892
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4158671035455005e-05,
+      "loss": 0.1838,
+      "step": 5893
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4156758470023793e-05,
+      "loss": 0.1823,
+      "step": 5894
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4154845720764653e-05,
+      "loss": 0.2058,
+      "step": 5895
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4152932787762183e-05,
+      "loss": 0.2176,
+      "step": 5896
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4151019671100976e-05,
+      "loss": 0.2078,
+      "step": 5897
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4149106370865637e-05,
+      "loss": 0.187,
+      "step": 5898
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4147192887140784e-05,
+      "loss": 0.2082,
+      "step": 5899
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4145279220011032e-05,
+      "loss": 0.2114,
+      "step": 5900
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4143365369561015e-05,
+      "loss": 0.1799,
+      "step": 5901
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.414145133587537e-05,
+      "loss": 0.2004,
+      "step": 5902
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4139537119038743e-05,
+      "loss": 0.1844,
+      "step": 5903
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4137622719135789e-05,
+      "loss": 0.2027,
+      "step": 5904
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4135708136251168e-05,
+      "loss": 0.2322,
+      "step": 5905
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4133793370469549e-05,
+      "loss": 0.1742,
+      "step": 5906
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4131878421875616e-05,
+      "loss": 0.1901,
+      "step": 5907
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.412996329055405e-05,
+      "loss": 0.1954,
+      "step": 5908
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.412804797658955e-05,
+      "loss": 0.2236,
+      "step": 5909
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4126132480066815e-05,
+      "loss": 0.1918,
+      "step": 5910
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4124216801070556e-05,
+      "loss": 0.2341,
+      "step": 5911
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4122300939685494e-05,
+      "loss": 0.2077,
+      "step": 5912
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4120384895996353e-05,
+      "loss": 0.1588,
+      "step": 5913
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4118468670087871e-05,
+      "loss": 0.1837,
+      "step": 5914
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4116552262044787e-05,
+      "loss": 0.1946,
+      "step": 5915
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4114635671951855e-05,
+      "loss": 0.203,
+      "step": 5916
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4112718899893833e-05,
+      "loss": 0.1961,
+      "step": 5917
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4110801945955492e-05,
+      "loss": 0.2181,
+      "step": 5918
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.41088848102216e-05,
+      "loss": 0.1958,
+      "step": 5919
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4106967492776942e-05,
+      "loss": 0.1953,
+      "step": 5920
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4105049993706315e-05,
+      "loss": 0.195,
+      "step": 5921
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4103132313094512e-05,
+      "loss": 0.1825,
+      "step": 5922
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4101214451026342e-05,
+      "loss": 0.1761,
+      "step": 5923
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4099296407586618e-05,
+      "loss": 0.2176,
+      "step": 5924
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.409737818286017e-05,
+      "loss": 0.2013,
+      "step": 5925
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4095459776931822e-05,
+      "loss": 0.1833,
+      "step": 5926
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4093541189886417e-05,
+      "loss": 0.2231,
+      "step": 5927
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4091622421808802e-05,
+      "loss": 0.1944,
+      "step": 5928
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4089703472783831e-05,
+      "loss": 0.2106,
+      "step": 5929
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.408778434289637e-05,
+      "loss": 0.1795,
+      "step": 5930
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.408586503223129e-05,
+      "loss": 0.2007,
+      "step": 5931
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4083945540873467e-05,
+      "loss": 0.1777,
+      "step": 5932
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4082025868907787e-05,
+      "loss": 0.2132,
+      "step": 5933
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4080106016419153e-05,
+      "loss": 0.182,
+      "step": 5934
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4078185983492465e-05,
+      "loss": 0.2097,
+      "step": 5935
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.407626577021263e-05,
+      "loss": 0.1842,
+      "step": 5936
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.407434537666457e-05,
+      "loss": 0.2044,
+      "step": 5937
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4072424802933217e-05,
+      "loss": 0.1952,
+      "step": 5938
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4070504049103499e-05,
+      "loss": 0.1821,
+      "step": 5939
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4068583115260364e-05,
+      "loss": 0.1954,
+      "step": 5940
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4066662001488758e-05,
+      "loss": 0.2166,
+      "step": 5941
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4064740707873646e-05,
+      "loss": 0.2072,
+      "step": 5942
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4062819234499991e-05,
+      "loss": 0.2029,
+      "step": 5943
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4060897581452768e-05,
+      "loss": 0.1923,
+      "step": 5944
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4058975748816965e-05,
+      "loss": 0.1932,
+      "step": 5945
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4057053736677566e-05,
+      "loss": 0.1749,
+      "step": 5946
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4055131545119573e-05,
+      "loss": 0.2104,
+      "step": 5947
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4053209174227995e-05,
+      "loss": 0.2118,
+      "step": 5948
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4051286624087843e-05,
+      "loss": 0.1984,
+      "step": 5949
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.404936389478414e-05,
+      "loss": 0.2425,
+      "step": 5950
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4047440986401919e-05,
+      "loss": 0.2182,
+      "step": 5951
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4045517899026216e-05,
+      "loss": 0.1663,
+      "step": 5952
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4043594632742078e-05,
+      "loss": 0.174,
+      "step": 5953
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4041671187634558e-05,
+      "loss": 0.2068,
+      "step": 5954
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4039747563788723e-05,
+      "loss": 0.2235,
+      "step": 5955
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4037823761289638e-05,
+      "loss": 0.1826,
+      "step": 5956
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4035899780222381e-05,
+      "loss": 0.1999,
+      "step": 5957
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.403397562067204e-05,
+      "loss": 0.2142,
+      "step": 5958
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4032051282723713e-05,
+      "loss": 0.1919,
+      "step": 5959
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4030126766462493e-05,
+      "loss": 0.2002,
+      "step": 5960
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4028202071973492e-05,
+      "loss": 0.1788,
+      "step": 5961
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4026277199341834e-05,
+      "loss": 0.178,
+      "step": 5962
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4024352148652635e-05,
+      "loss": 0.2195,
+      "step": 5963
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4022426919991032e-05,
+      "loss": 0.1573,
+      "step": 5964
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4020501513442169e-05,
+      "loss": 0.2407,
+      "step": 5965
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4018575929091192e-05,
+      "loss": 0.2114,
+      "step": 5966
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4016650167023258e-05,
+      "loss": 0.218,
+      "step": 5967
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.401472422732353e-05,
+      "loss": 0.2292,
+      "step": 5968
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4012798110077188e-05,
+      "loss": 0.209,
+      "step": 5969
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.40108718153694e-05,
+      "loss": 0.2042,
+      "step": 5970
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4008945343285363e-05,
+      "loss": 0.1783,
+      "step": 5971
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4007018693910274e-05,
+      "loss": 0.1876,
+      "step": 5972
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4005091867329332e-05,
+      "loss": 0.2336,
+      "step": 5973
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4003164863627749e-05,
+      "loss": 0.1746,
+      "step": 5974
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4001237682890744e-05,
+      "loss": 0.1968,
+      "step": 5975
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.3999310325203554e-05,
+      "loss": 0.1905,
+      "step": 5976
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.3997382790651402e-05,
+      "loss": 0.205,
+      "step": 5977
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.3995455079319535e-05,
+      "loss": 0.1719,
+      "step": 5978
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.3993527191293206e-05,
+      "loss": 0.1956,
+      "step": 5979
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.399159912665767e-05,
+      "loss": 0.2003,
+      "step": 5980
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.39896708854982e-05,
+      "loss": 0.1787,
+      "step": 5981
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3987742467900063e-05,
+      "loss": 0.1979,
+      "step": 5982
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3985813873948544e-05,
+      "loss": 0.1928,
+      "step": 5983
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3983885103728931e-05,
+      "loss": 0.2133,
+      "step": 5984
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3981956157326528e-05,
+      "loss": 0.2027,
+      "step": 5985
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3980027034826636e-05,
+      "loss": 0.2168,
+      "step": 5986
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3978097736314565e-05,
+      "loss": 0.1879,
+      "step": 5987
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3976168261875637e-05,
+      "loss": 0.1869,
+      "step": 5988
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.397423861159519e-05,
+      "loss": 0.2087,
+      "step": 5989
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.397230878555855e-05,
+      "loss": 0.209,
+      "step": 5990
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3970378783851066e-05,
+      "loss": 0.1576,
+      "step": 5991
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3968448606558087e-05,
+      "loss": 0.214,
+      "step": 5992
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3966518253764978e-05,
+      "loss": 0.2153,
+      "step": 5993
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3964587725557102e-05,
+      "loss": 0.1841,
+      "step": 5994
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3962657022019836e-05,
+      "loss": 0.1925,
+      "step": 5995
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3960726143238568e-05,
+      "loss": 0.1877,
+      "step": 5996
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3958795089298678e-05,
+      "loss": 0.1692,
+      "step": 5997
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3956863860285573e-05,
+      "loss": 0.1803,
+      "step": 5998
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3954932456284658e-05,
+      "loss": 0.198,
+      "step": 5999
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3953000877381348e-05,
+      "loss": 0.1928,
+      "step": 6000
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.395106912366106e-05,
+      "loss": 0.2139,
+      "step": 6001
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3949137195209228e-05,
+      "loss": 0.1998,
+      "step": 6002
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3947205092111292e-05,
+      "loss": 0.1874,
+      "step": 6003
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.394527281445269e-05,
+      "loss": 0.1792,
+      "step": 6004
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3943340362318877e-05,
+      "loss": 0.2214,
+      "step": 6005
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3941407735795318e-05,
+      "loss": 0.1739,
+      "step": 6006
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3939474934967476e-05,
+      "loss": 0.1864,
+      "step": 6007
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.393754195992083e-05,
+      "loss": 0.1901,
+      "step": 6008
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3935608810740861e-05,
+      "loss": 0.1841,
+      "step": 6009
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3933675487513065e-05,
+      "loss": 0.1793,
+      "step": 6010
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3931741990322934e-05,
+      "loss": 0.1882,
+      "step": 6011
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.392980831925598e-05,
+      "loss": 0.2073,
+      "step": 6012
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3927874474397717e-05,
+      "loss": 0.1927,
+      "step": 6013
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3925940455833665e-05,
+      "loss": 0.2065,
+      "step": 6014
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3924006263649354e-05,
+      "loss": 0.2005,
+      "step": 6015
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3922071897930321e-05,
+      "loss": 0.2215,
+      "step": 6016
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3920137358762116e-05,
+      "loss": 0.1684,
+      "step": 6017
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3918202646230288e-05,
+      "loss": 0.2018,
+      "step": 6018
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3916267760420397e-05,
+      "loss": 0.1753,
+      "step": 6019
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.391433270141801e-05,
+      "loss": 0.2075,
+      "step": 6020
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3912397469308705e-05,
+      "loss": 0.1885,
+      "step": 6021
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3910462064178065e-05,
+      "loss": 0.1839,
+      "step": 6022
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3908526486111683e-05,
+      "loss": 0.1711,
+      "step": 6023
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3906590735195156e-05,
+      "loss": 0.2154,
+      "step": 6024
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3904654811514086e-05,
+      "loss": 0.1888,
+      "step": 6025
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3902718715154096e-05,
+      "loss": 0.2137,
+      "step": 6026
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3900782446200802e-05,
+      "loss": 0.1564,
+      "step": 6027
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3898846004739834e-05,
+      "loss": 0.1982,
+      "step": 6028
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3896909390856828e-05,
+      "loss": 0.2157,
+      "step": 6029
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.389497260463743e-05,
+      "loss": 0.1598,
+      "step": 6030
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3893035646167293e-05,
+      "loss": 0.192,
+      "step": 6031
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3891098515532073e-05,
+      "loss": 0.1966,
+      "step": 6032
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3889161212817444e-05,
+      "loss": 0.2594,
+      "step": 6033
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3887223738109072e-05,
+      "loss": 0.1792,
+      "step": 6034
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3885286091492649e-05,
+      "loss": 0.1829,
+      "step": 6035
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.388334827305386e-05,
+      "loss": 0.1891,
+      "step": 6036
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3881410282878404e-05,
+      "loss": 0.1668,
+      "step": 6037
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3879472121051984e-05,
+      "loss": 0.1882,
+      "step": 6038
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3877533787660316e-05,
+      "loss": 0.1944,
+      "step": 6039
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3875595282789123e-05,
+      "loss": 0.2034,
+      "step": 6040
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3873656606524126e-05,
+      "loss": 0.1856,
+      "step": 6041
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3871717758951066e-05,
+      "loss": 0.1812,
+      "step": 6042
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3869778740155686e-05,
+      "loss": 0.196,
+      "step": 6043
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3867839550223734e-05,
+      "loss": 0.2056,
+      "step": 6044
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3865900189240971e-05,
+      "loss": 0.2083,
+      "step": 6045
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3863960657293163e-05,
+      "loss": 0.1965,
+      "step": 6046
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3862020954466082e-05,
+      "loss": 0.1971,
+      "step": 6047
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3860081080845511e-05,
+      "loss": 0.1922,
+      "step": 6048
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3858141036517236e-05,
+      "loss": 0.2049,
+      "step": 6049
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3856200821567059e-05,
+      "loss": 0.1893,
+      "step": 6050
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.385426043608078e-05,
+      "loss": 0.18,
+      "step": 6051
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3852319880144205e-05,
+      "loss": 0.1871,
+      "step": 6052
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3850379153843162e-05,
+      "loss": 0.1868,
+      "step": 6053
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3848438257263474e-05,
+      "loss": 0.1808,
+      "step": 6054
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3846497190490973e-05,
+      "loss": 0.1916,
+      "step": 6055
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3844555953611503e-05,
+      "loss": 0.1868,
+      "step": 6056
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.384261454671091e-05,
+      "loss": 0.2047,
+      "step": 6057
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3840672969875055e-05,
+      "loss": 0.2332,
+      "step": 6058
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3838731223189799e-05,
+      "loss": 0.1885,
+      "step": 6059
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3836789306741014e-05,
+      "loss": 0.1916,
+      "step": 6060
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3834847220614578e-05,
+      "loss": 0.1777,
+      "step": 6061
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3832904964896381e-05,
+      "loss": 0.1972,
+      "step": 6062
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3830962539672312e-05,
+      "loss": 0.1961,
+      "step": 6063
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3829019945028277e-05,
+      "loss": 0.166,
+      "step": 6064
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3827077181050184e-05,
+      "loss": 0.1818,
+      "step": 6065
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3825134247823946e-05,
+      "loss": 0.2061,
+      "step": 6066
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.382319114543549e-05,
+      "loss": 0.2063,
+      "step": 6067
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3821247873970754e-05,
+      "loss": 0.2179,
+      "step": 6068
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3819304433515663e-05,
+      "loss": 0.1944,
+      "step": 6069
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3817360824156174e-05,
+      "loss": 0.1798,
+      "step": 6070
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3815417045978237e-05,
+      "loss": 0.2054,
+      "step": 6071
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3813473099067813e-05,
+      "loss": 0.1922,
+      "step": 6072
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3811528983510873e-05,
+      "loss": 0.1956,
+      "step": 6073
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3809584699393392e-05,
+      "loss": 0.1762,
+      "step": 6074
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3807640246801354e-05,
+      "loss": 0.1956,
+      "step": 6075
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3805695625820752e-05,
+      "loss": 0.1835,
+      "step": 6076
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3803750836537582e-05,
+      "loss": 0.2039,
+      "step": 6077
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3801805879037851e-05,
+      "loss": 0.1852,
+      "step": 6078
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3799860753407572e-05,
+      "loss": 0.2053,
+      "step": 6079
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3797915459732765e-05,
+      "loss": 0.1898,
+      "step": 6080
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3795969998099464e-05,
+      "loss": 0.1828,
+      "step": 6081
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3794024368593695e-05,
+      "loss": 0.2069,
+      "step": 6082
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.379207857130151e-05,
+      "loss": 0.2375,
+      "step": 6083
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3790132606308956e-05,
+      "loss": 0.194,
+      "step": 6084
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3788186473702095e-05,
+      "loss": 0.1851,
+      "step": 6085
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3786240173566985e-05,
+      "loss": 0.1841,
+      "step": 6086
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3784293705989704e-05,
+      "loss": 0.1881,
+      "step": 6087
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3782347071056334e-05,
+      "loss": 0.1997,
+      "step": 6088
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3780400268852956e-05,
+      "loss": 0.2017,
+      "step": 6089
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3778453299465673e-05,
+      "loss": 0.1961,
+      "step": 6090
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3776506162980582e-05,
+      "loss": 0.1969,
+      "step": 6091
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3774558859483795e-05,
+      "loss": 0.2076,
+      "step": 6092
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3772611389061429e-05,
+      "loss": 0.2001,
+      "step": 6093
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3770663751799609e-05,
+      "loss": 0.2292,
+      "step": 6094
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3768715947784467e-05,
+      "loss": 0.2011,
+      "step": 6095
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3766767977102142e-05,
+      "loss": 0.2149,
+      "step": 6096
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.376481983983878e-05,
+      "loss": 0.2109,
+      "step": 6097
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3762871536080538e-05,
+      "loss": 0.1697,
+      "step": 6098
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3760923065913573e-05,
+      "loss": 0.2173,
+      "step": 6099
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.375897442942406e-05,
+      "loss": 0.2185,
+      "step": 6100
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3757025626698169e-05,
+      "loss": 0.1978,
+      "step": 6101
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3755076657822088e-05,
+      "loss": 0.1985,
+      "step": 6102
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3753127522882005e-05,
+      "loss": 0.2074,
+      "step": 6103
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.375117822196412e-05,
+      "loss": 0.1837,
+      "step": 6104
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3749228755154638e-05,
+      "loss": 0.1809,
+      "step": 6105
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3747279122539773e-05,
+      "loss": 0.1797,
+      "step": 6106
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3745329324205741e-05,
+      "loss": 0.2089,
+      "step": 6107
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3743379360238773e-05,
+      "loss": 0.222,
+      "step": 6108
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.374142923072511e-05,
+      "loss": 0.1974,
+      "step": 6109
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3739478935750983e-05,
+      "loss": 0.2183,
+      "step": 6110
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3737528475402648e-05,
+      "loss": 0.2023,
+      "step": 6111
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3735577849766361e-05,
+      "loss": 0.1861,
+      "step": 6112
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3733627058928388e-05,
+      "loss": 0.178,
+      "step": 6113
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3731676102974993e-05,
+      "loss": 0.1686,
+      "step": 6114
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3729724981992468e-05,
+      "loss": 0.2033,
+      "step": 6115
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3727773696067084e-05,
+      "loss": 0.2129,
+      "step": 6116
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3725822245285143e-05,
+      "loss": 0.1973,
+      "step": 6117
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3723870629732947e-05,
+      "loss": 0.1842,
+      "step": 6118
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3721918849496803e-05,
+      "loss": 0.211,
+      "step": 6119
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.371996690466302e-05,
+      "loss": 0.188,
+      "step": 6120
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3718014795317926e-05,
+      "loss": 0.2011,
+      "step": 6121
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3716062521547852e-05,
+      "loss": 0.1888,
+      "step": 6122
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3714110083439134e-05,
+      "loss": 0.1529,
+      "step": 6123
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3712157481078112e-05,
+      "loss": 0.2072,
+      "step": 6124
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3710204714551144e-05,
+      "loss": 0.2028,
+      "step": 6125
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3708251783944585e-05,
+      "loss": 0.178,
+      "step": 6126
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3706298689344802e-05,
+      "loss": 0.2193,
+      "step": 6127
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3704345430838169e-05,
+      "loss": 0.1888,
+      "step": 6128
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3702392008511068e-05,
+      "loss": 0.2066,
+      "step": 6129
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.370043842244988e-05,
+      "loss": 0.184,
+      "step": 6130
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3698484672741006e-05,
+      "loss": 0.1942,
+      "step": 6131
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3696530759470849e-05,
+      "loss": 0.2008,
+      "step": 6132
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.369457668272582e-05,
+      "loss": 0.1771,
+      "step": 6133
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3692622442592332e-05,
+      "loss": 0.1818,
+      "step": 6134
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3690668039156805e-05,
+      "loss": 0.1936,
+      "step": 6135
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3688713472505682e-05,
+      "loss": 0.1968,
+      "step": 6136
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.368675874272539e-05,
+      "loss": 0.1702,
+      "step": 6137
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3684803849902381e-05,
+      "loss": 0.2065,
+      "step": 6138
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3682848794123107e-05,
+      "loss": 0.2266,
+      "step": 6139
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3680893575474026e-05,
+      "loss": 0.1947,
+      "step": 6140
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3678938194041607e-05,
+      "loss": 0.2093,
+      "step": 6141
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3676982649912325e-05,
+      "loss": 0.1897,
+      "step": 6142
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3675026943172664e-05,
+      "loss": 0.2046,
+      "step": 6143
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3673071073909104e-05,
+      "loss": 0.1938,
+      "step": 6144
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.367111504220815e-05,
+      "loss": 0.2229,
+      "step": 6145
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3669158848156302e-05,
+      "loss": 0.1977,
+      "step": 6146
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3667202491840073e-05,
+      "loss": 0.2458,
+      "step": 6147
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3665245973345977e-05,
+      "loss": 0.1767,
+      "step": 6148
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3663289292760538e-05,
+      "loss": 0.1744,
+      "step": 6149
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3661332450170295e-05,
+      "loss": 0.1805,
+      "step": 6150
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3659375445661777e-05,
+      "loss": 0.1907,
+      "step": 6151
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3657418279321539e-05,
+      "loss": 0.1739,
+      "step": 6152
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.365546095123613e-05,
+      "loss": 0.1891,
+      "step": 6153
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3653503461492112e-05,
+      "loss": 0.1959,
+      "step": 6154
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3651545810176053e-05,
+      "loss": 0.2011,
+      "step": 6155
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.364958799737453e-05,
+      "loss": 0.2163,
+      "step": 6156
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3647630023174118e-05,
+      "loss": 0.2024,
+      "step": 6157
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.364567188766141e-05,
+      "loss": 0.2133,
+      "step": 6158
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3643713590923005e-05,
+      "loss": 0.2065,
+      "step": 6159
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3641755133045507e-05,
+      "loss": 0.1939,
+      "step": 6160
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.363979651411552e-05,
+      "loss": 0.1792,
+      "step": 6161
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3637837734219668e-05,
+      "loss": 0.2084,
+      "step": 6162
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3635878793444572e-05,
+      "loss": 0.1824,
+      "step": 6163
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3633919691876868e-05,
+      "loss": 0.2033,
+      "step": 6164
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3631960429603188e-05,
+      "loss": 0.1677,
+      "step": 6165
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3630001006710188e-05,
+      "loss": 0.186,
+      "step": 6166
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.362804142328451e-05,
+      "loss": 0.1905,
+      "step": 6167
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3626081679412824e-05,
+      "loss": 0.2053,
+      "step": 6168
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3624121775181793e-05,
+      "loss": 0.2151,
+      "step": 6169
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3622161710678095e-05,
+      "loss": 0.2127,
+      "step": 6170
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3620201485988404e-05,
+      "loss": 0.2012,
+      "step": 6171
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3618241101199414e-05,
+      "loss": 0.1877,
+      "step": 6172
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3616280556397822e-05,
+      "loss": 0.2091,
+      "step": 6173
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.361431985167033e-05,
+      "loss": 0.2022,
+      "step": 6174
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3612358987103647e-05,
+      "loss": 0.1722,
+      "step": 6175
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.361039796278449e-05,
+      "loss": 0.1937,
+      "step": 6176
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3608436778799583e-05,
+      "loss": 0.1904,
+      "step": 6177
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3606475435235655e-05,
+      "loss": 0.2042,
+      "step": 6178
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3604513932179449e-05,
+      "loss": 0.2171,
+      "step": 6179
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3602552269717708e-05,
+      "loss": 0.1884,
+      "step": 6180
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3600590447937183e-05,
+      "loss": 0.1838,
+      "step": 6181
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3598628466924634e-05,
+      "loss": 0.1847,
+      "step": 6182
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3596666326766829e-05,
+      "loss": 0.167,
+      "step": 6183
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3594704027550543e-05,
+      "loss": 0.1931,
+      "step": 6184
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3592741569362549e-05,
+      "loss": 0.2122,
+      "step": 6185
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.359077895228964e-05,
+      "loss": 0.1724,
+      "step": 6186
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3588816176418613e-05,
+      "loss": 0.1872,
+      "step": 6187
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3586853241836262e-05,
+      "loss": 0.2007,
+      "step": 6188
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3584890148629403e-05,
+      "loss": 0.2015,
+      "step": 6189
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3582926896884847e-05,
+      "loss": 0.1957,
+      "step": 6190
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3580963486689415e-05,
+      "loss": 0.179,
+      "step": 6191
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3578999918129944e-05,
+      "loss": 0.1815,
+      "step": 6192
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3577036191293264e-05,
+      "loss": 0.1877,
+      "step": 6193
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.357507230626622e-05,
+      "loss": 0.2211,
+      "step": 6194
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3573108263135666e-05,
+      "loss": 0.1955,
+      "step": 6195
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3571144061988454e-05,
+      "loss": 0.1994,
+      "step": 6196
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3569179702911455e-05,
+      "loss": 0.1622,
+      "step": 6197
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3567215185991536e-05,
+      "loss": 0.1771,
+      "step": 6198
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3565250511315572e-05,
+      "loss": 0.2089,
+      "step": 6199
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3563285678970457e-05,
+      "loss": 0.1816,
+      "step": 6200
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3561320689043081e-05,
+      "loss": 0.1801,
+      "step": 6201
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3559355541620339e-05,
+      "loss": 0.1884,
+      "step": 6202
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.355739023678914e-05,
+      "loss": 0.173,
+      "step": 6203
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3555424774636399e-05,
+      "loss": 0.1993,
+      "step": 6204
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3553459155249033e-05,
+      "loss": 0.2166,
+      "step": 6205
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3551493378713973e-05,
+      "loss": 0.1951,
+      "step": 6206
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3549527445118153e-05,
+      "loss": 0.1987,
+      "step": 6207
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.354756135454851e-05,
+      "loss": 0.2101,
+      "step": 6208
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3545595107091996e-05,
+      "loss": 0.1945,
+      "step": 6209
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3543628702835564e-05,
+      "loss": 0.1776,
+      "step": 6210
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3541662141866181e-05,
+      "loss": 0.1747,
+      "step": 6211
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3539695424270809e-05,
+      "loss": 0.1896,
+      "step": 6212
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3537728550136424e-05,
+      "loss": 0.2071,
+      "step": 6213
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3535761519550017e-05,
+      "loss": 0.1924,
+      "step": 6214
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3533794332598569e-05,
+      "loss": 0.1737,
+      "step": 6215
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.353182698936908e-05,
+      "loss": 0.197,
+      "step": 6216
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3529859489948553e-05,
+      "loss": 0.2043,
+      "step": 6217
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3527891834423998e-05,
+      "loss": 0.2274,
+      "step": 6218
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3525924022882435e-05,
+      "loss": 0.1904,
+      "step": 6219
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3523956055410885e-05,
+      "loss": 0.2049,
+      "step": 6220
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3521987932096383e-05,
+      "loss": 0.2324,
+      "step": 6221
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.352001965302596e-05,
+      "loss": 0.1888,
+      "step": 6222
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3518051218286668e-05,
+      "loss": 0.2025,
+      "step": 6223
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3516082627965555e-05,
+      "loss": 0.2126,
+      "step": 6224
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3514113882149684e-05,
+      "loss": 0.2117,
+      "step": 6225
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3512144980926114e-05,
+      "loss": 0.1942,
+      "step": 6226
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.351017592438192e-05,
+      "loss": 0.1898,
+      "step": 6227
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3508206712604186e-05,
+      "loss": 0.1747,
+      "step": 6228
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3506237345679991e-05,
+      "loss": 0.2173,
+      "step": 6229
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3504267823696432e-05,
+      "loss": 0.1887,
+      "step": 6230
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3502298146740608e-05,
+      "loss": 0.198,
+      "step": 6231
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3500328314899626e-05,
+      "loss": 0.1651,
+      "step": 6232
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3498358328260601e-05,
+      "loss": 0.2146,
+      "step": 6233
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.349638818691065e-05,
+      "loss": 0.1856,
+      "step": 6234
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3494417890936903e-05,
+      "loss": 0.1876,
+      "step": 6235
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3492447440426492e-05,
+      "loss": 0.1992,
+      "step": 6236
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.349047683546656e-05,
+      "loss": 0.1892,
+      "step": 6237
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3488506076144257e-05,
+      "loss": 0.1901,
+      "step": 6238
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.348653516254673e-05,
+      "loss": 0.2133,
+      "step": 6239
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3484564094761146e-05,
+      "loss": 0.2171,
+      "step": 6240
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3482592872874669e-05,
+      "loss": 0.2021,
+      "step": 6241
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3480621496974485e-05,
+      "loss": 0.1916,
+      "step": 6242
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3478649967147761e-05,
+      "loss": 0.2618,
+      "step": 6243
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3476678283481696e-05,
+      "loss": 0.196,
+      "step": 6244
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.347470644606348e-05,
+      "loss": 0.1781,
+      "step": 6245
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3472734454980319e-05,
+      "loss": 0.1869,
+      "step": 6246
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.347076231031942e-05,
+      "loss": 0.2007,
+      "step": 6247
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3468790012168003e-05,
+      "loss": 0.1915,
+      "step": 6248
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.346681756061328e-05,
+      "loss": 0.1882,
+      "step": 6249
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3464844955742491e-05,
+      "loss": 0.1911,
+      "step": 6250
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3462872197642871e-05,
+      "loss": 0.1939,
+      "step": 6251
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3460899286401659e-05,
+      "loss": 0.2099,
+      "step": 6252
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3458926222106108e-05,
+      "loss": 0.202,
+      "step": 6253
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.345695300484347e-05,
+      "loss": 0.1769,
+      "step": 6254
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3454979634701012e-05,
+      "loss": 0.2034,
+      "step": 6255
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3453006111766002e-05,
+      "loss": 0.2147,
+      "step": 6256
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3451032436125719e-05,
+      "loss": 0.1976,
+      "step": 6257
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.344905860786745e-05,
+      "loss": 0.1748,
+      "step": 6258
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3447084627078476e-05,
+      "loss": 0.2243,
+      "step": 6259
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3445110493846099e-05,
+      "loss": 0.1874,
+      "step": 6260
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3443136208257623e-05,
+      "loss": 0.1998,
+      "step": 6261
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3441161770400361e-05,
+      "loss": 0.1757,
+      "step": 6262
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3439187180361625e-05,
+      "loss": 0.2081,
+      "step": 6263
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3437212438228743e-05,
+      "loss": 0.1907,
+      "step": 6264
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3435237544089043e-05,
+      "loss": 0.2271,
+      "step": 6265
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3433262498029869e-05,
+      "loss": 0.1985,
+      "step": 6266
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3431287300138557e-05,
+      "loss": 0.1947,
+      "step": 6267
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3429311950502456e-05,
+      "loss": 0.2109,
+      "step": 6268
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3427336449208934e-05,
+      "loss": 0.2165,
+      "step": 6269
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3425360796345348e-05,
+      "loss": 0.2363,
+      "step": 6270
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.342338499199907e-05,
+      "loss": 0.1829,
+      "step": 6271
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.342140903625748e-05,
+      "loss": 0.1809,
+      "step": 6272
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.341943292920796e-05,
+      "loss": 0.1931,
+      "step": 6273
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3417456670937902e-05,
+      "loss": 0.2139,
+      "step": 6274
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3415480261534706e-05,
+      "loss": 0.1925,
+      "step": 6275
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3413503701085773e-05,
+      "loss": 0.2298,
+      "step": 6276
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3411526989678511e-05,
+      "loss": 0.2036,
+      "step": 6277
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3409550127400346e-05,
+      "loss": 0.2094,
+      "step": 6278
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.34075731143387e-05,
+      "loss": 0.1836,
+      "step": 6279
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3405595950581003e-05,
+      "loss": 0.1993,
+      "step": 6280
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.340361863621469e-05,
+      "loss": 0.1855,
+      "step": 6281
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3401641171327207e-05,
+      "loss": 0.2196,
+      "step": 6282
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.339966355600601e-05,
+      "loss": 0.1749,
+      "step": 6283
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.339768579033855e-05,
+      "loss": 0.1954,
+      "step": 6284
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3395707874412298e-05,
+      "loss": 0.215,
+      "step": 6285
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3393729808314717e-05,
+      "loss": 0.1817,
+      "step": 6286
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3391751592133293e-05,
+      "loss": 0.2123,
+      "step": 6287
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3389773225955505e-05,
+      "loss": 0.2107,
+      "step": 6288
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.338779470986885e-05,
+      "loss": 0.1983,
+      "step": 6289
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3385816043960815e-05,
+      "loss": 0.1849,
+      "step": 6290
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3383837228318912e-05,
+      "loss": 0.2032,
+      "step": 6291
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3381858263030649e-05,
+      "loss": 0.2168,
+      "step": 6292
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.337987914818355e-05,
+      "loss": 0.1837,
+      "step": 6293
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3377899883865131e-05,
+      "loss": 0.2153,
+      "step": 6294
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3375920470162927e-05,
+      "loss": 0.1947,
+      "step": 6295
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.337394090716447e-05,
+      "loss": 0.1908,
+      "step": 6296
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3371961194957313e-05,
+      "loss": 0.1815,
+      "step": 6297
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3369981333628999e-05,
+      "loss": 0.1789,
+      "step": 6298
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3368001323267088e-05,
+      "loss": 0.2002,
+      "step": 6299
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3366021163959144e-05,
+      "loss": 0.2053,
+      "step": 6300
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3364040855792736e-05,
+      "loss": 0.1959,
+      "step": 6301
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.336206039885544e-05,
+      "loss": 0.1945,
+      "step": 6302
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3360079793234846e-05,
+      "loss": 0.1714,
+      "step": 6303
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3358099039018534e-05,
+      "loss": 0.1573,
+      "step": 6304
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3356118136294106e-05,
+      "loss": 0.1817,
+      "step": 6305
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3354137085149169e-05,
+      "loss": 0.1915,
+      "step": 6306
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3352155885671325e-05,
+      "loss": 0.1902,
+      "step": 6307
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3350174537948192e-05,
+      "loss": 0.2049,
+      "step": 6308
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.33481930420674e-05,
+      "loss": 0.221,
+      "step": 6309
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3346211398116569e-05,
+      "loss": 0.184,
+      "step": 6310
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3344229606183341e-05,
+      "loss": 0.192,
+      "step": 6311
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3342247666355354e-05,
+      "loss": 0.1789,
+      "step": 6312
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3340265578720261e-05,
+      "loss": 0.2122,
+      "step": 6313
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3338283343365716e-05,
+      "loss": 0.2092,
+      "step": 6314
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3336300960379382e-05,
+      "loss": 0.1652,
+      "step": 6315
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3334318429848927e-05,
+      "loss": 0.1829,
+      "step": 6316
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3332335751862026e-05,
+      "loss": 0.2113,
+      "step": 6317
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3330352926506363e-05,
+      "loss": 0.1985,
+      "step": 6318
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3328369953869617e-05,
+      "loss": 0.1833,
+      "step": 6319
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3326386834039497e-05,
+      "loss": 0.1867,
+      "step": 6320
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3324403567103694e-05,
+      "loss": 0.1841,
+      "step": 6321
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3322420153149917e-05,
+      "loss": 0.2077,
+      "step": 6322
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3320436592265885e-05,
+      "loss": 0.194,
+      "step": 6323
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3318452884539312e-05,
+      "loss": 0.1856,
+      "step": 6324
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.331646903005793e-05,
+      "loss": 0.181,
+      "step": 6325
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3314485028909472e-05,
+      "loss": 0.2126,
+      "step": 6326
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3312500881181679e-05,
+      "loss": 0.1839,
+      "step": 6327
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3310516586962293e-05,
+      "loss": 0.2035,
+      "step": 6328
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3308532146339072e-05,
+      "loss": 0.1822,
+      "step": 6329
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3306547559399778e-05,
+      "loss": 0.1702,
+      "step": 6330
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.330456282623217e-05,
+      "loss": 0.1849,
+      "step": 6331
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3302577946924023e-05,
+      "loss": 0.2161,
+      "step": 6332
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3300592921563116e-05,
+      "loss": 0.2301,
+      "step": 6333
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3298607750237238e-05,
+      "loss": 0.1981,
+      "step": 6334
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3296622433034179e-05,
+      "loss": 0.1858,
+      "step": 6335
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3294636970041735e-05,
+      "loss": 0.1793,
+      "step": 6336
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3292651361347711e-05,
+      "loss": 0.1772,
+      "step": 6337
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3290665607039921e-05,
+      "loss": 0.2183,
+      "step": 6338
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3288679707206183e-05,
+      "loss": 0.1652,
+      "step": 6339
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3286693661934317e-05,
+      "loss": 0.2218,
+      "step": 6340
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.328470747131216e-05,
+      "loss": 0.2015,
+      "step": 6341
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.328272113542754e-05,
+      "loss": 0.1824,
+      "step": 6342
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.328073465436831e-05,
+      "loss": 0.1851,
+      "step": 6343
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3278748028222316e-05,
+      "loss": 0.2006,
+      "step": 6344
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3276761257077415e-05,
+      "loss": 0.2122,
+      "step": 6345
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3274774341021464e-05,
+      "loss": 0.2228,
+      "step": 6346
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3272787280142341e-05,
+      "loss": 0.2035,
+      "step": 6347
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3270800074527913e-05,
+      "loss": 0.2079,
+      "step": 6348
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3268812724266069e-05,
+      "loss": 0.1848,
+      "step": 6349
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3266825229444695e-05,
+      "loss": 0.2216,
+      "step": 6350
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3264837590151685e-05,
+      "loss": 0.1784,
+      "step": 6351
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3262849806474938e-05,
+      "loss": 0.1769,
+      "step": 6352
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3260861878502365e-05,
+      "loss": 0.2105,
+      "step": 6353
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3258873806321881e-05,
+      "loss": 0.1648,
+      "step": 6354
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.32568855900214e-05,
+      "loss": 0.1909,
+      "step": 6355
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3254897229688855e-05,
+      "loss": 0.1815,
+      "step": 6356
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3252908725412174e-05,
+      "loss": 0.1885,
+      "step": 6357
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3250920077279304e-05,
+      "loss": 0.207,
+      "step": 6358
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3248931285378181e-05,
+      "loss": 0.2028,
+      "step": 6359
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3246942349796759e-05,
+      "loss": 0.1909,
+      "step": 6360
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3244953270623004e-05,
+      "loss": 0.2005,
+      "step": 6361
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3242964047944875e-05,
+      "loss": 0.183,
+      "step": 6362
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3240974681850343e-05,
+      "loss": 0.2185,
+      "step": 6363
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3238985172427386e-05,
+      "loss": 0.1965,
+      "step": 6364
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3236995519763989e-05,
+      "loss": 0.2272,
+      "step": 6365
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.323500572394814e-05,
+      "loss": 0.1822,
+      "step": 6366
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3233015785067838e-05,
+      "loss": 0.1713,
+      "step": 6367
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3231025703211084e-05,
+      "loss": 0.206,
+      "step": 6368
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3229035478465885e-05,
+      "loss": 0.1698,
+      "step": 6369
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.322704511092026e-05,
+      "loss": 0.1871,
+      "step": 6370
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3225054600662234e-05,
+      "loss": 0.1976,
+      "step": 6371
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3223063947779828e-05,
+      "loss": 0.1922,
+      "step": 6372
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3221073152361079e-05,
+      "loss": 0.2134,
+      "step": 6373
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3219082214494025e-05,
+      "loss": 0.1655,
+      "step": 6374
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3217091134266722e-05,
+      "loss": 0.198,
+      "step": 6375
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3215099911767214e-05,
+      "loss": 0.1921,
+      "step": 6376
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3213108547083562e-05,
+      "loss": 0.1883,
+      "step": 6377
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3211117040303836e-05,
+      "loss": 0.1895,
+      "step": 6378
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3209125391516105e-05,
+      "loss": 0.1642,
+      "step": 6379
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3207133600808446e-05,
+      "loss": 0.2023,
+      "step": 6380
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3205141668268948e-05,
+      "loss": 0.2148,
+      "step": 6381
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3203149593985698e-05,
+      "loss": 0.1965,
+      "step": 6382
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3201157378046794e-05,
+      "loss": 0.1879,
+      "step": 6383
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.319916502054034e-05,
+      "loss": 0.2088,
+      "step": 6384
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3197172521554454e-05,
+      "loss": 0.1901,
+      "step": 6385
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3195179881177236e-05,
+      "loss": 0.1973,
+      "step": 6386
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3193187099496819e-05,
+      "loss": 0.1923,
+      "step": 6387
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3191194176601328e-05,
+      "loss": 0.2172,
+      "step": 6388
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3189201112578898e-05,
+      "loss": 0.1908,
+      "step": 6389
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3187207907517672e-05,
+      "loss": 0.1761,
+      "step": 6390
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3185214561505795e-05,
+      "loss": 0.2202,
+      "step": 6391
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3183221074631421e-05,
+      "loss": 0.209,
+      "step": 6392
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3181227446982709e-05,
+      "loss": 0.1733,
+      "step": 6393
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.317923367864783e-05,
+      "loss": 0.1949,
+      "step": 6394
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3177239769714951e-05,
+      "loss": 0.2292,
+      "step": 6395
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3175245720272248e-05,
+      "loss": 0.1718,
+      "step": 6396
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.317325153040791e-05,
+      "loss": 0.1947,
+      "step": 6397
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3171257200210128e-05,
+      "loss": 0.2094,
+      "step": 6398
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3169262729767098e-05,
+      "loss": 0.1838,
+      "step": 6399
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3167268119167023e-05,
+      "loss": 0.2155,
+      "step": 6400
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3165273368498111e-05,
+      "loss": 0.1916,
+      "step": 6401
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.316327847784858e-05,
+      "loss": 0.1782,
+      "step": 6402
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.316128344730665e-05,
+      "loss": 0.1773,
+      "step": 6403
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3159288276960551e-05,
+      "loss": 0.1934,
+      "step": 6404
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3157292966898514e-05,
+      "loss": 0.1784,
+      "step": 6405
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3155297517208783e-05,
+      "loss": 0.1903,
+      "step": 6406
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3153301927979601e-05,
+      "loss": 0.2084,
+      "step": 6407
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3151306199299221e-05,
+      "loss": 0.1811,
+      "step": 6408
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3149310331255909e-05,
+      "loss": 0.2053,
+      "step": 6409
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3147314323937917e-05,
+      "loss": 0.1885,
+      "step": 6410
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3145318177433526e-05,
+      "loss": 0.1827,
+      "step": 6411
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3143321891831011e-05,
+      "loss": 0.1939,
+      "step": 6412
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3141325467218655e-05,
+      "loss": 0.1939,
+      "step": 6413
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.313932890368475e-05,
+      "loss": 0.1774,
+      "step": 6414
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3137332201317583e-05,
+      "loss": 0.1687,
+      "step": 6415
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3135335360205467e-05,
+      "loss": 0.1823,
+      "step": 6416
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3133338380436707e-05,
+      "loss": 0.1678,
+      "step": 6417
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.313134126209961e-05,
+      "loss": 0.1662,
+      "step": 6418
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3129344005282504e-05,
+      "loss": 0.1905,
+      "step": 6419
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3127346610073715e-05,
+      "loss": 0.1992,
+      "step": 6420
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3125349076561574e-05,
+      "loss": 0.235,
+      "step": 6421
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.312335140483442e-05,
+      "loss": 0.191,
+      "step": 6422
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3121353594980595e-05,
+      "loss": 0.1951,
+      "step": 6423
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.311935564708845e-05,
+      "loss": 0.1916,
+      "step": 6424
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3117357561246346e-05,
+      "loss": 0.2052,
+      "step": 6425
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3115359337542646e-05,
+      "loss": 0.195,
+      "step": 6426
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3113360976065716e-05,
+      "loss": 0.1686,
+      "step": 6427
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3111362476903931e-05,
+      "loss": 0.213,
+      "step": 6428
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3109363840145674e-05,
+      "loss": 0.2094,
+      "step": 6429
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3107365065879334e-05,
+      "loss": 0.1664,
+      "step": 6430
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3105366154193302e-05,
+      "loss": 0.1947,
+      "step": 6431
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3103367105175976e-05,
+      "loss": 0.1809,
+      "step": 6432
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3101367918915767e-05,
+      "loss": 0.1902,
+      "step": 6433
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3099368595501085e-05,
+      "loss": 0.1815,
+      "step": 6434
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3097369135020346e-05,
+      "loss": 0.2028,
+      "step": 6435
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3095369537561974e-05,
+      "loss": 0.1997,
+      "step": 6436
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3093369803214399e-05,
+      "loss": 0.1844,
+      "step": 6437
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3091369932066056e-05,
+      "loss": 0.183,
+      "step": 6438
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3089369924205393e-05,
+      "loss": 0.2006,
+      "step": 6439
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.308736977972085e-05,
+      "loss": 0.201,
+      "step": 6440
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3085369498700884e-05,
+      "loss": 0.1724,
+      "step": 6441
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3083369081233957e-05,
+      "loss": 0.228,
+      "step": 6442
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3081368527408533e-05,
+      "loss": 0.1677,
+      "step": 6443
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3079367837313087e-05,
+      "loss": 0.1743,
+      "step": 6444
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3077367011036095e-05,
+      "loss": 0.181,
+      "step": 6445
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3075366048666041e-05,
+      "loss": 0.1793,
+      "step": 6446
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3073364950291416e-05,
+      "loss": 0.1958,
+      "step": 6447
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3071363716000715e-05,
+      "loss": 0.2069,
+      "step": 6448
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3069362345882443e-05,
+      "loss": 0.2196,
+      "step": 6449
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3067360840025112e-05,
+      "loss": 0.199,
+      "step": 6450
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3065359198517226e-05,
+      "loss": 0.1982,
+      "step": 6451
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.306335742144731e-05,
+      "loss": 0.1961,
+      "step": 6452
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3061355508903892e-05,
+      "loss": 0.2043,
+      "step": 6453
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3059353460975503e-05,
+      "loss": 0.198,
+      "step": 6454
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3057351277750683e-05,
+      "loss": 0.2106,
+      "step": 6455
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3055348959317972e-05,
+      "loss": 0.2066,
+      "step": 6456
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3053346505765924e-05,
+      "loss": 0.2297,
+      "step": 6457
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3051343917183097e-05,
+      "loss": 0.1828,
+      "step": 6458
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3049341193658047e-05,
+      "loss": 0.1908,
+      "step": 6459
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.304733833527935e-05,
+      "loss": 0.1816,
+      "step": 6460
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3045335342135572e-05,
+      "loss": 0.1837,
+      "step": 6461
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3043332214315299e-05,
+      "loss": 0.2099,
+      "step": 6462
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3041328951907115e-05,
+      "loss": 0.1866,
+      "step": 6463
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3039325554999612e-05,
+      "loss": 0.2056,
+      "step": 6464
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3037322023681383e-05,
+      "loss": 0.2232,
+      "step": 6465
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.303531835804104e-05,
+      "loss": 0.1622,
+      "step": 6466
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3033314558167193e-05,
+      "loss": 0.2133,
+      "step": 6467
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3031310624148452e-05,
+      "loss": 0.1684,
+      "step": 6468
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.302930655607344e-05,
+      "loss": 0.1632,
+      "step": 6469
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3027302354030788e-05,
+      "loss": 0.1872,
+      "step": 6470
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3025298018109126e-05,
+      "loss": 0.1963,
+      "step": 6471
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3023293548397096e-05,
+      "loss": 0.2142,
+      "step": 6472
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.302128894498334e-05,
+      "loss": 0.2167,
+      "step": 6473
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3019284207956511e-05,
+      "loss": 0.2164,
+      "step": 6474
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3017279337405269e-05,
+      "loss": 0.1901,
+      "step": 6475
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3015274333418275e-05,
+      "loss": 0.2226,
+      "step": 6476
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.30132691960842e-05,
+      "loss": 0.2087,
+      "step": 6477
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3011263925491713e-05,
+      "loss": 0.1919,
+      "step": 6478
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3009258521729498e-05,
+      "loss": 0.1855,
+      "step": 6479
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3007252984886245e-05,
+      "loss": 0.2042,
+      "step": 6480
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3005247315050643e-05,
+      "loss": 0.1978,
+      "step": 6481
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3003241512311391e-05,
+      "loss": 0.1672,
+      "step": 6482
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3001235576757195e-05,
+      "loss": 0.1826,
+      "step": 6483
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2999229508476763e-05,
+      "loss": 0.1845,
+      "step": 6484
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2997223307558811e-05,
+      "loss": 0.1922,
+      "step": 6485
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2995216974092061e-05,
+      "loss": 0.2323,
+      "step": 6486
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2993210508165248e-05,
+      "loss": 0.2047,
+      "step": 6487
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2991203909867091e-05,
+      "loss": 0.1785,
+      "step": 6488
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2989197179286344e-05,
+      "loss": 0.2278,
+      "step": 6489
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2987190316511744e-05,
+      "loss": 0.1875,
+      "step": 6490
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2985183321632046e-05,
+      "loss": 0.2043,
+      "step": 6491
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2983176194736008e-05,
+      "loss": 0.2202,
+      "step": 6492
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2981168935912385e-05,
+      "loss": 0.1902,
+      "step": 6493
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2979161545249956e-05,
+      "loss": 0.2014,
+      "step": 6494
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2977154022837488e-05,
+      "loss": 0.202,
+      "step": 6495
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2975146368763767e-05,
+      "loss": 0.1925,
+      "step": 6496
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2973138583117573e-05,
+      "loss": 0.191,
+      "step": 6497
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2971130665987706e-05,
+      "loss": 0.1785,
+      "step": 6498
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2969122617462955e-05,
+      "loss": 0.1779,
+      "step": 6499
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2967114437632132e-05,
+      "loss": 0.2335,
+      "step": 6500
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2965106126584041e-05,
+      "loss": 0.2021,
+      "step": 6501
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2963097684407497e-05,
+      "loss": 0.177,
+      "step": 6502
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2961089111191325e-05,
+      "loss": 0.2062,
+      "step": 6503
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2959080407024352e-05,
+      "loss": 0.1919,
+      "step": 6504
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2957071571995404e-05,
+      "loss": 0.1852,
+      "step": 6505
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2955062606193325e-05,
+      "loss": 0.2263,
+      "step": 6506
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2953053509706956e-05,
+      "loss": 0.213,
+      "step": 6507
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2951044282625156e-05,
+      "loss": 0.211,
+      "step": 6508
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.294903492503677e-05,
+      "loss": 0.1902,
+      "step": 6509
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2947025437030662e-05,
+      "loss": 0.1867,
+      "step": 6510
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2945015818695704e-05,
+      "loss": 0.1797,
+      "step": 6511
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2943006070120763e-05,
+      "loss": 0.1762,
+      "step": 6512
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2940996191394722e-05,
+      "loss": 0.1944,
+      "step": 6513
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2938986182606466e-05,
+      "loss": 0.2218,
+      "step": 6514
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2936976043844883e-05,
+      "loss": 0.1894,
+      "step": 6515
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2934965775198866e-05,
+      "loss": 0.2292,
+      "step": 6516
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2932955376757324e-05,
+      "loss": 0.2006,
+      "step": 6517
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2930944848609161e-05,
+      "loss": 0.2236,
+      "step": 6518
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2928934190843291e-05,
+      "loss": 0.1613,
+      "step": 6519
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2926923403548632e-05,
+      "loss": 0.1734,
+      "step": 6520
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2924912486814109e-05,
+      "loss": 0.1755,
+      "step": 6521
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2922901440728655e-05,
+      "loss": 0.1809,
+      "step": 6522
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.29208902653812e-05,
+      "loss": 0.2004,
+      "step": 6523
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2918878960860695e-05,
+      "loss": 0.2038,
+      "step": 6524
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2916867527256081e-05,
+      "loss": 0.2032,
+      "step": 6525
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2914855964656312e-05,
+      "loss": 0.193,
+      "step": 6526
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.291284427315035e-05,
+      "loss": 0.1877,
+      "step": 6527
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2910832452827157e-05,
+      "loss": 0.1908,
+      "step": 6528
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2908820503775704e-05,
+      "loss": 0.1986,
+      "step": 6529
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2906808426084967e-05,
+      "loss": 0.1988,
+      "step": 6530
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.290479621984393e-05,
+      "loss": 0.1962,
+      "step": 6531
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.290278388514158e-05,
+      "loss": 0.203,
+      "step": 6532
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2900771422066906e-05,
+      "loss": 0.1665,
+      "step": 6533
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.289875883070891e-05,
+      "loss": 0.1875,
+      "step": 6534
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.28967461111566e-05,
+      "loss": 0.1939,
+      "step": 6535
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2894733263498982e-05,
+      "loss": 0.1657,
+      "step": 6536
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2892720287825071e-05,
+      "loss": 0.2266,
+      "step": 6537
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2890707184223892e-05,
+      "loss": 0.2035,
+      "step": 6538
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.288869395278447e-05,
+      "loss": 0.209,
+      "step": 6539
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.288668059359584e-05,
+      "loss": 0.233,
+      "step": 6540
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2884667106747036e-05,
+      "loss": 0.2185,
+      "step": 6541
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2882653492327109e-05,
+      "loss": 0.2208,
+      "step": 6542
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2880639750425099e-05,
+      "loss": 0.2144,
+      "step": 6543
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2878625881130071e-05,
+      "loss": 0.181,
+      "step": 6544
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2876611884531085e-05,
+      "loss": 0.191,
+      "step": 6545
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2874597760717204e-05,
+      "loss": 0.1757,
+      "step": 6546
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2872583509777499e-05,
+      "loss": 0.203,
+      "step": 6547
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.287056913180105e-05,
+      "loss": 0.1649,
+      "step": 6548
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2868554626876946e-05,
+      "loss": 0.1975,
+      "step": 6549
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2866539995094267e-05,
+      "loss": 0.159,
+      "step": 6550
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2864525236542113e-05,
+      "loss": 0.1892,
+      "step": 6551
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2862510351309585e-05,
+      "loss": 0.1775,
+      "step": 6552
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2860495339485785e-05,
+      "loss": 0.2288,
+      "step": 6553
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.285848020115983e-05,
+      "loss": 0.1878,
+      "step": 6554
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2856464936420833e-05,
+      "loss": 0.1918,
+      "step": 6555
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2854449545357919e-05,
+      "loss": 0.2169,
+      "step": 6556
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2852434028060211e-05,
+      "loss": 0.1689,
+      "step": 6557
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2850418384616852e-05,
+      "loss": 0.1918,
+      "step": 6558
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2848402615116977e-05,
+      "loss": 0.1615,
+      "step": 6559
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2846386719649732e-05,
+      "loss": 0.212,
+      "step": 6560
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2844370698304265e-05,
+      "loss": 0.1655,
+      "step": 6561
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2842354551169733e-05,
+      "loss": 0.1794,
+      "step": 6562
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.28403382783353e-05,
+      "loss": 0.1917,
+      "step": 6563
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2838321879890132e-05,
+      "loss": 0.1645,
+      "step": 6564
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2836305355923402e-05,
+      "loss": 0.1968,
+      "step": 6565
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.283428870652429e-05,
+      "loss": 0.2118,
+      "step": 6566
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2832271931781976e-05,
+      "loss": 0.1833,
+      "step": 6567
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2830255031785655e-05,
+      "loss": 0.2006,
+      "step": 6568
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.282823800662452e-05,
+      "loss": 0.1665,
+      "step": 6569
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.282622085638777e-05,
+      "loss": 0.1695,
+      "step": 6570
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2824203581164609e-05,
+      "loss": 0.2009,
+      "step": 6571
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2822186181044255e-05,
+      "loss": 0.2129,
+      "step": 6572
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2820168656115923e-05,
+      "loss": 0.2028,
+      "step": 6573
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2818151006468833e-05,
+      "loss": 0.1845,
+      "step": 6574
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2816133232192217e-05,
+      "loss": 0.1612,
+      "step": 6575
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2814115333375306e-05,
+      "loss": 0.1831,
+      "step": 6576
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2812097310107338e-05,
+      "loss": 0.2064,
+      "step": 6577
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2810079162477562e-05,
+      "loss": 0.2161,
+      "step": 6578
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2808060890575227e-05,
+      "loss": 0.176,
+      "step": 6579
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2806042494489589e-05,
+      "loss": 0.2033,
+      "step": 6580
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2804023974309904e-05,
+      "loss": 0.2284,
+      "step": 6581
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2802005330125447e-05,
+      "loss": 0.1795,
+      "step": 6582
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2799986562025486e-05,
+      "loss": 0.1764,
+      "step": 6583
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2797967670099298e-05,
+      "loss": 0.1904,
+      "step": 6584
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2795948654436166e-05,
+      "loss": 0.1892,
+      "step": 6585
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2793929515125383e-05,
+      "loss": 0.1589,
+      "step": 6586
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2791910252256238e-05,
+      "loss": 0.1832,
+      "step": 6587
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2789890865918033e-05,
+      "loss": 0.1739,
+      "step": 6588
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2787871356200074e-05,
+      "loss": 0.1941,
+      "step": 6589
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2785851723191668e-05,
+      "loss": 0.1782,
+      "step": 6590
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2783831966982136e-05,
+      "loss": 0.1816,
+      "step": 6591
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2781812087660795e-05,
+      "loss": 0.2215,
+      "step": 6592
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2779792085316972e-05,
+      "loss": 0.2136,
+      "step": 6593
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2777771960040004e-05,
+      "loss": 0.1861,
+      "step": 6594
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2775751711919223e-05,
+      "loss": 0.1798,
+      "step": 6595
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.277373134104398e-05,
+      "loss": 0.2043,
+      "step": 6596
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2771710847503615e-05,
+      "loss": 0.1845,
+      "step": 6597
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2769690231387483e-05,
+      "loss": 0.1873,
+      "step": 6598
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.276766949278495e-05,
+      "loss": 0.1895,
+      "step": 6599
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2765648631785375e-05,
+      "loss": 0.1956,
+      "step": 6600
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.276362764847813e-05,
+      "loss": 0.1727,
+      "step": 6601
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.276160654295259e-05,
+      "loss": 0.2132,
+      "step": 6602
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2759585315298138e-05,
+      "loss": 0.1692,
+      "step": 6603
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2757563965604158e-05,
+      "loss": 0.2154,
+      "step": 6604
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2755542493960045e-05,
+      "loss": 0.1884,
+      "step": 6605
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2753520900455194e-05,
+      "loss": 0.1836,
+      "step": 6606
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2751499185179007e-05,
+      "loss": 0.194,
+      "step": 6607
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2749477348220894e-05,
+      "loss": 0.1891,
+      "step": 6608
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2747455389670267e-05,
+      "loss": 0.1755,
+      "step": 6609
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2745433309616548e-05,
+      "loss": 0.2046,
+      "step": 6610
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2743411108149157e-05,
+      "loss": 0.1724,
+      "step": 6611
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2741388785357524e-05,
+      "loss": 0.2138,
+      "step": 6612
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2739366341331087e-05,
+      "loss": 0.1799,
+      "step": 6613
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2737343776159285e-05,
+      "loss": 0.1975,
+      "step": 6614
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2735321089931562e-05,
+      "loss": 0.1592,
+      "step": 6615
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2733298282737372e-05,
+      "loss": 0.1781,
+      "step": 6616
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2731275354666166e-05,
+      "loss": 0.181,
+      "step": 6617
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.272925230580741e-05,
+      "loss": 0.2333,
+      "step": 6618
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2727229136250574e-05,
+      "loss": 0.1966,
+      "step": 6619
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2725205846085124e-05,
+      "loss": 0.1937,
+      "step": 6620
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2723182435400539e-05,
+      "loss": 0.1845,
+      "step": 6621
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2721158904286304e-05,
+      "loss": 0.1729,
+      "step": 6622
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2719135252831909e-05,
+      "loss": 0.2116,
+      "step": 6623
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2717111481126844e-05,
+      "loss": 0.2136,
+      "step": 6624
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.271508758926061e-05,
+      "loss": 0.2159,
+      "step": 6625
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.271306357732271e-05,
+      "loss": 0.214,
+      "step": 6626
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2711039445402656e-05,
+      "loss": 0.1973,
+      "step": 6627
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2709015193589957e-05,
+      "loss": 0.1681,
+      "step": 6628
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2706990821974141e-05,
+      "loss": 0.2027,
+      "step": 6629
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.270496633064473e-05,
+      "loss": 0.2148,
+      "step": 6630
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2702941719691253e-05,
+      "loss": 0.2099,
+      "step": 6631
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.270091698920325e-05,
+      "loss": 0.1835,
+      "step": 6632
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.269889213927026e-05,
+      "loss": 0.1967,
+      "step": 6633
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2696867169981832e-05,
+      "loss": 0.1708,
+      "step": 6634
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2694842081427509e-05,
+      "loss": 0.2015,
+      "step": 6635
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2692816873696862e-05,
+      "loss": 0.1914,
+      "step": 6636
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2690791546879449e-05,
+      "loss": 0.2059,
+      "step": 6637
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2688766101064829e-05,
+      "loss": 0.1926,
+      "step": 6638
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2686740536342585e-05,
+      "loss": 0.1946,
+      "step": 6639
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.268471485280229e-05,
+      "loss": 0.1847,
+      "step": 6640
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2682689050533531e-05,
+      "loss": 0.2014,
+      "step": 6641
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2680663129625896e-05,
+      "loss": 0.1793,
+      "step": 6642
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2678637090168979e-05,
+      "loss": 0.175,
+      "step": 6643
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2676610932252375e-05,
+      "loss": 0.1885,
+      "step": 6644
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2674584655965696e-05,
+      "loss": 0.1764,
+      "step": 6645
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2672558261398546e-05,
+      "loss": 0.2081,
+      "step": 6646
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2670531748640545e-05,
+      "loss": 0.2095,
+      "step": 6647
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2668505117781306e-05,
+      "loss": 0.1851,
+      "step": 6648
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.266647836891046e-05,
+      "loss": 0.1813,
+      "step": 6649
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2664451502117636e-05,
+      "loss": 0.2182,
+      "step": 6650
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2662424517492473e-05,
+      "loss": 0.1864,
+      "step": 6651
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2660397415124605e-05,
+      "loss": 0.1702,
+      "step": 6652
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2658370195103687e-05,
+      "loss": 0.2179,
+      "step": 6653
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2656342857519364e-05,
+      "loss": 0.1873,
+      "step": 6654
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2654315402461292e-05,
+      "loss": 0.1759,
+      "step": 6655
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2652287830019139e-05,
+      "loss": 0.2167,
+      "step": 6656
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2650260140282568e-05,
+      "loss": 0.1912,
+      "step": 6657
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2648232333341252e-05,
+      "loss": 0.2139,
+      "step": 6658
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2646204409284869e-05,
+      "loss": 0.1991,
+      "step": 6659
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.26441763682031e-05,
+      "loss": 0.2224,
+      "step": 6660
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2642148210185636e-05,
+      "loss": 0.1906,
+      "step": 6661
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2640119935322166e-05,
+      "loss": 0.2232,
+      "step": 6662
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.263809154370239e-05,
+      "loss": 0.1734,
+      "step": 6663
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2636063035416016e-05,
+      "loss": 0.2102,
+      "step": 6664
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2634034410552742e-05,
+      "loss": 0.1931,
+      "step": 6665
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2632005669202289e-05,
+      "loss": 0.2112,
+      "step": 6666
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2629976811454374e-05,
+      "loss": 0.1931,
+      "step": 6667
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2627947837398723e-05,
+      "loss": 0.1908,
+      "step": 6668
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2625918747125064e-05,
+      "loss": 0.1908,
+      "step": 6669
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2623889540723129e-05,
+      "loss": 0.2042,
+      "step": 6670
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2621860218282658e-05,
+      "loss": 0.168,
+      "step": 6671
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2619830779893397e-05,
+      "loss": 0.1697,
+      "step": 6672
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2617801225645094e-05,
+      "loss": 0.2253,
+      "step": 6673
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2615771555627507e-05,
+      "loss": 0.1654,
+      "step": 6674
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2613741769930394e-05,
+      "loss": 0.2041,
+      "step": 6675
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2611711868643517e-05,
+      "loss": 0.2018,
+      "step": 6676
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2609681851856648e-05,
+      "loss": 0.1844,
+      "step": 6677
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2607651719659565e-05,
+      "loss": 0.1736,
+      "step": 6678
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2605621472142047e-05,
+      "loss": 0.1983,
+      "step": 6679
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2603591109393876e-05,
+      "loss": 0.1906,
+      "step": 6680
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2601560631504843e-05,
+      "loss": 0.2173,
+      "step": 6681
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2599530038564752e-05,
+      "loss": 0.1952,
+      "step": 6682
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2597499330663392e-05,
+      "loss": 0.1884,
+      "step": 6683
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2595468507890577e-05,
+      "loss": 0.1551,
+      "step": 6684
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2593437570336111e-05,
+      "loss": 0.1703,
+      "step": 6685
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2591406518089817e-05,
+      "loss": 0.169,
+      "step": 6686
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.258937535124151e-05,
+      "loss": 0.1782,
+      "step": 6687
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2587344069881024e-05,
+      "loss": 0.1843,
+      "step": 6688
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.258531267409818e-05,
+      "loss": 0.2001,
+      "step": 6689
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2583281163982819e-05,
+      "loss": 0.1887,
+      "step": 6690
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2581249539624782e-05,
+      "loss": 0.1892,
+      "step": 6691
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2579217801113919e-05,
+      "loss": 0.1922,
+      "step": 6692
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2577185948540076e-05,
+      "loss": 0.1927,
+      "step": 6693
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2575153981993112e-05,
+      "loss": 0.1924,
+      "step": 6694
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2573121901562884e-05,
+      "loss": 0.2044,
+      "step": 6695
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2571089707339266e-05,
+      "loss": 0.1818,
+      "step": 6696
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2569057399412124e-05,
+      "loss": 0.2005,
+      "step": 6697
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2567024977871336e-05,
+      "loss": 0.1659,
+      "step": 6698
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2564992442806782e-05,
+      "loss": 0.1949,
+      "step": 6699
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.256295979430835e-05,
+      "loss": 0.2009,
+      "step": 6700
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2560927032465935e-05,
+      "loss": 0.2008,
+      "step": 6701
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2558894157369433e-05,
+      "loss": 0.1857,
+      "step": 6702
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2556861169108739e-05,
+      "loss": 0.2021,
+      "step": 6703
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.255482806777376e-05,
+      "loss": 0.1907,
+      "step": 6704
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2552794853454416e-05,
+      "loss": 0.1755,
+      "step": 6705
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2550761526240616e-05,
+      "loss": 0.2191,
+      "step": 6706
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2548728086222286e-05,
+      "loss": 0.183,
+      "step": 6707
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.254669453348935e-05,
+      "loss": 0.1969,
+      "step": 6708
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.254466086813174e-05,
+      "loss": 0.1854,
+      "step": 6709
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2542627090239395e-05,
+      "loss": 0.196,
+      "step": 6710
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2540593199902251e-05,
+      "loss": 0.2052,
+      "step": 6711
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2538559197210263e-05,
+      "loss": 0.2042,
+      "step": 6712
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.253652508225337e-05,
+      "loss": 0.1836,
+      "step": 6713
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2534490855121542e-05,
+      "loss": 0.173,
+      "step": 6714
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.253245651590473e-05,
+      "loss": 0.2297,
+      "step": 6715
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2530422064692908e-05,
+      "loss": 0.1877,
+      "step": 6716
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2528387501576045e-05,
+      "loss": 0.2006,
+      "step": 6717
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.252635282664411e-05,
+      "loss": 0.1869,
+      "step": 6718
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2524318039987096e-05,
+      "loss": 0.1838,
+      "step": 6719
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2522283141694982e-05,
+      "loss": 0.1811,
+      "step": 6720
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.252024813185776e-05,
+      "loss": 0.1966,
+      "step": 6721
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2518213010565425e-05,
+      "loss": 0.189,
+      "step": 6722
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.251617777790798e-05,
+      "loss": 0.2316,
+      "step": 6723
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2514142433975434e-05,
+      "loss": 0.191,
+      "step": 6724
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2512106978857789e-05,
+      "loss": 0.1799,
+      "step": 6725
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2510071412645068e-05,
+      "loss": 0.1837,
+      "step": 6726
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.250803573542729e-05,
+      "loss": 0.1931,
+      "step": 6727
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2505999947294478e-05,
+      "loss": 0.2129,
+      "step": 6728
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2503964048336668e-05,
+      "loss": 0.1745,
+      "step": 6729
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2501928038643889e-05,
+      "loss": 0.198,
+      "step": 6730
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2499891918306181e-05,
+      "loss": 0.1781,
+      "step": 6731
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2497855687413596e-05,
+      "loss": 0.1678,
+      "step": 6732
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.249581934605618e-05,
+      "loss": 0.1934,
+      "step": 6733
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2493782894323987e-05,
+      "loss": 0.2219,
+      "step": 6734
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2491746332307075e-05,
+      "loss": 0.1956,
+      "step": 6735
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2489709660095513e-05,
+      "loss": 0.181,
+      "step": 6736
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2487672877779368e-05,
+      "loss": 0.1709,
+      "step": 6737
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2485635985448716e-05,
+      "loss": 0.2146,
+      "step": 6738
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2483598983193637e-05,
+      "loss": 0.2101,
+      "step": 6739
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.248156187110421e-05,
+      "loss": 0.1941,
+      "step": 6740
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.247952464927053e-05,
+      "loss": 0.1993,
+      "step": 6741
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2477487317782686e-05,
+      "loss": 0.1966,
+      "step": 6742
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2475449876730783e-05,
+      "loss": 0.1963,
+      "step": 6743
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2473412326204918e-05,
+      "loss": 0.2008,
+      "step": 6744
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.24713746662952e-05,
+      "loss": 0.2039,
+      "step": 6745
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2469336897091747e-05,
+      "loss": 0.174,
+      "step": 6746
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2467299018684672e-05,
+      "loss": 0.194,
+      "step": 6747
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.24652610311641e-05,
+      "loss": 0.2144,
+      "step": 6748
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2463222934620158e-05,
+      "loss": 0.1828,
+      "step": 6749
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2461184729142981e-05,
+      "loss": 0.1665,
+      "step": 6750
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2459146414822702e-05,
+      "loss": 0.241,
+      "step": 6751
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2457107991749469e-05,
+      "loss": 0.184,
+      "step": 6752
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2455069460013425e-05,
+      "loss": 0.1855,
+      "step": 6753
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2453030819704716e-05,
+      "loss": 0.2035,
+      "step": 6754
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.245099207091351e-05,
+      "loss": 0.166,
+      "step": 6755
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2448953213729964e-05,
+      "loss": 0.1952,
+      "step": 6756
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2446914248244245e-05,
+      "loss": 0.2152,
+      "step": 6757
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2444875174546518e-05,
+      "loss": 0.1955,
+      "step": 6758
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.244283599272696e-05,
+      "loss": 0.1855,
+      "step": 6759
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2440796702875761e-05,
+      "loss": 0.1896,
+      "step": 6760
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2438757305083098e-05,
+      "loss": 0.1769,
+      "step": 6761
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2436717799439163e-05,
+      "loss": 0.1723,
+      "step": 6762
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2434678186034147e-05,
+      "loss": 0.1751,
+      "step": 6763
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2432638464958256e-05,
+      "loss": 0.1888,
+      "step": 6764
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.243059863630169e-05,
+      "loss": 0.1944,
+      "step": 6765
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2428558700154661e-05,
+      "loss": 0.2036,
+      "step": 6766
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2426518656607382e-05,
+      "loss": 0.1879,
+      "step": 6767
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2424478505750068e-05,
+      "loss": 0.204,
+      "step": 6768
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2422438247672948e-05,
+      "loss": 0.1931,
+      "step": 6769
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.242039788246625e-05,
+      "loss": 0.1758,
+      "step": 6770
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2418357410220201e-05,
+      "loss": 0.1709,
+      "step": 6771
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2416316831025043e-05,
+      "loss": 0.1885,
+      "step": 6772
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2414276144971014e-05,
+      "loss": 0.2056,
+      "step": 6773
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.241223535214837e-05,
+      "loss": 0.2014,
+      "step": 6774
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2410194452647353e-05,
+      "loss": 0.1909,
+      "step": 6775
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2408153446558223e-05,
+      "loss": 0.189,
+      "step": 6776
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2406112333971243e-05,
+      "loss": 0.2106,
+      "step": 6777
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2404071114976679e-05,
+      "loss": 0.2263,
+      "step": 6778
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2402029789664798e-05,
+      "loss": 0.2049,
+      "step": 6779
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2399988358125881e-05,
+      "loss": 0.1801,
+      "step": 6780
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.23979468204502e-05,
+      "loss": 0.1959,
+      "step": 6781
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2395905176728045e-05,
+      "loss": 0.1844,
+      "step": 6782
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2393863427049704e-05,
+      "loss": 0.1988,
+      "step": 6783
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2391821571505474e-05,
+      "loss": 0.1893,
+      "step": 6784
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2389779610185651e-05,
+      "loss": 0.2123,
+      "step": 6785
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2387737543180536e-05,
+      "loss": 0.2066,
+      "step": 6786
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2385695370580442e-05,
+      "loss": 0.1934,
+      "step": 6787
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2383653092475679e-05,
+      "loss": 0.2086,
+      "step": 6788
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2381610708956564e-05,
+      "loss": 0.2168,
+      "step": 6789
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.237956822011342e-05,
+      "loss": 0.1704,
+      "step": 6790
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2377525626036576e-05,
+      "loss": 0.2129,
+      "step": 6791
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.237548292681636e-05,
+      "loss": 0.1662,
+      "step": 6792
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2373440122543108e-05,
+      "loss": 0.1917,
+      "step": 6793
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2371397213307165e-05,
+      "loss": 0.1853,
+      "step": 6794
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2369354199198869e-05,
+      "loss": 0.1728,
+      "step": 6795
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2367311080308577e-05,
+      "loss": 0.1856,
+      "step": 6796
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2365267856726644e-05,
+      "loss": 0.1851,
+      "step": 6797
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.236322452854342e-05,
+      "loss": 0.2083,
+      "step": 6798
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.236118109584928e-05,
+      "loss": 0.1892,
+      "step": 6799
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2359137558734582e-05,
+      "loss": 0.1975,
+      "step": 6800
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.235709391728971e-05,
+      "loss": 0.1609,
+      "step": 6801
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2355050171605034e-05,
+      "loss": 0.1935,
+      "step": 6802
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.235300632177094e-05,
+      "loss": 0.1811,
+      "step": 6803
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2350962367877812e-05,
+      "loss": 0.1981,
+      "step": 6804
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2348918310016043e-05,
+      "loss": 0.1991,
+      "step": 6805
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2346874148276032e-05,
+      "loss": 0.2148,
+      "step": 6806
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2344829882748173e-05,
+      "loss": 0.1613,
+      "step": 6807
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.234278551352288e-05,
+      "loss": 0.2161,
+      "step": 6808
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2340741040690554e-05,
+      "loss": 0.2033,
+      "step": 6809
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2338696464341613e-05,
+      "loss": 0.1823,
+      "step": 6810
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2336651784566482e-05,
+      "loss": 0.1849,
+      "step": 6811
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2334607001455577e-05,
+      "loss": 0.1948,
+      "step": 6812
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2332562115099327e-05,
+      "loss": 0.199,
+      "step": 6813
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2330517125588165e-05,
+      "loss": 0.2042,
+      "step": 6814
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2328472033012536e-05,
+      "loss": 0.2028,
+      "step": 6815
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.232642683746287e-05,
+      "loss": 0.1901,
+      "step": 6816
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2324381539029621e-05,
+      "loss": 0.1957,
+      "step": 6817
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2322336137803239e-05,
+      "loss": 0.1689,
+      "step": 6818
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2320290633874178e-05,
+      "loss": 0.2026,
+      "step": 6819
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2318245027332898e-05,
+      "loss": 0.2032,
+      "step": 6820
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2316199318269868e-05,
+      "loss": 0.1792,
+      "step": 6821
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.231415350677555e-05,
+      "loss": 0.2118,
+      "step": 6822
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2312107592940418e-05,
+      "loss": 0.2017,
+      "step": 6823
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2310061576854957e-05,
+      "loss": 0.2043,
+      "step": 6824
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.230801545860965e-05,
+      "loss": 0.1688,
+      "step": 6825
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2305969238294977e-05,
+      "loss": 0.1696,
+      "step": 6826
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2303922916001432e-05,
+      "loss": 0.1783,
+      "step": 6827
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2301876491819514e-05,
+      "loss": 0.1922,
+      "step": 6828
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2299829965839721e-05,
+      "loss": 0.1916,
+      "step": 6829
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.229778333815256e-05,
+      "loss": 0.1987,
+      "step": 6830
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2295736608848541e-05,
+      "loss": 0.2238,
+      "step": 6831
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2293689778018177e-05,
+      "loss": 0.2127,
+      "step": 6832
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2291642845751987e-05,
+      "loss": 0.1984,
+      "step": 6833
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2289595812140495e-05,
+      "loss": 0.2133,
+      "step": 6834
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2287548677274233e-05,
+      "loss": 0.1752,
+      "step": 6835
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2285501441243724e-05,
+      "loss": 0.2247,
+      "step": 6836
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2283454104139505e-05,
+      "loss": 0.2095,
+      "step": 6837
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2281406666052129e-05,
+      "loss": 0.1711,
+      "step": 6838
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2279359127072131e-05,
+      "loss": 0.201,
+      "step": 6839
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2277311487290064e-05,
+      "loss": 0.2176,
+      "step": 6840
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2275263746796484e-05,
+      "loss": 0.1743,
+      "step": 6841
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2273215905681948e-05,
+      "loss": 0.1929,
+      "step": 6842
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2271167964037018e-05,
+      "loss": 0.171,
+      "step": 6843
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2269119921952265e-05,
+      "loss": 0.1784,
+      "step": 6844
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2267071779518264e-05,
+      "loss": 0.1937,
+      "step": 6845
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2265023536825584e-05,
+      "loss": 0.2167,
+      "step": 6846
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.226297519396481e-05,
+      "loss": 0.2073,
+      "step": 6847
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2260926751026529e-05,
+      "loss": 0.2063,
+      "step": 6848
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2258878208101335e-05,
+      "loss": 0.2297,
+      "step": 6849
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2256829565279813e-05,
+      "loss": 0.2019,
+      "step": 6850
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2254780822652563e-05,
+      "loss": 0.1985,
+      "step": 6851
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2252731980310197e-05,
+      "loss": 0.1758,
+      "step": 6852
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2250683038343316e-05,
+      "loss": 0.2147,
+      "step": 6853
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2248633996842534e-05,
+      "loss": 0.1762,
+      "step": 6854
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2246584855898467e-05,
+      "loss": 0.198,
+      "step": 6855
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2244535615601735e-05,
+      "loss": 0.1762,
+      "step": 6856
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2242486276042964e-05,
+      "loss": 0.1876,
+      "step": 6857
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2240436837312786e-05,
+      "loss": 0.173,
+      "step": 6858
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2238387299501833e-05,
+      "loss": 0.1881,
+      "step": 6859
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2236337662700744e-05,
+      "loss": 0.1999,
+      "step": 6860
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2234287927000159e-05,
+      "loss": 0.1925,
+      "step": 6861
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2232238092490731e-05,
+      "loss": 0.18,
+      "step": 6862
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2230188159263108e-05,
+      "loss": 0.2034,
+      "step": 6863
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2228138127407943e-05,
+      "loss": 0.1983,
+      "step": 6864
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2226087997015901e-05,
+      "loss": 0.1847,
+      "step": 6865
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2224037768177647e-05,
+      "loss": 0.1791,
+      "step": 6866
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2221987440983848e-05,
+      "loss": 0.2021,
+      "step": 6867
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2219937015525178e-05,
+      "loss": 0.1757,
+      "step": 6868
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2217886491892314e-05,
+      "loss": 0.1679,
+      "step": 6869
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.221583587017594e-05,
+      "loss": 0.1666,
+      "step": 6870
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.221378515046674e-05,
+      "loss": 0.1898,
+      "step": 6871
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2211734332855405e-05,
+      "loss": 0.1864,
+      "step": 6872
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2209683417432632e-05,
+      "loss": 0.226,
+      "step": 6873
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2207632404289118e-05,
+      "loss": 0.2171,
+      "step": 6874
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.220558129351557e-05,
+      "loss": 0.2224,
+      "step": 6875
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2203530085202693e-05,
+      "loss": 0.1893,
+      "step": 6876
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.22014787794412e-05,
+      "loss": 0.2104,
+      "step": 6877
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2199427376321806e-05,
+      "loss": 0.1835,
+      "step": 6878
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.219737587593524e-05,
+      "loss": 0.2087,
+      "step": 6879
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2195324278372215e-05,
+      "loss": 0.2226,
+      "step": 6880
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.219327258372347e-05,
+      "loss": 0.206,
+      "step": 6881
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2191220792079736e-05,
+      "loss": 0.1738,
+      "step": 6882
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.218916890353175e-05,
+      "loss": 0.1545,
+      "step": 6883
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2187116918170257e-05,
+      "loss": 0.1584,
+      "step": 6884
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2185064836086e-05,
+      "loss": 0.1929,
+      "step": 6885
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2183012657369736e-05,
+      "loss": 0.184,
+      "step": 6886
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.218096038211221e-05,
+      "loss": 0.1882,
+      "step": 6887
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2178908010404192e-05,
+      "loss": 0.1796,
+      "step": 6888
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2176855542336438e-05,
+      "loss": 0.1866,
+      "step": 6889
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2174802977999728e-05,
+      "loss": 0.1925,
+      "step": 6890
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2172750317484818e-05,
+      "loss": 0.2184,
+      "step": 6891
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2170697560882494e-05,
+      "loss": 0.1842,
+      "step": 6892
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2168644708283539e-05,
+      "loss": 0.1959,
+      "step": 6893
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2166591759778732e-05,
+      "loss": 0.2413,
+      "step": 6894
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2164538715458865e-05,
+      "loss": 0.2014,
+      "step": 6895
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2162485575414731e-05,
+      "loss": 0.2024,
+      "step": 6896
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.216043233973713e-05,
+      "loss": 0.2316,
+      "step": 6897
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.215837900851686e-05,
+      "loss": 0.194,
+      "step": 6898
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2156325581844732e-05,
+      "loss": 0.1755,
+      "step": 6899
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2154272059811554e-05,
+      "loss": 0.1857,
+      "step": 6900
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.215221844250814e-05,
+      "loss": 0.1927,
+      "step": 6901
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2150164730025308e-05,
+      "loss": 0.1903,
+      "step": 6902
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2148110922453885e-05,
+      "loss": 0.1761,
+      "step": 6903
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2146057019884694e-05,
+      "loss": 0.1872,
+      "step": 6904
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2144003022408567e-05,
+      "loss": 0.1916,
+      "step": 6905
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2141948930116343e-05,
+      "loss": 0.2032,
+      "step": 6906
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2139894743098863e-05,
+      "loss": 0.1849,
+      "step": 6907
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2137840461446965e-05,
+      "loss": 0.1797,
+      "step": 6908
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.21357860852515e-05,
+      "loss": 0.175,
+      "step": 6909
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.213373161460332e-05,
+      "loss": 0.1751,
+      "step": 6910
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2131677049593286e-05,
+      "loss": 0.2047,
+      "step": 6911
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2129622390312255e-05,
+      "loss": 0.2106,
+      "step": 6912
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2127567636851092e-05,
+      "loss": 0.1927,
+      "step": 6913
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2125512789300664e-05,
+      "loss": 0.1804,
+      "step": 6914
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2123457847751847e-05,
+      "loss": 0.1931,
+      "step": 6915
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2121402812295519e-05,
+      "loss": 0.1882,
+      "step": 6916
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2119347683022562e-05,
+      "loss": 0.173,
+      "step": 6917
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.211729246002386e-05,
+      "loss": 0.1907,
+      "step": 6918
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2115237143390305e-05,
+      "loss": 0.2177,
+      "step": 6919
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2113181733212784e-05,
+      "loss": 0.1743,
+      "step": 6920
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2111126229582207e-05,
+      "loss": 0.1962,
+      "step": 6921
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2109070632589465e-05,
+      "loss": 0.2052,
+      "step": 6922
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2107014942325473e-05,
+      "loss": 0.1711,
+      "step": 6923
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2104959158881136e-05,
+      "loss": 0.1672,
+      "step": 6924
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.210290328234737e-05,
+      "loss": 0.1879,
+      "step": 6925
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2100847312815096e-05,
+      "loss": 0.1747,
+      "step": 6926
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2098791250375237e-05,
+      "loss": 0.1735,
+      "step": 6927
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2096735095118715e-05,
+      "loss": 0.2071,
+      "step": 6928
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2094678847136466e-05,
+      "loss": 0.2092,
+      "step": 6929
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2092622506519424e-05,
+      "loss": 0.1988,
+      "step": 6930
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.209056607335853e-05,
+      "loss": 0.1669,
+      "step": 6931
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2088509547744722e-05,
+      "loss": 0.1865,
+      "step": 6932
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2086452929768951e-05,
+      "loss": 0.1756,
+      "step": 6933
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2084396219522173e-05,
+      "loss": 0.1978,
+      "step": 6934
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2082339417095337e-05,
+      "loss": 0.1757,
+      "step": 6935
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2080282522579405e-05,
+      "loss": 0.1868,
+      "step": 6936
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.207822553606534e-05,
+      "loss": 0.219,
+      "step": 6937
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2076168457644111e-05,
+      "loss": 0.1928,
+      "step": 6938
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2074111287406689e-05,
+      "loss": 0.2008,
+      "step": 6939
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2072054025444053e-05,
+      "loss": 0.1811,
+      "step": 6940
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2069996671847181e-05,
+      "loss": 0.1868,
+      "step": 6941
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2067939226707056e-05,
+      "loss": 0.19,
+      "step": 6942
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2065881690114665e-05,
+      "loss": 0.1875,
+      "step": 6943
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2063824062161007e-05,
+      "loss": 0.2001,
+      "step": 6944
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2061766342937071e-05,
+      "loss": 0.1688,
+      "step": 6945
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.205970853253386e-05,
+      "loss": 0.2008,
+      "step": 6946
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2057650631042374e-05,
+      "loss": 0.1899,
+      "step": 6947
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2055592638553633e-05,
+      "loss": 0.2099,
+      "step": 6948
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.205353455515864e-05,
+      "loss": 0.189,
+      "step": 6949
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.205147638094841e-05,
+      "loss": 0.1946,
+      "step": 6950
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2049418116013969e-05,
+      "loss": 0.2297,
+      "step": 6951
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.204735976044634e-05,
+      "loss": 0.1934,
+      "step": 6952
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2045301314336548e-05,
+      "loss": 0.1927,
+      "step": 6953
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2043242777775634e-05,
+      "loss": 0.1961,
+      "step": 6954
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2041184150854624e-05,
+      "loss": 0.1793,
+      "step": 6955
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.203912543366456e-05,
+      "loss": 0.1665,
+      "step": 6956
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2037066626296494e-05,
+      "loss": 0.1902,
+      "step": 6957
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.203500772884147e-05,
+      "loss": 0.2049,
+      "step": 6958
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.203294874139054e-05,
+      "loss": 0.2212,
+      "step": 6959
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2030889664034759e-05,
+      "loss": 0.1698,
+      "step": 6960
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.202883049686519e-05,
+      "loss": 0.1674,
+      "step": 6961
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2026771239972895e-05,
+      "loss": 0.2153,
+      "step": 6962
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2024711893448944e-05,
+      "loss": 0.2054,
+      "step": 6963
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.202265245738441e-05,
+      "loss": 0.1818,
+      "step": 6964
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2020592931870367e-05,
+      "loss": 0.2097,
+      "step": 6965
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2018533316997898e-05,
+      "loss": 0.1989,
+      "step": 6966
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2016473612858084e-05,
+      "loss": 0.2362,
+      "step": 6967
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2014413819542017e-05,
+      "loss": 0.1788,
+      "step": 6968
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2012353937140783e-05,
+      "loss": 0.198,
+      "step": 6969
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.201029396574548e-05,
+      "loss": 0.197,
+      "step": 6970
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2008233905447212e-05,
+      "loss": 0.1864,
+      "step": 6971
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2006173756337082e-05,
+      "loss": 0.2089,
+      "step": 6972
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2004113518506194e-05,
+      "loss": 0.1787,
+      "step": 6973
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2002053192045662e-05,
+      "loss": 0.1875,
+      "step": 6974
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1999992777046601e-05,
+      "loss": 0.2094,
+      "step": 6975
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1997932273600131e-05,
+      "loss": 0.2043,
+      "step": 6976
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1995871681797374e-05,
+      "loss": 0.2028,
+      "step": 6977
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.199381100172946e-05,
+      "loss": 0.1742,
+      "step": 6978
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1991750233487516e-05,
+      "loss": 0.2078,
+      "step": 6979
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1989689377162682e-05,
+      "loss": 0.191,
+      "step": 6980
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1987628432846094e-05,
+      "loss": 0.1752,
+      "step": 6981
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1985567400628899e-05,
+      "loss": 0.208,
+      "step": 6982
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1983506280602237e-05,
+      "loss": 0.2008,
+      "step": 6983
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1981445072857261e-05,
+      "loss": 0.1796,
+      "step": 6984
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1979383777485131e-05,
+      "loss": 0.2049,
+      "step": 6985
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1977322394577e-05,
+      "loss": 0.185,
+      "step": 6986
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.197526092422403e-05,
+      "loss": 0.173,
+      "step": 6987
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1973199366517392e-05,
+      "loss": 0.2175,
+      "step": 6988
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.197113772154825e-05,
+      "loss": 0.1713,
+      "step": 6989
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1969075989407781e-05,
+      "loss": 0.2016,
+      "step": 6990
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1967014170187163e-05,
+      "loss": 0.1997,
+      "step": 6991
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1964952263977578e-05,
+      "loss": 0.1956,
+      "step": 6992
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.196289027087021e-05,
+      "loss": 0.2098,
+      "step": 6993
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1960828190956248e-05,
+      "loss": 0.1907,
+      "step": 6994
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.195876602432689e-05,
+      "loss": 0.1942,
+      "step": 6995
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1956703771073326e-05,
+      "loss": 0.1725,
+      "step": 6996
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1954641431286757e-05,
+      "loss": 0.1954,
+      "step": 6997
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1952579005058393e-05,
+      "loss": 0.1785,
+      "step": 6998
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1950516492479442e-05,
+      "loss": 0.1659,
+      "step": 6999
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1948453893641112e-05,
+      "loss": 0.1903,
+      "step": 7000
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1946391208634622e-05,
+      "loss": 0.2035,
+      "step": 7001
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1944328437551191e-05,
+      "loss": 0.1773,
+      "step": 7002
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1942265580482041e-05,
+      "loss": 0.1866,
+      "step": 7003
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1940202637518406e-05,
+      "loss": 0.2182,
+      "step": 7004
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.193813960875151e-05,
+      "loss": 0.172,
+      "step": 7005
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.193607649427259e-05,
+      "loss": 0.2073,
+      "step": 7006
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1934013294172886e-05,
+      "loss": 0.2014,
+      "step": 7007
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1931950008543642e-05,
+      "loss": 0.2018,
+      "step": 7008
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1929886637476104e-05,
+      "loss": 0.1846,
+      "step": 7009
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.192782318106152e-05,
+      "loss": 0.22,
+      "step": 7010
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1925759639391141e-05,
+      "loss": 0.2074,
+      "step": 7011
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1923696012556233e-05,
+      "loss": 0.1916,
+      "step": 7012
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1921632300648053e-05,
+      "loss": 0.17,
+      "step": 7013
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1919568503757866e-05,
+      "loss": 0.19,
+      "step": 7014
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1917504621976944e-05,
+      "loss": 0.2027,
+      "step": 7015
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1915440655396555e-05,
+      "loss": 0.2218,
+      "step": 7016
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.191337660410798e-05,
+      "loss": 0.2104,
+      "step": 7017
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1911312468202496e-05,
+      "loss": 0.1875,
+      "step": 7018
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1909248247771392e-05,
+      "loss": 0.1917,
+      "step": 7019
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1907183942905947e-05,
+      "loss": 0.2001,
+      "step": 7020
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1905119553697461e-05,
+      "loss": 0.1907,
+      "step": 7021
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1903055080237227e-05,
+      "loss": 0.1891,
+      "step": 7022
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1900990522616544e-05,
+      "loss": 0.1867,
+      "step": 7023
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1898925880926712e-05,
+      "loss": 0.2025,
+      "step": 7024
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1896861155259038e-05,
+      "loss": 0.1902,
+      "step": 7025
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1894796345704838e-05,
+      "loss": 0.1794,
+      "step": 7026
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.189273145235542e-05,
+      "loss": 0.1799,
+      "step": 7027
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1890666475302101e-05,
+      "loss": 0.1928,
+      "step": 7028
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1888601414636205e-05,
+      "loss": 0.2091,
+      "step": 7029
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1886536270449055e-05,
+      "loss": 0.1838,
+      "step": 7030
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1884471042831985e-05,
+      "loss": 0.2061,
+      "step": 7031
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1882405731876321e-05,
+      "loss": 0.1833,
+      "step": 7032
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1880340337673401e-05,
+      "loss": 0.1912,
+      "step": 7033
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1878274860314563e-05,
+      "loss": 0.181,
+      "step": 7034
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1876209299891156e-05,
+      "loss": 0.2,
+      "step": 7035
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1874143656494522e-05,
+      "loss": 0.2177,
+      "step": 7036
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1872077930216012e-05,
+      "loss": 0.2122,
+      "step": 7037
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1870012121146983e-05,
+      "loss": 0.1912,
+      "step": 7038
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1867946229378788e-05,
+      "loss": 0.1887,
+      "step": 7039
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1865880255002797e-05,
+      "loss": 0.1957,
+      "step": 7040
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.186381419811037e-05,
+      "loss": 0.2117,
+      "step": 7041
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1861748058792875e-05,
+      "loss": 0.1839,
+      "step": 7042
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1859681837141688e-05,
+      "loss": 0.1999,
+      "step": 7043
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1857615533248182e-05,
+      "loss": 0.1829,
+      "step": 7044
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.185554914720374e-05,
+      "loss": 0.2143,
+      "step": 7045
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1853482679099746e-05,
+      "loss": 0.1873,
+      "step": 7046
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1851416129027582e-05,
+      "loss": 0.2002,
+      "step": 7047
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1849349497078642e-05,
+      "loss": 0.182,
+      "step": 7048
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.184728278334432e-05,
+      "loss": 0.1999,
+      "step": 7049
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.184521598791602e-05,
+      "loss": 0.1796,
+      "step": 7050
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1843149110885135e-05,
+      "loss": 0.1999,
+      "step": 7051
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1841082152343073e-05,
+      "loss": 0.1976,
+      "step": 7052
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1839015112381244e-05,
+      "loss": 0.178,
+      "step": 7053
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1836947991091061e-05,
+      "loss": 0.1732,
+      "step": 7054
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1834880788563941e-05,
+      "loss": 0.206,
+      "step": 7055
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1832813504891298e-05,
+      "loss": 0.2069,
+      "step": 7056
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1830746140164562e-05,
+      "loss": 0.1805,
+      "step": 7057
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1828678694475155e-05,
+      "loss": 0.2142,
+      "step": 7058
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.182661116791451e-05,
+      "loss": 0.1781,
+      "step": 7059
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1824543560574064e-05,
+      "loss": 0.2189,
+      "step": 7060
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1822475872545245e-05,
+      "loss": 0.1641,
+      "step": 7061
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1820408103919502e-05,
+      "loss": 0.1706,
+      "step": 7062
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1818340254788282e-05,
+      "loss": 0.1818,
+      "step": 7063
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1816272325243027e-05,
+      "loss": 0.1977,
+      "step": 7064
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.181420431537519e-05,
+      "loss": 0.21,
+      "step": 7065
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1812136225276227e-05,
+      "loss": 0.2161,
+      "step": 7066
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.18100680550376e-05,
+      "loss": 0.1845,
+      "step": 7067
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1807999804750767e-05,
+      "loss": 0.1866,
+      "step": 7068
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1805931474507198e-05,
+      "loss": 0.2441,
+      "step": 7069
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1803863064398358e-05,
+      "loss": 0.1975,
+      "step": 7070
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1801794574515722e-05,
+      "loss": 0.1721,
+      "step": 7071
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1799726004950766e-05,
+      "loss": 0.1786,
+      "step": 7072
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1797657355794976e-05,
+      "loss": 0.1834,
+      "step": 7073
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.179558862713983e-05,
+      "loss": 0.1909,
+      "step": 7074
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1793519819076813e-05,
+      "loss": 0.1957,
+      "step": 7075
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.179145093169742e-05,
+      "loss": 0.1881,
+      "step": 7076
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1789381965093146e-05,
+      "loss": 0.1972,
+      "step": 7077
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1787312919355484e-05,
+      "loss": 0.1674,
+      "step": 7078
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.178524379457594e-05,
+      "loss": 0.1811,
+      "step": 7079
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1783174590846014e-05,
+      "loss": 0.1869,
+      "step": 7080
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.178110530825722e-05,
+      "loss": 0.1988,
+      "step": 7081
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1779035946901063e-05,
+      "loss": 0.1866,
+      "step": 7082
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1776966506869065e-05,
+      "loss": 0.1758,
+      "step": 7083
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.177489698825274e-05,
+      "loss": 0.1816,
+      "step": 7084
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1772827391143609e-05,
+      "loss": 0.183,
+      "step": 7085
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1770757715633204e-05,
+      "loss": 0.1965,
+      "step": 7086
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1768687961813048e-05,
+      "loss": 0.1757,
+      "step": 7087
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1766618129774676e-05,
+      "loss": 0.1672,
+      "step": 7088
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.176454821960962e-05,
+      "loss": 0.1797,
+      "step": 7089
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1762478231409426e-05,
+      "loss": 0.184,
+      "step": 7090
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1760408165265635e-05,
+      "loss": 0.1808,
+      "step": 7091
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1758338021269791e-05,
+      "loss": 0.1786,
+      "step": 7092
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1756267799513443e-05,
+      "loss": 0.1873,
+      "step": 7093
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1754197500088149e-05,
+      "loss": 0.2088,
+      "step": 7094
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1752127123085462e-05,
+      "loss": 0.2055,
+      "step": 7095
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.175005666859694e-05,
+      "loss": 0.2117,
+      "step": 7096
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1747986136714152e-05,
+      "loss": 0.2047,
+      "step": 7097
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1745915527528662e-05,
+      "loss": 0.2168,
+      "step": 7098
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.174384484113204e-05,
+      "loss": 0.1946,
+      "step": 7099
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1741774077615859e-05,
+      "loss": 0.1843,
+      "step": 7100
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1739703237071702e-05,
+      "loss": 0.2216,
+      "step": 7101
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1737632319591142e-05,
+      "loss": 0.1855,
+      "step": 7102
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1735561325265763e-05,
+      "loss": 0.1967,
+      "step": 7103
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.173349025418716e-05,
+      "loss": 0.1888,
+      "step": 7104
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1731419106446918e-05,
+      "loss": 0.1769,
+      "step": 7105
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.172934788213663e-05,
+      "loss": 0.1889,
+      "step": 7106
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.17272765813479e-05,
+      "loss": 0.1928,
+      "step": 7107
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1725205204172321e-05,
+      "loss": 0.1995,
+      "step": 7108
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1723133750701501e-05,
+      "loss": 0.1691,
+      "step": 7109
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1721062221027049e-05,
+      "loss": 0.1786,
+      "step": 7110
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1718990615240577e-05,
+      "loss": 0.2261,
+      "step": 7111
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1716918933433695e-05,
+      "loss": 0.1902,
+      "step": 7112
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.171484717569802e-05,
+      "loss": 0.1943,
+      "step": 7113
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.171277534212518e-05,
+      "loss": 0.1942,
+      "step": 7114
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1710703432806798e-05,
+      "loss": 0.186,
+      "step": 7115
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1708631447834495e-05,
+      "loss": 0.1983,
+      "step": 7116
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1706559387299905e-05,
+      "loss": 0.1818,
+      "step": 7117
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1704487251294669e-05,
+      "loss": 0.1919,
+      "step": 7118
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1702415039910418e-05,
+      "loss": 0.1873,
+      "step": 7119
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1700342753238796e-05,
+      "loss": 0.187,
+      "step": 7120
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1698270391371445e-05,
+      "loss": 0.1563,
+      "step": 7121
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1696197954400015e-05,
+      "loss": 0.225,
+      "step": 7122
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1694125442416158e-05,
+      "loss": 0.1897,
+      "step": 7123
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1692052855511527e-05,
+      "loss": 0.1778,
+      "step": 7124
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1689980193777779e-05,
+      "loss": 0.1939,
+      "step": 7125
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1687907457306576e-05,
+      "loss": 0.2049,
+      "step": 7126
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1685834646189584e-05,
+      "loss": 0.2004,
+      "step": 7127
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.168376176051847e-05,
+      "loss": 0.1862,
+      "step": 7128
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1681688800384902e-05,
+      "loss": 0.1624,
+      "step": 7129
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1679615765880557e-05,
+      "loss": 0.2228,
+      "step": 7130
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1677542657097111e-05,
+      "loss": 0.2095,
+      "step": 7131
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.167546947412625e-05,
+      "loss": 0.1944,
+      "step": 7132
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1673396217059652e-05,
+      "loss": 0.1792,
+      "step": 7133
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1671322885989007e-05,
+      "loss": 0.1989,
+      "step": 7134
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1669249481006005e-05,
+      "loss": 0.1702,
+      "step": 7135
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1667176002202338e-05,
+      "loss": 0.187,
+      "step": 7136
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.166510244966971e-05,
+      "loss": 0.2045,
+      "step": 7137
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1663028823499816e-05,
+      "loss": 0.1835,
+      "step": 7138
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1660955123784359e-05,
+      "loss": 0.1632,
+      "step": 7139
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1658881350615047e-05,
+      "loss": 0.1536,
+      "step": 7140
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1656807504083594e-05,
+      "loss": 0.2056,
+      "step": 7141
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1654733584281712e-05,
+      "loss": 0.1957,
+      "step": 7142
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1652659591301114e-05,
+      "loss": 0.2062,
+      "step": 7143
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1650585525233522e-05,
+      "loss": 0.1715,
+      "step": 7144
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.164851138617066e-05,
+      "loss": 0.2129,
+      "step": 7145
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1646437174204256e-05,
+      "loss": 0.2141,
+      "step": 7146
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1644362889426036e-05,
+      "loss": 0.1881,
+      "step": 7147
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1642288531927735e-05,
+      "loss": 0.1529,
+      "step": 7148
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1640214101801088e-05,
+      "loss": 0.1801,
+      "step": 7149
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1638139599137835e-05,
+      "loss": 0.1808,
+      "step": 7150
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.163606502402972e-05,
+      "loss": 0.2041,
+      "step": 7151
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1633990376568487e-05,
+      "loss": 0.2042,
+      "step": 7152
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1631915656845882e-05,
+      "loss": 0.1946,
+      "step": 7153
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1629840864953663e-05,
+      "loss": 0.1831,
+      "step": 7154
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1627766000983582e-05,
+      "loss": 0.1632,
+      "step": 7155
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1625691065027402e-05,
+      "loss": 0.2057,
+      "step": 7156
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.162361605717688e-05,
+      "loss": 0.191,
+      "step": 7157
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1621540977523775e-05,
+      "loss": 0.204,
+      "step": 7158
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1619465826159869e-05,
+      "loss": 0.197,
+      "step": 7159
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1617390603176921e-05,
+      "loss": 0.1969,
+      "step": 7160
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1615315308666712e-05,
+      "loss": 0.1849,
+      "step": 7161
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.161323994272102e-05,
+      "loss": 0.181,
+      "step": 7162
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1611164505431622e-05,
+      "loss": 0.2057,
+      "step": 7163
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1609088996890303e-05,
+      "loss": 0.2026,
+      "step": 7164
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1607013417188851e-05,
+      "loss": 0.2082,
+      "step": 7165
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1604937766419057e-05,
+      "loss": 0.1647,
+      "step": 7166
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1602862044672709e-05,
+      "loss": 0.1849,
+      "step": 7167
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1600786252041612e-05,
+      "loss": 0.1983,
+      "step": 7168
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.159871038861756e-05,
+      "loss": 0.1985,
+      "step": 7169
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1596634454492355e-05,
+      "loss": 0.2087,
+      "step": 7170
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1594558449757804e-05,
+      "loss": 0.1946,
+      "step": 7171
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1592482374505716e-05,
+      "loss": 0.1859,
+      "step": 7172
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1590406228827905e-05,
+      "loss": 0.2099,
+      "step": 7173
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1588330012816187e-05,
+      "loss": 0.1886,
+      "step": 7174
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1586253726562376e-05,
+      "loss": 0.2077,
+      "step": 7175
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1584177370158294e-05,
+      "loss": 0.1929,
+      "step": 7176
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1582100943695767e-05,
+      "loss": 0.2137,
+      "step": 7177
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1580024447266622e-05,
+      "loss": 0.1999,
+      "step": 7178
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1577947880962695e-05,
+      "loss": 0.197,
+      "step": 7179
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1575871244875812e-05,
+      "loss": 0.184,
+      "step": 7180
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1573794539097807e-05,
+      "loss": 0.1716,
+      "step": 7181
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1571717763720531e-05,
+      "loss": 0.2057,
+      "step": 7182
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1569640918835825e-05,
+      "loss": 0.1817,
+      "step": 7183
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1567564004535527e-05,
+      "loss": 0.198,
+      "step": 7184
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1565487020911494e-05,
+      "loss": 0.1991,
+      "step": 7185
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1563409968055575e-05,
+      "loss": 0.1902,
+      "step": 7186
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1561332846059623e-05,
+      "loss": 0.1612,
+      "step": 7187
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.15592556550155e-05,
+      "loss": 0.1745,
+      "step": 7188
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1557178395015067e-05,
+      "loss": 0.1853,
+      "step": 7189
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.155510106615019e-05,
+      "loss": 0.1697,
+      "step": 7190
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1553023668512731e-05,
+      "loss": 0.1927,
+      "step": 7191
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1550946202194566e-05,
+      "loss": 0.1958,
+      "step": 7192
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1548868667287566e-05,
+      "loss": 0.1793,
+      "step": 7193
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1546791063883607e-05,
+      "loss": 0.1984,
+      "step": 7194
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1544713392074572e-05,
+      "loss": 0.2124,
+      "step": 7195
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1542635651952341e-05,
+      "loss": 0.2233,
+      "step": 7196
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.15405578436088e-05,
+      "loss": 0.196,
+      "step": 7197
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1538479967135837e-05,
+      "loss": 0.1921,
+      "step": 7198
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1536402022625347e-05,
+      "loss": 0.191,
+      "step": 7199
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1534324010169218e-05,
+      "loss": 0.2082,
+      "step": 7200
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1532245929859353e-05,
+      "loss": 0.1725,
+      "step": 7201
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1530167781787656e-05,
+      "loss": 0.1793,
+      "step": 7202
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1528089566046024e-05,
+      "loss": 0.1856,
+      "step": 7203
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1526011282726363e-05,
+      "loss": 0.187,
+      "step": 7204
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1523932931920591e-05,
+      "loss": 0.202,
+      "step": 7205
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1521854513720614e-05,
+      "loss": 0.2486,
+      "step": 7206
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1519776028218352e-05,
+      "loss": 0.1847,
+      "step": 7207
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1517697475505716e-05,
+      "loss": 0.1907,
+      "step": 7208
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1515618855674634e-05,
+      "loss": 0.1968,
+      "step": 7209
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1513540168817033e-05,
+      "loss": 0.1915,
+      "step": 7210
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1511461415024834e-05,
+      "loss": 0.1912,
+      "step": 7211
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1509382594389971e-05,
+      "loss": 0.1887,
+      "step": 7212
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1507303707004373e-05,
+      "loss": 0.1582,
+      "step": 7213
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1505224752959988e-05,
+      "loss": 0.1877,
+      "step": 7214
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1503145732348743e-05,
+      "loss": 0.1751,
+      "step": 7215
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1501066645262587e-05,
+      "loss": 0.1687,
+      "step": 7216
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.149898749179346e-05,
+      "loss": 0.1792,
+      "step": 7217
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1496908272033315e-05,
+      "loss": 0.217,
+      "step": 7218
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1494828986074102e-05,
+      "loss": 0.1541,
+      "step": 7219
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1492749634007778e-05,
+      "loss": 0.1953,
+      "step": 7220
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1490670215926294e-05,
+      "loss": 0.2274,
+      "step": 7221
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.148859073192161e-05,
+      "loss": 0.1746,
+      "step": 7222
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1486511182085693e-05,
+      "loss": 0.1892,
+      "step": 7223
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.148443156651051e-05,
+      "loss": 0.187,
+      "step": 7224
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1482351885288023e-05,
+      "loss": 0.1773,
+      "step": 7225
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1480272138510208e-05,
+      "loss": 0.1619,
+      "step": 7226
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1478192326269042e-05,
+      "loss": 0.1945,
+      "step": 7227
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1476112448656496e-05,
+      "loss": 0.1947,
+      "step": 7228
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1474032505764553e-05,
+      "loss": 0.2121,
+      "step": 7229
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1471952497685198e-05,
+      "loss": 0.2117,
+      "step": 7230
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1469872424510413e-05,
+      "loss": 0.1915,
+      "step": 7231
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1467792286332192e-05,
+      "loss": 0.2011,
+      "step": 7232
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1465712083242523e-05,
+      "loss": 0.1873,
+      "step": 7233
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1463631815333403e-05,
+      "loss": 0.2154,
+      "step": 7234
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1461551482696827e-05,
+      "loss": 0.207,
+      "step": 7235
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1459471085424794e-05,
+      "loss": 0.2096,
+      "step": 7236
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1457390623609314e-05,
+      "loss": 0.1807,
+      "step": 7237
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1455310097342388e-05,
+      "loss": 0.1772,
+      "step": 7238
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1453229506716022e-05,
+      "loss": 0.1929,
+      "step": 7239
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1451148851822233e-05,
+      "loss": 0.1927,
+      "step": 7240
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1449068132753035e-05,
+      "loss": 0.189,
+      "step": 7241
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1446987349600445e-05,
+      "loss": 0.2273,
+      "step": 7242
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1444906502456481e-05,
+      "loss": 0.2092,
+      "step": 7243
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1442825591413171e-05,
+      "loss": 0.1851,
+      "step": 7244
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1440744616562534e-05,
+      "loss": 0.1981,
+      "step": 7245
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1438663577996603e-05,
+      "loss": 0.2332,
+      "step": 7246
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1436582475807411e-05,
+      "loss": 0.2296,
+      "step": 7247
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1434501310086991e-05,
+      "loss": 0.1755,
+      "step": 7248
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1432420080927378e-05,
+      "loss": 0.1721,
+      "step": 7249
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1430338788420612e-05,
+      "loss": 0.1745,
+      "step": 7250
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.142825743265874e-05,
+      "loss": 0.1931,
+      "step": 7251
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1426176013733806e-05,
+      "loss": 0.196,
+      "step": 7252
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1424094531737855e-05,
+      "loss": 0.1828,
+      "step": 7253
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1422012986762942e-05,
+      "loss": 0.1798,
+      "step": 7254
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.141993137890112e-05,
+      "loss": 0.169,
+      "step": 7255
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1417849708244442e-05,
+      "loss": 0.212,
+      "step": 7256
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1415767974884973e-05,
+      "loss": 0.1827,
+      "step": 7257
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1413686178914774e-05,
+      "loss": 0.1916,
+      "step": 7258
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1411604320425908e-05,
+      "loss": 0.1959,
+      "step": 7259
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1409522399510443e-05,
+      "loss": 0.1853,
+      "step": 7260
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1407440416260455e-05,
+      "loss": 0.1747,
+      "step": 7261
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.140535837076801e-05,
+      "loss": 0.1929,
+      "step": 7262
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1403276263125185e-05,
+      "loss": 0.1839,
+      "step": 7263
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1401194093424062e-05,
+      "loss": 0.1954,
+      "step": 7264
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1399111861756724e-05,
+      "loss": 0.1957,
+      "step": 7265
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1397029568215251e-05,
+      "loss": 0.1798,
+      "step": 7266
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1394947212891733e-05,
+      "loss": 0.1758,
+      "step": 7267
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1392864795878258e-05,
+      "loss": 0.2001,
+      "step": 7268
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1390782317266919e-05,
+      "loss": 0.1828,
+      "step": 7269
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1388699777149812e-05,
+      "loss": 0.18,
+      "step": 7270
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1386617175619034e-05,
+      "loss": 0.1631,
+      "step": 7271
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1384534512766686e-05,
+      "loss": 0.2046,
+      "step": 7272
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.138245178868487e-05,
+      "loss": 0.2022,
+      "step": 7273
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1380369003465695e-05,
+      "loss": 0.1714,
+      "step": 7274
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.137828615720127e-05,
+      "loss": 0.2095,
+      "step": 7275
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1376203249983703e-05,
+      "loss": 0.1857,
+      "step": 7276
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.137412028190511e-05,
+      "loss": 0.1678,
+      "step": 7277
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.137203725305761e-05,
+      "loss": 0.2052,
+      "step": 7278
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.136995416353332e-05,
+      "loss": 0.1838,
+      "step": 7279
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1367871013424362e-05,
+      "loss": 0.2018,
+      "step": 7280
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.136578780282286e-05,
+      "loss": 0.1856,
+      "step": 7281
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1363704531820947e-05,
+      "loss": 0.1967,
+      "step": 7282
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1361621200510748e-05,
+      "loss": 0.1914,
+      "step": 7283
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1359537808984398e-05,
+      "loss": 0.1534,
+      "step": 7284
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1357454357334034e-05,
+      "loss": 0.2126,
+      "step": 7285
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1355370845651788e-05,
+      "loss": 0.2183,
+      "step": 7286
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.135328727402981e-05,
+      "loss": 0.1702,
+      "step": 7287
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1351203642560238e-05,
+      "loss": 0.2021,
+      "step": 7288
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1349119951335219e-05,
+      "loss": 0.2026,
+      "step": 7289
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1347036200446903e-05,
+      "loss": 0.1867,
+      "step": 7290
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1344952389987438e-05,
+      "loss": 0.2058,
+      "step": 7291
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1342868520048987e-05,
+      "loss": 0.2048,
+      "step": 7292
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1340784590723694e-05,
+      "loss": 0.2308,
+      "step": 7293
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.133870060210373e-05,
+      "loss": 0.1876,
+      "step": 7294
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1336616554281249e-05,
+      "loss": 0.1954,
+      "step": 7295
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.133453244734842e-05,
+      "loss": 0.1892,
+      "step": 7296
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1332448281397408e-05,
+      "loss": 0.2044,
+      "step": 7297
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1330364056520387e-05,
+      "loss": 0.1945,
+      "step": 7298
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1328279772809526e-05,
+      "loss": 0.2159,
+      "step": 7299
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1326195430356996e-05,
+      "loss": 0.191,
+      "step": 7300
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1324111029254982e-05,
+      "loss": 0.1729,
+      "step": 7301
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1322026569595663e-05,
+      "loss": 0.1569,
+      "step": 7302
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.131994205147122e-05,
+      "loss": 0.2001,
+      "step": 7303
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1317857474973837e-05,
+      "loss": 0.221,
+      "step": 7304
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1315772840195702e-05,
+      "loss": 0.1765,
+      "step": 7305
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1313688147229012e-05,
+      "loss": 0.1859,
+      "step": 7306
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1311603396165953e-05,
+      "loss": 0.1968,
+      "step": 7307
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1309518587098722e-05,
+      "loss": 0.1978,
+      "step": 7308
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1307433720119522e-05,
+      "loss": 0.2099,
+      "step": 7309
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1305348795320547e-05,
+      "loss": 0.1869,
+      "step": 7310
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1303263812794008e-05,
+      "loss": 0.1914,
+      "step": 7311
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1301178772632107e-05,
+      "loss": 0.1892,
+      "step": 7312
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.129909367492705e-05,
+      "loss": 0.1875,
+      "step": 7313
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1297008519771047e-05,
+      "loss": 0.1922,
+      "step": 7314
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.129492330725632e-05,
+      "loss": 0.1844,
+      "step": 7315
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.129283803747508e-05,
+      "loss": 0.1767,
+      "step": 7316
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1290752710519546e-05,
+      "loss": 0.1641,
+      "step": 7317
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1288667326481937e-05,
+      "loss": 0.1889,
+      "step": 7318
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1286581885454481e-05,
+      "loss": 0.2031,
+      "step": 7319
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1284496387529402e-05,
+      "loss": 0.1629,
+      "step": 7320
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.128241083279893e-05,
+      "loss": 0.1919,
+      "step": 7321
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1280325221355292e-05,
+      "loss": 0.1984,
+      "step": 7322
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1278239553290727e-05,
+      "loss": 0.1852,
+      "step": 7323
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.127615382869747e-05,
+      "loss": 0.1963,
+      "step": 7324
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1274068047667757e-05,
+      "loss": 0.1768,
+      "step": 7325
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1271982210293834e-05,
+      "loss": 0.2094,
+      "step": 7326
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.126989631666794e-05,
+      "loss": 0.1874,
+      "step": 7327
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1267810366882326e-05,
+      "loss": 0.2196,
+      "step": 7328
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1265724361029237e-05,
+      "loss": 0.2025,
+      "step": 7329
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1263638299200928e-05,
+      "loss": 0.2,
+      "step": 7330
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1261552181489646e-05,
+      "loss": 0.1963,
+      "step": 7331
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1259466007987654e-05,
+      "loss": 0.1944,
+      "step": 7332
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1257379778787207e-05,
+      "loss": 0.1704,
+      "step": 7333
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1255293493980568e-05,
+      "loss": 0.2296,
+      "step": 7334
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.125320715366e-05,
+      "loss": 0.2046,
+      "step": 7335
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1251120757917769e-05,
+      "loss": 0.2293,
+      "step": 7336
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.124903430684614e-05,
+      "loss": 0.1796,
+      "step": 7337
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.124694780053739e-05,
+      "loss": 0.2289,
+      "step": 7338
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.124486123908379e-05,
+      "loss": 0.1912,
+      "step": 7339
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1242774622577616e-05,
+      "loss": 0.1951,
+      "step": 7340
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1240687951111145e-05,
+      "loss": 0.1821,
+      "step": 7341
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1238601224776658e-05,
+      "loss": 0.1703,
+      "step": 7342
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.123651444366644e-05,
+      "loss": 0.1743,
+      "step": 7343
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1234427607872774e-05,
+      "loss": 0.1886,
+      "step": 7344
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1232340717487945e-05,
+      "loss": 0.1757,
+      "step": 7345
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1230253772604252e-05,
+      "loss": 0.1618,
+      "step": 7346
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1228166773313982e-05,
+      "loss": 0.1821,
+      "step": 7347
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1226079719709431e-05,
+      "loss": 0.1877,
+      "step": 7348
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1223992611882897e-05,
+      "loss": 0.17,
+      "step": 7349
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1221905449926678e-05,
+      "loss": 0.1832,
+      "step": 7350
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.121981823393308e-05,
+      "loss": 0.1983,
+      "step": 7351
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1217730963994404e-05,
+      "loss": 0.1945,
+      "step": 7352
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1215643640202962e-05,
+      "loss": 0.1796,
+      "step": 7353
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.121355626265106e-05,
+      "loss": 0.2043,
+      "step": 7354
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1211468831431007e-05,
+      "loss": 0.1996,
+      "step": 7355
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1209381346635124e-05,
+      "loss": 0.2015,
+      "step": 7356
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1207293808355727e-05,
+      "loss": 0.213,
+      "step": 7357
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1205206216685126e-05,
+      "loss": 0.2236,
+      "step": 7358
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1203118571715652e-05,
+      "loss": 0.1703,
+      "step": 7359
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1201030873539624e-05,
+      "loss": 0.2159,
+      "step": 7360
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1198943122249373e-05,
+      "loss": 0.1796,
+      "step": 7361
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1196855317937222e-05,
+      "loss": 0.2027,
+      "step": 7362
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1194767460695502e-05,
+      "loss": 0.1964,
+      "step": 7363
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.119267955061655e-05,
+      "loss": 0.1871,
+      "step": 7364
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.11905915877927e-05,
+      "loss": 0.1876,
+      "step": 7365
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1188503572316284e-05,
+      "loss": 0.1825,
+      "step": 7366
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1186415504279655e-05,
+      "loss": 0.1956,
+      "step": 7367
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1184327383775144e-05,
+      "loss": 0.2085,
+      "step": 7368
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1182239210895096e-05,
+      "loss": 0.1904,
+      "step": 7369
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1180150985731868e-05,
+      "loss": 0.2055,
+      "step": 7370
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1178062708377799e-05,
+      "loss": 0.195,
+      "step": 7371
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1175974378925246e-05,
+      "loss": 0.1982,
+      "step": 7372
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1173885997466558e-05,
+      "loss": 0.1734,
+      "step": 7373
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1171797564094097e-05,
+      "loss": 0.1991,
+      "step": 7374
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1169709078900222e-05,
+      "loss": 0.166,
+      "step": 7375
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1167620541977288e-05,
+      "loss": 0.1713,
+      "step": 7376
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1165531953417665e-05,
+      "loss": 0.1705,
+      "step": 7377
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1163443313313711e-05,
+      "loss": 0.1757,
+      "step": 7378
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1161354621757801e-05,
+      "loss": 0.2038,
+      "step": 7379
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.11592658788423e-05,
+      "loss": 0.179,
+      "step": 7380
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1157177084659585e-05,
+      "loss": 0.2059,
+      "step": 7381
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1155088239302024e-05,
+      "loss": 0.1861,
+      "step": 7382
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1152999342861997e-05,
+      "loss": 0.1921,
+      "step": 7383
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1150910395431886e-05,
+      "loss": 0.1846,
+      "step": 7384
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1148821397104068e-05,
+      "loss": 0.2029,
+      "step": 7385
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.114673234797093e-05,
+      "loss": 0.2185,
+      "step": 7386
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1144643248124855e-05,
+      "loss": 0.212,
+      "step": 7387
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1142554097658232e-05,
+      "loss": 0.2083,
+      "step": 7388
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1140464896663451e-05,
+      "loss": 0.1921,
+      "step": 7389
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1138375645232906e-05,
+      "loss": 0.2165,
+      "step": 7390
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1136286343458992e-05,
+      "loss": 0.205,
+      "step": 7391
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1134196991434103e-05,
+      "loss": 0.178,
+      "step": 7392
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1132107589250642e-05,
+      "loss": 0.1894,
+      "step": 7393
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1130018137001009e-05,
+      "loss": 0.1693,
+      "step": 7394
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1127928634777608e-05,
+      "loss": 0.2234,
+      "step": 7395
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1125839082672841e-05,
+      "loss": 0.18,
+      "step": 7396
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1123749480779115e-05,
+      "loss": 0.1811,
+      "step": 7397
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1121659829188853e-05,
+      "loss": 0.1831,
+      "step": 7398
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1119570127994455e-05,
+      "loss": 0.1606,
+      "step": 7399
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.111748037728834e-05,
+      "loss": 0.1966,
+      "step": 7400
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1115390577162924e-05,
+      "loss": 0.1944,
+      "step": 7401
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1113300727710627e-05,
+      "loss": 0.1686,
+      "step": 7402
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1111210829023868e-05,
+      "loss": 0.2075,
+      "step": 7403
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1109120881195077e-05,
+      "loss": 0.16,
+      "step": 7404
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1107030884316667e-05,
+      "loss": 0.1857,
+      "step": 7405
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1104940838481078e-05,
+      "loss": 0.1673,
+      "step": 7406
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1102850743780734e-05,
+      "loss": 0.1771,
+      "step": 7407
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.110076060030807e-05,
+      "loss": 0.1839,
+      "step": 7408
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1098670408155518e-05,
+      "loss": 0.1804,
+      "step": 7409
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1096580167415512e-05,
+      "loss": 0.1952,
+      "step": 7410
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1094489878180497e-05,
+      "loss": 0.1698,
+      "step": 7411
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1092399540542908e-05,
+      "loss": 0.2001,
+      "step": 7412
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1090309154595191e-05,
+      "loss": 0.1865,
+      "step": 7413
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1088218720429788e-05,
+      "loss": 0.1923,
+      "step": 7414
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1086128238139151e-05,
+      "loss": 0.1891,
+      "step": 7415
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1084037707815724e-05,
+      "loss": 0.1815,
+      "step": 7416
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.108194712955196e-05,
+      "loss": 0.208,
+      "step": 7417
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1079856503440316e-05,
+      "loss": 0.1702,
+      "step": 7418
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1077765829573239e-05,
+      "loss": 0.1928,
+      "step": 7419
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1075675108043196e-05,
+      "loss": 0.1598,
+      "step": 7420
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1073584338942643e-05,
+      "loss": 0.1792,
+      "step": 7421
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1071493522364043e-05,
+      "loss": 0.1944,
+      "step": 7422
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1069402658399857e-05,
+      "loss": 0.1991,
+      "step": 7423
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.106731174714255e-05,
+      "loss": 0.1842,
+      "step": 7424
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1065220788684598e-05,
+      "loss": 0.1898,
+      "step": 7425
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1063129783118465e-05,
+      "loss": 0.1928,
+      "step": 7426
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1061038730536623e-05,
+      "loss": 0.1917,
+      "step": 7427
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1058947631031548e-05,
+      "loss": 0.1945,
+      "step": 7428
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1056856484695719e-05,
+      "loss": 0.1907,
+      "step": 7429
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.105476529162161e-05,
+      "loss": 0.2066,
+      "step": 7430
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1052674051901706e-05,
+      "loss": 0.2118,
+      "step": 7431
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1050582765628488e-05,
+      "loss": 0.2207,
+      "step": 7432
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1048491432894434e-05,
+      "loss": 0.1671,
+      "step": 7433
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1046400053792043e-05,
+      "loss": 0.21,
+      "step": 7434
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1044308628413798e-05,
+      "loss": 0.2189,
+      "step": 7435
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1042217156852188e-05,
+      "loss": 0.1818,
+      "step": 7436
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1040125639199708e-05,
+      "loss": 0.2168,
+      "step": 7437
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1038034075548852e-05,
+      "loss": 0.1876,
+      "step": 7438
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.103594246599212e-05,
+      "loss": 0.1724,
+      "step": 7439
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1033850810622009e-05,
+      "loss": 0.1798,
+      "step": 7440
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1031759109531017e-05,
+      "loss": 0.1635,
+      "step": 7441
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1029667362811654e-05,
+      "loss": 0.2018,
+      "step": 7442
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1027575570556417e-05,
+      "loss": 0.1876,
+      "step": 7443
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.102548373285782e-05,
+      "loss": 0.1927,
+      "step": 7444
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1023391849808372e-05,
+      "loss": 0.1973,
+      "step": 7445
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1021299921500579e-05,
+      "loss": 0.1722,
+      "step": 7446
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1019207948026953e-05,
+      "loss": 0.1857,
+      "step": 7447
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1017115929480015e-05,
+      "loss": 0.1869,
+      "step": 7448
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1015023865952285e-05,
+      "loss": 0.188,
+      "step": 7449
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1012931757536273e-05,
+      "loss": 0.1677,
+      "step": 7450
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1010839604324506e-05,
+      "loss": 0.1897,
+      "step": 7451
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1008747406409504e-05,
+      "loss": 0.237,
+      "step": 7452
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1006655163883795e-05,
+      "loss": 0.2089,
+      "step": 7453
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1004562876839902e-05,
+      "loss": 0.2004,
+      "step": 7454
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1002470545370359e-05,
+      "loss": 0.1679,
+      "step": 7455
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1000378169567694e-05,
+      "loss": 0.1829,
+      "step": 7456
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.099828574952444e-05,
+      "loss": 0.17,
+      "step": 7457
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0996193285333132e-05,
+      "loss": 0.1927,
+      "step": 7458
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.099410077708631e-05,
+      "loss": 0.219,
+      "step": 7459
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.099200822487651e-05,
+      "loss": 0.2101,
+      "step": 7460
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0989915628796266e-05,
+      "loss": 0.1693,
+      "step": 7461
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0987822988938136e-05,
+      "loss": 0.1748,
+      "step": 7462
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0985730305394654e-05,
+      "loss": 0.2029,
+      "step": 7463
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0983637578258365e-05,
+      "loss": 0.1799,
+      "step": 7464
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0981544807621824e-05,
+      "loss": 0.2309,
+      "step": 7465
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0979451993577577e-05,
+      "loss": 0.1795,
+      "step": 7466
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0977359136218179e-05,
+      "loss": 0.1609,
+      "step": 7467
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0975266235636184e-05,
+      "loss": 0.1922,
+      "step": 7468
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0973173291924147e-05,
+      "loss": 0.1827,
+      "step": 7469
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0971080305174623e-05,
+      "loss": 0.198,
+      "step": 7470
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0968987275480179e-05,
+      "loss": 0.1872,
+      "step": 7471
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0966894202933372e-05,
+      "loss": 0.1827,
+      "step": 7472
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0964801087626768e-05,
+      "loss": 0.2164,
+      "step": 7473
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0962707929652929e-05,
+      "loss": 0.1735,
+      "step": 7474
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0960614729104429e-05,
+      "loss": 0.1839,
+      "step": 7475
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0958521486073832e-05,
+      "loss": 0.2298,
+      "step": 7476
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.095642820065371e-05,
+      "loss": 0.1941,
+      "step": 7477
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0954334872936637e-05,
+      "loss": 0.2199,
+      "step": 7478
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0952241503015188e-05,
+      "loss": 0.1932,
+      "step": 7479
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.095014809098194e-05,
+      "loss": 0.1915,
+      "step": 7480
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0948054636929473e-05,
+      "loss": 0.2031,
+      "step": 7481
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0945961140950363e-05,
+      "loss": 0.1654,
+      "step": 7482
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.09438676031372e-05,
+      "loss": 0.1989,
+      "step": 7483
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.094177402358256e-05,
+      "loss": 0.2056,
+      "step": 7484
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0939680402379036e-05,
+      "loss": 0.1807,
+      "step": 7485
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0937586739619215e-05,
+      "loss": 0.1983,
+      "step": 7486
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0935493035395684e-05,
+      "loss": 0.181,
+      "step": 7487
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0933399289801032e-05,
+      "loss": 0.1809,
+      "step": 7488
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0931305502927857e-05,
+      "loss": 0.1976,
+      "step": 7489
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0929211674868761e-05,
+      "loss": 0.174,
+      "step": 7490
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0927117805716327e-05,
+      "loss": 0.1721,
+      "step": 7491
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0925023895563162e-05,
+      "loss": 0.1658,
+      "step": 7492
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0922929944501868e-05,
+      "loss": 0.2246,
+      "step": 7493
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0920835952625045e-05,
+      "loss": 0.1973,
+      "step": 7494
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0918741920025295e-05,
+      "loss": 0.2067,
+      "step": 7495
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.091664784679523e-05,
+      "loss": 0.2005,
+      "step": 7496
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0914553733027453e-05,
+      "loss": 0.2128,
+      "step": 7497
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0912459578814576e-05,
+      "loss": 0.1805,
+      "step": 7498
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0910365384249211e-05,
+      "loss": 0.1969,
+      "step": 7499
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0908271149423972e-05,
+      "loss": 0.1715,
+      "step": 7500
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0906176874431471e-05,
+      "loss": 0.1811,
+      "step": 7501
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0904082559364326e-05,
+      "loss": 0.1893,
+      "step": 7502
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0901988204315159e-05,
+      "loss": 0.1883,
+      "step": 7503
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0899893809376588e-05,
+      "loss": 0.1973,
+      "step": 7504
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0897799374641234e-05,
+      "loss": 0.2035,
+      "step": 7505
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.089570490020172e-05,
+      "loss": 0.1791,
+      "step": 7506
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0893610386150677e-05,
+      "loss": 0.1831,
+      "step": 7507
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0891515832580727e-05,
+      "loss": 0.1692,
+      "step": 7508
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0889421239584506e-05,
+      "loss": 0.2002,
+      "step": 7509
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.088732660725464e-05,
+      "loss": 0.2046,
+      "step": 7510
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0885231935683758e-05,
+      "loss": 0.1927,
+      "step": 7511
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0883137224964502e-05,
+      "loss": 0.169,
+      "step": 7512
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0881042475189503e-05,
+      "loss": 0.1944,
+      "step": 7513
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0878947686451406e-05,
+      "loss": 0.2005,
+      "step": 7514
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0876852858842841e-05,
+      "loss": 0.1924,
+      "step": 7515
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0874757992456455e-05,
+      "loss": 0.2256,
+      "step": 7516
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0872663087384893e-05,
+      "loss": 0.203,
+      "step": 7517
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0870568143720794e-05,
+      "loss": 0.152,
+      "step": 7518
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0868473161556809e-05,
+      "loss": 0.1666,
+      "step": 7519
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0866378140985584e-05,
+      "loss": 0.207,
+      "step": 7520
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0864283082099768e-05,
+      "loss": 0.1772,
+      "step": 7521
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0862187984992016e-05,
+      "loss": 0.2188,
+      "step": 7522
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0860092849754982e-05,
+      "loss": 0.177,
+      "step": 7523
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0857997676481315e-05,
+      "loss": 0.1776,
+      "step": 7524
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0855902465263678e-05,
+      "loss": 0.1882,
+      "step": 7525
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0853807216194723e-05,
+      "loss": 0.179,
+      "step": 7526
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0851711929367116e-05,
+      "loss": 0.2026,
+      "step": 7527
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0849616604873516e-05,
+      "loss": 0.217,
+      "step": 7528
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0847521242806585e-05,
+      "loss": 0.1715,
+      "step": 7529
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0845425843258988e-05,
+      "loss": 0.1897,
+      "step": 7530
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0843330406323398e-05,
+      "loss": 0.2041,
+      "step": 7531
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0841234932092475e-05,
+      "loss": 0.1796,
+      "step": 7532
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0839139420658895e-05,
+      "loss": 0.2052,
+      "step": 7533
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0837043872115325e-05,
+      "loss": 0.1935,
+      "step": 7534
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0834948286554441e-05,
+      "loss": 0.1905,
+      "step": 7535
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0832852664068916e-05,
+      "loss": 0.1873,
+      "step": 7536
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0830757004751432e-05,
+      "loss": 0.193,
+      "step": 7537
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0828661308694656e-05,
+      "loss": 0.2009,
+      "step": 7538
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.082656557599128e-05,
+      "loss": 0.1825,
+      "step": 7539
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0824469806733977e-05,
+      "loss": 0.1815,
+      "step": 7540
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0822374001015437e-05,
+      "loss": 0.1929,
+      "step": 7541
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.082027815892834e-05,
+      "loss": 0.1651,
+      "step": 7542
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0818182280565368e-05,
+      "loss": 0.205,
+      "step": 7543
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0816086366019218e-05,
+      "loss": 0.18,
+      "step": 7544
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0813990415382577e-05,
+      "loss": 0.1797,
+      "step": 7545
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0811894428748131e-05,
+      "loss": 0.2015,
+      "step": 7546
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0809798406208576e-05,
+      "loss": 0.1749,
+      "step": 7547
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0807702347856606e-05,
+      "loss": 0.2038,
+      "step": 7548
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0805606253784917e-05,
+      "loss": 0.1934,
+      "step": 7549
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0803510124086207e-05,
+      "loss": 0.1791,
+      "step": 7550
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0801413958853177e-05,
+      "loss": 0.1911,
+      "step": 7551
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.079931775817852e-05,
+      "loss": 0.1946,
+      "step": 7552
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0797221522154944e-05,
+      "loss": 0.175,
+      "step": 7553
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0795125250875156e-05,
+      "loss": 0.215,
+      "step": 7554
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0793028944431851e-05,
+      "loss": 0.2127,
+      "step": 7555
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0790932602917745e-05,
+      "loss": 0.1668,
+      "step": 7556
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0788836226425538e-05,
+      "loss": 0.1961,
+      "step": 7557
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0786739815047951e-05,
+      "loss": 0.185,
+      "step": 7558
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0784643368877688e-05,
+      "loss": 0.2166,
+      "step": 7559
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0782546888007461e-05,
+      "loss": 0.2068,
+      "step": 7560
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0780450372529987e-05,
+      "loss": 0.1808,
+      "step": 7561
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.077835382253798e-05,
+      "loss": 0.1744,
+      "step": 7562
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0776257238124161e-05,
+      "loss": 0.1757,
+      "step": 7563
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0774160619381248e-05,
+      "loss": 0.1872,
+      "step": 7564
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0772063966401962e-05,
+      "loss": 0.1961,
+      "step": 7565
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.076996727927902e-05,
+      "loss": 0.1787,
+      "step": 7566
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.076787055810515e-05,
+      "loss": 0.2226,
+      "step": 7567
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.076577380297308e-05,
+      "loss": 0.1948,
+      "step": 7568
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.076367701397553e-05,
+      "loss": 0.1815,
+      "step": 7569
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0761580191205234e-05,
+      "loss": 0.2107,
+      "step": 7570
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0759483334754915e-05,
+      "loss": 0.1892,
+      "step": 7571
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0757386444717313e-05,
+      "loss": 0.189,
+      "step": 7572
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0755289521185153e-05,
+      "loss": 0.192,
+      "step": 7573
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0753192564251171e-05,
+      "loss": 0.1578,
+      "step": 7574
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0751095574008103e-05,
+      "loss": 0.1706,
+      "step": 7575
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0748998550548686e-05,
+      "loss": 0.1902,
+      "step": 7576
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.074690149396566e-05,
+      "loss": 0.1736,
+      "step": 7577
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0744804404351765e-05,
+      "loss": 0.2069,
+      "step": 7578
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.074270728179974e-05,
+      "loss": 0.1612,
+      "step": 7579
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0740610126402326e-05,
+      "loss": 0.1753,
+      "step": 7580
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0738512938252272e-05,
+      "loss": 0.2173,
+      "step": 7581
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0736415717442326e-05,
+      "loss": 0.1692,
+      "step": 7582
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0734318464065227e-05,
+      "loss": 0.1847,
+      "step": 7583
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.073222117821373e-05,
+      "loss": 0.1836,
+      "step": 7584
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.073012385998058e-05,
+      "loss": 0.1971,
+      "step": 7585
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0728026509458533e-05,
+      "loss": 0.1938,
+      "step": 7586
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.072592912674034e-05,
+      "loss": 0.1655,
+      "step": 7587
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0723831711918758e-05,
+      "loss": 0.1788,
+      "step": 7588
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0721734265086536e-05,
+      "loss": 0.1741,
+      "step": 7589
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.071963678633644e-05,
+      "loss": 0.1926,
+      "step": 7590
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0717539275761222e-05,
+      "loss": 0.1811,
+      "step": 7591
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0715441733453647e-05,
+      "loss": 0.1899,
+      "step": 7592
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0713344159506472e-05,
+      "loss": 0.201,
+      "step": 7593
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0711246554012459e-05,
+      "loss": 0.2047,
+      "step": 7594
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.070914891706438e-05,
+      "loss": 0.1783,
+      "step": 7595
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0707051248754993e-05,
+      "loss": 0.1801,
+      "step": 7596
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0704953549177069e-05,
+      "loss": 0.1906,
+      "step": 7597
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0702855818423372e-05,
+      "loss": 0.2031,
+      "step": 7598
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0700758056586676e-05,
+      "loss": 0.1698,
+      "step": 7599
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.069866026375975e-05,
+      "loss": 0.2143,
+      "step": 7600
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0696562440035369e-05,
+      "loss": 0.1841,
+      "step": 7601
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0694464585506304e-05,
+      "loss": 0.1975,
+      "step": 7602
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.069236670026533e-05,
+      "loss": 0.1497,
+      "step": 7603
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.069026878440523e-05,
+      "loss": 0.1838,
+      "step": 7604
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0688170838018772e-05,
+      "loss": 0.1904,
+      "step": 7605
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0686072861198744e-05,
+      "loss": 0.1791,
+      "step": 7606
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0683974854037917e-05,
+      "loss": 0.1565,
+      "step": 7607
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0681876816629084e-05,
+      "loss": 0.2111,
+      "step": 7608
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0679778749065022e-05,
+      "loss": 0.1792,
+      "step": 7609
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0677680651438518e-05,
+      "loss": 0.1812,
+      "step": 7610
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0675582523842353e-05,
+      "loss": 0.2026,
+      "step": 7611
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0673484366369322e-05,
+      "loss": 0.1859,
+      "step": 7612
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0671386179112207e-05,
+      "loss": 0.1734,
+      "step": 7613
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.06692879621638e-05,
+      "loss": 0.2095,
+      "step": 7614
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0667189715616897e-05,
+      "loss": 0.1782,
+      "step": 7615
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0665091439564283e-05,
+      "loss": 0.1885,
+      "step": 7616
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0662993134098757e-05,
+      "loss": 0.2021,
+      "step": 7617
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.066089479931311e-05,
+      "loss": 0.1732,
+      "step": 7618
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0658796435300146e-05,
+      "loss": 0.2243,
+      "step": 7619
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0656698042152656e-05,
+      "loss": 0.1786,
+      "step": 7620
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0654599619963436e-05,
+      "loss": 0.2011,
+      "step": 7621
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0652501168825296e-05,
+      "loss": 0.1814,
+      "step": 7622
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0650402688831034e-05,
+      "loss": 0.1804,
+      "step": 7623
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.064830418007345e-05,
+      "loss": 0.1872,
+      "step": 7624
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0646205642645349e-05,
+      "loss": 0.198,
+      "step": 7625
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0644107076639539e-05,
+      "loss": 0.1988,
+      "step": 7626
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0642008482148822e-05,
+      "loss": 0.1785,
+      "step": 7627
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0639909859266011e-05,
+      "loss": 0.1834,
+      "step": 7628
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0637811208083913e-05,
+      "loss": 0.1821,
+      "step": 7629
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.063571252869534e-05,
+      "loss": 0.197,
+      "step": 7630
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0633613821193099e-05,
+      "loss": 0.2156,
+      "step": 7631
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0631515085670007e-05,
+      "loss": 0.19,
+      "step": 7632
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0629416322218883e-05,
+      "loss": 0.2085,
+      "step": 7633
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0627317530932532e-05,
+      "loss": 0.1717,
+      "step": 7634
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0625218711903772e-05,
+      "loss": 0.235,
+      "step": 7635
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0623119865225433e-05,
+      "loss": 0.2084,
+      "step": 7636
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.062102099099032e-05,
+      "loss": 0.1795,
+      "step": 7637
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0618922089291262e-05,
+      "loss": 0.2227,
+      "step": 7638
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0616823160221075e-05,
+      "loss": 0.1763,
+      "step": 7639
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0614724203872586e-05,
+      "loss": 0.2054,
+      "step": 7640
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0612625220338615e-05,
+      "loss": 0.1724,
+      "step": 7641
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.061052620971199e-05,
+      "loss": 0.2062,
+      "step": 7642
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.060842717208554e-05,
+      "loss": 0.1764,
+      "step": 7643
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0606328107552083e-05,
+      "loss": 0.1988,
+      "step": 7644
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0604229016204459e-05,
+      "loss": 0.1764,
+      "step": 7645
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0602129898135491e-05,
+      "loss": 0.178,
+      "step": 7646
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0600030753438015e-05,
+      "loss": 0.2023,
+      "step": 7647
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.059793158220486e-05,
+      "loss": 0.1623,
+      "step": 7648
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0595832384528855e-05,
+      "loss": 0.2007,
+      "step": 7649
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0593733160502844e-05,
+      "loss": 0.1801,
+      "step": 7650
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0591633910219656e-05,
+      "loss": 0.2191,
+      "step": 7651
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0589534633772129e-05,
+      "loss": 0.1684,
+      "step": 7652
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0587435331253104e-05,
+      "loss": 0.1903,
+      "step": 7653
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.058533600275542e-05,
+      "loss": 0.1738,
+      "step": 7654
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0583236648371912e-05,
+      "loss": 0.2028,
+      "step": 7655
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.058113726819543e-05,
+      "loss": 0.1666,
+      "step": 7656
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0579037862318813e-05,
+      "loss": 0.1828,
+      "step": 7657
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.05769384308349e-05,
+      "loss": 0.1866,
+      "step": 7658
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.057483897383654e-05,
+      "loss": 0.1798,
+      "step": 7659
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0572739491416584e-05,
+      "loss": 0.1836,
+      "step": 7660
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0570639983667873e-05,
+      "loss": 0.196,
+      "step": 7661
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0568540450683253e-05,
+      "loss": 0.193,
+      "step": 7662
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0566440892555576e-05,
+      "loss": 0.1849,
+      "step": 7663
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0564341309377698e-05,
+      "loss": 0.1732,
+      "step": 7664
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0562241701242469e-05,
+      "loss": 0.1838,
+      "step": 7665
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0560142068242733e-05,
+      "loss": 0.2069,
+      "step": 7666
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0558042410471354e-05,
+      "loss": 0.2177,
+      "step": 7667
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0555942728021183e-05,
+      "loss": 0.1862,
+      "step": 7668
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0553843020985075e-05,
+      "loss": 0.1908,
+      "step": 7669
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0551743289455892e-05,
+      "loss": 0.1865,
+      "step": 7670
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0549643533526481e-05,
+      "loss": 0.1828,
+      "step": 7671
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0547543753289713e-05,
+      "loss": 0.22,
+      "step": 7672
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0545443948838445e-05,
+      "loss": 0.207,
+      "step": 7673
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.054334412026554e-05,
+      "loss": 0.2038,
+      "step": 7674
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0541244267663855e-05,
+      "loss": 0.2311,
+      "step": 7675
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0539144391126256e-05,
+      "loss": 0.1933,
+      "step": 7676
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0537044490745612e-05,
+      "loss": 0.1948,
+      "step": 7677
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0534944566614784e-05,
+      "loss": 0.1862,
+      "step": 7678
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0532844618826639e-05,
+      "loss": 0.1837,
+      "step": 7679
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0530744647474048e-05,
+      "loss": 0.2028,
+      "step": 7680
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0528644652649875e-05,
+      "loss": 0.1831,
+      "step": 7681
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0526544634446994e-05,
+      "loss": 0.1738,
+      "step": 7682
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0524444592958273e-05,
+      "loss": 0.1868,
+      "step": 7683
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0522344528276588e-05,
+      "loss": 0.1856,
+      "step": 7684
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0520244440494807e-05,
+      "loss": 0.2189,
+      "step": 7685
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0518144329705808e-05,
+      "loss": 0.1779,
+      "step": 7686
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0516044196002466e-05,
+      "loss": 0.1883,
+      "step": 7687
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0513944039477652e-05,
+      "loss": 0.1903,
+      "step": 7688
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.051184386022425e-05,
+      "loss": 0.2256,
+      "step": 7689
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.050974365833513e-05,
+      "loss": 0.1798,
+      "step": 7690
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0507643433903178e-05,
+      "loss": 0.1695,
+      "step": 7691
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0505543187021274e-05,
+      "loss": 0.183,
+      "step": 7692
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0503442917782294e-05,
+      "loss": 0.2165,
+      "step": 7693
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0501342626279123e-05,
+      "loss": 0.1862,
+      "step": 7694
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0499242312604642e-05,
+      "loss": 0.2012,
+      "step": 7695
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.049714197685174e-05,
+      "loss": 0.2131,
+      "step": 7696
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0495041619113297e-05,
+      "loss": 0.1827,
+      "step": 7697
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0492941239482204e-05,
+      "loss": 0.1909,
+      "step": 7698
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0490840838051338e-05,
+      "loss": 0.1985,
+      "step": 7699
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0488740414913597e-05,
+      "loss": 0.1813,
+      "step": 7700
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0486639970161867e-05,
+      "loss": 0.1869,
+      "step": 7701
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0484539503889036e-05,
+      "loss": 0.1875,
+      "step": 7702
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0482439016187995e-05,
+      "loss": 0.1901,
+      "step": 7703
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0480338507151635e-05,
+      "loss": 0.1712,
+      "step": 7704
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0478237976872854e-05,
+      "loss": 0.1823,
+      "step": 7705
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.047613742544454e-05,
+      "loss": 0.1916,
+      "step": 7706
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0474036852959588e-05,
+      "loss": 0.1656,
+      "step": 7707
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0471936259510896e-05,
+      "loss": 0.1857,
+      "step": 7708
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0469835645191357e-05,
+      "loss": 0.1975,
+      "step": 7709
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0467735010093872e-05,
+      "loss": 0.185,
+      "step": 7710
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0465634354311339e-05,
+      "loss": 0.1838,
+      "step": 7711
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0463533677936652e-05,
+      "loss": 0.2188,
+      "step": 7712
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0461432981062713e-05,
+      "loss": 0.202,
+      "step": 7713
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0459332263782427e-05,
+      "loss": 0.1746,
+      "step": 7714
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0457231526188695e-05,
+      "loss": 0.1709,
+      "step": 7715
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0455130768374415e-05,
+      "loss": 0.2015,
+      "step": 7716
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0453029990432495e-05,
+      "loss": 0.18,
+      "step": 7717
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.045092919245584e-05,
+      "loss": 0.1599,
+      "step": 7718
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.044882837453735e-05,
+      "loss": 0.204,
+      "step": 7719
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0446727536769935e-05,
+      "loss": 0.1881,
+      "step": 7720
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0444626679246506e-05,
+      "loss": 0.1751,
+      "step": 7721
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0442525802059964e-05,
+      "loss": 0.1817,
+      "step": 7722
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0440424905303218e-05,
+      "loss": 0.1752,
+      "step": 7723
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0438323989069186e-05,
+      "loss": 0.1812,
+      "step": 7724
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0436223053450774e-05,
+      "loss": 0.2027,
+      "step": 7725
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0434122098540892e-05,
+      "loss": 0.1993,
+      "step": 7726
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0432021124432447e-05,
+      "loss": 0.1822,
+      "step": 7727
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0429920131218364e-05,
+      "loss": 0.188,
+      "step": 7728
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0427819118991552e-05,
+      "loss": 0.1618,
+      "step": 7729
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0425718087844923e-05,
+      "loss": 0.179,
+      "step": 7730
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0423617037871396e-05,
+      "loss": 0.1881,
+      "step": 7731
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0421515969163886e-05,
+      "loss": 0.1842,
+      "step": 7732
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.041941488181531e-05,
+      "loss": 0.2024,
+      "step": 7733
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.041731377591859e-05,
+      "loss": 0.1852,
+      "step": 7734
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.041521265156664e-05,
+      "loss": 0.1918,
+      "step": 7735
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0413111508852382e-05,
+      "loss": 0.1718,
+      "step": 7736
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0411010347868737e-05,
+      "loss": 0.2037,
+      "step": 7737
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0408909168708627e-05,
+      "loss": 0.195,
+      "step": 7738
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0406807971464976e-05,
+      "loss": 0.2019,
+      "step": 7739
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0404706756230699e-05,
+      "loss": 0.1747,
+      "step": 7740
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0402605523098728e-05,
+      "loss": 0.215,
+      "step": 7741
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0400504272161987e-05,
+      "loss": 0.2004,
+      "step": 7742
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0398403003513395e-05,
+      "loss": 0.203,
+      "step": 7743
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0396301717245885e-05,
+      "loss": 0.1959,
+      "step": 7744
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.039420041345238e-05,
+      "loss": 0.1841,
+      "step": 7745
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0392099092225812e-05,
+      "loss": 0.1543,
+      "step": 7746
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0389997753659105e-05,
+      "loss": 0.1833,
+      "step": 7747
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.038789639784519e-05,
+      "loss": 0.1625,
+      "step": 7748
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0385795024877e-05,
+      "loss": 0.1937,
+      "step": 7749
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0383693634847458e-05,
+      "loss": 0.1791,
+      "step": 7750
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0381592227849503e-05,
+      "loss": 0.1726,
+      "step": 7751
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0379490803976071e-05,
+      "loss": 0.1642,
+      "step": 7752
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0377389363320083e-05,
+      "loss": 0.1799,
+      "step": 7753
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0375287905974478e-05,
+      "loss": 0.1788,
+      "step": 7754
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0373186432032195e-05,
+      "loss": 0.1867,
+      "step": 7755
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0371084941586165e-05,
+      "loss": 0.1646,
+      "step": 7756
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0368983434729325e-05,
+      "loss": 0.2001,
+      "step": 7757
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0366881911554612e-05,
+      "loss": 0.1999,
+      "step": 7758
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0364780372154962e-05,
+      "loss": 0.192,
+      "step": 7759
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0362678816623312e-05,
+      "loss": 0.1818,
+      "step": 7760
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0360577245052607e-05,
+      "loss": 0.1957,
+      "step": 7761
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.035847565753578e-05,
+      "loss": 0.1734,
+      "step": 7762
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0356374054165777e-05,
+      "loss": 0.1976,
+      "step": 7763
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0354272435035536e-05,
+      "loss": 0.2048,
+      "step": 7764
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0352170800237997e-05,
+      "loss": 0.1852,
+      "step": 7765
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0350069149866106e-05,
+      "loss": 0.1698,
+      "step": 7766
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0347967484012805e-05,
+      "loss": 0.1928,
+      "step": 7767
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0345865802771032e-05,
+      "loss": 0.1885,
+      "step": 7768
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0343764106233743e-05,
+      "loss": 0.1849,
+      "step": 7769
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0341662394493875e-05,
+      "loss": 0.1764,
+      "step": 7770
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0339560667644373e-05,
+      "loss": 0.1596,
+      "step": 7771
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0337458925778187e-05,
+      "loss": 0.1981,
+      "step": 7772
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0335357168988265e-05,
+      "loss": 0.1998,
+      "step": 7773
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0333255397367552e-05,
+      "loss": 0.1753,
+      "step": 7774
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0331153611008997e-05,
+      "loss": 0.187,
+      "step": 7775
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0329051810005553e-05,
+      "loss": 0.1717,
+      "step": 7776
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0326949994450162e-05,
+      "loss": 0.1759,
+      "step": 7777
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.032484816443578e-05,
+      "loss": 0.182,
+      "step": 7778
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0322746320055358e-05,
+      "loss": 0.1918,
+      "step": 7779
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0320644461401844e-05,
+      "loss": 0.1925,
+      "step": 7780
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0318542588568197e-05,
+      "loss": 0.1813,
+      "step": 7781
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0316440701647361e-05,
+      "loss": 0.1846,
+      "step": 7782
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0314338800732298e-05,
+      "loss": 0.1619,
+      "step": 7783
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0312236885915959e-05,
+      "loss": 0.195,
+      "step": 7784
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0310134957291295e-05,
+      "loss": 0.166,
+      "step": 7785
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0308033014951267e-05,
+      "loss": 0.1772,
+      "step": 7786
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0305931058988827e-05,
+      "loss": 0.1528,
+      "step": 7787
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0303829089496934e-05,
+      "loss": 0.1702,
+      "step": 7788
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0301727106568544e-05,
+      "loss": 0.1767,
+      "step": 7789
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0299625110296621e-05,
+      "loss": 0.2173,
+      "step": 7790
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0297523100774109e-05,
+      "loss": 0.2074,
+      "step": 7791
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0295421078093978e-05,
+      "loss": 0.1731,
+      "step": 7792
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.029331904234919e-05,
+      "loss": 0.1869,
+      "step": 7793
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0291216993632698e-05,
+      "loss": 0.1687,
+      "step": 7794
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0289114932037465e-05,
+      "loss": 0.185,
+      "step": 7795
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0287012857656449e-05,
+      "loss": 0.2296,
+      "step": 7796
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0284910770582621e-05,
+      "loss": 0.1767,
+      "step": 7797
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0282808670908935e-05,
+      "loss": 0.1928,
+      "step": 7798
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0280706558728358e-05,
+      "loss": 0.2082,
+      "step": 7799
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.027860443413385e-05,
+      "loss": 0.189,
+      "step": 7800
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0276502297218379e-05,
+      "loss": 0.1782,
+      "step": 7801
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0274400148074908e-05,
+      "loss": 0.2228,
+      "step": 7802
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0272297986796402e-05,
+      "loss": 0.2022,
+      "step": 7803
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0270195813475824e-05,
+      "loss": 0.2111,
+      "step": 7804
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0268093628206146e-05,
+      "loss": 0.2139,
+      "step": 7805
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.026599143108033e-05,
+      "loss": 0.1788,
+      "step": 7806
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0263889222191348e-05,
+      "loss": 0.1819,
+      "step": 7807
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0261787001632165e-05,
+      "loss": 0.201,
+      "step": 7808
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0259684769495743e-05,
+      "loss": 0.1688,
+      "step": 7809
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0257582525875063e-05,
+      "loss": 0.1969,
+      "step": 7810
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0255480270863085e-05,
+      "loss": 0.1889,
+      "step": 7811
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0253378004552786e-05,
+      "loss": 0.1892,
+      "step": 7812
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0251275727037128e-05,
+      "loss": 0.2132,
+      "step": 7813
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0249173438409088e-05,
+      "loss": 0.197,
+      "step": 7814
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0247071138761635e-05,
+      "loss": 0.2189,
+      "step": 7815
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0244968828187742e-05,
+      "loss": 0.1582,
+      "step": 7816
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0242866506780385e-05,
+      "loss": 0.1652,
+      "step": 7817
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0240764174632527e-05,
+      "loss": 0.1746,
+      "step": 7818
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0238661831837146e-05,
+      "loss": 0.1848,
+      "step": 7819
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0236559478487222e-05,
+      "loss": 0.1625,
+      "step": 7820
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0234457114675722e-05,
+      "loss": 0.1948,
+      "step": 7821
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.023235474049562e-05,
+      "loss": 0.2167,
+      "step": 7822
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.023025235603989e-05,
+      "loss": 0.2005,
+      "step": 7823
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0228149961401521e-05,
+      "loss": 0.1824,
+      "step": 7824
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0226047556673474e-05,
+      "loss": 0.1873,
+      "step": 7825
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0223945141948728e-05,
+      "loss": 0.17,
+      "step": 7826
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0221842717320266e-05,
+      "loss": 0.1952,
+      "step": 7827
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0219740282881058e-05,
+      "loss": 0.2046,
+      "step": 7828
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.021763783872409e-05,
+      "loss": 0.1869,
+      "step": 7829
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0215535384942332e-05,
+      "loss": 0.2279,
+      "step": 7830
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0213432921628773e-05,
+      "loss": 0.1853,
+      "step": 7831
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0211330448876376e-05,
+      "loss": 0.1746,
+      "step": 7832
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0209227966778135e-05,
+      "loss": 0.1693,
+      "step": 7833
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0207125475427026e-05,
+      "loss": 0.1753,
+      "step": 7834
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0205022974916025e-05,
+      "loss": 0.1887,
+      "step": 7835
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0202920465338117e-05,
+      "loss": 0.1947,
+      "step": 7836
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0200817946786282e-05,
+      "loss": 0.1962,
+      "step": 7837
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0198715419353502e-05,
+      "loss": 0.19,
+      "step": 7838
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0196612883132757e-05,
+      "loss": 0.1754,
+      "step": 7839
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.019451033821703e-05,
+      "loss": 0.1727,
+      "step": 7840
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0192407784699305e-05,
+      "loss": 0.1545,
+      "step": 7841
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0190305222672565e-05,
+      "loss": 0.1844,
+      "step": 7842
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0188202652229793e-05,
+      "loss": 0.1685,
+      "step": 7843
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0186100073463971e-05,
+      "loss": 0.1959,
+      "step": 7844
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0183997486468087e-05,
+      "loss": 0.1634,
+      "step": 7845
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0181894891335119e-05,
+      "loss": 0.2148,
+      "step": 7846
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0179792288158058e-05,
+      "loss": 0.179,
+      "step": 7847
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.017768967702989e-05,
+      "loss": 0.2131,
+      "step": 7848
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0175587058043594e-05,
+      "loss": 0.1767,
+      "step": 7849
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.017348443129216e-05,
+      "loss": 0.1745,
+      "step": 7850
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0171381796868577e-05,
+      "loss": 0.2006,
+      "step": 7851
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0169279154865826e-05,
+      "loss": 0.1758,
+      "step": 7852
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0167176505376896e-05,
+      "loss": 0.1743,
+      "step": 7853
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0165073848494777e-05,
+      "loss": 0.1926,
+      "step": 7854
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0162971184312453e-05,
+      "loss": 0.1926,
+      "step": 7855
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0160868512922914e-05,
+      "loss": 0.1545,
+      "step": 7856
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0158765834419149e-05,
+      "loss": 0.1946,
+      "step": 7857
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0156663148894146e-05,
+      "loss": 0.1882,
+      "step": 7858
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.015456045644089e-05,
+      "loss": 0.1744,
+      "step": 7859
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0152457757152373e-05,
+      "loss": 0.1905,
+      "step": 7860
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0150355051121588e-05,
+      "loss": 0.1682,
+      "step": 7861
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.014825233844152e-05,
+      "loss": 0.1756,
+      "step": 7862
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.014614961920516e-05,
+      "loss": 0.1781,
+      "step": 7863
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0144046893505502e-05,
+      "loss": 0.175,
+      "step": 7864
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.014194416143553e-05,
+      "loss": 0.191,
+      "step": 7865
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0139841423088242e-05,
+      "loss": 0.1587,
+      "step": 7866
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0137738678556624e-05,
+      "loss": 0.1787,
+      "step": 7867
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0135635927933666e-05,
+      "loss": 0.1687,
+      "step": 7868
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0133533171312369e-05,
+      "loss": 0.1933,
+      "step": 7869
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0131430408785717e-05,
+      "loss": 0.1873,
+      "step": 7870
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0129327640446702e-05,
+      "loss": 0.1824,
+      "step": 7871
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0127224866388325e-05,
+      "loss": 0.2049,
+      "step": 7872
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0125122086703567e-05,
+      "loss": 0.2099,
+      "step": 7873
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0123019301485428e-05,
+      "loss": 0.1931,
+      "step": 7874
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0120916510826904e-05,
+      "loss": 0.2265,
+      "step": 7875
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0118813714820982e-05,
+      "loss": 0.1876,
+      "step": 7876
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0116710913560656e-05,
+      "loss": 0.1839,
+      "step": 7877
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0114608107138924e-05,
+      "loss": 0.1603,
+      "step": 7878
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0112505295648779e-05,
+      "loss": 0.1959,
+      "step": 7879
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0110402479183214e-05,
+      "loss": 0.1896,
+      "step": 7880
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0108299657835225e-05,
+      "loss": 0.1896,
+      "step": 7881
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0106196831697808e-05,
+      "loss": 0.219,
+      "step": 7882
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0104094000863956e-05,
+      "loss": 0.1923,
+      "step": 7883
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0101991165426666e-05,
+      "loss": 0.1663,
+      "step": 7884
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0099888325478932e-05,
+      "loss": 0.1954,
+      "step": 7885
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0097785481113751e-05,
+      "loss": 0.1783,
+      "step": 7886
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0095682632424114e-05,
+      "loss": 0.2136,
+      "step": 7887
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0093579779503024e-05,
+      "loss": 0.1842,
+      "step": 7888
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0091476922443477e-05,
+      "loss": 0.2199,
+      "step": 7889
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0089374061338464e-05,
+      "loss": 0.1881,
+      "step": 7890
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0087271196280986e-05,
+      "loss": 0.204,
+      "step": 7891
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0085168327364037e-05,
+      "loss": 0.2077,
+      "step": 7892
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0083065454680616e-05,
+      "loss": 0.1983,
+      "step": 7893
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.008096257832372e-05,
+      "loss": 0.1841,
+      "step": 7894
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0078859698386348e-05,
+      "loss": 0.1938,
+      "step": 7895
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0076756814961491e-05,
+      "loss": 0.171,
+      "step": 7896
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0074653928142152e-05,
+      "loss": 0.1958,
+      "step": 7897
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.007255103802133e-05,
+      "loss": 0.188,
+      "step": 7898
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0070448144692022e-05,
+      "loss": 0.1521,
+      "step": 7899
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0068345248247221e-05,
+      "loss": 0.1561,
+      "step": 7900
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.006624234877993e-05,
+      "loss": 0.1928,
+      "step": 7901
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0064139446383147e-05,
+      "loss": 0.2046,
+      "step": 7902
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0062036541149871e-05,
+      "loss": 0.1839,
+      "step": 7903
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0059933633173098e-05,
+      "loss": 0.191,
+      "step": 7904
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0057830722545827e-05,
+      "loss": 0.1906,
+      "step": 7905
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.005572780936106e-05,
+      "loss": 0.1702,
+      "step": 7906
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0053624893711793e-05,
+      "loss": 0.2089,
+      "step": 7907
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0051521975691026e-05,
+      "loss": 0.1854,
+      "step": 7908
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.004941905539176e-05,
+      "loss": 0.1988,
+      "step": 7909
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.004731613290699e-05,
+      "loss": 0.2003,
+      "step": 7910
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0045213208329719e-05,
+      "loss": 0.1603,
+      "step": 7911
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0043110281752943e-05,
+      "loss": 0.1993,
+      "step": 7912
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0041007353269666e-05,
+      "loss": 0.1759,
+      "step": 7913
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0038904422972886e-05,
+      "loss": 0.2057,
+      "step": 7914
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0036801490955598e-05,
+      "loss": 0.1844,
+      "step": 7915
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0034698557310812e-05,
+      "loss": 0.1657,
+      "step": 7916
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0032595622131518e-05,
+      "loss": 0.2167,
+      "step": 7917
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.003049268551072e-05,
+      "loss": 0.2009,
+      "step": 7918
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0028389747541418e-05,
+      "loss": 0.1847,
+      "step": 7919
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.002628680831661e-05,
+      "loss": 0.187,
+      "step": 7920
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.00241838679293e-05,
+      "loss": 0.171,
+      "step": 7921
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0022080926472483e-05,
+      "loss": 0.2039,
+      "step": 7922
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0019977984039167e-05,
+      "loss": 0.1703,
+      "step": 7923
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0017875040722342e-05,
+      "loss": 0.2314,
+      "step": 7924
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0015772096615017e-05,
+      "loss": 0.1892,
+      "step": 7925
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.001366915181019e-05,
+      "loss": 0.1736,
+      "step": 7926
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0011566206400858e-05,
+      "loss": 0.1983,
+      "step": 7927
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0009463260480023e-05,
+      "loss": 0.1532,
+      "step": 7928
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0007360314140685e-05,
+      "loss": 0.1967,
+      "step": 7929
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0005257367475848e-05,
+      "loss": 0.2109,
+      "step": 7930
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0003154420578511e-05,
+      "loss": 0.2029,
+      "step": 7931
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0001051473541671e-05,
+      "loss": 0.1905,
+      "step": 7932
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.998948526458332e-06,
+      "loss": 0.176,
+      "step": 7933
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.996845579421492e-06,
+      "loss": 0.2311,
+      "step": 7934
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.994742632524157e-06,
+      "loss": 0.1804,
+      "step": 7935
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.992639685859316e-06,
+      "loss": 0.2039,
+      "step": 7936
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.990536739519979e-06,
+      "loss": 0.1632,
+      "step": 7937
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.988433793599147e-06,
+      "loss": 0.1875,
+      "step": 7938
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.986330848189813e-06,
+      "loss": 0.1817,
+      "step": 7939
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.984227903384986e-06,
+      "loss": 0.2027,
+      "step": 7940
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.982124959277658e-06,
+      "loss": 0.1709,
+      "step": 7941
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.980022015960834e-06,
+      "loss": 0.1839,
+      "step": 7942
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.977919073527518e-06,
+      "loss": 0.2068,
+      "step": 7943
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.975816132070701e-06,
+      "loss": 0.1837,
+      "step": 7944
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.973713191683393e-06,
+      "loss": 0.1859,
+      "step": 7945
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.971610252458585e-06,
+      "loss": 0.195,
+      "step": 7946
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.969507314489284e-06,
+      "loss": 0.1611,
+      "step": 7947
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.967404377868485e-06,
+      "loss": 0.1952,
+      "step": 7948
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.965301442689193e-06,
+      "loss": 0.1747,
+      "step": 7949
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.963198509044402e-06,
+      "loss": 0.1839,
+      "step": 7950
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.961095577027118e-06,
+      "loss": 0.1641,
+      "step": 7951
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.958992646730339e-06,
+      "loss": 0.1857,
+      "step": 7952
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.956889718247059e-06,
+      "loss": 0.1867,
+      "step": 7953
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.954786791670286e-06,
+      "loss": 0.1945,
+      "step": 7954
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.952683867093014e-06,
+      "loss": 0.1715,
+      "step": 7955
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.950580944608242e-06,
+      "loss": 0.1822,
+      "step": 7956
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.948478024308976e-06,
+      "loss": 0.1985,
+      "step": 7957
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.946375106288207e-06,
+      "loss": 0.1729,
+      "step": 7958
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.944272190638943e-06,
+      "loss": 0.1835,
+      "step": 7959
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.942169277454175e-06,
+      "loss": 0.2037,
+      "step": 7960
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.940066366826903e-06,
+      "loss": 0.1985,
+      "step": 7961
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.937963458850132e-06,
+      "loss": 0.2049,
+      "step": 7962
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.935860553616856e-06,
+      "loss": 0.1954,
+      "step": 7963
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.933757651220072e-06,
+      "loss": 0.1989,
+      "step": 7964
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.93165475175278e-06,
+      "loss": 0.1752,
+      "step": 7965
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.929551855307983e-06,
+      "loss": 0.2111,
+      "step": 7966
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.927448961978671e-06,
+      "loss": 0.186,
+      "step": 7967
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.925346071857849e-06,
+      "loss": 0.1884,
+      "step": 7968
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.92324318503851e-06,
+      "loss": 0.157,
+      "step": 7969
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.921140301613654e-06,
+      "loss": 0.1604,
+      "step": 7970
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.919037421676283e-06,
+      "loss": 0.1693,
+      "step": 7971
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.91693454531939e-06,
+      "loss": 0.1863,
+      "step": 7972
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.914831672635965e-06,
+      "loss": 0.1686,
+      "step": 7973
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.912728803719017e-06,
+      "loss": 0.1901,
+      "step": 7974
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.910625938661537e-06,
+      "loss": 0.1899,
+      "step": 7975
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.908523077556525e-06,
+      "loss": 0.1831,
+      "step": 7976
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.906420220496978e-06,
+      "loss": 0.1799,
+      "step": 7977
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.904317367575886e-06,
+      "loss": 0.1696,
+      "step": 7978
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.902214518886252e-06,
+      "loss": 0.1986,
+      "step": 7979
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.900111674521073e-06,
+      "loss": 0.2439,
+      "step": 7980
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.898008834573336e-06,
+      "loss": 0.1767,
+      "step": 7981
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.895905999136048e-06,
+      "loss": 0.1934,
+      "step": 7982
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.893803168302196e-06,
+      "loss": 0.1704,
+      "step": 7983
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.891700342164775e-06,
+      "loss": 0.1834,
+      "step": 7984
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.889597520816788e-06,
+      "loss": 0.168,
+      "step": 7985
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.887494704351226e-06,
+      "loss": 0.1909,
+      "step": 7986
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.88539189286108e-06,
+      "loss": 0.2011,
+      "step": 7987
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.883289086439347e-06,
+      "loss": 0.1649,
+      "step": 7988
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.881186285179025e-06,
+      "loss": 0.222,
+      "step": 7989
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.879083489173101e-06,
+      "loss": 0.198,
+      "step": 7990
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.876980698514576e-06,
+      "loss": 0.2087,
+      "step": 7991
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.874877913296434e-06,
+      "loss": 0.1652,
+      "step": 7992
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.872775133611678e-06,
+      "loss": 0.2036,
+      "step": 7993
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.8706723595533e-06,
+      "loss": 0.1696,
+      "step": 7994
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.868569591214283e-06,
+      "loss": 0.2201,
+      "step": 7995
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.866466828687635e-06,
+      "loss": 0.1622,
+      "step": 7996
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.864364072066335e-06,
+      "loss": 0.2138,
+      "step": 7997
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.862261321443378e-06,
+      "loss": 0.2001,
+      "step": 7998
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.860158576911763e-06,
+      "loss": 0.1947,
+      "step": 7999
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.858055838564474e-06,
+      "loss": 0.1704,
+      "step": 8000
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.855953106494503e-06,
+      "loss": 0.2156,
+      "step": 8001
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.853850380794841e-06,
+      "loss": 0.1872,
+      "step": 8002
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.851747661558484e-06,
+      "loss": 0.176,
+      "step": 8003
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.849644948878414e-06,
+      "loss": 0.151,
+      "step": 8004
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.847542242847629e-06,
+      "loss": 0.1914,
+      "step": 8005
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.845439543559114e-06,
+      "loss": 0.1962,
+      "step": 8006
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.843336851105857e-06,
+      "loss": 0.1934,
+      "step": 8007
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.841234165580855e-06,
+      "loss": 0.1972,
+      "step": 8008
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.839131487077086e-06,
+      "loss": 0.1715,
+      "step": 8009
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.837028815687549e-06,
+      "loss": 0.1858,
+      "step": 8010
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.834926151505227e-06,
+      "loss": 0.2002,
+      "step": 8011
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.832823494623104e-06,
+      "loss": 0.1844,
+      "step": 8012
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.830720845134178e-06,
+      "loss": 0.1837,
+      "step": 8013
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.828618203131428e-06,
+      "loss": 0.206,
+      "step": 8014
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.826515568707842e-06,
+      "loss": 0.1703,
+      "step": 8015
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.82441294195641e-06,
+      "loss": 0.1646,
+      "step": 8016
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.822310322970117e-06,
+      "loss": 0.2275,
+      "step": 8017
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.820207711841944e-06,
+      "loss": 0.2085,
+      "step": 8018
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.818105108664885e-06,
+      "loss": 0.1729,
+      "step": 8019
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.81600251353192e-06,
+      "loss": 0.2164,
+      "step": 8020
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.81389992653603e-06,
+      "loss": 0.1856,
+      "step": 8021
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.811797347770212e-06,
+      "loss": 0.1845,
+      "step": 8022
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.80969477732744e-06,
+      "loss": 0.1721,
+      "step": 8023
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.807592215300697e-06,
+      "loss": 0.212,
+      "step": 8024
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.805489661782972e-06,
+      "loss": 0.1727,
+      "step": 8025
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.803387116867243e-06,
+      "loss": 0.1689,
+      "step": 8026
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.801284580646501e-06,
+      "loss": 0.1781,
+      "step": 8027
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.799182053213723e-06,
+      "loss": 0.1796,
+      "step": 8028
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.797079534661885e-06,
+      "loss": 0.2014,
+      "step": 8029
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.794977025083977e-06,
+      "loss": 0.2024,
+      "step": 8030
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.792874524572979e-06,
+      "loss": 0.2244,
+      "step": 8031
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.790772033221865e-06,
+      "loss": 0.172,
+      "step": 8032
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.788669551123625e-06,
+      "loss": 0.2127,
+      "step": 8033
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.786567078371234e-06,
+      "loss": 0.1927,
+      "step": 8034
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.784464615057668e-06,
+      "loss": 0.1602,
+      "step": 8035
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.782362161275914e-06,
+      "loss": 0.192,
+      "step": 8036
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.780259717118945e-06,
+      "loss": 0.194,
+      "step": 8037
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.778157282679738e-06,
+      "loss": 0.1838,
+      "step": 8038
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.776054858051274e-06,
+      "loss": 0.1991,
+      "step": 8039
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.773952443326532e-06,
+      "loss": 0.1921,
+      "step": 8040
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.771850038598482e-06,
+      "loss": 0.1924,
+      "step": 8041
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.769747643960111e-06,
+      "loss": 0.1815,
+      "step": 8042
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.767645259504382e-06,
+      "loss": 0.1964,
+      "step": 8043
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.765542885324281e-06,
+      "loss": 0.1608,
+      "step": 8044
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.763440521512783e-06,
+      "loss": 0.1947,
+      "step": 8045
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.761338168162854e-06,
+      "loss": 0.1911,
+      "step": 8046
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.759235825367476e-06,
+      "loss": 0.2131,
+      "step": 8047
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.75713349321962e-06,
+      "loss": 0.1779,
+      "step": 8048
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.755031171812257e-06,
+      "loss": 0.2104,
+      "step": 8049
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.752928861238367e-06,
+      "loss": 0.1786,
+      "step": 8050
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.750826561590917e-06,
+      "loss": 0.19,
+      "step": 8051
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.748724272962875e-06,
+      "loss": 0.2154,
+      "step": 8052
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.74662199544722e-06,
+      "loss": 0.2052,
+      "step": 8053
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.744519729136918e-06,
+      "loss": 0.1963,
+      "step": 8054
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.74241747412494e-06,
+      "loss": 0.1968,
+      "step": 8055
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.74031523050426e-06,
+      "loss": 0.2114,
+      "step": 8056
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.738212998367842e-06,
+      "loss": 0.1945,
+      "step": 8057
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.736110777808653e-06,
+      "loss": 0.1852,
+      "step": 8058
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.734008568919673e-06,
+      "loss": 0.2111,
+      "step": 8059
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.731906371793856e-06,
+      "loss": 0.1857,
+      "step": 8060
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.729804186524179e-06,
+      "loss": 0.1927,
+      "step": 8061
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.727702013203603e-06,
+      "loss": 0.1788,
+      "step": 8062
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.725599851925094e-06,
+      "loss": 0.178,
+      "step": 8063
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.723497702781623e-06,
+      "loss": 0.1997,
+      "step": 8064
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.721395565866152e-06,
+      "loss": 0.1785,
+      "step": 8065
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.719293441271646e-06,
+      "loss": 0.1973,
+      "step": 8066
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.717191329091066e-06,
+      "loss": 0.203,
+      "step": 8067
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.715089229417384e-06,
+      "loss": 0.163,
+      "step": 8068
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.71298714234355e-06,
+      "loss": 0.1862,
+      "step": 8069
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.710885067962538e-06,
+      "loss": 0.1893,
+      "step": 8070
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.708783006367307e-06,
+      "loss": 0.1789,
+      "step": 8071
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.706680957650812e-06,
+      "loss": 0.2013,
+      "step": 8072
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.704578921906023e-06,
+      "loss": 0.1793,
+      "step": 8073
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.702476899225891e-06,
+      "loss": 0.1954,
+      "step": 8074
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.700374889703384e-06,
+      "loss": 0.2005,
+      "step": 8075
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.698272893431457e-06,
+      "loss": 0.1745,
+      "step": 8076
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.696170910503067e-06,
+      "loss": 0.1808,
+      "step": 8077
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.694068941011175e-06,
+      "loss": 0.1567,
+      "step": 8078
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.691966985048736e-06,
+      "loss": 0.1707,
+      "step": 8079
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.689865042708708e-06,
+      "loss": 0.1776,
+      "step": 8080
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.687763114084045e-06,
+      "loss": 0.1908,
+      "step": 8081
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.685661199267707e-06,
+      "loss": 0.1633,
+      "step": 8082
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.68355929835264e-06,
+      "loss": 0.1752,
+      "step": 8083
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.681457411431807e-06,
+      "loss": 0.1783,
+      "step": 8084
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.679355538598159e-06,
+      "loss": 0.1831,
+      "step": 8085
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.677253679944644e-06,
+      "loss": 0.1886,
+      "step": 8086
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.675151835564224e-06,
+      "loss": 0.2067,
+      "step": 8087
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.673050005549842e-06,
+      "loss": 0.2093,
+      "step": 8088
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.67094818999445e-06,
+      "loss": 0.1886,
+      "step": 8089
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.668846388991006e-06,
+      "loss": 0.1669,
+      "step": 8090
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.666744602632448e-06,
+      "loss": 0.1717,
+      "step": 8091
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.664642831011738e-06,
+      "loss": 0.1776,
+      "step": 8092
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.662541074221816e-06,
+      "loss": 0.1935,
+      "step": 8093
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.660439332355628e-06,
+      "loss": 0.2041,
+      "step": 8094
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.658337605506129e-06,
+      "loss": 0.2113,
+      "step": 8095
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.656235893766262e-06,
+      "loss": 0.1817,
+      "step": 8096
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.654134197228968e-06,
+      "loss": 0.1917,
+      "step": 8097
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.652032515987199e-06,
+      "loss": 0.188,
+      "step": 8098
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.649930850133899e-06,
+      "loss": 0.1721,
+      "step": 8099
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.647829199762004e-06,
+      "loss": 0.2362,
+      "step": 8100
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.645727564964468e-06,
+      "loss": 0.2302,
+      "step": 8101
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.643625945834227e-06,
+      "loss": 0.2035,
+      "step": 8102
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.64152434246422e-06,
+      "loss": 0.1806,
+      "step": 8103
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.639422754947396e-06,
+      "loss": 0.1754,
+      "step": 8104
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.637321183376693e-06,
+      "loss": 0.1726,
+      "step": 8105
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.635219627845041e-06,
+      "loss": 0.2097,
+      "step": 8106
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.633118088445392e-06,
+      "loss": 0.1892,
+      "step": 8107
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.631016565270675e-06,
+      "loss": 0.1753,
+      "step": 8108
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.628915058413836e-06,
+      "loss": 0.1838,
+      "step": 8109
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.626813567967808e-06,
+      "loss": 0.1983,
+      "step": 8110
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.624712094025522e-06,
+      "loss": 0.1738,
+      "step": 8111
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.62261063667992e-06,
+      "loss": 0.1767,
+      "step": 8112
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.620509196023934e-06,
+      "loss": 0.2043,
+      "step": 8113
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.618407772150495e-06,
+      "loss": 0.1749,
+      "step": 8114
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.616306365152544e-06,
+      "loss": 0.1828,
+      "step": 8115
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.614204975123005e-06,
+      "loss": 0.1875,
+      "step": 8116
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.612103602154811e-06,
+      "loss": 0.1806,
+      "step": 8117
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.610002246340898e-06,
+      "loss": 0.2105,
+      "step": 8118
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.607900907774193e-06,
+      "loss": 0.2035,
+      "step": 8119
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.605799586547622e-06,
+      "loss": 0.1727,
+      "step": 8120
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.603698282754118e-06,
+      "loss": 0.1686,
+      "step": 8121
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.601596996486609e-06,
+      "loss": 0.1861,
+      "step": 8122
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.599495727838017e-06,
+      "loss": 0.1622,
+      "step": 8123
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.597394476901277e-06,
+      "loss": 0.1827,
+      "step": 8124
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.595293243769303e-06,
+      "loss": 0.1934,
+      "step": 8125
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.593192028535029e-06,
+      "loss": 0.1903,
+      "step": 8126
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.591090831291375e-06,
+      "loss": 0.1809,
+      "step": 8127
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.588989652131263e-06,
+      "loss": 0.1624,
+      "step": 8128
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.58688849114762e-06,
+      "loss": 0.1886,
+      "step": 8129
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.584787348433362e-06,
+      "loss": 0.1855,
+      "step": 8130
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.582686224081411e-06,
+      "loss": 0.1883,
+      "step": 8131
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.580585118184691e-06,
+      "loss": 0.1875,
+      "step": 8132
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.578484030836119e-06,
+      "loss": 0.2134,
+      "step": 8133
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.576382962128607e-06,
+      "loss": 0.1926,
+      "step": 8134
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.574281912155079e-06,
+      "loss": 0.1919,
+      "step": 8135
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.572180881008453e-06,
+      "loss": 0.2062,
+      "step": 8136
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.570079868781637e-06,
+      "loss": 0.1825,
+      "step": 8137
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.567978875567555e-06,
+      "loss": 0.1878,
+      "step": 8138
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.565877901459115e-06,
+      "loss": 0.2127,
+      "step": 8139
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.56377694654923e-06,
+      "loss": 0.1871,
+      "step": 8140
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.561676010930816e-06,
+      "loss": 0.2358,
+      "step": 8141
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.55957509469678e-06,
+      "loss": 0.185,
+      "step": 8142
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.55747419794004e-06,
+      "loss": 0.156,
+      "step": 8143
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.555373320753497e-06,
+      "loss": 0.1961,
+      "step": 8144
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.553272463230065e-06,
+      "loss": 0.1686,
+      "step": 8145
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.551171625462654e-06,
+      "loss": 0.1721,
+      "step": 8146
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.549070807544166e-06,
+      "loss": 0.1727,
+      "step": 8147
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.546970009567506e-06,
+      "loss": 0.1799,
+      "step": 8148
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.544869231625587e-06,
+      "loss": 0.1773,
+      "step": 8149
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.54276847381131e-06,
+      "loss": 0.2011,
+      "step": 8150
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.540667736217573e-06,
+      "loss": 0.1979,
+      "step": 8151
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.53856701893729e-06,
+      "loss": 0.1815,
+      "step": 8152
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.536466322063354e-06,
+      "loss": 0.1996,
+      "step": 8153
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.534365645688666e-06,
+      "loss": 0.1923,
+      "step": 8154
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.532264989906131e-06,
+      "loss": 0.1917,
+      "step": 8155
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.530164354808648e-06,
+      "loss": 0.1961,
+      "step": 8156
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.528063740489108e-06,
+      "loss": 0.1786,
+      "step": 8157
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.525963147040414e-06,
+      "loss": 0.1827,
+      "step": 8158
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.523862574555462e-06,
+      "loss": 0.1846,
+      "step": 8159
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.521762023127149e-06,
+      "loss": 0.2121,
+      "step": 8160
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.519661492848368e-06,
+      "loss": 0.1854,
+      "step": 8161
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.517560983812008e-06,
+      "loss": 0.1721,
+      "step": 8162
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.515460496110966e-06,
+      "loss": 0.1982,
+      "step": 8163
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.513360029838138e-06,
+      "loss": 0.1999,
+      "step": 8164
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.511259585086403e-06,
+      "loss": 0.1587,
+      "step": 8165
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.509159161948664e-06,
+      "loss": 0.1949,
+      "step": 8166
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.507058760517803e-06,
+      "loss": 0.1733,
+      "step": 8167
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.504958380886704e-06,
+      "loss": 0.1872,
+      "step": 8168
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.502858023148262e-06,
+      "loss": 0.196,
+      "step": 8169
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.500757687395361e-06,
+      "loss": 0.1598,
+      "step": 8170
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.49865737372088e-06,
+      "loss": 0.1905,
+      "step": 8171
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.49655708221771e-06,
+      "loss": 0.1717,
+      "step": 8172
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.494456812978731e-06,
+      "loss": 0.1846,
+      "step": 8173
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.492356566096824e-06,
+      "loss": 0.1614,
+      "step": 8174
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.490256341664874e-06,
+      "loss": 0.1986,
+      "step": 8175
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.488156139775754e-06,
+      "loss": 0.2144,
+      "step": 8176
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.48605596052235e-06,
+      "loss": 0.1761,
+      "step": 8177
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.48395580399754e-06,
+      "loss": 0.2004,
+      "step": 8178
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.481855670294193e-06,
+      "loss": 0.1704,
+      "step": 8179
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.479755559505196e-06,
+      "loss": 0.2186,
+      "step": 8180
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.477655471723415e-06,
+      "loss": 0.2083,
+      "step": 8181
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.475555407041728e-06,
+      "loss": 0.1978,
+      "step": 8182
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.47345536555301e-06,
+      "loss": 0.202,
+      "step": 8183
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.47135534735013e-06,
+      "loss": 0.1917,
+      "step": 8184
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.469255352525957e-06,
+      "loss": 0.1882,
+      "step": 8185
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.467155381173364e-06,
+      "loss": 0.2032,
+      "step": 8186
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.465055433385221e-06,
+      "loss": 0.2044,
+      "step": 8187
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.46295550925439e-06,
+      "loss": 0.1656,
+      "step": 8188
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.460855608873747e-06,
+      "loss": 0.2079,
+      "step": 8189
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.458755732336148e-06,
+      "loss": 0.212,
+      "step": 8190
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.456655879734462e-06,
+      "loss": 0.2066,
+      "step": 8191
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.454556051161558e-06,
+      "loss": 0.1729,
+      "step": 8192
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.452456246710287e-06,
+      "loss": 0.2012,
+      "step": 8193
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.45035646647352e-06,
+      "loss": 0.1744,
+      "step": 8194
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.448256710544115e-06,
+      "loss": 0.1814,
+      "step": 8195
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.446156979014925e-06,
+      "loss": 0.2076,
+      "step": 8196
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.44405727197882e-06,
+      "loss": 0.1739,
+      "step": 8197
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.441957589528649e-06,
+      "loss": 0.2273,
+      "step": 8198
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.439857931757269e-06,
+      "loss": 0.1918,
+      "step": 8199
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.437758298757536e-06,
+      "loss": 0.1935,
+      "step": 8200
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.435658690622305e-06,
+      "loss": 0.2004,
+      "step": 8201
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.433559107444424e-06,
+      "loss": 0.2009,
+      "step": 8202
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.43145954931675e-06,
+      "loss": 0.1611,
+      "step": 8203
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.429360016332134e-06,
+      "loss": 0.1743,
+      "step": 8204
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.42726050858342e-06,
+      "loss": 0.1812,
+      "step": 8205
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.425161026163461e-06,
+      "loss": 0.1901,
+      "step": 8206
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.423061569165101e-06,
+      "loss": 0.2023,
+      "step": 8207
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.42096213768119e-06,
+      "loss": 0.1854,
+      "step": 8208
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.418862731804572e-06,
+      "loss": 0.1888,
+      "step": 8209
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.416763351628086e-06,
+      "loss": 0.1718,
+      "step": 8210
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.414663997244583e-06,
+      "loss": 0.2059,
+      "step": 8211
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.412564668746898e-06,
+      "loss": 0.2304,
+      "step": 8212
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.410465366227873e-06,
+      "loss": 0.1579,
+      "step": 8213
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.408366089780347e-06,
+      "loss": 0.1717,
+      "step": 8214
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.406266839497163e-06,
+      "loss": 0.1852,
+      "step": 8215
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.404167615471146e-06,
+      "loss": 0.1782,
+      "step": 8216
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.402068417795146e-06,
+      "loss": 0.2,
+      "step": 8217
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.399969246561989e-06,
+      "loss": 0.1889,
+      "step": 8218
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.397870101864509e-06,
+      "loss": 0.1758,
+      "step": 8219
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.395770983795543e-06,
+      "loss": 0.193,
+      "step": 8220
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.393671892447918e-06,
+      "loss": 0.2051,
+      "step": 8221
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.391572827914462e-06,
+      "loss": 0.1821,
+      "step": 8222
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.389473790288011e-06,
+      "loss": 0.1851,
+      "step": 8223
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.387374779661386e-06,
+      "loss": 0.1537,
+      "step": 8224
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.385275796127417e-06,
+      "loss": 0.1991,
+      "step": 8225
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.383176839778928e-06,
+      "loss": 0.1932,
+      "step": 8226
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.381077910708738e-06,
+      "loss": 0.2072,
+      "step": 8227
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.378979009009681e-06,
+      "loss": 0.1653,
+      "step": 8228
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.376880134774572e-06,
+      "loss": 0.1754,
+      "step": 8229
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.374781288096226e-06,
+      "loss": 0.1849,
+      "step": 8230
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.372682469067471e-06,
+      "loss": 0.1808,
+      "step": 8231
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.370583677781122e-06,
+      "loss": 0.1854,
+      "step": 8232
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.368484914329992e-06,
+      "loss": 0.1659,
+      "step": 8233
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.366386178806902e-06,
+      "loss": 0.1711,
+      "step": 8234
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.364287471304665e-06,
+      "loss": 0.2011,
+      "step": 8235
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.362188791916089e-06,
+      "loss": 0.2021,
+      "step": 8236
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.360090140733992e-06,
+      "loss": 0.1519,
+      "step": 8237
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.357991517851183e-06,
+      "loss": 0.1837,
+      "step": 8238
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.355892923360466e-06,
+      "loss": 0.1865,
+      "step": 8239
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.353794357354654e-06,
+      "loss": 0.1754,
+      "step": 8240
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.35169581992655e-06,
+      "loss": 0.2046,
+      "step": 8241
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.349597311168969e-06,
+      "loss": 0.1673,
+      "step": 8242
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.347498831174706e-06,
+      "loss": 0.1756,
+      "step": 8243
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.345400380036563e-06,
+      "loss": 0.2006,
+      "step": 8244
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.343301957847347e-06,
+      "loss": 0.1728,
+      "step": 8245
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.341203564699857e-06,
+      "loss": 0.1892,
+      "step": 8246
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.339105200686888e-06,
+      "loss": 0.1978,
+      "step": 8247
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.337006865901245e-06,
+      "loss": 0.1786,
+      "step": 8248
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.334908560435719e-06,
+      "loss": 0.2146,
+      "step": 8249
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.332810284383105e-06,
+      "loss": 0.1587,
+      "step": 8250
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.330712037836201e-06,
+      "loss": 0.1732,
+      "step": 8251
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.328613820887798e-06,
+      "loss": 0.1793,
+      "step": 8252
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.326515633630681e-06,
+      "loss": 0.2161,
+      "step": 8253
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.324417476157648e-06,
+      "loss": 0.1906,
+      "step": 8254
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.322319348561489e-06,
+      "loss": 0.2042,
+      "step": 8255
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.32022125093498e-06,
+      "loss": 0.1724,
+      "step": 8256
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.31812318337092e-06,
+      "loss": 0.2004,
+      "step": 8257
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.316025145962083e-06,
+      "loss": 0.1826,
+      "step": 8258
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.313927138801259e-06,
+      "loss": 0.1736,
+      "step": 8259
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.311829161981231e-06,
+      "loss": 0.1957,
+      "step": 8260
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.309731215594772e-06,
+      "loss": 0.1905,
+      "step": 8261
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.307633299734671e-06,
+      "loss": 0.1927,
+      "step": 8262
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.305535414493697e-06,
+      "loss": 0.1706,
+      "step": 8263
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.303437559964633e-06,
+      "loss": 0.1915,
+      "step": 8264
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.301339736240252e-06,
+      "loss": 0.1882,
+      "step": 8265
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.29924194341333e-06,
+      "loss": 0.2284,
+      "step": 8266
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.29714418157663e-06,
+      "loss": 0.187,
+      "step": 8267
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.295046450822935e-06,
+      "loss": 0.1736,
+      "step": 8268
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.292948751245012e-06,
+      "loss": 0.1774,
+      "step": 8269
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.290851082935622e-06,
+      "loss": 0.1682,
+      "step": 8270
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.288753445987542e-06,
+      "loss": 0.1688,
+      "step": 8271
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.286655840493533e-06,
+      "loss": 0.1784,
+      "step": 8272
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.284558266546355e-06,
+      "loss": 0.1888,
+      "step": 8273
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.282460724238781e-06,
+      "loss": 0.1983,
+      "step": 8274
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.280363213663562e-06,
+      "loss": 0.2016,
+      "step": 8275
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.278265734913466e-06,
+      "loss": 0.2038,
+      "step": 8276
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.276168288081247e-06,
+      "loss": 0.1979,
+      "step": 8277
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.27407087325966e-06,
+      "loss": 0.1712,
+      "step": 8278
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.27197349054147e-06,
+      "loss": 0.1894,
+      "step": 8279
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.269876140019425e-06,
+      "loss": 0.217,
+      "step": 8280
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.267778821786274e-06,
+      "loss": 0.1853,
+      "step": 8281
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.265681535934776e-06,
+      "loss": 0.1947,
+      "step": 8282
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.263584282557679e-06,
+      "loss": 0.1908,
+      "step": 8283
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.261487061747728e-06,
+      "loss": 0.1758,
+      "step": 8284
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.259389873597676e-06,
+      "loss": 0.1977,
+      "step": 8285
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.257292718200264e-06,
+      "loss": 0.171,
+      "step": 8286
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.255195595648236e-06,
+      "loss": 0.2056,
+      "step": 8287
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.253098506034343e-06,
+      "loss": 0.1812,
+      "step": 8288
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.251001449451317e-06,
+      "loss": 0.1892,
+      "step": 8289
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.2489044259919e-06,
+      "loss": 0.2075,
+      "step": 8290
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.246807435748832e-06,
+      "loss": 0.1965,
+      "step": 8291
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.244710478814849e-06,
+      "loss": 0.2189,
+      "step": 8292
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.242613555282692e-06,
+      "loss": 0.1995,
+      "step": 8293
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.240516665245089e-06,
+      "loss": 0.1854,
+      "step": 8294
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.23841980879477e-06,
+      "loss": 0.1837,
+      "step": 8295
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.236322986024473e-06,
+      "loss": 0.1903,
+      "step": 8296
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.234226197026924e-06,
+      "loss": 0.1624,
+      "step": 8297
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.232129441894851e-06,
+      "loss": 0.2154,
+      "step": 8298
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.230032720720984e-06,
+      "loss": 0.2112,
+      "step": 8299
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.227936033598043e-06,
+      "loss": 0.1866,
+      "step": 8300
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.225839380618753e-06,
+      "loss": 0.2114,
+      "step": 8301
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.223742761875842e-06,
+      "loss": 0.1545,
+      "step": 8302
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.221646177462024e-06,
+      "loss": 0.1669,
+      "step": 8303
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.219549627470016e-06,
+      "loss": 0.1647,
+      "step": 8304
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.217453111992542e-06,
+      "loss": 0.1931,
+      "step": 8305
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.215356631122319e-06,
+      "loss": 0.212,
+      "step": 8306
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.21326018495205e-06,
+      "loss": 0.1599,
+      "step": 8307
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.211163773574463e-06,
+      "loss": 0.1805,
+      "step": 8308
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.209067397082257e-06,
+      "loss": 0.2107,
+      "step": 8309
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.20697105556815e-06,
+      "loss": 0.189,
+      "step": 8310
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.20487474912485e-06,
+      "loss": 0.1865,
+      "step": 8311
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.202778477845056e-06,
+      "loss": 0.2205,
+      "step": 8312
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.200682241821483e-06,
+      "loss": 0.1784,
+      "step": 8313
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.198586041146828e-06,
+      "loss": 0.1763,
+      "step": 8314
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.196489875913793e-06,
+      "loss": 0.1895,
+      "step": 8315
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.194393746215085e-06,
+      "loss": 0.1889,
+      "step": 8316
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.192297652143399e-06,
+      "loss": 0.2038,
+      "step": 8317
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.190201593791428e-06,
+      "loss": 0.1983,
+      "step": 8318
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.188105571251872e-06,
+      "loss": 0.2067,
+      "step": 8319
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.18600958461743e-06,
+      "loss": 0.2177,
+      "step": 8320
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.183913633980784e-06,
+      "loss": 0.2125,
+      "step": 8321
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.181817719434634e-06,
+      "loss": 0.1984,
+      "step": 8322
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.179721841071665e-06,
+      "loss": 0.1768,
+      "step": 8323
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.177625998984565e-06,
+      "loss": 0.1741,
+      "step": 8324
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.175530193266026e-06,
+      "loss": 0.1784,
+      "step": 8325
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.17343442400872e-06,
+      "loss": 0.1787,
+      "step": 8326
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.171338691305346e-06,
+      "loss": 0.1633,
+      "step": 8327
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.169242995248573e-06,
+      "loss": 0.1669,
+      "step": 8328
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.167147335931084e-06,
+      "loss": 0.2156,
+      "step": 8329
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.165051713445562e-06,
+      "loss": 0.188,
+      "step": 8330
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.162956127884678e-06,
+      "loss": 0.198,
+      "step": 8331
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.16086057934111e-06,
+      "loss": 0.1992,
+      "step": 8332
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.158765067907526e-06,
+      "loss": 0.1878,
+      "step": 8333
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.156669593676607e-06,
+      "loss": 0.1991,
+      "step": 8334
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.154574156741012e-06,
+      "loss": 0.1895,
+      "step": 8335
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.152478757193417e-06,
+      "loss": 0.1932,
+      "step": 8336
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.15038339512649e-06,
+      "loss": 0.1671,
+      "step": 8337
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.148288070632886e-06,
+      "loss": 0.2066,
+      "step": 8338
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.14619278380528e-06,
+      "loss": 0.1792,
+      "step": 8339
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.144097534736327e-06,
+      "loss": 0.1925,
+      "step": 8340
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.142002323518688e-06,
+      "loss": 0.1872,
+      "step": 8341
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.139907150245021e-06,
+      "loss": 0.1866,
+      "step": 8342
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.137812015007982e-06,
+      "loss": 0.1696,
+      "step": 8343
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.135716917900233e-06,
+      "loss": 0.2005,
+      "step": 8344
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.133621859014419e-06,
+      "loss": 0.1959,
+      "step": 8345
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.131526838443195e-06,
+      "loss": 0.1954,
+      "step": 8346
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.129431856279209e-06,
+      "loss": 0.1771,
+      "step": 8347
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.127336912615112e-06,
+      "loss": 0.1829,
+      "step": 8348
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.125242007543545e-06,
+      "loss": 0.1698,
+      "step": 8349
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.12314714115716e-06,
+      "loss": 0.17,
+      "step": 8350
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.1210523135486e-06,
+      "loss": 0.1849,
+      "step": 8351
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.118957524810497e-06,
+      "loss": 0.1819,
+      "step": 8352
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.116862775035503e-06,
+      "loss": 0.1992,
+      "step": 8353
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.114768064316245e-06,
+      "loss": 0.1688,
+      "step": 8354
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.112673392745364e-06,
+      "loss": 0.2047,
+      "step": 8355
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.110578760415499e-06,
+      "loss": 0.2032,
+      "step": 8356
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.108484167419273e-06,
+      "loss": 0.1877,
+      "step": 8357
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.106389613849325e-06,
+      "loss": 0.1736,
+      "step": 8358
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.104295099798282e-06,
+      "loss": 0.1939,
+      "step": 8359
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.102200625358767e-06,
+      "loss": 0.1876,
+      "step": 8360
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.100106190623417e-06,
+      "loss": 0.1914,
+      "step": 8361
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.098011795684846e-06,
+      "loss": 0.2033,
+      "step": 8362
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.095917440635676e-06,
+      "loss": 0.1807,
+      "step": 8363
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.093823125568532e-06,
+      "loss": 0.163,
+      "step": 8364
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.091728850576033e-06,
+      "loss": 0.2054,
+      "step": 8365
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.08963461575079e-06,
+      "loss": 0.1833,
+      "step": 8366
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.087540421185427e-06,
+      "loss": 0.1969,
+      "step": 8367
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.08544626697255e-06,
+      "loss": 0.1844,
+      "step": 8368
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.083352153204772e-06,
+      "loss": 0.1625,
+      "step": 8369
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.081258079974707e-06,
+      "loss": 0.1918,
+      "step": 8370
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.07916404737496e-06,
+      "loss": 0.1708,
+      "step": 8371
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.077070055498135e-06,
+      "loss": 0.1781,
+      "step": 8372
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.07497610443684e-06,
+      "loss": 0.1856,
+      "step": 8373
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.072882194283673e-06,
+      "loss": 0.2217,
+      "step": 8374
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.070788325131244e-06,
+      "loss": 0.1573,
+      "step": 8375
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.068694497072144e-06,
+      "loss": 0.1733,
+      "step": 8376
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.06660071019897e-06,
+      "loss": 0.1968,
+      "step": 8377
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.064506964604321e-06,
+      "loss": 0.1887,
+      "step": 8378
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.06241326038079e-06,
+      "loss": 0.1831,
+      "step": 8379
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.060319597620964e-06,
+      "loss": 0.1678,
+      "step": 8380
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.058225976417443e-06,
+      "loss": 0.1941,
+      "step": 8381
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.056132396862805e-06,
+      "loss": 0.1454,
+      "step": 8382
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.054038859049637e-06,
+      "loss": 0.1811,
+      "step": 8383
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.05194536307053e-06,
+      "loss": 0.1876,
+      "step": 8384
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.049851909018065e-06,
+      "loss": 0.1964,
+      "step": 8385
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.047758496984815e-06,
+      "loss": 0.1563,
+      "step": 8386
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.045665127063365e-06,
+      "loss": 0.1891,
+      "step": 8387
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.043571799346295e-06,
+      "loss": 0.188,
+      "step": 8388
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.041478513926171e-06,
+      "loss": 0.1961,
+      "step": 8389
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.039385270895576e-06,
+      "loss": 0.2012,
+      "step": 8390
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.037292070347071e-06,
+      "loss": 0.1902,
+      "step": 8391
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.035198912373234e-06,
+      "loss": 0.1788,
+      "step": 8392
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.03310579706663e-06,
+      "loss": 0.1799,
+      "step": 8393
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.031012724519821e-06,
+      "loss": 0.2078,
+      "step": 8394
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.028919694825378e-06,
+      "loss": 0.1685,
+      "step": 8395
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.026826708075858e-06,
+      "loss": 0.2112,
+      "step": 8396
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.024733764363818e-06,
+      "loss": 0.1961,
+      "step": 8397
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.022640863781823e-06,
+      "loss": 0.1928,
+      "step": 8398
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.020548006422426e-06,
+      "loss": 0.1825,
+      "step": 8399
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.01845519237818e-06,
+      "loss": 0.1715,
+      "step": 8400
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.016362421741638e-06,
+      "loss": 0.1888,
+      "step": 8401
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.014269694605353e-06,
+      "loss": 0.1978,
+      "step": 8402
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.012177011061866e-06,
+      "loss": 0.1974,
+      "step": 8403
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.010084371203736e-06,
+      "loss": 0.2026,
+      "step": 8404
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.007991775123497e-06,
+      "loss": 0.1651,
+      "step": 8405
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.005899222913691e-06,
+      "loss": 0.1584,
+      "step": 8406
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.003806714666871e-06,
+      "loss": 0.1909,
+      "step": 8407
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.001714250475561e-06,
+      "loss": 0.2047,
+      "step": 8408
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.99962183043231e-06,
+      "loss": 0.1712,
+      "step": 8409
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.997529454629645e-06,
+      "loss": 0.2,
+      "step": 8410
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.995437123160098e-06,
+      "loss": 0.1709,
+      "step": 8411
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.993344836116208e-06,
+      "loss": 0.1889,
+      "step": 8412
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.991252593590501e-06,
+      "loss": 0.1617,
+      "step": 8413
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.989160395675498e-06,
+      "loss": 0.2032,
+      "step": 8414
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.987068242463728e-06,
+      "loss": 0.2049,
+      "step": 8415
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.98497613404772e-06,
+      "loss": 0.1731,
+      "step": 8416
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.982884070519983e-06,
+      "loss": 0.1728,
+      "step": 8417
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.980792051973049e-06,
+      "loss": 0.2259,
+      "step": 8418
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.978700078499426e-06,
+      "loss": 0.1922,
+      "step": 8419
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.976608150191632e-06,
+      "loss": 0.1835,
+      "step": 8420
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.974516267142183e-06,
+      "loss": 0.2042,
+      "step": 8421
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.972424429443586e-06,
+      "loss": 0.2024,
+      "step": 8422
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.97033263718835e-06,
+      "loss": 0.2017,
+      "step": 8423
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.968240890468984e-06,
+      "loss": 0.2143,
+      "step": 8424
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.966149189377991e-06,
+      "loss": 0.2086,
+      "step": 8425
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.964057534007881e-06,
+      "loss": 0.1678,
+      "step": 8426
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.961965924451151e-06,
+      "loss": 0.1983,
+      "step": 8427
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.959874360800293e-06,
+      "loss": 0.2052,
+      "step": 8428
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.957782843147814e-06,
+      "loss": 0.1864,
+      "step": 8429
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.955691371586207e-06,
+      "loss": 0.1857,
+      "step": 8430
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.953599946207958e-06,
+      "loss": 0.2077,
+      "step": 8431
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.951508567105567e-06,
+      "loss": 0.182,
+      "step": 8432
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.949417234371519e-06,
+      "loss": 0.193,
+      "step": 8433
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.947325948098296e-06,
+      "loss": 0.1898,
+      "step": 8434
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.945234708378392e-06,
+      "loss": 0.2012,
+      "step": 8435
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.943143515304286e-06,
+      "loss": 0.2158,
+      "step": 8436
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.941052368968453e-06,
+      "loss": 0.2097,
+      "step": 8437
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.93896126946338e-06,
+      "loss": 0.171,
+      "step": 8438
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.93687021688154e-06,
+      "loss": 0.1813,
+      "step": 8439
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.934779211315403e-06,
+      "loss": 0.182,
+      "step": 8440
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.932688252857452e-06,
+      "loss": 0.1942,
+      "step": 8441
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.930597341600145e-06,
+      "loss": 0.174,
+      "step": 8442
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.92850647763596e-06,
+      "loss": 0.2187,
+      "step": 8443
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.92641566105736e-06,
+      "loss": 0.2381,
+      "step": 8444
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.924324891956804e-06,
+      "loss": 0.1826,
+      "step": 8445
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.922234170426763e-06,
+      "loss": 0.1838,
+      "step": 8446
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.92014349655969e-06,
+      "loss": 0.179,
+      "step": 8447
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.91805287044804e-06,
+      "loss": 0.1843,
+      "step": 8448
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.91596229218428e-06,
+      "loss": 0.1675,
+      "step": 8449
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.913871761860856e-06,
+      "loss": 0.1927,
+      "step": 8450
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.911781279570214e-06,
+      "loss": 0.1737,
+      "step": 8451
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.909690845404812e-06,
+      "loss": 0.2178,
+      "step": 8452
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.907600459457097e-06,
+      "loss": 0.1739,
+      "step": 8453
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.905510121819506e-06,
+      "loss": 0.1811,
+      "step": 8454
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.903419832584492e-06,
+      "loss": 0.2066,
+      "step": 8455
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.901329591844487e-06,
+      "loss": 0.1795,
+      "step": 8456
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.899239399691932e-06,
+      "loss": 0.1816,
+      "step": 8457
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.89714925621927e-06,
+      "loss": 0.1875,
+      "step": 8458
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.895059161518922e-06,
+      "loss": 0.1974,
+      "step": 8459
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.892969115683335e-06,
+      "loss": 0.2314,
+      "step": 8460
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.89087911880493e-06,
+      "loss": 0.1897,
+      "step": 8461
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.888789170976132e-06,
+      "loss": 0.1878,
+      "step": 8462
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.886699272289376e-06,
+      "loss": 0.1791,
+      "step": 8463
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.884609422837078e-06,
+      "loss": 0.2079,
+      "step": 8464
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.882519622711662e-06,
+      "loss": 0.1796,
+      "step": 8465
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.880429872005547e-06,
+      "loss": 0.2092,
+      "step": 8466
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.878340170811152e-06,
+      "loss": 0.2311,
+      "step": 8467
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.876250519220883e-06,
+      "loss": 0.1877,
+      "step": 8468
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.874160917327162e-06,
+      "loss": 0.1599,
+      "step": 8469
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.872071365222399e-06,
+      "loss": 0.1917,
+      "step": 8470
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.869981862998993e-06,
+      "loss": 0.2126,
+      "step": 8471
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.867892410749361e-06,
+      "loss": 0.2161,
+      "step": 8472
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.865803008565899e-06,
+      "loss": 0.1839,
+      "step": 8473
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.863713656541011e-06,
+      "loss": 0.1801,
+      "step": 8474
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.861624354767096e-06,
+      "loss": 0.2073,
+      "step": 8475
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.859535103336549e-06,
+      "loss": 0.1705,
+      "step": 8476
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.85744590234177e-06,
+      "loss": 0.1884,
+      "step": 8477
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.855356751875148e-06,
+      "loss": 0.2017,
+      "step": 8478
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.853267652029073e-06,
+      "loss": 0.2169,
+      "step": 8479
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.851178602895935e-06,
+      "loss": 0.1902,
+      "step": 8480
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.849089604568119e-06,
+      "loss": 0.2242,
+      "step": 8481
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.847000657138005e-06,
+      "loss": 0.2087,
+      "step": 8482
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.844911760697978e-06,
+      "loss": 0.2013,
+      "step": 8483
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.842822915340421e-06,
+      "loss": 0.1781,
+      "step": 8484
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.840734121157702e-06,
+      "loss": 0.177,
+      "step": 8485
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.838645378242204e-06,
+      "loss": 0.1875,
+      "step": 8486
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.836556686686292e-06,
+      "loss": 0.206,
+      "step": 8487
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.834468046582337e-06,
+      "loss": 0.151,
+      "step": 8488
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.832379458022713e-06,
+      "loss": 0.1756,
+      "step": 8489
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.830290921099778e-06,
+      "loss": 0.1989,
+      "step": 8490
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.828202435905904e-06,
+      "loss": 0.1865,
+      "step": 8491
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.826114002533443e-06,
+      "loss": 0.1717,
+      "step": 8492
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.824025621074756e-06,
+      "loss": 0.1915,
+      "step": 8493
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.821937291622204e-06,
+      "loss": 0.1947,
+      "step": 8494
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.819849014268138e-06,
+      "loss": 0.1914,
+      "step": 8495
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.817760789104904e-06,
+      "loss": 0.1881,
+      "step": 8496
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.81567261622486e-06,
+      "loss": 0.1811,
+      "step": 8497
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.813584495720352e-06,
+      "loss": 0.1647,
+      "step": 8498
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.811496427683716e-06,
+      "loss": 0.2001,
+      "step": 8499
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.809408412207306e-06,
+      "loss": 0.1882,
+      "step": 8500
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.807320449383455e-06,
+      "loss": 0.1751,
+      "step": 8501
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.8052325393045e-06,
+      "loss": 0.1987,
+      "step": 8502
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.803144682062782e-06,
+      "loss": 0.2046,
+      "step": 8503
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.801056877750633e-06,
+      "loss": 0.1973,
+      "step": 8504
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.798969126460378e-06,
+      "loss": 0.1788,
+      "step": 8505
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.796881428284351e-06,
+      "loss": 0.1697,
+      "step": 8506
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.794793783314875e-06,
+      "loss": 0.1894,
+      "step": 8507
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.792706191644278e-06,
+      "loss": 0.1701,
+      "step": 8508
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.790618653364879e-06,
+      "loss": 0.2227,
+      "step": 8509
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.788531168568993e-06,
+      "loss": 0.1984,
+      "step": 8510
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.786443737348944e-06,
+      "loss": 0.1768,
+      "step": 8511
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.784356359797041e-06,
+      "loss": 0.1826,
+      "step": 8512
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.782269036005596e-06,
+      "loss": 0.2037,
+      "step": 8513
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.780181766066924e-06,
+      "loss": 0.187,
+      "step": 8514
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.778094550073325e-06,
+      "loss": 0.1766,
+      "step": 8515
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.776007388117104e-06,
+      "loss": 0.1955,
+      "step": 8516
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.773920280290572e-06,
+      "loss": 0.1805,
+      "step": 8517
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.771833226686023e-06,
+      "loss": 0.2007,
+      "step": 8518
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.769746227395752e-06,
+      "loss": 0.1845,
+      "step": 8519
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.767659282512056e-06,
+      "loss": 0.1889,
+      "step": 8520
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.765572392127233e-06,
+      "loss": 0.1888,
+      "step": 8521
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.763485556333565e-06,
+      "loss": 0.1917,
+      "step": 8522
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.761398775223346e-06,
+      "loss": 0.144,
+      "step": 8523
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.759312048888857e-06,
+      "loss": 0.1875,
+      "step": 8524
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.757225377422386e-06,
+      "loss": 0.2042,
+      "step": 8525
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.755138760916211e-06,
+      "loss": 0.1842,
+      "step": 8526
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.75305219946261e-06,
+      "loss": 0.1854,
+      "step": 8527
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.750965693153862e-06,
+      "loss": 0.1898,
+      "step": 8528
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.748879242082236e-06,
+      "loss": 0.1956,
+      "step": 8529
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.74679284634e-06,
+      "loss": 0.1972,
+      "step": 8530
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.744706506019435e-06,
+      "loss": 0.1753,
+      "step": 8531
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.742620221212798e-06,
+      "loss": 0.1798,
+      "step": 8532
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.740533992012349e-06,
+      "loss": 0.2133,
+      "step": 8533
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.738447818510356e-06,
+      "loss": 0.191,
+      "step": 8534
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.736361700799078e-06,
+      "loss": 0.1966,
+      "step": 8535
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.734275638970764e-06,
+      "loss": 0.1835,
+      "step": 8536
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.732189633117677e-06,
+      "loss": 0.1635,
+      "step": 8537
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.730103683332063e-06,
+      "loss": 0.2095,
+      "step": 8538
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.728017789706167e-06,
+      "loss": 0.1937,
+      "step": 8539
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.725931952332245e-06,
+      "loss": 0.2204,
+      "step": 8540
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.72384617130253e-06,
+      "loss": 0.2114,
+      "step": 8541
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.721760446709277e-06,
+      "loss": 0.1489,
+      "step": 8542
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.719674778644711e-06,
+      "loss": 0.1824,
+      "step": 8543
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.717589167201072e-06,
+      "loss": 0.1856,
+      "step": 8544
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.715503612470601e-06,
+      "loss": 0.1864,
+      "step": 8545
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.713418114545524e-06,
+      "loss": 0.1724,
+      "step": 8546
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.711332673518065e-06,
+      "loss": 0.1905,
+      "step": 8547
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.709247289480457e-06,
+      "loss": 0.1914,
+      "step": 8548
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.707161962524924e-06,
+      "loss": 0.206,
+      "step": 8549
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.705076692743682e-06,
+      "loss": 0.1926,
+      "step": 8550
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.702991480228955e-06,
+      "loss": 0.1862,
+      "step": 8551
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.700906325072956e-06,
+      "loss": 0.1531,
+      "step": 8552
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.698821227367897e-06,
+      "loss": 0.1518,
+      "step": 8553
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.696736187205997e-06,
+      "loss": 0.193,
+      "step": 8554
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.694651204679456e-06,
+      "loss": 0.184,
+      "step": 8555
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.692566279880481e-06,
+      "loss": 0.2059,
+      "step": 8556
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.69048141290128e-06,
+      "loss": 0.1791,
+      "step": 8557
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.688396603834049e-06,
+      "loss": 0.1793,
+      "step": 8558
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.686311852770991e-06,
+      "loss": 0.2151,
+      "step": 8559
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.684227159804302e-06,
+      "loss": 0.2084,
+      "step": 8560
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.682142525026165e-06,
+      "loss": 0.2232,
+      "step": 8561
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.680057948528785e-06,
+      "loss": 0.1847,
+      "step": 8562
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.677973430404342e-06,
+      "loss": 0.1978,
+      "step": 8563
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.67588897074502e-06,
+      "loss": 0.1797,
+      "step": 8564
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.673804569643007e-06,
+      "loss": 0.2318,
+      "step": 8565
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.67172022719048e-06,
+      "loss": 0.1837,
+      "step": 8566
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.669635943479615e-06,
+      "loss": 0.1856,
+      "step": 8567
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.667551718602594e-06,
+      "loss": 0.1646,
+      "step": 8568
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.665467552651585e-06,
+      "loss": 0.2053,
+      "step": 8569
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.663383445718753e-06,
+      "loss": 0.2154,
+      "step": 8570
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.661299397896275e-06,
+      "loss": 0.1754,
+      "step": 8571
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.659215409276311e-06,
+      "loss": 0.1838,
+      "step": 8572
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.657131479951018e-06,
+      "loss": 0.193,
+      "step": 8573
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.655047610012564e-06,
+      "loss": 0.1863,
+      "step": 8574
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.6529637995531e-06,
+      "loss": 0.1776,
+      "step": 8575
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.650880048664783e-06,
+      "loss": 0.2009,
+      "step": 8576
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.648796357439765e-06,
+      "loss": 0.2021,
+      "step": 8577
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.646712725970192e-06,
+      "loss": 0.2039,
+      "step": 8578
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.644629154348213e-06,
+      "loss": 0.1891,
+      "step": 8579
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.642545642665971e-06,
+      "loss": 0.1907,
+      "step": 8580
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.640462191015603e-06,
+      "loss": 0.1773,
+      "step": 8581
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.638378799489256e-06,
+      "loss": 0.1985,
+      "step": 8582
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.63629546817906e-06,
+      "loss": 0.1969,
+      "step": 8583
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.634212197177141e-06,
+      "loss": 0.1872,
+      "step": 8584
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.632128986575642e-06,
+      "loss": 0.1674,
+      "step": 8585
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.630045836466686e-06,
+      "loss": 0.1715,
+      "step": 8586
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.627962746942393e-06,
+      "loss": 0.1899,
+      "step": 8587
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.625879718094894e-06,
+      "loss": 0.1762,
+      "step": 8588
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.623796750016302e-06,
+      "loss": 0.1732,
+      "step": 8589
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.621713842798733e-06,
+      "loss": 0.1953,
+      "step": 8590
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.619630996534308e-06,
+      "loss": 0.2075,
+      "step": 8591
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.617548211315132e-06,
+      "loss": 0.1799,
+      "step": 8592
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.61546548723332e-06,
+      "loss": 0.1531,
+      "step": 8593
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.61338282438097e-06,
+      "loss": 0.1699,
+      "step": 8594
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.61130022285019e-06,
+      "loss": 0.1898,
+      "step": 8595
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.609217682733084e-06,
+      "loss": 0.2244,
+      "step": 8596
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.607135204121745e-06,
+      "loss": 0.1714,
+      "step": 8597
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.60505278710827e-06,
+      "loss": 0.1671,
+      "step": 8598
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.60297043178475e-06,
+      "loss": 0.1903,
+      "step": 8599
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.60088813824328e-06,
+      "loss": 0.1982,
+      "step": 8600
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.598805906575938e-06,
+      "loss": 0.1996,
+      "step": 8601
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.596723736874818e-06,
+      "loss": 0.1818,
+      "step": 8602
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.594641629231994e-06,
+      "loss": 0.1933,
+      "step": 8603
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.592559583739546e-06,
+      "loss": 0.1924,
+      "step": 8604
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.590477600489558e-06,
+      "loss": 0.1835,
+      "step": 8605
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.588395679574095e-06,
+      "loss": 0.1832,
+      "step": 8606
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.58631382108523e-06,
+      "loss": 0.1593,
+      "step": 8607
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.584232025115029e-06,
+      "loss": 0.206,
+      "step": 8608
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.582150291755558e-06,
+      "loss": 0.183,
+      "step": 8609
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.580068621098884e-06,
+      "loss": 0.1623,
+      "step": 8610
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.577987013237061e-06,
+      "loss": 0.1998,
+      "step": 8611
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.575905468262147e-06,
+      "loss": 0.2144,
+      "step": 8612
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.573823986266196e-06,
+      "loss": 0.1846,
+      "step": 8613
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.571742567341263e-06,
+      "loss": 0.1727,
+      "step": 8614
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.56966121157939e-06,
+      "loss": 0.1782,
+      "step": 8615
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.567579919072623e-06,
+      "loss": 0.1887,
+      "step": 8616
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.565498689913014e-06,
+      "loss": 0.1695,
+      "step": 8617
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.56341752419259e-06,
+      "loss": 0.1709,
+      "step": 8618
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.561336422003399e-06,
+      "loss": 0.2394,
+      "step": 8619
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.55925538343747e-06,
+      "loss": 0.1657,
+      "step": 8620
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.55717440858683e-06,
+      "loss": 0.2158,
+      "step": 8621
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.55509349754352e-06,
+      "loss": 0.195,
+      "step": 8622
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.553012650399556e-06,
+      "loss": 0.2183,
+      "step": 8623
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.550931867246968e-06,
+      "loss": 0.1756,
+      "step": 8624
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.54885114817777e-06,
+      "loss": 0.1652,
+      "step": 8625
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.546770493283978e-06,
+      "loss": 0.1687,
+      "step": 8626
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.544689902657617e-06,
+      "loss": 0.181,
+      "step": 8627
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.542609376390691e-06,
+      "loss": 0.1746,
+      "step": 8628
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.540528914575208e-06,
+      "loss": 0.1806,
+      "step": 8629
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.538448517303176e-06,
+      "loss": 0.2174,
+      "step": 8630
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.536368184666602e-06,
+      "loss": 0.1709,
+      "step": 8631
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.534287916757477e-06,
+      "loss": 0.2024,
+      "step": 8632
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.532207713667812e-06,
+      "loss": 0.2326,
+      "step": 8633
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.530127575489589e-06,
+      "loss": 0.1871,
+      "step": 8634
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.528047502314804e-06,
+      "loss": 0.1647,
+      "step": 8635
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.52596749423545e-06,
+      "loss": 0.1827,
+      "step": 8636
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.52388755134351e-06,
+      "loss": 0.1634,
+      "step": 8637
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.521807673730961e-06,
+      "loss": 0.1892,
+      "step": 8638
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.519727861489794e-06,
+      "loss": 0.1851,
+      "step": 8639
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.517648114711975e-06,
+      "loss": 0.1761,
+      "step": 8640
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.515568433489493e-06,
+      "loss": 0.1573,
+      "step": 8641
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.513488817914309e-06,
+      "loss": 0.1585,
+      "step": 8642
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.511409268078392e-06,
+      "loss": 0.1825,
+      "step": 8643
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.50932978407371e-06,
+      "loss": 0.1956,
+      "step": 8644
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.507250365992227e-06,
+      "loss": 0.1945,
+      "step": 8645
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.505171013925898e-06,
+      "loss": 0.1712,
+      "step": 8646
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.503091727966687e-06,
+      "loss": 0.1774,
+      "step": 8647
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.501012508206543e-06,
+      "loss": 0.1734,
+      "step": 8648
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.498933354737415e-06,
+      "loss": 0.1563,
+      "step": 8649
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.49685426765126e-06,
+      "loss": 0.2579,
+      "step": 8650
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.494775247040017e-06,
+      "loss": 0.1871,
+      "step": 8651
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.492696292995625e-06,
+      "loss": 0.2132,
+      "step": 8652
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.49061740561003e-06,
+      "loss": 0.2145,
+      "step": 8653
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.48853858497517e-06,
+      "loss": 0.2005,
+      "step": 8654
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.486459831182969e-06,
+      "loss": 0.1941,
+      "step": 8655
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.484381144325369e-06,
+      "loss": 0.1903,
+      "step": 8656
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.482302524494286e-06,
+      "loss": 0.2133,
+      "step": 8657
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.480223971781651e-06,
+      "loss": 0.199,
+      "step": 8658
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.478145486279388e-06,
+      "loss": 0.1923,
+      "step": 8659
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.47606706807941e-06,
+      "loss": 0.1911,
+      "step": 8660
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.473988717273638e-06,
+      "loss": 0.1811,
+      "step": 8661
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.471910433953981e-06,
+      "loss": 0.1762,
+      "step": 8662
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.469832218212346e-06,
+      "loss": 0.1959,
+      "step": 8663
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.467754070140648e-06,
+      "loss": 0.1658,
+      "step": 8664
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.465675989830787e-06,
+      "loss": 0.2011,
+      "step": 8665
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.463597977374658e-06,
+      "loss": 0.1868,
+      "step": 8666
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.461520032864165e-06,
+      "loss": 0.189,
+      "step": 8667
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.459442156391205e-06,
+      "loss": 0.1843,
+      "step": 8668
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.457364348047662e-06,
+      "loss": 0.208,
+      "step": 8669
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.455286607925433e-06,
+      "loss": 0.185,
+      "step": 8670
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.453208936116396e-06,
+      "loss": 0.1967,
+      "step": 8671
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.451131332712435e-06,
+      "loss": 0.1673,
+      "step": 8672
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.449053797805437e-06,
+      "loss": 0.1923,
+      "step": 8673
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.44697633148727e-06,
+      "loss": 0.1826,
+      "step": 8674
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.444898933849815e-06,
+      "loss": 0.2106,
+      "step": 8675
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.442821604984935e-06,
+      "loss": 0.2039,
+      "step": 8676
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.440744344984499e-06,
+      "loss": 0.2044,
+      "step": 8677
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.43866715394038e-06,
+      "loss": 0.174,
+      "step": 8678
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.436590031944432e-06,
+      "loss": 0.2167,
+      "step": 8679
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.434512979088509e-06,
+      "loss": 0.1806,
+      "step": 8680
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.432435995464475e-06,
+      "loss": 0.19,
+      "step": 8681
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.43035908116418e-06,
+      "loss": 0.2086,
+      "step": 8682
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.428282236279469e-06,
+      "loss": 0.2202,
+      "step": 8683
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.426205460902195e-06,
+      "loss": 0.165,
+      "step": 8684
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.424128755124195e-06,
+      "loss": 0.1828,
+      "step": 8685
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.422052119037308e-06,
+      "loss": 0.1736,
+      "step": 8686
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.41997555273338e-06,
+      "loss": 0.1866,
+      "step": 8687
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.417899056304238e-06,
+      "loss": 0.1811,
+      "step": 8688
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.41582262984171e-06,
+      "loss": 0.1818,
+      "step": 8689
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.413746273437629e-06,
+      "loss": 0.1799,
+      "step": 8690
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.411669987183815e-06,
+      "loss": 0.1627,
+      "step": 8691
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.409593771172096e-06,
+      "loss": 0.1908,
+      "step": 8692
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.407517625494287e-06,
+      "loss": 0.1923,
+      "step": 8693
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.405441550242197e-06,
+      "loss": 0.175,
+      "step": 8694
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.403365545507649e-06,
+      "loss": 0.1898,
+      "step": 8695
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.401289611382446e-06,
+      "loss": 0.1964,
+      "step": 8696
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.39921374795839e-06,
+      "loss": 0.1875,
+      "step": 8697
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.397137955327293e-06,
+      "loss": 0.177,
+      "step": 8698
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.395062233580948e-06,
+      "loss": 0.1737,
+      "step": 8699
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.392986582811149e-06,
+      "loss": 0.1853,
+      "step": 8700
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.3909110031097e-06,
+      "loss": 0.1898,
+      "step": 8701
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.388835494568383e-06,
+      "loss": 0.1766,
+      "step": 8702
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.386760057278983e-06,
+      "loss": 0.2166,
+      "step": 8703
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.38468469133329e-06,
+      "loss": 0.1815,
+      "step": 8704
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.382609396823082e-06,
+      "loss": 0.1806,
+      "step": 8705
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.380534173840135e-06,
+      "loss": 0.1792,
+      "step": 8706
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.378459022476228e-06,
+      "loss": 0.1718,
+      "step": 8707
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.376383942823126e-06,
+      "loss": 0.1957,
+      "step": 8708
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.3743089349726e-06,
+      "loss": 0.1695,
+      "step": 8709
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.372233999016419e-06,
+      "loss": 0.1802,
+      "step": 8710
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.370159135046337e-06,
+      "loss": 0.2108,
+      "step": 8711
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.368084343154121e-06,
+      "loss": 0.1618,
+      "step": 8712
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.366009623431518e-06,
+      "loss": 0.1994,
+      "step": 8713
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.363934975970282e-06,
+      "loss": 0.1728,
+      "step": 8714
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.361860400862167e-06,
+      "loss": 0.178,
+      "step": 8715
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.359785898198918e-06,
+      "loss": 0.1789,
+      "step": 8716
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.35771146807227e-06,
+      "loss": 0.2016,
+      "step": 8717
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.355637110573968e-06,
+      "loss": 0.1827,
+      "step": 8718
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.35356282579575e-06,
+      "loss": 0.2036,
+      "step": 8719
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.351488613829342e-06,
+      "loss": 0.1626,
+      "step": 8720
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.349414474766482e-06,
+      "loss": 0.1861,
+      "step": 8721
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.34734040869889e-06,
+      "loss": 0.1987,
+      "step": 8722
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.34526641571829e-06,
+      "loss": 0.1663,
+      "step": 8723
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.343192495916408e-06,
+      "loss": 0.1641,
+      "step": 8724
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.341118649384951e-06,
+      "loss": 0.1689,
+      "step": 8725
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.339044876215644e-06,
+      "loss": 0.1905,
+      "step": 8726
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.336971176500188e-06,
+      "loss": 0.1724,
+      "step": 8727
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.334897550330291e-06,
+      "loss": 0.1809,
+      "step": 8728
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.332823997797663e-06,
+      "loss": 0.1955,
+      "step": 8729
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.330750518993998e-06,
+      "loss": 0.2083,
+      "step": 8730
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.328677114010996e-06,
+      "loss": 0.193,
+      "step": 8731
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.326603782940352e-06,
+      "loss": 0.1997,
+      "step": 8732
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.324530525873754e-06,
+      "loss": 0.1924,
+      "step": 8733
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.32245734290289e-06,
+      "loss": 0.213,
+      "step": 8734
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.320384234119448e-06,
+      "loss": 0.2138,
+      "step": 8735
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.318311199615103e-06,
+      "loss": 0.1822,
+      "step": 8736
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.316238239481533e-06,
+      "loss": 0.1825,
+      "step": 8737
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.314165353810419e-06,
+      "loss": 0.1794,
+      "step": 8738
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.312092542693426e-06,
+      "loss": 0.2039,
+      "step": 8739
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.310019806222224e-06,
+      "loss": 0.202,
+      "step": 8740
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.307947144488477e-06,
+      "loss": 0.1911,
+      "step": 8741
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.305874557583842e-06,
+      "loss": 0.2032,
+      "step": 8742
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.303802045599987e-06,
+      "loss": 0.2013,
+      "step": 8743
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.301729608628556e-06,
+      "loss": 0.1886,
+      "step": 8744
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.299657246761206e-06,
+      "loss": 0.2147,
+      "step": 8745
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.297584960089584e-06,
+      "loss": 0.1869,
+      "step": 8746
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.295512748705336e-06,
+      "loss": 0.2185,
+      "step": 8747
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.293440612700095e-06,
+      "loss": 0.1781,
+      "step": 8748
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.291368552165507e-06,
+      "loss": 0.1895,
+      "step": 8749
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.289296567193209e-06,
+      "loss": 0.2046,
+      "step": 8750
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.28722465787482e-06,
+      "loss": 0.1853,
+      "step": 8751
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.285152824301982e-06,
+      "loss": 0.1692,
+      "step": 8752
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.28308106656631e-06,
+      "loss": 0.1677,
+      "step": 8753
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.281009384759424e-06,
+      "loss": 0.1973,
+      "step": 8754
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.278937778972953e-06,
+      "loss": 0.1735,
+      "step": 8755
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.276866249298499e-06,
+      "loss": 0.1585,
+      "step": 8756
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.274794795827682e-06,
+      "loss": 0.1767,
+      "step": 8757
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.272723418652105e-06,
+      "loss": 0.1998,
+      "step": 8758
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.27065211786337e-06,
+      "loss": 0.1875,
+      "step": 8759
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.268580893553085e-06,
+      "loss": 0.189,
+      "step": 8760
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.266509745812845e-06,
+      "loss": 0.1699,
+      "step": 8761
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.264438674734237e-06,
+      "loss": 0.1879,
+      "step": 8762
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.262367680408861e-06,
+      "loss": 0.1758,
+      "step": 8763
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.260296762928303e-06,
+      "loss": 0.1943,
+      "step": 8764
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.258225922384141e-06,
+      "loss": 0.2244,
+      "step": 8765
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.256155158867964e-06,
+      "loss": 0.1602,
+      "step": 8766
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.254084472471342e-06,
+      "loss": 0.2056,
+      "step": 8767
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.252013863285848e-06,
+      "loss": 0.2153,
+      "step": 8768
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.249943331403062e-06,
+      "loss": 0.2042,
+      "step": 8769
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.247872876914545e-06,
+      "loss": 0.1691,
+      "step": 8770
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.245802499911853e-06,
+      "loss": 0.1708,
+      "step": 8771
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.243732200486559e-06,
+      "loss": 0.2066,
+      "step": 8772
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.24166197873021e-06,
+      "loss": 0.2027,
+      "step": 8773
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.239591834734368e-06,
+      "loss": 0.2088,
+      "step": 8774
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.237521768590577e-06,
+      "loss": 0.1931,
+      "step": 8775
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.23545178039038e-06,
+      "loss": 0.1904,
+      "step": 8776
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.233381870225327e-06,
+      "loss": 0.203,
+      "step": 8777
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.231312038186956e-06,
+      "loss": 0.1939,
+      "step": 8778
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.229242284366798e-06,
+      "loss": 0.1757,
+      "step": 8779
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.227172608856393e-06,
+      "loss": 0.1806,
+      "step": 8780
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.225103011747265e-06,
+      "loss": 0.18,
+      "step": 8781
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.223033493130936e-06,
+      "loss": 0.1758,
+      "step": 8782
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.220964053098938e-06,
+      "loss": 0.1789,
+      "step": 8783
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.218894691742786e-06,
+      "loss": 0.1723,
+      "step": 8784
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.216825409153987e-06,
+      "loss": 0.1778,
+      "step": 8785
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.214756205424064e-06,
+      "loss": 0.17,
+      "step": 8786
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.21268708064452e-06,
+      "loss": 0.1976,
+      "step": 8787
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.210618034906856e-06,
+      "loss": 0.1926,
+      "step": 8788
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.208549068302584e-06,
+      "loss": 0.1677,
+      "step": 8789
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.206480180923189e-06,
+      "loss": 0.1771,
+      "step": 8790
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.204411372860172e-06,
+      "loss": 0.1794,
+      "step": 8791
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.202342644205027e-06,
+      "loss": 0.1595,
+      "step": 8792
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.200273995049232e-06,
+      "loss": 0.1825,
+      "step": 8793
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.198205425484281e-06,
+      "loss": 0.1816,
+      "step": 8794
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.196136935601646e-06,
+      "loss": 0.2103,
+      "step": 8795
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.194068525492804e-06,
+      "loss": 0.2058,
+      "step": 8796
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.192000195249235e-06,
+      "loss": 0.1673,
+      "step": 8797
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.189931944962405e-06,
+      "loss": 0.1705,
+      "step": 8798
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.187863774723774e-06,
+      "loss": 0.1582,
+      "step": 8799
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.185795684624812e-06,
+      "loss": 0.1776,
+      "step": 8800
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.183727674756978e-06,
+      "loss": 0.187,
+      "step": 8801
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.18165974521172e-06,
+      "loss": 0.2416,
+      "step": 8802
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.1795918960805e-06,
+      "loss": 0.1716,
+      "step": 8803
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.177524127454757e-06,
+      "loss": 0.1999,
+      "step": 8804
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.175456439425939e-06,
+      "loss": 0.2036,
+      "step": 8805
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.173388832085492e-06,
+      "loss": 0.1857,
+      "step": 8806
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.171321305524845e-06,
+      "loss": 0.1647,
+      "step": 8807
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.169253859835442e-06,
+      "loss": 0.2051,
+      "step": 8808
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.167186495108705e-06,
+      "loss": 0.179,
+      "step": 8809
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.16511921143606e-06,
+      "loss": 0.157,
+      "step": 8810
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.16305200890894e-06,
+      "loss": 0.1728,
+      "step": 8811
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.16098488761876e-06,
+      "loss": 0.1923,
+      "step": 8812
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.158917847656928e-06,
+      "loss": 0.221,
+      "step": 8813
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.156850889114869e-06,
+      "loss": 0.1766,
+      "step": 8814
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.154784012083985e-06,
+      "loss": 0.2105,
+      "step": 8815
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.152717216655679e-06,
+      "loss": 0.1875,
+      "step": 8816
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.150650502921362e-06,
+      "loss": 0.1858,
+      "step": 8817
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.148583870972423e-06,
+      "loss": 0.1861,
+      "step": 8818
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.146517320900257e-06,
+      "loss": 0.1889,
+      "step": 8819
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.144450852796264e-06,
+      "loss": 0.1763,
+      "step": 8820
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.142384466751823e-06,
+      "loss": 0.1978,
+      "step": 8821
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.140318162858315e-06,
+      "loss": 0.1874,
+      "step": 8822
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.138251941207126e-06,
+      "loss": 0.2005,
+      "step": 8823
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.13618580188963e-06,
+      "loss": 0.1909,
+      "step": 8824
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.134119744997204e-06,
+      "loss": 0.1949,
+      "step": 8825
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.132053770621215e-06,
+      "loss": 0.188,
+      "step": 8826
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.12998787885302e-06,
+      "loss": 0.1674,
+      "step": 8827
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.12792206978399e-06,
+      "loss": 0.1808,
+      "step": 8828
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.125856343505483e-06,
+      "loss": 0.2242,
+      "step": 8829
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.123790700108846e-06,
+      "loss": 0.1867,
+      "step": 8830
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.121725139685439e-06,
+      "loss": 0.2069,
+      "step": 8831
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.119659662326604e-06,
+      "loss": 0.1812,
+      "step": 8832
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.11759426812368e-06,
+      "loss": 0.1835,
+      "step": 8833
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.115528957168019e-06,
+      "loss": 0.2076,
+      "step": 8834
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.113463729550948e-06,
+      "loss": 0.1947,
+      "step": 8835
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.111398585363798e-06,
+      "loss": 0.1686,
+      "step": 8836
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.109333524697902e-06,
+      "loss": 0.1852,
+      "step": 8837
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.107268547644585e-06,
+      "loss": 0.1828,
+      "step": 8838
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.105203654295164e-06,
+      "loss": 0.1973,
+      "step": 8839
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.103138844740964e-06,
+      "loss": 0.1974,
+      "step": 8840
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.10107411907329e-06,
+      "loss": 0.2098,
+      "step": 8841
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.099009477383459e-06,
+      "loss": 0.2181,
+      "step": 8842
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.096944919762776e-06,
+      "loss": 0.1808,
+      "step": 8843
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.09488044630254e-06,
+      "loss": 0.18,
+      "step": 8844
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.092816057094054e-06,
+      "loss": 0.1793,
+      "step": 8845
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.090751752228613e-06,
+      "loss": 0.1886,
+      "step": 8846
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.088687531797505e-06,
+      "loss": 0.1776,
+      "step": 8847
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.086623395892023e-06,
+      "loss": 0.2041,
+      "step": 8848
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.08455934460345e-06,
+      "loss": 0.1788,
+      "step": 8849
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.08249537802306e-06,
+      "loss": 0.1892,
+      "step": 8850
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.080431496242136e-06,
+      "loss": 0.1961,
+      "step": 8851
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.078367699351952e-06,
+      "loss": 0.2215,
+      "step": 8852
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.076303987443768e-06,
+      "loss": 0.1859,
+      "step": 8853
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.074240360608862e-06,
+      "loss": 0.1954,
+      "step": 8854
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.072176818938487e-06,
+      "loss": 0.1756,
+      "step": 8855
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.0701133625239e-06,
+      "loss": 0.1835,
+      "step": 8856
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.068049991456361e-06,
+      "loss": 0.1982,
+      "step": 8857
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.065986705827114e-06,
+      "loss": 0.1926,
+      "step": 8858
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.063923505727412e-06,
+      "loss": 0.1857,
+      "step": 8859
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.061860391248493e-06,
+      "loss": 0.1617,
+      "step": 8860
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.059797362481596e-06,
+      "loss": 0.1797,
+      "step": 8861
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.05773441951796e-06,
+      "loss": 0.192,
+      "step": 8862
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.055671562448812e-06,
+      "loss": 0.1624,
+      "step": 8863
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.05360879136538e-06,
+      "loss": 0.1884,
+      "step": 8864
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.05154610635889e-06,
+      "loss": 0.2097,
+      "step": 8865
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.049483507520563e-06,
+      "loss": 0.1888,
+      "step": 8866
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.047420994941608e-06,
+      "loss": 0.1708,
+      "step": 8867
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.045358568713246e-06,
+      "loss": 0.1824,
+      "step": 8868
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.04329622892668e-06,
+      "loss": 0.1978,
+      "step": 8869
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.041233975673112e-06,
+      "loss": 0.1746,
+      "step": 8870
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.039171809043755e-06,
+      "loss": 0.1779,
+      "step": 8871
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.037109729129792e-06,
+      "loss": 0.185,
+      "step": 8872
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.035047736022426e-06,
+      "loss": 0.1728,
+      "step": 8873
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.03298582981284e-06,
+      "loss": 0.2019,
+      "step": 8874
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.030924010592218e-06,
+      "loss": 0.2154,
+      "step": 8875
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.028862278451752e-06,
+      "loss": 0.1843,
+      "step": 8876
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.026800633482611e-06,
+      "loss": 0.2147,
+      "step": 8877
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.024739075775972e-06,
+      "loss": 0.1999,
+      "step": 8878
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.022677605423003e-06,
+      "loss": 0.1703,
+      "step": 8879
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.020616222514874e-06,
+      "loss": 0.1613,
+      "step": 8880
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.018554927142739e-06,
+      "loss": 0.1956,
+      "step": 8881
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.016493719397765e-06,
+      "loss": 0.1994,
+      "step": 8882
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.014432599371106e-06,
+      "loss": 0.1904,
+      "step": 8883
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.012371567153907e-06,
+      "loss": 0.1912,
+      "step": 8884
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.01031062283732e-06,
+      "loss": 0.1892,
+      "step": 8885
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.008249766512487e-06,
+      "loss": 0.1965,
+      "step": 8886
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.006188998270543e-06,
+      "loss": 0.164,
+      "step": 8887
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.00412831820263e-06,
+      "loss": 0.1828,
+      "step": 8888
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.002067726399876e-06,
+      "loss": 0.1771,
+      "step": 8889
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.000007222953404e-06,
+      "loss": 0.2137,
+      "step": 8890
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.997946807954343e-06,
+      "loss": 0.21,
+      "step": 8891
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.995886481493806e-06,
+      "loss": 0.1797,
+      "step": 8892
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.993826243662921e-06,
+      "loss": 0.1793,
+      "step": 8893
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.991766094552791e-06,
+      "loss": 0.1809,
+      "step": 8894
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.989706034254521e-06,
+      "loss": 0.1802,
+      "step": 8895
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.98764606285922e-06,
+      "loss": 0.1931,
+      "step": 8896
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.98558618045799e-06,
+      "loss": 0.2029,
+      "step": 8897
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.983526387141918e-06,
+      "loss": 0.1842,
+      "step": 8898
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.981466683002107e-06,
+      "loss": 0.1912,
+      "step": 8899
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.979407068129636e-06,
+      "loss": 0.158,
+      "step": 8900
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.977347542615592e-06,
+      "loss": 0.1797,
+      "step": 8901
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.975288106551058e-06,
+      "loss": 0.1993,
+      "step": 8902
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.97322876002711e-06,
+      "loss": 0.1601,
+      "step": 8903
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.971169503134814e-06,
+      "loss": 0.1718,
+      "step": 8904
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.969110335965244e-06,
+      "loss": 0.2044,
+      "step": 8905
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.967051258609462e-06,
+      "loss": 0.1979,
+      "step": 8906
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.964992271158532e-06,
+      "loss": 0.1796,
+      "step": 8907
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.96293337370351e-06,
+      "loss": 0.1801,
+      "step": 8908
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.96087456633544e-06,
+      "loss": 0.1886,
+      "step": 8909
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.95881584914538e-06,
+      "loss": 0.1648,
+      "step": 8910
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.956757222224371e-06,
+      "loss": 0.1821,
+      "step": 8911
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.954698685663452e-06,
+      "loss": 0.2319,
+      "step": 8912
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.952640239553662e-06,
+      "loss": 0.1857,
+      "step": 8913
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.950581883986033e-06,
+      "loss": 0.1769,
+      "step": 8914
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.94852361905159e-06,
+      "loss": 0.1795,
+      "step": 8915
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.946465444841363e-06,
+      "loss": 0.1832,
+      "step": 8916
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.94440736144637e-06,
+      "loss": 0.185,
+      "step": 8917
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.942349368957625e-06,
+      "loss": 0.2205,
+      "step": 8918
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.940291467466143e-06,
+      "loss": 0.1935,
+      "step": 8919
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.938233657062935e-06,
+      "loss": 0.1872,
+      "step": 8920
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.936175937838996e-06,
+      "loss": 0.1781,
+      "step": 8921
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.934118309885337e-06,
+      "loss": 0.1849,
+      "step": 8922
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.932060773292946e-06,
+      "loss": 0.178,
+      "step": 8923
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.93000332815282e-06,
+      "loss": 0.1828,
+      "step": 8924
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.92794597455595e-06,
+      "loss": 0.1747,
+      "step": 8925
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.92588871259331e-06,
+      "loss": 0.1957,
+      "step": 8926
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.923831542355892e-06,
+      "loss": 0.1633,
+      "step": 8927
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.921774463934663e-06,
+      "loss": 0.1838,
+      "step": 8928
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.919717477420596e-06,
+      "loss": 0.1716,
+      "step": 8929
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.917660582904667e-06,
+      "loss": 0.2162,
+      "step": 8930
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.915603780477832e-06,
+      "loss": 0.1748,
+      "step": 8931
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.913547070231049e-06,
+      "loss": 0.1623,
+      "step": 8932
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.91149045225528e-06,
+      "loss": 0.1671,
+      "step": 8933
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.909433926641475e-06,
+      "loss": 0.1621,
+      "step": 8934
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.907377493480577e-06,
+      "loss": 0.1623,
+      "step": 8935
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.905321152863537e-06,
+      "loss": 0.1826,
+      "step": 8936
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.903264904881289e-06,
+      "loss": 0.2222,
+      "step": 8937
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.901208749624767e-06,
+      "loss": 0.1891,
+      "step": 8938
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.899152687184907e-06,
+      "loss": 0.167,
+      "step": 8939
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.89709671765263e-06,
+      "loss": 0.1938,
+      "step": 8940
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.895040841118868e-06,
+      "loss": 0.1901,
+      "step": 8941
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.892985057674532e-06,
+      "loss": 0.1848,
+      "step": 8942
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.890929367410536e-06,
+      "loss": 0.2223,
+      "step": 8943
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.888873770417797e-06,
+      "loss": 0.2242,
+      "step": 8944
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.88681826678722e-06,
+      "loss": 0.1623,
+      "step": 8945
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.8847628566097e-06,
+      "loss": 0.1838,
+      "step": 8946
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.882707539976144e-06,
+      "loss": 0.1905,
+      "step": 8947
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.880652316977443e-06,
+      "loss": 0.1777,
+      "step": 8948
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.878597187704483e-06,
+      "loss": 0.2158,
+      "step": 8949
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.876542152248156e-06,
+      "loss": 0.1601,
+      "step": 8950
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.874487210699339e-06,
+      "loss": 0.1821,
+      "step": 8951
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.872432363148912e-06,
+      "loss": 0.1972,
+      "step": 8952
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.87037760968775e-06,
+      "loss": 0.1592,
+      "step": 8953
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.868322950406719e-06,
+      "loss": 0.1688,
+      "step": 8954
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.866268385396681e-06,
+      "loss": 0.157,
+      "step": 8955
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.864213914748502e-06,
+      "loss": 0.1856,
+      "step": 8956
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.862159538553037e-06,
+      "loss": 0.1772,
+      "step": 8957
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.86010525690114e-06,
+      "loss": 0.1794,
+      "step": 8958
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.85805106988366e-06,
+      "loss": 0.1801,
+      "step": 8959
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.855996977591434e-06,
+      "loss": 0.2046,
+      "step": 8960
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.853942980115309e-06,
+      "loss": 0.1827,
+      "step": 8961
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.85188907754612e-06,
+      "loss": 0.1714,
+      "step": 8962
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.849835269974694e-06,
+      "loss": 0.1661,
+      "step": 8963
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.847781557491864e-06,
+      "loss": 0.1722,
+      "step": 8964
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.84572794018845e-06,
+      "loss": 0.19,
+      "step": 8965
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.84367441815527e-06,
+      "loss": 0.1939,
+      "step": 8966
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.841620991483142e-06,
+      "loss": 0.1971,
+      "step": 8967
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.839567660262875e-06,
+      "loss": 0.1653,
+      "step": 8968
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.83751442458527e-06,
+      "loss": 0.1937,
+      "step": 8969
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.835461284541137e-06,
+      "loss": 0.1868,
+      "step": 8970
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.833408240221273e-06,
+      "loss": 0.2082,
+      "step": 8971
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.831355291716464e-06,
+      "loss": 0.2297,
+      "step": 8972
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.82930243911751e-06,
+      "loss": 0.1789,
+      "step": 8973
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.827249682515184e-06,
+      "loss": 0.1658,
+      "step": 8974
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.825197022000277e-06,
+      "loss": 0.1882,
+      "step": 8975
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.823144457663563e-06,
+      "loss": 0.2033,
+      "step": 8976
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.82109198959581e-06,
+      "loss": 0.1785,
+      "step": 8977
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.819039617887793e-06,
+      "loss": 0.2427,
+      "step": 8978
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.81698734263027e-06,
+      "loss": 0.1849,
+      "step": 8979
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.814935163914002e-06,
+      "loss": 0.1964,
+      "step": 8980
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.812883081829747e-06,
+      "loss": 0.1803,
+      "step": 8981
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.810831096468255e-06,
+      "loss": 0.2414,
+      "step": 8982
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.808779207920266e-06,
+      "loss": 0.1936,
+      "step": 8983
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.806727416276532e-06,
+      "loss": 0.198,
+      "step": 8984
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.804675721627788e-06,
+      "loss": 0.1635,
+      "step": 8985
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.802624124064764e-06,
+      "loss": 0.1529,
+      "step": 8986
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.800572623678197e-06,
+      "loss": 0.1738,
+      "step": 8987
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.798521220558805e-06,
+      "loss": 0.1831,
+      "step": 8988
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.796469914797308e-06,
+      "loss": 0.1838,
+      "step": 8989
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.794418706484435e-06,
+      "loss": 0.1984,
+      "step": 8990
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.792367595710884e-06,
+      "loss": 0.187,
+      "step": 8991
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.790316582567371e-06,
+      "loss": 0.1776,
+      "step": 8992
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.788265667144598e-06,
+      "loss": 0.1851,
+      "step": 8993
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.786214849533262e-06,
+      "loss": 0.1647,
+      "step": 8994
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.784164129824063e-06,
+      "loss": 0.1826,
+      "step": 8995
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.78211350810769e-06,
+      "loss": 0.2007,
+      "step": 8996
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.780062984474824e-06,
+      "loss": 0.1999,
+      "step": 8997
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.778012559016153e-06,
+      "loss": 0.1672,
+      "step": 8998
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.775962231822356e-06,
+      "loss": 0.186,
+      "step": 8999
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.773912002984098e-06,
+      "loss": 0.1896,
+      "step": 9000
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.77186187259206e-06,
+      "loss": 0.1989,
+      "step": 9001
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.769811840736897e-06,
+      "loss": 0.1825,
+      "step": 9002
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.76776190750927e-06,
+      "loss": 0.1732,
+      "step": 9003
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.765712072999844e-06,
+      "loss": 0.1894,
+      "step": 9004
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.763662337299261e-06,
+      "loss": 0.207,
+      "step": 9005
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.76161270049817e-06,
+      "loss": 0.1949,
+      "step": 9006
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.759563162687216e-06,
+      "loss": 0.1802,
+      "step": 9007
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.757513723957036e-06,
+      "loss": 0.1769,
+      "step": 9008
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.755464384398268e-06,
+      "loss": 0.1965,
+      "step": 9009
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.753415144101536e-06,
+      "loss": 0.1706,
+      "step": 9010
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.751366003157468e-06,
+      "loss": 0.1843,
+      "step": 9011
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.749316961656686e-06,
+      "loss": 0.1639,
+      "step": 9012
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.747268019689806e-06,
+      "loss": 0.1698,
+      "step": 9013
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.745219177347437e-06,
+      "loss": 0.1798,
+      "step": 9014
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.74317043472019e-06,
+      "loss": 0.2036,
+      "step": 9015
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.741121791898672e-06,
+      "loss": 0.1827,
+      "step": 9016
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.73907324897347e-06,
+      "loss": 0.1912,
+      "step": 9017
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.737024806035192e-06,
+      "loss": 0.1635,
+      "step": 9018
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.73497646317442e-06,
+      "loss": 0.1769,
+      "step": 9019
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.73292822048174e-06,
+      "loss": 0.17,
+      "step": 9020
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.730880078047736e-06,
+      "loss": 0.1709,
+      "step": 9021
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.728832035962987e-06,
+      "loss": 0.1864,
+      "step": 9022
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.726784094318055e-06,
+      "loss": 0.1996,
+      "step": 9023
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.72473625320352e-06,
+      "loss": 0.1996,
+      "step": 9024
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.722688512709938e-06,
+      "loss": 0.1877,
+      "step": 9025
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.72064087292787e-06,
+      "loss": 0.1877,
+      "step": 9026
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.718593333947876e-06,
+      "loss": 0.2439,
+      "step": 9027
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.716545895860495e-06,
+      "loss": 0.1895,
+      "step": 9028
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.71449855875628e-06,
+      "loss": 0.1653,
+      "step": 9029
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.712451322725774e-06,
+      "loss": 0.176,
+      "step": 9030
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.710404187859505e-06,
+      "loss": 0.193,
+      "step": 9031
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.708357154248015e-06,
+      "loss": 0.1881,
+      "step": 9032
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.706310221981825e-06,
+      "loss": 0.1662,
+      "step": 9033
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.70426339115146e-06,
+      "loss": 0.1776,
+      "step": 9034
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.702216661847443e-06,
+      "loss": 0.1973,
+      "step": 9035
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.700170034160284e-06,
+      "loss": 0.2139,
+      "step": 9036
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.69812350818049e-06,
+      "loss": 0.1996,
+      "step": 9037
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.696077083998571e-06,
+      "loss": 0.1563,
+      "step": 9038
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.694030761705025e-06,
+      "loss": 0.1771,
+      "step": 9039
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.691984541390354e-06,
+      "loss": 0.1786,
+      "step": 9040
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.689938423145044e-06,
+      "loss": 0.1889,
+      "step": 9041
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.687892407059582e-06,
+      "loss": 0.1636,
+      "step": 9042
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.685846493224454e-06,
+      "loss": 0.1721,
+      "step": 9043
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.683800681730138e-06,
+      "loss": 0.1871,
+      "step": 9044
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.681754972667102e-06,
+      "loss": 0.1775,
+      "step": 9045
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.679709366125824e-06,
+      "loss": 0.169,
+      "step": 9046
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.677663862196763e-06,
+      "loss": 0.1925,
+      "step": 9047
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.675618460970379e-06,
+      "loss": 0.1667,
+      "step": 9048
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.673573162537132e-06,
+      "loss": 0.1818,
+      "step": 9049
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.67152796698747e-06,
+      "loss": 0.2026,
+      "step": 9050
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.669482874411836e-06,
+      "loss": 0.1912,
+      "step": 9051
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.667437884900677e-06,
+      "loss": 0.1647,
+      "step": 9052
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.665392998544428e-06,
+      "loss": 0.1609,
+      "step": 9053
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.66334821543352e-06,
+      "loss": 0.1964,
+      "step": 9054
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.661303535658388e-06,
+      "loss": 0.196,
+      "step": 9055
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.659258959309448e-06,
+      "loss": 0.2096,
+      "step": 9056
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.657214486477124e-06,
+      "loss": 0.1601,
+      "step": 9057
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.65517011725183e-06,
+      "loss": 0.17,
+      "step": 9058
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.65312585172397e-06,
+      "loss": 0.1545,
+      "step": 9059
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.651081689983959e-06,
+      "loss": 0.1995,
+      "step": 9060
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.649037632122191e-06,
+      "loss": 0.1819,
+      "step": 9061
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.646993678229061e-06,
+      "loss": 0.2076,
+      "step": 9062
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.644949828394969e-06,
+      "loss": 0.1594,
+      "step": 9063
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.642906082710295e-06,
+      "loss": 0.1907,
+      "step": 9064
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.640862441265418e-06,
+      "loss": 0.1714,
+      "step": 9065
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.638818904150723e-06,
+      "loss": 0.1594,
+      "step": 9066
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.636775471456583e-06,
+      "loss": 0.1838,
+      "step": 9067
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.63473214327336e-06,
+      "loss": 0.1975,
+      "step": 9068
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.632688919691426e-06,
+      "loss": 0.1718,
+      "step": 9069
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.630645800801133e-06,
+      "loss": 0.1975,
+      "step": 9070
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.628602786692837e-06,
+      "loss": 0.2057,
+      "step": 9071
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.626559877456894e-06,
+      "loss": 0.1856,
+      "step": 9072
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.624517073183642e-06,
+      "loss": 0.1845,
+      "step": 9073
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.6224743739634275e-06,
+      "loss": 0.2043,
+      "step": 9074
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.620431779886581e-06,
+      "loss": 0.1734,
+      "step": 9075
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.618389291043437e-06,
+      "loss": 0.2063,
+      "step": 9076
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.616346907524325e-06,
+      "loss": 0.1924,
+      "step": 9077
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.614304629419563e-06,
+      "loss": 0.1937,
+      "step": 9078
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.6122624568194655e-06,
+      "loss": 0.1785,
+      "step": 9079
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.6102203898143525e-06,
+      "loss": 0.1994,
+      "step": 9080
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.60817842849453e-06,
+      "loss": 0.2129,
+      "step": 9081
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.6061365729502964e-06,
+      "loss": 0.1851,
+      "step": 9082
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.6040948232719595e-06,
+      "loss": 0.183,
+      "step": 9083
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.602053179549804e-06,
+      "loss": 0.1725,
+      "step": 9084
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.600011641874122e-06,
+      "loss": 0.2011,
+      "step": 9085
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.597970210335205e-06,
+      "loss": 0.191,
+      "step": 9086
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.595928885023327e-06,
+      "loss": 0.179,
+      "step": 9087
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.593887666028759e-06,
+      "loss": 0.1583,
+      "step": 9088
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.591846553441778e-06,
+      "loss": 0.1727,
+      "step": 9089
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.589805547352647e-06,
+      "loss": 0.1833,
+      "step": 9090
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.587764647851635e-06,
+      "loss": 0.1618,
+      "step": 9091
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.585723855028988e-06,
+      "loss": 0.1778,
+      "step": 9092
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.58368316897496e-06,
+      "loss": 0.1974,
+      "step": 9093
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.5816425897798e-06,
+      "loss": 0.2022,
+      "step": 9094
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.5796021175337555e-06,
+      "loss": 0.1712,
+      "step": 9095
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.5775617523270515e-06,
+      "loss": 0.1558,
+      "step": 9096
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.575521494249933e-06,
+      "loss": 0.1557,
+      "step": 9097
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.573481343392621e-06,
+      "loss": 0.1687,
+      "step": 9098
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.571441299845339e-06,
+      "loss": 0.1828,
+      "step": 9099
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.5694013636983125e-06,
+      "loss": 0.1842,
+      "step": 9100
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.56736153504175e-06,
+      "loss": 0.1906,
+      "step": 9101
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.565321813965856e-06,
+      "loss": 0.2156,
+      "step": 9102
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.5632822005608416e-06,
+      "loss": 0.1645,
+      "step": 9103
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.561242694916908e-06,
+      "loss": 0.2126,
+      "step": 9104
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.559203297124241e-06,
+      "loss": 0.1973,
+      "step": 9105
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.557164007273041e-06,
+      "loss": 0.1574,
+      "step": 9106
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.555124825453486e-06,
+      "loss": 0.1816,
+      "step": 9107
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.5530857517557595e-06,
+      "loss": 0.1658,
+      "step": 9108
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.5510467862700395e-06,
+      "loss": 0.1995,
+      "step": 9109
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.549007929086489e-06,
+      "loss": 0.1612,
+      "step": 9110
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.5469691802952845e-06,
+      "loss": 0.2007,
+      "step": 9111
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.54493053998658e-06,
+      "loss": 0.1986,
+      "step": 9112
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.542892008250533e-06,
+      "loss": 0.177,
+      "step": 9113
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.540853585177299e-06,
+      "loss": 0.2045,
+      "step": 9114
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.538815270857024e-06,
+      "loss": 0.1981,
+      "step": 9115
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.536777065379843e-06,
+      "loss": 0.1862,
+      "step": 9116
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.534738968835903e-06,
+      "loss": 0.2021,
+      "step": 9117
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.5327009813153325e-06,
+      "loss": 0.188,
+      "step": 9118
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.5306631029082564e-06,
+      "loss": 0.1755,
+      "step": 9119
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.528625333704804e-06,
+      "loss": 0.1924,
+      "step": 9120
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.5265876737950875e-06,
+      "loss": 0.1803,
+      "step": 9121
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.52455012326922e-06,
+      "loss": 0.1675,
+      "step": 9122
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.522512682217315e-06,
+      "loss": 0.1616,
+      "step": 9123
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.5204753507294715e-06,
+      "loss": 0.1681,
+      "step": 9124
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.518438128895792e-06,
+      "loss": 0.2243,
+      "step": 9125
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.516401016806367e-06,
+      "loss": 0.195,
+      "step": 9126
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.514364014551284e-06,
+      "loss": 0.1927,
+      "step": 9127
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.512327122220634e-06,
+      "loss": 0.2058,
+      "step": 9128
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.510290339904491e-06,
+      "loss": 0.2001,
+      "step": 9129
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.5082536676929275e-06,
+      "loss": 0.2031,
+      "step": 9130
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.506217105676017e-06,
+      "loss": 0.1683,
+      "step": 9131
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.504180653943825e-06,
+      "loss": 0.1823,
+      "step": 9132
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.5021443125864055e-06,
+      "loss": 0.1932,
+      "step": 9133
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.500108081693821e-06,
+      "loss": 0.1957,
+      "step": 9134
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.498071961356116e-06,
+      "loss": 0.1843,
+      "step": 9135
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.496035951663334e-06,
+      "loss": 0.1894,
+      "step": 9136
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.494000052705524e-06,
+      "loss": 0.1806,
+      "step": 9137
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.491964264572713e-06,
+      "loss": 0.1646,
+      "step": 9138
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.489928587354934e-06,
+      "loss": 0.1711,
+      "step": 9139
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.487893021142213e-06,
+      "loss": 0.1992,
+      "step": 9140
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.485857566024568e-06,
+      "loss": 0.1817,
+      "step": 9141
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.4838222220920206e-06,
+      "loss": 0.2167,
+      "step": 9142
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.481786989434576e-06,
+      "loss": 0.1998,
+      "step": 9143
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.4797518681422425e-06,
+      "loss": 0.2074,
+      "step": 9144
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.477716858305021e-06,
+      "loss": 0.1965,
+      "step": 9145
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.475681960012908e-06,
+      "loss": 0.2041,
+      "step": 9146
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.47364717335589e-06,
+      "loss": 0.1698,
+      "step": 9147
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.4716124984239595e-06,
+      "loss": 0.1815,
+      "step": 9148
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.469577935307095e-06,
+      "loss": 0.1732,
+      "step": 9149
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.46754348409527e-06,
+      "loss": 0.1745,
+      "step": 9150
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.465509144878462e-06,
+      "loss": 0.1779,
+      "step": 9151
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.463474917746633e-06,
+      "loss": 0.1759,
+      "step": 9152
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.461440802789741e-06,
+      "loss": 0.184,
+      "step": 9153
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.459406800097753e-06,
+      "loss": 0.1847,
+      "step": 9154
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.457372909760612e-06,
+      "loss": 0.1879,
+      "step": 9155
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.455339131868263e-06,
+      "loss": 0.203,
+      "step": 9156
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.453305466510654e-06,
+      "loss": 0.2123,
+      "step": 9157
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.451271913777718e-06,
+      "loss": 0.2029,
+      "step": 9158
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.449238473759386e-06,
+      "loss": 0.2255,
+      "step": 9159
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.4472051465455894e-06,
+      "loss": 0.1799,
+      "step": 9160
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.445171932226241e-06,
+      "loss": 0.2155,
+      "step": 9161
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.443138830891265e-06,
+      "loss": 0.204,
+      "step": 9162
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.441105842630574e-06,
+      "loss": 0.1617,
+      "step": 9163
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.439072967534065e-06,
+      "loss": 0.1812,
+      "step": 9164
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.43704020569165e-06,
+      "loss": 0.1937,
+      "step": 9165
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.43500755719322e-06,
+      "loss": 0.2007,
+      "step": 9166
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.432975022128665e-06,
+      "loss": 0.1669,
+      "step": 9167
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.4309426005878805e-06,
+      "loss": 0.1989,
+      "step": 9168
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.4289102926607405e-06,
+      "loss": 0.1887,
+      "step": 9169
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.426878098437118e-06,
+      "loss": 0.1864,
+      "step": 9170
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.424846018006893e-06,
+      "loss": 0.1925,
+      "step": 9171
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.422814051459926e-06,
+      "loss": 0.176,
+      "step": 9172
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.420782198886083e-06,
+      "loss": 0.1777,
+      "step": 9173
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.418750460375221e-06,
+      "loss": 0.1857,
+      "step": 9174
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.416718836017183e-06,
+      "loss": 0.1861,
+      "step": 9175
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.4146873259018224e-06,
+      "loss": 0.2049,
+      "step": 9176
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.412655930118981e-06,
+      "loss": 0.1857,
+      "step": 9177
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.410624648758489e-06,
+      "loss": 0.1906,
+      "step": 9178
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.408593481910187e-06,
+      "loss": 0.2146,
+      "step": 9179
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.40656242966389e-06,
+      "loss": 0.1914,
+      "step": 9180
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.404531492109424e-06,
+      "loss": 0.1951,
+      "step": 9181
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.40250066933661e-06,
+      "loss": 0.2088,
+      "step": 9182
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.400469961435254e-06,
+      "loss": 0.1753,
+      "step": 9183
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.398439368495156e-06,
+      "loss": 0.2024,
+      "step": 9184
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.396408890606126e-06,
+      "loss": 0.1855,
+      "step": 9185
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.394378527857958e-06,
+      "loss": 0.2203,
+      "step": 9186
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.392348280340437e-06,
+      "loss": 0.1854,
+      "step": 9187
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.390318148143355e-06,
+      "loss": 0.1761,
+      "step": 9188
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.388288131356484e-06,
+      "loss": 0.1948,
+      "step": 9189
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.386258230069609e-06,
+      "loss": 0.1631,
+      "step": 9190
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.384228444372497e-06,
+      "loss": 0.1882,
+      "step": 9191
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.382198774354906e-06,
+      "loss": 0.2108,
+      "step": 9192
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.380169220106606e-06,
+      "loss": 0.1896,
+      "step": 9193
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.378139781717346e-06,
+      "loss": 0.1705,
+      "step": 9194
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.3761104592768725e-06,
+      "loss": 0.2063,
+      "step": 9195
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.374081252874941e-06,
+      "loss": 0.2183,
+      "step": 9196
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.372052162601282e-06,
+      "loss": 0.1925,
+      "step": 9197
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.370023188545628e-06,
+      "loss": 0.1553,
+      "step": 9198
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.367994330797713e-06,
+      "loss": 0.1834,
+      "step": 9199
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.365965589447263e-06,
+      "loss": 0.1832,
+      "step": 9200
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.363936964583988e-06,
+      "loss": 0.2282,
+      "step": 9201
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.361908456297613e-06,
+      "loss": 0.1834,
+      "step": 9202
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.359880064677838e-06,
+      "loss": 0.1702,
+      "step": 9203
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.357851789814365e-06,
+      "loss": 0.1678,
+      "step": 9204
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.355823631796904e-06,
+      "loss": 0.1838,
+      "step": 9205
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.353795590715131e-06,
+      "loss": 0.2117,
+      "step": 9206
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.351767666658751e-06,
+      "loss": 0.1804,
+      "step": 9207
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.349739859717435e-06,
+      "loss": 0.162,
+      "step": 9208
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.347712169980862e-06,
+      "loss": 0.201,
+      "step": 9209
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.345684597538709e-06,
+      "loss": 0.1689,
+      "step": 9210
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.343657142480642e-06,
+      "loss": 0.1447,
+      "step": 9211
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.341629804896318e-06,
+      "loss": 0.1898,
+      "step": 9212
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.339602584875396e-06,
+      "loss": 0.1858,
+      "step": 9213
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.337575482507533e-06,
+      "loss": 0.171,
+      "step": 9214
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.335548497882366e-06,
+      "loss": 0.1887,
+      "step": 9215
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.3335216310895445e-06,
+      "loss": 0.1728,
+      "step": 9216
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.331494882218698e-06,
+      "loss": 0.1672,
+      "step": 9217
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.329468251359459e-06,
+      "loss": 0.1716,
+      "step": 9218
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.327441738601458e-06,
+      "loss": 0.1671,
+      "step": 9219
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.32541534403431e-06,
+      "loss": 0.1688,
+      "step": 9220
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.323389067747626e-06,
+      "loss": 0.1845,
+      "step": 9221
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.321362909831025e-06,
+      "loss": 0.1842,
+      "step": 9222
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.3193368703741045e-06,
+      "loss": 0.1998,
+      "step": 9223
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.3173109494664695e-06,
+      "loss": 0.1717,
+      "step": 9224
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.315285147197712e-06,
+      "loss": 0.1845,
+      "step": 9225
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.313259463657417e-06,
+      "loss": 0.1793,
+      "step": 9226
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.311233898935173e-06,
+      "loss": 0.1682,
+      "step": 9227
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.309208453120558e-06,
+      "loss": 0.1732,
+      "step": 9228
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.307183126303139e-06,
+      "loss": 0.1616,
+      "step": 9229
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.305157918572492e-06,
+      "loss": 0.1814,
+      "step": 9230
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.303132830018173e-06,
+      "loss": 0.1916,
+      "step": 9231
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.301107860729741e-06,
+      "loss": 0.1642,
+      "step": 9232
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.299083010796752e-06,
+      "loss": 0.1593,
+      "step": 9233
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.2970582803087505e-06,
+      "loss": 0.194,
+      "step": 9234
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.295033669355274e-06,
+      "loss": 0.1658,
+      "step": 9235
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.293009178025862e-06,
+      "loss": 0.1711,
+      "step": 9236
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.290984806410046e-06,
+      "loss": 0.2015,
+      "step": 9237
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.288960554597349e-06,
+      "loss": 0.1754,
+      "step": 9238
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.286936422677295e-06,
+      "loss": 0.2034,
+      "step": 9239
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.284912410739393e-06,
+      "loss": 0.1798,
+      "step": 9240
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.2828885188731565e-06,
+      "loss": 0.1862,
+      "step": 9241
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.280864747168095e-06,
+      "loss": 0.1907,
+      "step": 9242
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.278841095713696e-06,
+      "loss": 0.1615,
+      "step": 9243
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.276817564599463e-06,
+      "loss": 0.1934,
+      "step": 9244
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.274794153914879e-06,
+      "loss": 0.194,
+      "step": 9245
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.2727708637494276e-06,
+      "loss": 0.2044,
+      "step": 9246
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.27074769419259e-06,
+      "loss": 0.2012,
+      "step": 9247
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.268724645333838e-06,
+      "loss": 0.1902,
+      "step": 9248
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.266701717262632e-06,
+      "loss": 0.174,
+      "step": 9249
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.26467891006844e-06,
+      "loss": 0.1916,
+      "step": 9250
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.262656223840718e-06,
+      "loss": 0.1752,
+      "step": 9251
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.260633658668914e-06,
+      "loss": 0.2055,
+      "step": 9252
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.258611214642479e-06,
+      "loss": 0.1837,
+      "step": 9253
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.256588891850847e-06,
+      "loss": 0.2101,
+      "step": 9254
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.2545666903834536e-06,
+      "loss": 0.2066,
+      "step": 9255
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.252544610329735e-06,
+      "loss": 0.1931,
+      "step": 9256
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.250522651779106e-06,
+      "loss": 0.1639,
+      "step": 9257
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.248500814820996e-06,
+      "loss": 0.1842,
+      "step": 9258
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.2464790995448096e-06,
+      "loss": 0.1717,
+      "step": 9259
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.244457506039956e-06,
+      "loss": 0.1766,
+      "step": 9260
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.242436034395844e-06,
+      "loss": 0.1548,
+      "step": 9261
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.240414684701866e-06,
+      "loss": 0.1592,
+      "step": 9262
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.238393457047412e-06,
+      "loss": 0.1918,
+      "step": 9263
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.236372351521873e-06,
+      "loss": 0.2025,
+      "step": 9264
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.234351368214629e-06,
+      "loss": 0.1592,
+      "step": 9265
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.232330507215053e-06,
+      "loss": 0.2108,
+      "step": 9266
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.23030976861252e-06,
+      "loss": 0.1711,
+      "step": 9267
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.2282891524963905e-06,
+      "loss": 0.2205,
+      "step": 9268
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.2262686589560226e-06,
+      "loss": 0.2005,
+      "step": 9269
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.224248288080777e-06,
+      "loss": 0.1806,
+      "step": 9270
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.222228039959997e-06,
+      "loss": 0.1619,
+      "step": 9271
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.220207914683028e-06,
+      "loss": 0.1837,
+      "step": 9272
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.218187912339207e-06,
+      "loss": 0.2109,
+      "step": 9273
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.216168033017865e-06,
+      "loss": 0.1785,
+      "step": 9274
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.214148276808334e-06,
+      "loss": 0.1817,
+      "step": 9275
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.212128643799929e-06,
+      "loss": 0.1861,
+      "step": 9276
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.210109134081968e-06,
+      "loss": 0.1718,
+      "step": 9277
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.208089747743765e-06,
+      "loss": 0.1925,
+      "step": 9278
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.206070484874622e-06,
+      "loss": 0.175,
+      "step": 9279
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.204051345563835e-06,
+      "loss": 0.1801,
+      "step": 9280
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.202032329900705e-06,
+      "loss": 0.1911,
+      "step": 9281
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.200013437974519e-06,
+      "loss": 0.1608,
+      "step": 9282
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.197994669874555e-06,
+      "loss": 0.1863,
+      "step": 9283
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.195976025690099e-06,
+      "loss": 0.1706,
+      "step": 9284
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.193957505510416e-06,
+      "loss": 0.1938,
+      "step": 9285
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.191939109424777e-06,
+      "loss": 0.1692,
+      "step": 9286
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.18992083752244e-06,
+      "loss": 0.2073,
+      "step": 9287
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.187902689892666e-06,
+      "loss": 0.1672,
+      "step": 9288
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.185884666624699e-06,
+      "loss": 0.1713,
+      "step": 9289
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.183866767807787e-06,
+      "loss": 0.1921,
+      "step": 9290
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.181848993531169e-06,
+      "loss": 0.1842,
+      "step": 9291
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.179831343884081e-06,
+      "loss": 0.2033,
+      "step": 9292
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.177813818955749e-06,
+      "loss": 0.1776,
+      "step": 9293
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.1757964188353925e-06,
+      "loss": 0.1952,
+      "step": 9294
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.173779143612234e-06,
+      "loss": 0.2042,
+      "step": 9295
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.171761993375486e-06,
+      "loss": 0.1863,
+      "step": 9296
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.169744968214346e-06,
+      "loss": 0.1993,
+      "step": 9297
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.167728068218026e-06,
+      "loss": 0.1773,
+      "step": 9298
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.165711293475714e-06,
+      "loss": 0.1844,
+      "step": 9299
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.163694644076599e-06,
+      "loss": 0.1754,
+      "step": 9300
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.161678120109871e-06,
+      "loss": 0.174,
+      "step": 9301
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.1596617216647055e-06,
+      "loss": 0.1687,
+      "step": 9302
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.157645448830271e-06,
+      "loss": 0.1914,
+      "step": 9303
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.155629301695739e-06,
+      "loss": 0.2027,
+      "step": 9304
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.153613280350269e-06,
+      "loss": 0.1646,
+      "step": 9305
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.151597384883024e-06,
+      "loss": 0.1803,
+      "step": 9306
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.1495816153831495e-06,
+      "loss": 0.1959,
+      "step": 9307
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.1475659719397884e-06,
+      "loss": 0.1878,
+      "step": 9308
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.145550454642083e-06,
+      "loss": 0.1868,
+      "step": 9309
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.14353506357917e-06,
+      "loss": 0.1841,
+      "step": 9310
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.1415197988401705e-06,
+      "loss": 0.1983,
+      "step": 9311
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.139504660514217e-06,
+      "loss": 0.1841,
+      "step": 9312
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.137489648690419e-06,
+      "loss": 0.1851,
+      "step": 9313
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.135474763457887e-06,
+      "loss": 0.1841,
+      "step": 9314
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.133460004905735e-06,
+      "loss": 0.1641,
+      "step": 9315
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.1314453731230605e-06,
+      "loss": 0.1935,
+      "step": 9316
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.12943086819895e-06,
+      "loss": 0.1747,
+      "step": 9317
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.127416490222503e-06,
+      "loss": 0.182,
+      "step": 9318
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.125402239282801e-06,
+      "loss": 0.1782,
+      "step": 9319
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.123388115468916e-06,
+      "loss": 0.1784,
+      "step": 9320
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.12137411886993e-06,
+      "loss": 0.2123,
+      "step": 9321
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.1193602495749e-06,
+      "loss": 0.1988,
+      "step": 9322
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.117346507672894e-06,
+      "loss": 0.1879,
+      "step": 9323
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.115332893252967e-06,
+      "loss": 0.1687,
+      "step": 9324
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.113319406404162e-06,
+      "loss": 0.1956,
+      "step": 9325
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.111306047215534e-06,
+      "loss": 0.1735,
+      "step": 9326
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.109292815776111e-06,
+      "loss": 0.1785,
+      "step": 9327
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.107279712174929e-06,
+      "loss": 0.1689,
+      "step": 9328
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.105266736501022e-06,
+      "loss": 0.1996,
+      "step": 9329
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.103253888843404e-06,
+      "loss": 0.1808,
+      "step": 9330
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.101241169291091e-06,
+      "loss": 0.207,
+      "step": 9331
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.099228577933097e-06,
+      "loss": 0.1843,
+      "step": 9332
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.097216114858427e-06,
+      "loss": 0.1833,
+      "step": 9333
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.095203780156073e-06,
+      "loss": 0.2048,
+      "step": 9334
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.093191573915037e-06,
+      "loss": 0.1811,
+      "step": 9335
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.0911794962243e-06,
+      "loss": 0.1975,
+      "step": 9336
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.089167547172846e-06,
+      "loss": 0.176,
+      "step": 9337
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.087155726849655e-06,
+      "loss": 0.1819,
+      "step": 9338
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.0851440353436894e-06,
+      "loss": 0.1766,
+      "step": 9339
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.083132472743923e-06,
+      "loss": 0.2071,
+      "step": 9340
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.081121039139308e-06,
+      "loss": 0.165,
+      "step": 9341
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.079109734618799e-06,
+      "loss": 0.1928,
+      "step": 9342
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.077098559271348e-06,
+      "loss": 0.1939,
+      "step": 9343
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.075087513185894e-06,
+      "loss": 0.1872,
+      "step": 9344
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.073076596451371e-06,
+      "loss": 0.168,
+      "step": 9345
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.071065809156712e-06,
+      "loss": 0.1827,
+      "step": 9346
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.0690551513908424e-06,
+      "loss": 0.1859,
+      "step": 9347
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.067044623242677e-06,
+      "loss": 0.1678,
+      "step": 9348
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.065034224801137e-06,
+      "loss": 0.2025,
+      "step": 9349
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.063023956155123e-06,
+      "loss": 0.1805,
+      "step": 9350
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.061013817393536e-06,
+      "loss": 0.1804,
+      "step": 9351
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.05900380860528e-06,
+      "loss": 0.1854,
+      "step": 9352
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.056993929879241e-06,
+      "loss": 0.1693,
+      "step": 9353
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.054984181304301e-06,
+      "loss": 0.2009,
+      "step": 9354
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.05297456296934e-06,
+      "loss": 0.2024,
+      "step": 9355
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.0509650749632315e-06,
+      "loss": 0.1907,
+      "step": 9356
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.048955717374848e-06,
+      "loss": 0.22,
+      "step": 9357
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.046946490293045e-06,
+      "loss": 0.1867,
+      "step": 9358
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.044937393806677e-06,
+      "loss": 0.1953,
+      "step": 9359
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.042928428004599e-06,
+      "loss": 0.1688,
+      "step": 9360
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.040919592975655e-06,
+      "loss": 0.1909,
+      "step": 9361
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.038910888808677e-06,
+      "loss": 0.165,
+      "step": 9362
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.036902315592507e-06,
+      "loss": 0.2005,
+      "step": 9363
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.034893873415965e-06,
+      "loss": 0.1964,
+      "step": 9364
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.032885562367871e-06,
+      "loss": 0.1898,
+      "step": 9365
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.030877382537047e-06,
+      "loss": 0.1973,
+      "step": 9366
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.0288693340123e-06,
+      "loss": 0.2099,
+      "step": 9367
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.026861416882429e-06,
+      "loss": 0.1886,
+      "step": 9368
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.024853631236237e-06,
+      "loss": 0.1784,
+      "step": 9369
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.022845977162516e-06,
+      "loss": 0.1783,
+      "step": 9370
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.020838454750047e-06,
+      "loss": 0.1722,
+      "step": 9371
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.018831064087619e-06,
+      "loss": 0.1975,
+      "step": 9372
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.016823805263996e-06,
+      "loss": 0.1821,
+      "step": 9373
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.014816678367955e-06,
+      "loss": 0.1624,
+      "step": 9374
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.012809683488257e-06,
+      "loss": 0.1785,
+      "step": 9375
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.010802820713657e-06,
+      "loss": 0.1929,
+      "step": 9376
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.00879609013291e-06,
+      "loss": 0.1669,
+      "step": 9377
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.006789491834758e-06,
+      "loss": 0.1959,
+      "step": 9378
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.004783025907937e-06,
+      "loss": 0.1918,
+      "step": 9379
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.002776692441192e-06,
+      "loss": 0.1995,
+      "step": 9380
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.000770491523243e-06,
+      "loss": 0.1883,
+      "step": 9381
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.99876442324281e-06,
+      "loss": 0.1984,
+      "step": 9382
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.996758487688612e-06,
+      "loss": 0.1549,
+      "step": 9383
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.9947526849493615e-06,
+      "loss": 0.1676,
+      "step": 9384
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.992747015113758e-06,
+      "loss": 0.1791,
+      "step": 9385
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.9907414782705065e-06,
+      "loss": 0.173,
+      "step": 9386
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.9887360745082925e-06,
+      "loss": 0.1833,
+      "step": 9387
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.986730803915804e-06,
+      "loss": 0.2009,
+      "step": 9388
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.984725666581729e-06,
+      "loss": 0.2324,
+      "step": 9389
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.9827206625947305e-06,
+      "loss": 0.1753,
+      "step": 9390
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.9807157920434905e-06,
+      "loss": 0.2135,
+      "step": 9391
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.978711055016663e-06,
+      "loss": 0.1808,
+      "step": 9392
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.976706451602906e-06,
+      "loss": 0.1795,
+      "step": 9393
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.974701981890877e-06,
+      "loss": 0.1497,
+      "step": 9394
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.9726976459692174e-06,
+      "loss": 0.1737,
+      "step": 9395
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.970693443926562e-06,
+      "loss": 0.2064,
+      "step": 9396
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.968689375851551e-06,
+      "loss": 0.207,
+      "step": 9397
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.966685441832812e-06,
+      "loss": 0.2123,
+      "step": 9398
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.964681641958959e-06,
+      "loss": 0.1734,
+      "step": 9399
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.962677976318619e-06,
+      "loss": 0.1995,
+      "step": 9400
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.960674445000395e-06,
+      "loss": 0.2035,
+      "step": 9401
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.958671048092889e-06,
+      "loss": 0.1961,
+      "step": 9402
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.956667785684705e-06,
+      "loss": 0.1899,
+      "step": 9403
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.9546646578644315e-06,
+      "loss": 0.218,
+      "step": 9404
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.952661664720655e-06,
+      "loss": 0.1778,
+      "step": 9405
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.950658806341954e-06,
+      "loss": 0.1748,
+      "step": 9406
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.948656082816904e-06,
+      "loss": 0.1849,
+      "step": 9407
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.946653494234077e-06,
+      "loss": 0.1973,
+      "step": 9408
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.944651040682029e-06,
+      "loss": 0.1975,
+      "step": 9409
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.94264872224932e-06,
+      "loss": 0.2168,
+      "step": 9410
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.940646539024498e-06,
+      "loss": 0.1891,
+      "step": 9411
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.938644491096112e-06,
+      "loss": 0.1632,
+      "step": 9412
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.936642578552692e-06,
+      "loss": 0.1902,
+      "step": 9413
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.934640801482777e-06,
+      "loss": 0.1774,
+      "step": 9414
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.932639159974894e-06,
+      "loss": 0.192,
+      "step": 9415
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.930637654117556e-06,
+      "loss": 0.1956,
+      "step": 9416
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.928636283999286e-06,
+      "loss": 0.1839,
+      "step": 9417
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.926635049708587e-06,
+      "loss": 0.1582,
+      "step": 9418
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.924633951333962e-06,
+      "loss": 0.1932,
+      "step": 9419
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.9226329889639085e-06,
+      "loss": 0.1712,
+      "step": 9420
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.920632162686917e-06,
+      "loss": 0.1536,
+      "step": 9421
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.918631472591469e-06,
+      "loss": 0.1796,
+      "step": 9422
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.916630918766045e-06,
+      "loss": 0.1651,
+      "step": 9423
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.914630501299119e-06,
+      "loss": 0.1951,
+      "step": 9424
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.912630220279154e-06,
+      "loss": 0.161,
+      "step": 9425
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.910630075794613e-06,
+      "loss": 0.1992,
+      "step": 9426
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.908630067933945e-06,
+      "loss": 0.1821,
+      "step": 9427
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.9066301967856035e-06,
+      "loss": 0.1773,
+      "step": 9428
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.90463046243803e-06,
+      "loss": 0.1954,
+      "step": 9429
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.902630864979656e-06,
+      "loss": 0.1663,
+      "step": 9430
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.900631404498917e-06,
+      "loss": 0.1962,
+      "step": 9431
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.898632081084233e-06,
+      "loss": 0.1941,
+      "step": 9432
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.896632894824023e-06,
+      "loss": 0.1981,
+      "step": 9433
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.8946338458067005e-06,
+      "loss": 0.1465,
+      "step": 9434
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.8926349341206715e-06,
+      "loss": 0.1667,
+      "step": 9435
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.890636159854328e-06,
+      "loss": 0.1511,
+      "step": 9436
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.888637523096071e-06,
+      "loss": 0.2041,
+      "step": 9437
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.886639023934289e-06,
+      "loss": 0.1669,
+      "step": 9438
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.884640662457357e-06,
+      "loss": 0.1858,
+      "step": 9439
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.882642438753657e-06,
+      "loss": 0.1699,
+      "step": 9440
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.880644352911552e-06,
+      "loss": 0.2093,
+      "step": 9441
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.878646405019409e-06,
+      "loss": 0.1824,
+      "step": 9442
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.8766485951655845e-06,
+      "loss": 0.1781,
+      "step": 9443
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.874650923438427e-06,
+      "loss": 0.1944,
+      "step": 9444
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.872653389926287e-06,
+      "loss": 0.1855,
+      "step": 9445
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.870655994717497e-06,
+      "loss": 0.205,
+      "step": 9446
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.868658737900389e-06,
+      "loss": 0.1986,
+      "step": 9447
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.866661619563298e-06,
+      "loss": 0.1889,
+      "step": 9448
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.864664639794537e-06,
+      "loss": 0.1821,
+      "step": 9449
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.862667798682419e-06,
+      "loss": 0.1963,
+      "step": 9450
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.860671096315254e-06,
+      "loss": 0.1615,
+      "step": 9451
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.8586745327813485e-06,
+      "loss": 0.2128,
+      "step": 9452
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.85667810816899e-06,
+      "loss": 0.1695,
+      "step": 9453
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.854681822566478e-06,
+      "loss": 0.1947,
+      "step": 9454
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.852685676062084e-06,
+      "loss": 0.1978,
+      "step": 9455
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.850689668744095e-06,
+      "loss": 0.187,
+      "step": 9456
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.8486938007007806e-06,
+      "loss": 0.1733,
+      "step": 9457
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.8466980720204014e-06,
+      "loss": 0.1703,
+      "step": 9458
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.844702482791221e-06,
+      "loss": 0.1902,
+      "step": 9459
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.842707033101489e-06,
+      "loss": 0.1833,
+      "step": 9460
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.84071172303945e-06,
+      "loss": 0.1614,
+      "step": 9461
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.838716552693352e-06,
+      "loss": 0.2085,
+      "step": 9462
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.836721522151424e-06,
+      "loss": 0.1681,
+      "step": 9463
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.834726631501891e-06,
+      "loss": 0.1935,
+      "step": 9464
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.832731880832979e-06,
+      "loss": 0.1688,
+      "step": 9465
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.830737270232906e-06,
+      "loss": 0.1792,
+      "step": 9466
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.828742799789874e-06,
+      "loss": 0.202,
+      "step": 9467
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.826748469592093e-06,
+      "loss": 0.1877,
+      "step": 9468
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.824754279727756e-06,
+      "loss": 0.2007,
+      "step": 9469
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.822760230285053e-06,
+      "loss": 0.1574,
+      "step": 9470
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.8207663213521735e-06,
+      "loss": 0.1761,
+      "step": 9471
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.81877255301729e-06,
+      "loss": 0.1708,
+      "step": 9472
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.8167789253685815e-06,
+      "loss": 0.1749,
+      "step": 9473
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.8147854384942084e-06,
+      "loss": 0.1757,
+      "step": 9474
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.81279209248233e-06,
+      "loss": 0.2111,
+      "step": 9475
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.810798887421104e-06,
+      "loss": 0.1982,
+      "step": 9476
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.808805823398678e-06,
+      "loss": 0.1647,
+      "step": 9477
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.8068129005031846e-06,
+      "loss": 0.1868,
+      "step": 9478
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.804820118822767e-06,
+      "loss": 0.1744,
+      "step": 9479
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.802827478445553e-06,
+      "loss": 0.1893,
+      "step": 9480
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.800834979459659e-06,
+      "loss": 0.2032,
+      "step": 9481
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.798842621953208e-06,
+      "loss": 0.173,
+      "step": 9482
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.7968504060143055e-06,
+      "loss": 0.1856,
+      "step": 9483
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.794858331731054e-06,
+      "loss": 0.1939,
+      "step": 9484
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.792866399191558e-06,
+      "loss": 0.206,
+      "step": 9485
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.790874608483901e-06,
+      "loss": 0.1812,
+      "step": 9486
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.788882959696168e-06,
+      "loss": 0.1765,
+      "step": 9487
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.78689145291644e-06,
+      "loss": 0.1842,
+      "step": 9488
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.784900088232788e-06,
+      "loss": 0.2273,
+      "step": 9489
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.782908865733282e-06,
+      "loss": 0.2004,
+      "step": 9490
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.780917785505977e-06,
+      "loss": 0.2147,
+      "step": 9491
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.778926847638924e-06,
+      "loss": 0.1892,
+      "step": 9492
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.7769360522201746e-06,
+      "loss": 0.1707,
+      "step": 9493
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.7749453993377715e-06,
+      "loss": 0.198,
+      "step": 9494
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.772954889079739e-06,
+      "loss": 0.1788,
+      "step": 9495
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.770964521534118e-06,
+      "loss": 0.2043,
+      "step": 9496
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.76897429678892e-06,
+      "loss": 0.1788,
+      "step": 9497
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.766984214932164e-06,
+      "loss": 0.209,
+      "step": 9498
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.764994276051864e-06,
+      "loss": 0.1848,
+      "step": 9499
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.763004480236018e-06,
+      "loss": 0.1893,
+      "step": 9500
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.761014827572616e-06,
+      "loss": 0.1799,
+      "step": 9501
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.75902531814966e-06,
+      "loss": 0.2,
+      "step": 9502
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.75703595205513e-06,
+      "loss": 0.1667,
+      "step": 9503
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.7550467293769975e-06,
+      "loss": 0.1705,
+      "step": 9504
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.753057650203243e-06,
+      "loss": 0.1801,
+      "step": 9505
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.751068714621822e-06,
+      "loss": 0.1966,
+      "step": 9506
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.749079922720701e-06,
+      "loss": 0.2038,
+      "step": 9507
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.7470912745878295e-06,
+      "loss": 0.1931,
+      "step": 9508
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.745102770311147e-06,
+      "loss": 0.2253,
+      "step": 9509
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.7431144099786025e-06,
+      "loss": 0.1464,
+      "step": 9510
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.741126193678124e-06,
+      "loss": 0.151,
+      "step": 9511
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.7391381214976345e-06,
+      "loss": 0.1729,
+      "step": 9512
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.7371501935250636e-06,
+      "loss": 0.1947,
+      "step": 9513
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.735162409848321e-06,
+      "loss": 0.1903,
+      "step": 9514
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.733174770555309e-06,
+      "loss": 0.175,
+      "step": 9515
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.731187275733933e-06,
+      "loss": 0.1817,
+      "step": 9516
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.729199925472091e-06,
+      "loss": 0.1966,
+      "step": 9517
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.727212719857663e-06,
+      "loss": 0.1768,
+      "step": 9518
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.72522565897854e-06,
+      "loss": 0.1994,
+      "step": 9519
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.723238742922591e-06,
+      "loss": 0.1783,
+      "step": 9520
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.7212519717776845e-06,
+      "loss": 0.1789,
+      "step": 9521
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.719265345631693e-06,
+      "loss": 0.1884,
+      "step": 9522
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.717278864572458e-06,
+      "loss": 0.1814,
+      "step": 9523
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.715292528687845e-06,
+      "loss": 0.1768,
+      "step": 9524
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.713306338065686e-06,
+      "loss": 0.16,
+      "step": 9525
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.711320292793819e-06,
+      "loss": 0.1783,
+      "step": 9526
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.709334392960082e-06,
+      "loss": 0.2094,
+      "step": 9527
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.707348638652293e-06,
+      "loss": 0.1904,
+      "step": 9528
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.705363029958269e-06,
+      "loss": 0.1758,
+      "step": 9529
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.703377566965825e-06,
+      "loss": 0.1895,
+      "step": 9530
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.701392249762766e-06,
+      "loss": 0.1998,
+      "step": 9531
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.699407078436886e-06,
+      "loss": 0.1865,
+      "step": 9532
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.697422053075981e-06,
+      "loss": 0.2039,
+      "step": 9533
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.695437173767836e-06,
+      "loss": 0.1805,
+      "step": 9534
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.693452440600225e-06,
+      "loss": 0.2141,
+      "step": 9535
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.691467853660929e-06,
+      "loss": 0.1728,
+      "step": 9536
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.68948341303771e-06,
+      "loss": 0.2141,
+      "step": 9537
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.687499118818325e-06,
+      "loss": 0.185,
+      "step": 9538
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.68551497109053e-06,
+      "loss": 0.1714,
+      "step": 9539
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6835309699420705e-06,
+      "loss": 0.2015,
+      "step": 9540
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.681547115460691e-06,
+      "loss": 0.1766,
+      "step": 9541
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6795634077341195e-06,
+      "loss": 0.1697,
+      "step": 9542
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.677579846850085e-06,
+      "loss": 0.1947,
+      "step": 9543
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.67559643289631e-06,
+      "loss": 0.2102,
+      "step": 9544
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.673613165960508e-06,
+      "loss": 0.1751,
+      "step": 9545
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.671630046130383e-06,
+      "loss": 0.16,
+      "step": 9546
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.669647073493641e-06,
+      "loss": 0.1912,
+      "step": 9547
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.667664248137977e-06,
+      "loss": 0.2215,
+      "step": 9548
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.665681570151074e-06,
+      "loss": 0.2169,
+      "step": 9549
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.66369903962062e-06,
+      "loss": 0.17,
+      "step": 9550
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.661716656634285e-06,
+      "loss": 0.2062,
+      "step": 9551
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.659734421279741e-06,
+      "loss": 0.1716,
+      "step": 9552
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.657752333644648e-06,
+      "loss": 0.1906,
+      "step": 9553
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6557703938166655e-06,
+      "loss": 0.1762,
+      "step": 9554
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.653788601883433e-06,
+      "loss": 0.1792,
+      "step": 9555
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.651806957932604e-06,
+      "loss": 0.1834,
+      "step": 9556
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6498254620518084e-06,
+      "loss": 0.1798,
+      "step": 9557
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.647844114328677e-06,
+      "loss": 0.1819,
+      "step": 9558
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.645862914850835e-06,
+      "loss": 0.1869,
+      "step": 9559
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.643881863705894e-06,
+      "loss": 0.1738,
+      "step": 9560
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.641900960981466e-06,
+      "loss": 0.1925,
+      "step": 9561
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.639920206765159e-06,
+      "loss": 0.1868,
+      "step": 9562
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6379396011445584e-06,
+      "loss": 0.188,
+      "step": 9563
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.635959144207267e-06,
+      "loss": 0.1812,
+      "step": 9564
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.633978836040859e-06,
+      "loss": 0.1781,
+      "step": 9565
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.631998676732911e-06,
+      "loss": 0.1831,
+      "step": 9566
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.630018666371003e-06,
+      "loss": 0.1613,
+      "step": 9567
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.628038805042691e-06,
+      "loss": 0.1879,
+      "step": 9568
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6260590928355305e-06,
+      "loss": 0.171,
+      "step": 9569
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6240795298370765e-06,
+      "loss": 0.1665,
+      "step": 9570
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.622100116134873e-06,
+      "loss": 0.1874,
+      "step": 9571
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6201208518164515e-06,
+      "loss": 0.1849,
+      "step": 9572
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.618141736969352e-06,
+      "loss": 0.1912,
+      "step": 9573
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.616162771681089e-06,
+      "loss": 0.1629,
+      "step": 9574
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.614183956039187e-06,
+      "loss": 0.1898,
+      "step": 9575
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.6122052901311574e-06,
+      "loss": 0.2063,
+      "step": 9576
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.610226774044494e-06,
+      "loss": 0.1866,
+      "step": 9577
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.60824840786671e-06,
+      "loss": 0.1638,
+      "step": 9578
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.606270191685284e-06,
+      "loss": 0.1775,
+      "step": 9579
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.604292125587703e-06,
+      "loss": 0.1547,
+      "step": 9580
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.602314209661451e-06,
+      "loss": 0.1874,
+      "step": 9581
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.600336443993995e-06,
+      "loss": 0.1792,
+      "step": 9582
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.598358828672794e-06,
+      "loss": 0.179,
+      "step": 9583
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.5963813637853125e-06,
+      "loss": 0.1916,
+      "step": 9584
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.594404049419003e-06,
+      "loss": 0.1782,
+      "step": 9585
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.592426885661301e-06,
+      "loss": 0.194,
+      "step": 9586
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.590449872599655e-06,
+      "loss": 0.1821,
+      "step": 9587
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.588473010321488e-06,
+      "loss": 0.1917,
+      "step": 9588
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.58649629891423e-06,
+      "loss": 0.1742,
+      "step": 9589
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.5845197384652985e-06,
+      "loss": 0.1684,
+      "step": 9590
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.5825433290620986e-06,
+      "loss": 0.1782,
+      "step": 9591
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.580567070792042e-06,
+      "loss": 0.1661,
+      "step": 9592
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.578590963742523e-06,
+      "loss": 0.1823,
+      "step": 9593
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.576615008000929e-06,
+      "loss": 0.1667,
+      "step": 9594
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.5746392036546545e-06,
+      "loss": 0.1893,
+      "step": 9595
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.572663550791071e-06,
+      "loss": 0.165,
+      "step": 9596
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.570688049497544e-06,
+      "loss": 0.165,
+      "step": 9597
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.5687126998614485e-06,
+      "loss": 0.1637,
+      "step": 9598
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.566737501970137e-06,
+      "loss": 0.1748,
+      "step": 9599
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.564762455910956e-06,
+      "loss": 0.1941,
+      "step": 9600
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.56278756177126e-06,
+      "loss": 0.1808,
+      "step": 9601
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.560812819638378e-06,
+      "loss": 0.1903,
+      "step": 9602
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.55883822959964e-06,
+      "loss": 0.1623,
+      "step": 9603
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.556863791742378e-06,
+      "loss": 0.1556,
+      "step": 9604
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.554889506153902e-06,
+      "loss": 0.1752,
+      "step": 9605
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.552915372921528e-06,
+      "loss": 0.1861,
+      "step": 9606
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.550941392132555e-06,
+      "loss": 0.1914,
+      "step": 9607
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.54896756387428e-06,
+      "loss": 0.1757,
+      "step": 9608
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.546993888234e-06,
+      "loss": 0.2111,
+      "step": 9609
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.545020365298993e-06,
+      "loss": 0.1999,
+      "step": 9610
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.5430469951565336e-06,
+      "loss": 0.1974,
+      "step": 9611
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.541073777893896e-06,
+      "loss": 0.1669,
+      "step": 9612
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.539100713598345e-06,
+      "loss": 0.1854,
+      "step": 9613
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.537127802357131e-06,
+      "loss": 0.1733,
+      "step": 9614
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.53515504425751e-06,
+      "loss": 0.1824,
+      "step": 9615
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.533182439386721e-06,
+      "loss": 0.1671,
+      "step": 9616
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.531209987832e-06,
+      "loss": 0.188,
+      "step": 9617
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.529237689680582e-06,
+      "loss": 0.188,
+      "step": 9618
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.527265545019685e-06,
+      "loss": 0.2012,
+      "step": 9619
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.525293553936523e-06,
+      "loss": 0.1705,
+      "step": 9620
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.523321716518307e-06,
+      "loss": 0.1959,
+      "step": 9621
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.52135003285224e-06,
+      "loss": 0.1954,
+      "step": 9622
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.51937850302552e-06,
+      "loss": 0.187,
+      "step": 9623
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.5174071271253335e-06,
+      "loss": 0.1775,
+      "step": 9624
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.515435905238859e-06,
+      "loss": 0.1896,
+      "step": 9625
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.513464837453274e-06,
+      "loss": 0.1748,
+      "step": 9626
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.511493923855749e-06,
+      "loss": 0.1945,
+      "step": 9627
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.509523164533441e-06,
+      "loss": 0.1614,
+      "step": 9628
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.507552559573511e-06,
+      "loss": 0.1768,
+      "step": 9629
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.5055821090631e-06,
+      "loss": 0.1815,
+      "step": 9630
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.50361181308935e-06,
+      "loss": 0.1916,
+      "step": 9631
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.501641671739403e-06,
+      "loss": 0.172,
+      "step": 9632
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.499671685100377e-06,
+      "loss": 0.1863,
+      "step": 9633
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.497701853259394e-06,
+      "loss": 0.1873,
+      "step": 9634
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.49573217630357e-06,
+      "loss": 0.1761,
+      "step": 9635
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.493762654320013e-06,
+      "loss": 0.1849,
+      "step": 9636
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.491793287395817e-06,
+      "loss": 0.1673,
+      "step": 9637
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.489824075618083e-06,
+      "loss": 0.1976,
+      "step": 9638
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.487855019073887e-06,
+      "loss": 0.1789,
+      "step": 9639
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.485886117850318e-06,
+      "loss": 0.1861,
+      "step": 9640
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.483917372034447e-06,
+      "loss": 0.1736,
+      "step": 9641
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.481948781713333e-06,
+      "loss": 0.1792,
+      "step": 9642
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.479980346974042e-06,
+      "loss": 0.1868,
+      "step": 9643
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.478012067903622e-06,
+      "loss": 0.2108,
+      "step": 9644
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.476043944589115e-06,
+      "loss": 0.196,
+      "step": 9645
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.474075977117567e-06,
+      "loss": 0.1944,
+      "step": 9646
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.472108165576005e-06,
+      "loss": 0.166,
+      "step": 9647
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.47014051005145e-06,
+      "loss": 0.1934,
+      "step": 9648
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.468173010630923e-06,
+      "loss": 0.2024,
+      "step": 9649
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.466205667401435e-06,
+      "loss": 0.167,
+      "step": 9650
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.464238480449986e-06,
+      "loss": 0.1821,
+      "step": 9651
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.462271449863577e-06,
+      "loss": 0.2042,
+      "step": 9652
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.4603045757291945e-06,
+      "loss": 0.1957,
+      "step": 9653
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.458337858133821e-06,
+      "loss": 0.2123,
+      "step": 9654
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.456371297164436e-06,
+      "loss": 0.1895,
+      "step": 9655
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.454404892908004e-06,
+      "loss": 0.1997,
+      "step": 9656
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.452438645451492e-06,
+      "loss": 0.1852,
+      "step": 9657
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.45047255488185e-06,
+      "loss": 0.2026,
+      "step": 9658
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.448506621286027e-06,
+      "loss": 0.1961,
+      "step": 9659
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.446540844750968e-06,
+      "loss": 0.1714,
+      "step": 9660
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.444575225363606e-06,
+      "loss": 0.194,
+      "step": 9661
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.442609763210863e-06,
+      "loss": 0.1911,
+      "step": 9662
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.440644458379665e-06,
+      "loss": 0.19,
+      "step": 9663
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.438679310956926e-06,
+      "loss": 0.178,
+      "step": 9664
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.436714321029545e-06,
+      "loss": 0.1936,
+      "step": 9665
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.434749488684431e-06,
+      "loss": 0.2082,
+      "step": 9666
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.43278481400847e-06,
+      "loss": 0.1855,
+      "step": 9667
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.430820297088547e-06,
+      "loss": 0.1831,
+      "step": 9668
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.428855938011548e-06,
+      "loss": 0.1899,
+      "step": 9669
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.426891736864337e-06,
+      "loss": 0.1885,
+      "step": 9670
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.424927693733781e-06,
+      "loss": 0.2047,
+      "step": 9671
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.422963808706738e-06,
+      "loss": 0.1887,
+      "step": 9672
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.421000081870057e-06,
+      "loss": 0.1942,
+      "step": 9673
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.4190365133105855e-06,
+      "loss": 0.1617,
+      "step": 9674
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.417073103115157e-06,
+      "loss": 0.1916,
+      "step": 9675
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.4151098513706e-06,
+      "loss": 0.1962,
+      "step": 9676
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.41314675816374e-06,
+      "loss": 0.221,
+      "step": 9677
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.411183823581392e-06,
+      "loss": 0.2121,
+      "step": 9678
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.409221047710361e-06,
+      "loss": 0.1955,
+      "step": 9679
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.407258430637456e-06,
+      "loss": 0.2044,
+      "step": 9680
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.4052959724494625e-06,
+      "loss": 0.2066,
+      "step": 9681
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.403333673233171e-06,
+      "loss": 0.2202,
+      "step": 9682
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.4013715330753675e-06,
+      "loss": 0.2006,
+      "step": 9683
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.399409552062819e-06,
+      "loss": 0.1677,
+      "step": 9684
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.397447730282295e-06,
+      "loss": 0.1781,
+      "step": 9685
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.395486067820553e-06,
+      "loss": 0.2007,
+      "step": 9686
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.393524564764349e-06,
+      "loss": 0.1726,
+      "step": 9687
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.391563221200421e-06,
+      "loss": 0.2032,
+      "step": 9688
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.389602037215513e-06,
+      "loss": 0.1858,
+      "step": 9689
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.387641012896355e-06,
+      "loss": 0.1862,
+      "step": 9690
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.385680148329671e-06,
+      "loss": 0.1833,
+      "step": 9691
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.383719443602181e-06,
+      "loss": 0.1983,
+      "step": 9692
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.381758898800586e-06,
+      "loss": 0.2007,
+      "step": 9693
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.379798514011597e-06,
+      "loss": 0.1659,
+      "step": 9694
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.377838289321911e-06,
+      "loss": 0.1838,
+      "step": 9695
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.3758782248182084e-06,
+      "loss": 0.1696,
+      "step": 9696
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.373918320587178e-06,
+      "loss": 0.1905,
+      "step": 9697
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.371958576715491e-06,
+      "loss": 0.1667,
+      "step": 9698
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.369998993289814e-06,
+      "loss": 0.19,
+      "step": 9699
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.368039570396814e-06,
+      "loss": 0.1968,
+      "step": 9700
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.366080308123138e-06,
+      "loss": 0.1945,
+      "step": 9701
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.364121206555431e-06,
+      "loss": 0.2135,
+      "step": 9702
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.362162265780335e-06,
+      "loss": 0.166,
+      "step": 9703
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.360203485884484e-06,
+      "loss": 0.2021,
+      "step": 9704
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.358244866954496e-06,
+      "loss": 0.2012,
+      "step": 9705
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.356286409076997e-06,
+      "loss": 0.2204,
+      "step": 9706
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.35432811233859e-06,
+      "loss": 0.1853,
+      "step": 9707
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.352369976825885e-06,
+      "loss": 0.204,
+      "step": 9708
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.350412002625476e-06,
+      "loss": 0.1849,
+      "step": 9709
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.348454189823947e-06,
+      "loss": 0.1838,
+      "step": 9710
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.34649653850789e-06,
+      "loss": 0.196,
+      "step": 9711
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.3445390487638714e-06,
+      "loss": 0.2105,
+      "step": 9712
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.342581720678461e-06,
+      "loss": 0.1855,
+      "step": 9713
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.3406245543382235e-06,
+      "loss": 0.1908,
+      "step": 9714
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.338667549829711e-06,
+      "loss": 0.1533,
+      "step": 9715
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.336710707239463e-06,
+      "loss": 0.1919,
+      "step": 9716
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.334754026654026e-06,
+      "loss": 0.1926,
+      "step": 9717
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.332797508159931e-06,
+      "loss": 0.1888,
+      "step": 9718
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.3308411518436985e-06,
+      "loss": 0.1985,
+      "step": 9719
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.328884957791852e-06,
+      "loss": 0.1747,
+      "step": 9720
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.326928926090896e-06,
+      "loss": 0.1759,
+      "step": 9721
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.3249730568273395e-06,
+      "loss": 0.172,
+      "step": 9722
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.323017350087677e-06,
+      "loss": 0.1901,
+      "step": 9723
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.321061805958393e-06,
+      "loss": 0.1955,
+      "step": 9724
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.319106424525977e-06,
+      "loss": 0.1929,
+      "step": 9725
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.317151205876897e-06,
+      "loss": 0.2001,
+      "step": 9726
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.31519615009762e-06,
+      "loss": 0.1675,
+      "step": 9727
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.313241257274612e-06,
+      "loss": 0.1777,
+      "step": 9728
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.311286527494325e-06,
+      "loss": 0.1945,
+      "step": 9729
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.309331960843195e-06,
+      "loss": 0.1955,
+      "step": 9730
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.3073775574076725e-06,
+      "loss": 0.1816,
+      "step": 9731
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.305423317274184e-06,
+      "loss": 0.1827,
+      "step": 9732
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.30346924052915e-06,
+      "loss": 0.1795,
+      "step": 9733
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.3015153272589955e-06,
+      "loss": 0.169,
+      "step": 9734
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.299561577550124e-06,
+      "loss": 0.1771,
+      "step": 9735
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.2976079914889366e-06,
+      "loss": 0.1642,
+      "step": 9736
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.295654569161835e-06,
+      "loss": 0.1921,
+      "step": 9737
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.293701310655199e-06,
+      "loss": 0.1983,
+      "step": 9738
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.291748216055418e-06,
+      "loss": 0.1945,
+      "step": 9739
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.289795285448859e-06,
+      "loss": 0.2001,
+      "step": 9740
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.287842518921887e-06,
+      "loss": 0.201,
+      "step": 9741
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.28588991656087e-06,
+      "loss": 0.1808,
+      "step": 9742
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.283937478452151e-06,
+      "loss": 0.1751,
+      "step": 9743
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.281985204682075e-06,
+      "loss": 0.2093,
+      "step": 9744
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.280033095336981e-06,
+      "loss": 0.2006,
+      "step": 9745
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.278081150503203e-06,
+      "loss": 0.171,
+      "step": 9746
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.276129370267055e-06,
+      "loss": 0.1705,
+      "step": 9747
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.274177754714859e-06,
+      "loss": 0.1819,
+      "step": 9748
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.2722263039329205e-06,
+      "loss": 0.1823,
+      "step": 9749
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.270275018007537e-06,
+      "loss": 0.1838,
+      "step": 9750
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.268323897025008e-06,
+      "loss": 0.2026,
+      "step": 9751
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.266372941071619e-06,
+      "loss": 0.1703,
+      "step": 9752
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.264422150233642e-06,
+      "loss": 0.1674,
+      "step": 9753
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.262471524597354e-06,
+      "loss": 0.1674,
+      "step": 9754
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.2605210642490165e-06,
+      "loss": 0.1744,
+      "step": 9755
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.258570769274894e-06,
+      "loss": 0.1714,
+      "step": 9756
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.256620639761228e-06,
+      "loss": 0.1871,
+      "step": 9757
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.254670675794261e-06,
+      "loss": 0.1963,
+      "step": 9758
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.2527208774602325e-06,
+      "loss": 0.1695,
+      "step": 9759
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.2507712448453675e-06,
+      "loss": 0.1825,
+      "step": 9760
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.248821778035883e-06,
+      "loss": 0.1783,
+      "step": 9761
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.2468724771180004e-06,
+      "loss": 0.1746,
+      "step": 9762
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.244923342177916e-06,
+      "loss": 0.1722,
+      "step": 9763
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.242974373301832e-06,
+      "loss": 0.1675,
+      "step": 9764
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.241025570575944e-06,
+      "loss": 0.1829,
+      "step": 9765
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.239076934086431e-06,
+      "loss": 0.1795,
+      "step": 9766
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.237128463919465e-06,
+      "loss": 0.1918,
+      "step": 9767
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.235180160161222e-06,
+      "loss": 0.1944,
+      "step": 9768
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.233232022897862e-06,
+      "loss": 0.1724,
+      "step": 9769
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.231284052215535e-06,
+      "loss": 0.1824,
+      "step": 9770
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.229336248200395e-06,
+      "loss": 0.2063,
+      "step": 9771
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.227388610938572e-06,
+      "loss": 0.1762,
+      "step": 9772
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.225441140516206e-06,
+      "loss": 0.1799,
+      "step": 9773
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.223493837019421e-06,
+      "loss": 0.1956,
+      "step": 9774
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.221546700534329e-06,
+      "loss": 0.1965,
+      "step": 9775
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.2195997311470455e-06,
+      "loss": 0.2024,
+      "step": 9776
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.21765292894367e-06,
+      "loss": 0.2255,
+      "step": 9777
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.215706294010296e-06,
+      "loss": 0.1831,
+      "step": 9778
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.213759826433018e-06,
+      "loss": 0.1689,
+      "step": 9779
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.211813526297912e-06,
+      "loss": 0.1871,
+      "step": 9780
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.209867393691046e-06,
+      "loss": 0.2156,
+      "step": 9781
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.207921428698493e-06,
+      "loss": 0.1781,
+      "step": 9782
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.205975631406309e-06,
+      "loss": 0.1962,
+      "step": 9783
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.204030001900541e-06,
+      "loss": 0.2109,
+      "step": 9784
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.2020845402672394e-06,
+      "loss": 0.1912,
+      "step": 9785
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.200139246592434e-06,
+      "loss": 0.1601,
+      "step": 9786
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.198194120962152e-06,
+      "loss": 0.1868,
+      "step": 9787
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.196249163462422e-06,
+      "loss": 0.1612,
+      "step": 9788
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.19430437417925e-06,
+      "loss": 0.2014,
+      "step": 9789
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.192359753198648e-06,
+      "loss": 0.1979,
+      "step": 9790
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.19041530060661e-06,
+      "loss": 0.2072,
+      "step": 9791
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.188471016489127e-06,
+      "loss": 0.1889,
+      "step": 9792
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.186526900932189e-06,
+      "loss": 0.2134,
+      "step": 9793
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.184582954021769e-06,
+      "loss": 0.187,
+      "step": 9794
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.18263917584383e-06,
+      "loss": 0.1723,
+      "step": 9795
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.1806955664843396e-06,
+      "loss": 0.2059,
+      "step": 9796
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.1787521260292525e-06,
+      "loss": 0.2038,
+      "step": 9797
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.1768088545645085e-06,
+      "loss": 0.1827,
+      "step": 9798
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.174865752176057e-06,
+      "loss": 0.2057,
+      "step": 9799
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.172922818949821e-06,
+      "loss": 0.1854,
+      "step": 9800
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.170980054971724e-06,
+      "loss": 0.1912,
+      "step": 9801
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.169037460327691e-06,
+      "loss": 0.1742,
+      "step": 9802
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.167095035103623e-06,
+      "loss": 0.1991,
+      "step": 9803
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.1651527793854235e-06,
+      "loss": 0.1738,
+      "step": 9804
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.16321069325899e-06,
+      "loss": 0.1951,
+      "step": 9805
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.161268776810202e-06,
+      "loss": 0.1905,
+      "step": 9806
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.159327030124947e-06,
+      "loss": 0.2044,
+      "step": 9807
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.157385453289091e-06,
+      "loss": 0.1924,
+      "step": 9808
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.155444046388499e-06,
+      "loss": 0.1711,
+      "step": 9809
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.153502809509029e-06,
+      "loss": 0.1883,
+      "step": 9810
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.151561742736531e-06,
+      "loss": 0.1876,
+      "step": 9811
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.149620846156838e-06,
+      "loss": 0.1893,
+      "step": 9812
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.147680119855796e-06,
+      "loss": 0.1883,
+      "step": 9813
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.1457395639192255e-06,
+      "loss": 0.1822,
+      "step": 9814
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.143799178432942e-06,
+      "loss": 0.1834,
+      "step": 9815
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.1418589634827655e-06,
+      "loss": 0.2008,
+      "step": 9816
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.139918919154492e-06,
+      "loss": 0.1888,
+      "step": 9817
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.137979045533921e-06,
+      "loss": 0.1835,
+      "step": 9818
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.13603934270684e-06,
+      "loss": 0.1828,
+      "step": 9819
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.134099810759033e-06,
+      "loss": 0.2091,
+      "step": 9820
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.132160449776269e-06,
+      "loss": 0.1724,
+      "step": 9821
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.130221259844318e-06,
+      "loss": 0.1839,
+      "step": 9822
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.128282241048936e-06,
+      "loss": 0.1817,
+      "step": 9823
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.126343393475875e-06,
+      "loss": 0.1803,
+      "step": 9824
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.124404717210883e-06,
+      "loss": 0.1903,
+      "step": 9825
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.122466212339684e-06,
+      "loss": 0.1751,
+      "step": 9826
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.120527878948017e-06,
+      "loss": 0.1781,
+      "step": 9827
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.1185897171216e-06,
+      "loss": 0.1967,
+      "step": 9828
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.1166517269461416e-06,
+      "loss": 0.1637,
+      "step": 9829
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.114713908507353e-06,
+      "loss": 0.1863,
+      "step": 9830
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.112776261890929e-06,
+      "loss": 0.1919,
+      "step": 9831
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.110838787182558e-06,
+      "loss": 0.1657,
+      "step": 9832
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.108901484467928e-06,
+      "loss": 0.1812,
+      "step": 9833
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.106964353832713e-06,
+      "loss": 0.1965,
+      "step": 9834
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.105027395362573e-06,
+      "loss": 0.1942,
+      "step": 9835
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.103090609143174e-06,
+      "loss": 0.1949,
+      "step": 9836
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.1011539952601715e-06,
+      "loss": 0.1751,
+      "step": 9837
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.099217553799201e-06,
+      "loss": 0.162,
+      "step": 9838
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0972812848459084e-06,
+      "loss": 0.1745,
+      "step": 9839
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.095345188485914e-06,
+      "loss": 0.1765,
+      "step": 9840
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.093409264804846e-06,
+      "loss": 0.1884,
+      "step": 9841
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.09147351388832e-06,
+      "loss": 0.1857,
+      "step": 9842
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0895379358219345e-06,
+      "loss": 0.1942,
+      "step": 9843
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.087602530691298e-06,
+      "loss": 0.2056,
+      "step": 9844
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.085667298581993e-06,
+      "loss": 0.181,
+      "step": 9845
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.083732239579605e-06,
+      "loss": 0.1885,
+      "step": 9846
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.081797353769716e-06,
+      "loss": 0.1878,
+      "step": 9847
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.079862641237888e-06,
+      "loss": 0.2083,
+      "step": 9848
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.077928102069679e-06,
+      "loss": 0.1898,
+      "step": 9849
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.075993736350648e-06,
+      "loss": 0.1773,
+      "step": 9850
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.07405954416634e-06,
+      "loss": 0.1766,
+      "step": 9851
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.072125525602286e-06,
+      "loss": 0.1989,
+      "step": 9852
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.070191680744024e-06,
+      "loss": 0.1853,
+      "step": 9853
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.068258009677069e-06,
+      "loss": 0.19,
+      "step": 9854
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.066324512486939e-06,
+      "loss": 0.1689,
+      "step": 9855
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.064391189259142e-06,
+      "loss": 0.1881,
+      "step": 9856
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.062458040079171e-06,
+      "loss": 0.1973,
+      "step": 9857
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0605250650325265e-06,
+      "loss": 0.1994,
+      "step": 9858
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.058592264204685e-06,
+      "loss": 0.1734,
+      "step": 9859
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.056659637681121e-06,
+      "loss": 0.1957,
+      "step": 9860
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.054727185547313e-06,
+      "loss": 0.1721,
+      "step": 9861
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.052794907888714e-06,
+      "loss": 0.1782,
+      "step": 9862
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.050862804790772e-06,
+      "loss": 0.187,
+      "step": 9863
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.048930876338941e-06,
+      "loss": 0.2026,
+      "step": 9864
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0469991226186575e-06,
+      "loss": 0.1848,
+      "step": 9865
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0450675437153435e-06,
+      "loss": 0.1778,
+      "step": 9866
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.04313613971443e-06,
+      "loss": 0.1904,
+      "step": 9867
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.041204910701326e-06,
+      "loss": 0.1722,
+      "step": 9868
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.039273856761436e-06,
+      "loss": 0.2058,
+      "step": 9869
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0373429779801655e-06,
+      "loss": 0.1896,
+      "step": 9870
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0354122744428975e-06,
+      "loss": 0.2039,
+      "step": 9871
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.033481746235026e-06,
+      "loss": 0.1911,
+      "step": 9872
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.031551393441914e-06,
+      "loss": 0.1846,
+      "step": 9873
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.029621216148935e-06,
+      "loss": 0.162,
+      "step": 9874
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.027691214441454e-06,
+      "loss": 0.1862,
+      "step": 9875
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.025761388404815e-06,
+      "loss": 0.1729,
+      "step": 9876
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.023831738124363e-06,
+      "loss": 0.2054,
+      "step": 9877
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.021902263685437e-06,
+      "loss": 0.188,
+      "step": 9878
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.019972965173371e-06,
+      "loss": 0.1808,
+      "step": 9879
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.018043842673474e-06,
+      "loss": 0.1587,
+      "step": 9880
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0161148962710706e-06,
+      "loss": 0.1914,
+      "step": 9881
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.01418612605146e-06,
+      "loss": 0.1869,
+      "step": 9882
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.012257532099939e-06,
+      "loss": 0.1844,
+      "step": 9883
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.010329114501804e-06,
+      "loss": 0.1827,
+      "step": 9884
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.008400873342333e-06,
+      "loss": 0.189,
+      "step": 9885
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.006472808706797e-06,
+      "loss": 0.1991,
+      "step": 9886
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.0045449206804675e-06,
+      "loss": 0.2017,
+      "step": 9887
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.002617209348599e-06,
+      "loss": 0.1857,
+      "step": 9888
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.0006896747964494e-06,
+      "loss": 0.1942,
+      "step": 9889
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.998762317109257e-06,
+      "loss": 0.1931,
+      "step": 9890
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.996835136372252e-06,
+      "loss": 0.1819,
+      "step": 9891
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.994908132670672e-06,
+      "loss": 0.1783,
+      "step": 9892
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.99298130608973e-06,
+      "loss": 0.1989,
+      "step": 9893
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9910546567146366e-06,
+      "loss": 0.1851,
+      "step": 9894
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.989128184630603e-06,
+      "loss": 0.168,
+      "step": 9895
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.987201889922818e-06,
+      "loss": 0.1867,
+      "step": 9896
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.985275772676468e-06,
+      "loss": 0.2149,
+      "step": 9897
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9833498329767445e-06,
+      "loss": 0.1689,
+      "step": 9898
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9814240709088124e-06,
+      "loss": 0.1691,
+      "step": 9899
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.979498486557833e-06,
+      "loss": 0.1843,
+      "step": 9900
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.977573080008969e-06,
+      "loss": 0.1719,
+      "step": 9901
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.97564785134737e-06,
+      "loss": 0.1852,
+      "step": 9902
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.973722800658171e-06,
+      "loss": 0.175,
+      "step": 9903
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9717979280265115e-06,
+      "loss": 0.2026,
+      "step": 9904
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.969873233537511e-06,
+      "loss": 0.1731,
+      "step": 9905
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.967948717276291e-06,
+      "loss": 0.1982,
+      "step": 9906
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.966024379327962e-06,
+      "loss": 0.2271,
+      "step": 9907
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.96410021977762e-06,
+      "loss": 0.1945,
+      "step": 9908
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9621762387103666e-06,
+      "loss": 0.1981,
+      "step": 9909
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9602524362112815e-06,
+      "loss": 0.1962,
+      "step": 9910
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9583288123654415e-06,
+      "loss": 0.2008,
+      "step": 9911
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.956405367257926e-06,
+      "loss": 0.1801,
+      "step": 9912
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.95448210097379e-06,
+      "loss": 0.1665,
+      "step": 9913
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.952559013598085e-06,
+      "loss": 0.1945,
+      "step": 9914
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.950636105215862e-06,
+      "loss": 0.1753,
+      "step": 9915
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9487133759121616e-06,
+      "loss": 0.2048,
+      "step": 9916
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9467908257720075e-06,
+      "loss": 0.1998,
+      "step": 9917
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.94486845488043e-06,
+      "loss": 0.1756,
+      "step": 9918
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.942946263322438e-06,
+      "loss": 0.191,
+      "step": 9919
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9410242511830366e-06,
+      "loss": 0.1969,
+      "step": 9920
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.939102418547234e-06,
+      "loss": 0.188,
+      "step": 9921
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.93718076550001e-06,
+      "loss": 0.2157,
+      "step": 9922
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9352592921263586e-06,
+      "loss": 0.1844,
+      "step": 9923
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.933337998511245e-06,
+      "loss": 0.1945,
+      "step": 9924
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.931416884739638e-06,
+      "loss": 0.1917,
+      "step": 9925
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.929495950896504e-06,
+      "loss": 0.1767,
+      "step": 9926
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.927575197066788e-06,
+      "loss": 0.1885,
+      "step": 9927
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.92565462333543e-06,
+      "loss": 0.2073,
+      "step": 9928
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.923734229787372e-06,
+      "loss": 0.1846,
+      "step": 9929
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.921814016507541e-06,
+      "loss": 0.1724,
+      "step": 9930
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.919893983580849e-06,
+      "loss": 0.1986,
+      "step": 9931
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.917974131092216e-06,
+      "loss": 0.2272,
+      "step": 9932
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.916054459126539e-06,
+      "loss": 0.1657,
+      "step": 9933
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.914134967768714e-06,
+      "loss": 0.1858,
+      "step": 9934
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.912215657103634e-06,
+      "loss": 0.206,
+      "step": 9935
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.910296527216173e-06,
+      "loss": 0.1873,
+      "step": 9936
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.908377578191202e-06,
+      "loss": 0.1661,
+      "step": 9937
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.906458810113587e-06,
+      "loss": 0.1671,
+      "step": 9938
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.90454022306818e-06,
+      "loss": 0.1963,
+      "step": 9939
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.902621817139834e-06,
+      "loss": 0.1426,
+      "step": 9940
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.900703592413386e-06,
+      "loss": 0.1941,
+      "step": 9941
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.8987855489736625e-06,
+      "loss": 0.1903,
+      "step": 9942
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.896867686905492e-06,
+      "loss": 0.1868,
+      "step": 9943
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.89495000629369e-06,
+      "loss": 0.1913,
+      "step": 9944
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.893032507223057e-06,
+      "loss": 0.1627,
+      "step": 9945
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.891115189778405e-06,
+      "loss": 0.2068,
+      "step": 9946
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.889198054044512e-06,
+      "loss": 0.1802,
+      "step": 9947
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.887281100106165e-06,
+      "loss": 0.1624,
+      "step": 9948
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.885364328048147e-06,
+      "loss": 0.1872,
+      "step": 9949
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.8834477379552144e-06,
+      "loss": 0.2041,
+      "step": 9950
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.8815313299121325e-06,
+      "loss": 0.1869,
+      "step": 9951
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.87961510400365e-06,
+      "loss": 0.188,
+      "step": 9952
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.877699060314511e-06,
+      "loss": 0.1795,
+      "step": 9953
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.875783198929447e-06,
+      "loss": 0.2014,
+      "step": 9954
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.873867519933188e-06,
+      "loss": 0.2106,
+      "step": 9955
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.8719520234104535e-06,
+      "loss": 0.1893,
+      "step": 9956
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.8700367094459525e-06,
+      "loss": 0.1723,
+      "step": 9957
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.868121578124388e-06,
+      "loss": 0.2059,
+      "step": 9958
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.866206629530453e-06,
+      "loss": 0.1778,
+      "step": 9959
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.864291863748837e-06,
+      "loss": 0.1861,
+      "step": 9960
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.8623772808642164e-06,
+      "loss": 0.2008,
+      "step": 9961
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.86046288096126e-06,
+      "loss": 0.1772,
+      "step": 9962
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.858548664124634e-06,
+      "loss": 0.1879,
+      "step": 9963
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.856634630438989e-06,
+      "loss": 0.1931,
+      "step": 9964
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.8547207799889695e-06,
+      "loss": 0.1787,
+      "step": 9965
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.85280711285922e-06,
+      "loss": 0.1695,
+      "step": 9966
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.850893629134366e-06,
+      "loss": 0.1989,
+      "step": 9967
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.848980328899026e-06,
+      "loss": 0.2129,
+      "step": 9968
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.847067212237819e-06,
+      "loss": 0.1727,
+      "step": 9969
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.84515427923535e-06,
+      "loss": 0.2254,
+      "step": 9970
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.843241529976211e-06,
+      "loss": 0.1629,
+      "step": 9971
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.841328964544998e-06,
+      "loss": 0.2096,
+      "step": 9972
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.839416583026285e-06,
+      "loss": 0.168,
+      "step": 9973
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.8375043855046535e-06,
+      "loss": 0.1934,
+      "step": 9974
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.83559237206466e-06,
+      "loss": 0.1774,
+      "step": 9975
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.833680542790865e-06,
+      "loss": 0.1734,
+      "step": 9976
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.831768897767816e-06,
+      "loss": 0.18,
+      "step": 9977
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.829857437080055e-06,
+      "loss": 0.1773,
+      "step": 9978
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.827946160812113e-06,
+      "loss": 0.153,
+      "step": 9979
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.826035069048512e-06,
+      "loss": 0.1875,
+      "step": 9980
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.824124161873772e-06,
+      "loss": 0.1792,
+      "step": 9981
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.822213439372398e-06,
+      "loss": 0.1696,
+      "step": 9982
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.820302901628892e-06,
+      "loss": 0.1813,
+      "step": 9983
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.818392548727745e-06,
+      "loss": 0.1781,
+      "step": 9984
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.816482380753431e-06,
+      "loss": 0.1919,
+      "step": 9985
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.814572397790437e-06,
+      "loss": 0.1744,
+      "step": 9986
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.81266259992323e-06,
+      "loss": 0.1791,
+      "step": 9987
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.8107529872362554e-06,
+      "loss": 0.2084,
+      "step": 9988
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.808843559813981e-06,
+      "loss": 0.1853,
+      "step": 9989
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.806934317740833e-06,
+      "loss": 0.1845,
+      "step": 9990
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.8050252611012555e-06,
+      "loss": 0.2155,
+      "step": 9991
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.803116389979677e-06,
+      "loss": 0.1832,
+      "step": 9992
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.801207704460502e-06,
+      "loss": 0.1765,
+      "step": 9993
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.799299204628156e-06,
+      "loss": 0.1762,
+      "step": 9994
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.797390890567027e-06,
+      "loss": 0.2518,
+      "step": 9995
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.795482762361511e-06,
+      "loss": 0.1912,
+      "step": 9996
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7935748200960015e-06,
+      "loss": 0.1821,
+      "step": 9997
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.791667063854864e-06,
+      "loss": 0.1872,
+      "step": 9998
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.789759493722473e-06,
+      "loss": 0.1652,
+      "step": 9999
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.787852109783188e-06,
+      "loss": 0.1803,
+      "step": 10000
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.785944912121357e-06,
+      "loss": 0.1718,
+      "step": 10001
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.784037900821329e-06,
+      "loss": 0.1972,
+      "step": 10002
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.782131075967436e-06,
+      "loss": 0.1929,
+      "step": 10003
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7802244376440065e-06,
+      "loss": 0.1755,
+      "step": 10004
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.778317985935358e-06,
+      "loss": 0.1504,
+      "step": 10005
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.776411720925803e-06,
+      "loss": 0.174,
+      "step": 10006
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7745056426996436e-06,
+      "loss": 0.187,
+      "step": 10007
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.772599751341172e-06,
+      "loss": 0.1674,
+      "step": 10008
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7706940469346774e-06,
+      "loss": 0.1892,
+      "step": 10009
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.768788529564436e-06,
+      "loss": 0.1852,
+      "step": 10010
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.766883199314715e-06,
+      "loss": 0.1714,
+      "step": 10011
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7649780562697765e-06,
+      "loss": 0.2111,
+      "step": 10012
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.763073100513875e-06,
+      "loss": 0.181,
+      "step": 10013
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.761168332131255e-06,
+      "loss": 0.2068,
+      "step": 10014
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.759263751206155e-06,
+      "loss": 0.1766,
+      "step": 10015
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.757359357822791e-06,
+      "loss": 0.1669,
+      "step": 10016
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7554551520653945e-06,
+      "loss": 0.1711,
+      "step": 10017
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.753551134018179e-06,
+      "loss": 0.1903,
+      "step": 10018
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.751647303765334e-06,
+      "loss": 0.221,
+      "step": 10019
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.749743661391071e-06,
+      "loss": 0.1774,
+      "step": 10020
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.747840206979558e-06,
+      "loss": 0.1916,
+      "step": 10021
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7459369406149934e-06,
+      "loss": 0.1884,
+      "step": 10022
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7440338623815305e-06,
+      "loss": 0.2264,
+      "step": 10023
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.742130972363334e-06,
+      "loss": 0.2009,
+      "step": 10024
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.740228270644566e-06,
+      "loss": 0.1792,
+      "step": 10025
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.738325757309362e-06,
+      "loss": 0.1673,
+      "step": 10026
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7364234324418625e-06,
+      "loss": 0.1759,
+      "step": 10027
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7345212961261944e-06,
+      "loss": 0.1861,
+      "step": 10028
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.732619348446477e-06,
+      "loss": 0.1703,
+      "step": 10029
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.730717589486823e-06,
+      "loss": 0.1681,
+      "step": 10030
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7288160193313354e-06,
+      "loss": 0.1849,
+      "step": 10031
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7269146380641095e-06,
+      "loss": 0.1705,
+      "step": 10032
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.725013445769231e-06,
+      "loss": 0.1942,
+      "step": 10033
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7231124425307774e-06,
+      "loss": 0.1836,
+      "step": 10034
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.721211628432819e-06,
+      "loss": 0.1889,
+      "step": 10035
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.719311003559417e-06,
+      "loss": 0.1969,
+      "step": 10036
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.717410567994625e-06,
+      "loss": 0.1824,
+      "step": 10037
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.715510321822486e-06,
+      "loss": 0.2302,
+      "step": 10038
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.713610265127039e-06,
+      "loss": 0.1688,
+      "step": 10039
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.7117103979923095e-06,
+      "loss": 0.1668,
+      "step": 10040
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.709810720502317e-06,
+      "loss": 0.184,
+      "step": 10041
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.7079112327410735e-06,
+      "loss": 0.1776,
+      "step": 10042
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.706011934792586e-06,
+      "loss": 0.1987,
+      "step": 10043
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.704112826740835e-06,
+      "loss": 0.1752,
+      "step": 10044
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.702213908669821e-06,
+      "loss": 0.1789,
+      "step": 10045
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.700315180663521e-06,
+      "loss": 0.1898,
+      "step": 10046
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.698416642805889e-06,
+      "loss": 0.2223,
+      "step": 10047
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.696518295180904e-06,
+      "loss": 0.1669,
+      "step": 10048
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.694620137872508e-06,
+      "loss": 0.1949,
+      "step": 10049
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.692722170964642e-06,
+      "loss": 0.1903,
+      "step": 10050
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.690824394541255e-06,
+      "loss": 0.1859,
+      "step": 10051
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.688926808686259e-06,
+      "loss": 0.1758,
+      "step": 10052
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.687029413483583e-06,
+      "loss": 0.1903,
+      "step": 10053
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.68513220901713e-06,
+      "loss": 0.1867,
+      "step": 10054
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.683235195370802e-06,
+      "loss": 0.174,
+      "step": 10055
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.681338372628502e-06,
+      "loss": 0.1645,
+      "step": 10056
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.679441740874105e-06,
+      "loss": 0.1695,
+      "step": 10057
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.677545300191491e-06,
+      "loss": 0.1755,
+      "step": 10058
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.675649050664527e-06,
+      "loss": 0.1713,
+      "step": 10059
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.673752992377073e-06,
+      "loss": 0.1983,
+      "step": 10060
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.6718571254129795e-06,
+      "loss": 0.186,
+      "step": 10061
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.66996144985609e-06,
+      "loss": 0.1911,
+      "step": 10062
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.668065965790237e-06,
+      "loss": 0.1843,
+      "step": 10063
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.666170673299247e-06,
+      "loss": 0.1709,
+      "step": 10064
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.664275572466937e-06,
+      "loss": 0.1831,
+      "step": 10065
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.66238066337712e-06,
+      "loss": 0.1764,
+      "step": 10066
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.660485946113586e-06,
+      "loss": 0.1855,
+      "step": 10067
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.658591420760136e-06,
+      "loss": 0.2033,
+      "step": 10068
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.656697087400553e-06,
+      "loss": 0.2049,
+      "step": 10069
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.654802946118602e-06,
+      "loss": 0.1915,
+      "step": 10070
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.652908996998063e-06,
+      "loss": 0.1936,
+      "step": 10071
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.651015240122682e-06,
+      "loss": 0.1749,
+      "step": 10072
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.6491216755762134e-06,
+      "loss": 0.1747,
+      "step": 10073
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.6472283034424025e-06,
+      "loss": 0.2083,
+      "step": 10074
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.645335123804969e-06,
+      "loss": 0.1958,
+      "step": 10075
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.643442136747652e-06,
+      "loss": 0.18,
+      "step": 10076
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.641549342354155e-06,
+      "loss": 0.2,
+      "step": 10077
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.639656740708186e-06,
+      "loss": 0.1857,
+      "step": 10078
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.637764331893454e-06,
+      "loss": 0.1749,
+      "step": 10079
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.635872115993634e-06,
+      "loss": 0.1914,
+      "step": 10080
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.633980093092415e-06,
+      "loss": 0.1874,
+      "step": 10081
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.632088263273469e-06,
+      "loss": 0.1687,
+      "step": 10082
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.630196626620458e-06,
+      "loss": 0.1745,
+      "step": 10083
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.6283051832170385e-06,
+      "loss": 0.1988,
+      "step": 10084
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.626413933146858e-06,
+      "loss": 0.2104,
+      "step": 10085
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.624522876493555e-06,
+      "loss": 0.1927,
+      "step": 10086
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.622632013340757e-06,
+      "loss": 0.1958,
+      "step": 10087
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.620741343772089e-06,
+      "loss": 0.1923,
+      "step": 10088
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.618850867871161e-06,
+      "loss": 0.1991,
+      "step": 10089
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.616960585721578e-06,
+      "loss": 0.1745,
+      "step": 10090
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.615070497406936e-06,
+      "loss": 0.1497,
+      "step": 10091
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.613180603010819e-06,
+      "loss": 0.1854,
+      "step": 10092
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.6112909026168104e-06,
+      "loss": 0.1567,
+      "step": 10093
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.609401396308482e-06,
+      "loss": 0.1838,
+      "step": 10094
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.607512084169382e-06,
+      "loss": 0.1827,
+      "step": 10095
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.605622966283075e-06,
+      "loss": 0.2279,
+      "step": 10096
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.603734042733107e-06,
+      "loss": 0.1945,
+      "step": 10097
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.601845313602999e-06,
+      "loss": 0.1766,
+      "step": 10098
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.599956778976296e-06,
+      "loss": 0.1866,
+      "step": 10099
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.598068438936503e-06,
+      "loss": 0.2029,
+      "step": 10100
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.596180293567132e-06,
+      "loss": 0.1644,
+      "step": 10101
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.594292342951693e-06,
+      "loss": 0.1718,
+      "step": 10102
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.592404587173668e-06,
+      "loss": 0.204,
+      "step": 10103
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.5905170263165445e-06,
+      "loss": 0.2012,
+      "step": 10104
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.588629660463798e-06,
+      "loss": 0.1812,
+      "step": 10105
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.586742489698891e-06,
+      "loss": 0.1852,
+      "step": 10106
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.584855514105294e-06,
+      "loss": 0.2339,
+      "step": 10107
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.5829687337664426e-06,
+      "loss": 0.1691,
+      "step": 10108
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.581082148765784e-06,
+      "loss": 0.1801,
+      "step": 10109
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.579195759186748e-06,
+      "loss": 0.1879,
+      "step": 10110
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.577309565112761e-06,
+      "loss": 0.1805,
+      "step": 10111
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.575423566627234e-06,
+      "loss": 0.1851,
+      "step": 10112
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.573537763813574e-06,
+      "loss": 0.1883,
+      "step": 10113
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.571652156755183e-06,
+      "loss": 0.2084,
+      "step": 10114
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.569766745535443e-06,
+      "loss": 0.176,
+      "step": 10115
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.5678815302377395e-06,
+      "loss": 0.1916,
+      "step": 10116
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.565996510945446e-06,
+      "loss": 0.1818,
+      "step": 10117
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.564111687741912e-06,
+      "loss": 0.2067,
+      "step": 10118
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.562227060710506e-06,
+      "loss": 0.1986,
+      "step": 10119
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.5603426299345716e-06,
+      "loss": 0.2079,
+      "step": 10120
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.5584583954974355e-06,
+      "loss": 0.1543,
+      "step": 10121
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.556574357482441e-06,
+      "loss": 0.1907,
+      "step": 10122
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.55469051597289e-06,
+      "loss": 0.1759,
+      "step": 10123
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.552806871052107e-06,
+      "loss": 0.1834,
+      "step": 10124
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.550923422803396e-06,
+      "loss": 0.1796,
+      "step": 10125
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.549040171310034e-06,
+      "loss": 0.1744,
+      "step": 10126
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.547157116655324e-06,
+      "loss": 0.2002,
+      "step": 10127
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.545274258922529e-06,
+      "loss": 0.1685,
+      "step": 10128
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.543391598194917e-06,
+      "loss": 0.1856,
+      "step": 10129
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.541509134555759e-06,
+      "loss": 0.2033,
+      "step": 10130
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.539626868088293e-06,
+      "loss": 0.17,
+      "step": 10131
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.537744798875761e-06,
+      "loss": 0.1701,
+      "step": 10132
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.535862927001401e-06,
+      "loss": 0.1627,
+      "step": 10133
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.53398125254843e-06,
+      "loss": 0.1852,
+      "step": 10134
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.532099775600068e-06,
+      "loss": 0.188,
+      "step": 10135
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.53021849623952e-06,
+      "loss": 0.187,
+      "step": 10136
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.528337414549982e-06,
+      "loss": 0.1876,
+      "step": 10137
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.5264565306146435e-06,
+      "loss": 0.2165,
+      "step": 10138
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.524575844516686e-06,
+      "loss": 0.1578,
+      "step": 10139
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.522695356339278e-06,
+      "loss": 0.1746,
+      "step": 10140
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.520815066165584e-06,
+      "loss": 0.1891,
+      "step": 10141
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.518934974078755e-06,
+      "loss": 0.1883,
+      "step": 10142
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.51705508016194e-06,
+      "loss": 0.1894,
+      "step": 10143
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.515175384498273e-06,
+      "loss": 0.1814,
+      "step": 10144
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.51329588717088e-06,
+      "loss": 0.1935,
+      "step": 10145
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.511416588262881e-06,
+      "loss": 0.182,
+      "step": 10146
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.5095374878573884e-06,
+      "loss": 0.1784,
+      "step": 10147
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.507658586037503e-06,
+      "loss": 0.1974,
+      "step": 10148
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.505779882886306e-06,
+      "loss": 0.1911,
+      "step": 10149
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.503901378486895e-06,
+      "loss": 0.1822,
+      "step": 10150
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.502023072922344e-06,
+      "loss": 0.2214,
+      "step": 10151
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.500144966275707e-06,
+      "loss": 0.2126,
+      "step": 10152
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.4982670586300545e-06,
+      "loss": 0.1835,
+      "step": 10153
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.496389350068422e-06,
+      "loss": 0.1736,
+      "step": 10154
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.494511840673864e-06,
+      "loss": 0.1655,
+      "step": 10155
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.492634530529399e-06,
+      "loss": 0.2057,
+      "step": 10156
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.490757419718049e-06,
+      "loss": 0.1891,
+      "step": 10157
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.48888050832284e-06,
+      "loss": 0.1938,
+      "step": 10158
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.487003796426763e-06,
+      "loss": 0.1851,
+      "step": 10159
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.4851272841128185e-06,
+      "loss": 0.1667,
+      "step": 10160
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.483250971463992e-06,
+      "loss": 0.169,
+      "step": 10161
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.481374858563262e-06,
+      "loss": 0.1741,
+      "step": 10162
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.479498945493596e-06,
+      "loss": 0.192,
+      "step": 10163
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.477623232337957e-06,
+      "loss": 0.1981,
+      "step": 10164
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.475747719179294e-06,
+      "loss": 0.2045,
+      "step": 10165
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.473872406100551e-06,
+      "loss": 0.1852,
+      "step": 10166
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.471997293184661e-06,
+      "loss": 0.2002,
+      "step": 10167
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.470122380514548e-06,
+      "loss": 0.1937,
+      "step": 10168
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.468247668173128e-06,
+      "loss": 0.1605,
+      "step": 10169
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.466373156243307e-06,
+      "loss": 0.1543,
+      "step": 10170
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.464498844807987e-06,
+      "loss": 0.1827,
+      "step": 10171
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.462624733950055e-06,
+      "loss": 0.1882,
+      "step": 10172
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.46075082375239e-06,
+      "loss": 0.162,
+      "step": 10173
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.458877114297864e-06,
+      "loss": 0.1819,
+      "step": 10174
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.457003605669342e-06,
+      "loss": 0.1778,
+      "step": 10175
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.45513029794968e-06,
+      "loss": 0.1532,
+      "step": 10176
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.453257191221711e-06,
+      "loss": 0.1821,
+      "step": 10177
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.451384285568284e-06,
+      "loss": 0.201,
+      "step": 10178
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.4495115810722244e-06,
+      "loss": 0.2099,
+      "step": 10179
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.44763907781634e-06,
+      "loss": 0.1953,
+      "step": 10180
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.445766775883454e-06,
+      "loss": 0.2049,
+      "step": 10181
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.4438946753563586e-06,
+      "loss": 0.1896,
+      "step": 10182
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.442022776317842e-06,
+      "loss": 0.1691,
+      "step": 10183
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.4401510788506996e-06,
+      "loss": 0.1822,
+      "step": 10184
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.438279583037693e-06,
+      "loss": 0.1835,
+      "step": 10185
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.436408288961592e-06,
+      "loss": 0.1952,
+      "step": 10186
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.434537196705152e-06,
+      "loss": 0.208,
+      "step": 10187
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.4326663063511155e-06,
+      "loss": 0.2104,
+      "step": 10188
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.430795617982234e-06,
+      "loss": 0.227,
+      "step": 10189
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.428925131681222e-06,
+      "loss": 0.1778,
+      "step": 10190
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.427054847530804e-06,
+      "loss": 0.1906,
+      "step": 10191
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.425184765613694e-06,
+      "loss": 0.1773,
+      "step": 10192
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.4233148860125905e-06,
+      "loss": 0.1725,
+      "step": 10193
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.4214452088101895e-06,
+      "loss": 0.1763,
+      "step": 10194
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.419575734089176e-06,
+      "loss": 0.1852,
+      "step": 10195
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.417706461932222e-06,
+      "loss": 0.1998,
+      "step": 10196
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.415837392421996e-06,
+      "loss": 0.1857,
+      "step": 10197
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.413968525641156e-06,
+      "loss": 0.1934,
+      "step": 10198
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.412099861672353e-06,
+      "loss": 0.1836,
+      "step": 10199
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.410231400598216e-06,
+      "loss": 0.1667,
+      "step": 10200
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.408363142501387e-06,
+      "loss": 0.1725,
+      "step": 10201
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.406495087464487e-06,
+      "loss": 0.201,
+      "step": 10202
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.404627235570118e-06,
+      "loss": 0.1843,
+      "step": 10203
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.402759586900898e-06,
+      "loss": 0.1745,
+      "step": 10204
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.400892141539406e-06,
+      "loss": 0.193,
+      "step": 10205
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.3990248995682415e-06,
+      "loss": 0.1779,
+      "step": 10206
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.3971578610699794e-06,
+      "loss": 0.2065,
+      "step": 10207
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.395291026127175e-06,
+      "loss": 0.196,
+      "step": 10208
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.393424394822404e-06,
+      "loss": 0.1836,
+      "step": 10209
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.391557967238203e-06,
+      "loss": 0.169,
+      "step": 10210
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.389691743457115e-06,
+      "loss": 0.1499,
+      "step": 10211
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.387825723561682e-06,
+      "loss": 0.17,
+      "step": 10212
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.3859599076344126e-06,
+      "loss": 0.1821,
+      "step": 10213
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.384094295757828e-06,
+      "loss": 0.1872,
+      "step": 10214
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.3822288880144306e-06,
+      "loss": 0.1982,
+      "step": 10215
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.3803636844867155e-06,
+      "loss": 0.1788,
+      "step": 10216
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.3784986852571715e-06,
+      "loss": 0.1802,
+      "step": 10217
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.3766338904082735e-06,
+      "loss": 0.1817,
+      "step": 10218
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.374769300022494e-06,
+      "loss": 0.2066,
+      "step": 10219
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.372904914182286e-06,
+      "loss": 0.2145,
+      "step": 10220
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.371040732970105e-06,
+      "loss": 0.1914,
+      "step": 10221
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.369176756468391e-06,
+      "loss": 0.2051,
+      "step": 10222
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.367312984759576e-06,
+      "loss": 0.1814,
+      "step": 10223
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.365449417926083e-06,
+      "loss": 0.1796,
+      "step": 10224
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.363586056050325e-06,
+      "loss": 0.1891,
+      "step": 10225
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.361722899214709e-06,
+      "loss": 0.1828,
+      "step": 10226
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.3598599475016356e-06,
+      "loss": 0.1799,
+      "step": 10227
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.357997200993479e-06,
+      "loss": 0.1737,
+      "step": 10228
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.356134659772627e-06,
+      "loss": 0.144,
+      "step": 10229
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.35427232392145e-06,
+      "loss": 0.1764,
+      "step": 10230
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.352410193522295e-06,
+      "loss": 0.1807,
+      "step": 10231
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.35054826865753e-06,
+      "loss": 0.1756,
+      "step": 10232
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.348686549409483e-06,
+      "loss": 0.1938,
+      "step": 10233
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.346825035860488e-06,
+      "loss": 0.1576,
+      "step": 10234
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.344963728092879e-06,
+      "loss": 0.173,
+      "step": 10235
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.343102626188958e-06,
+      "loss": 0.1972,
+      "step": 10236
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.341241730231036e-06,
+      "loss": 0.1843,
+      "step": 10237
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.339381040301408e-06,
+      "loss": 0.1893,
+      "step": 10238
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.337520556482355e-06,
+      "loss": 0.1671,
+      "step": 10239
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.33566027885617e-06,
+      "loss": 0.1857,
+      "step": 10240
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.333800207505106e-06,
+      "loss": 0.184,
+      "step": 10241
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.331940342511431e-06,
+      "loss": 0.1984,
+      "step": 10242
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.330080683957392e-06,
+      "loss": 0.2058,
+      "step": 10243
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.328221231925232e-06,
+      "loss": 0.1526,
+      "step": 10244
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.3263619864971816e-06,
+      "loss": 0.1731,
+      "step": 10245
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.3245029477554655e-06,
+      "loss": 0.2013,
+      "step": 10246
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.3226441157822965e-06,
+      "loss": 0.1764,
+      "step": 10247
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.32078549065988e-06,
+      "loss": 0.1827,
+      "step": 10248
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.3189270724704116e-06,
+      "loss": 0.1893,
+      "step": 10249
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.317068861296079e-06,
+      "loss": 0.1781,
+      "step": 10250
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.315210857219053e-06,
+      "loss": 0.169,
+      "step": 10251
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.3133530603215085e-06,
+      "loss": 0.1717,
+      "step": 10252
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.311495470685608e-06,
+      "loss": 0.1854,
+      "step": 10253
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.3096380883934875e-06,
+      "loss": 0.1865,
+      "step": 10254
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.3077809135273025e-06,
+      "loss": 0.1729,
+      "step": 10255
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.305923946169172e-06,
+      "loss": 0.1893,
+      "step": 10256
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.304067186401226e-06,
+      "loss": 0.1906,
+      "step": 10257
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.30221063430558e-06,
+      "loss": 0.1851,
+      "step": 10258
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.3003542899643265e-06,
+      "loss": 0.1825,
+      "step": 10259
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.298498153459577e-06,
+      "loss": 0.1616,
+      "step": 10260
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.2966422248734e-06,
+      "loss": 0.1625,
+      "step": 10261
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.294786504287878e-06,
+      "loss": 0.1472,
+      "step": 10262
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.292930991785086e-06,
+      "loss": 0.1913,
+      "step": 10263
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.291075687447071e-06,
+      "loss": 0.2214,
+      "step": 10264
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.2892205913558885e-06,
+      "loss": 0.1755,
+      "step": 10265
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.287365703593573e-06,
+      "loss": 0.192,
+      "step": 10266
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.285511024242158e-06,
+      "loss": 0.1673,
+      "step": 10267
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.283656553383665e-06,
+      "loss": 0.1478,
+      "step": 10268
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.281802291100104e-06,
+      "loss": 0.1413,
+      "step": 10269
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.2799482374734775e-06,
+      "loss": 0.1735,
+      "step": 10270
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.278094392585782e-06,
+      "loss": 0.1451,
+      "step": 10271
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.276240756518998e-06,
+      "loss": 0.1737,
+      "step": 10272
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.274387329355103e-06,
+      "loss": 0.1584,
+      "step": 10273
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.2725341111760606e-06,
+      "loss": 0.1698,
+      "step": 10274
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.2706811020638295e-06,
+      "loss": 0.1735,
+      "step": 10275
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.2688283021003554e-06,
+      "loss": 0.1639,
+      "step": 10276
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.266975711367577e-06,
+      "loss": 0.1786,
+      "step": 10277
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.265123329947422e-06,
+      "loss": 0.1704,
+      "step": 10278
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.26327115792181e-06,
+      "loss": 0.1722,
+      "step": 10279
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.261419195372653e-06,
+      "loss": 0.1698,
+      "step": 10280
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.259567442381854e-06,
+      "loss": 0.1806,
+      "step": 10281
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.257715899031295e-06,
+      "loss": 0.1724,
+      "step": 10282
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.255864565402871e-06,
+      "loss": 0.1789,
+      "step": 10283
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.254013441578446e-06,
+      "loss": 0.1598,
+      "step": 10284
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.252162527639882e-06,
+      "loss": 0.1812,
+      "step": 10285
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.250311823669047e-06,
+      "loss": 0.1501,
+      "step": 10286
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.24846132974777e-06,
+      "loss": 0.2013,
+      "step": 10287
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.246611045957901e-06,
+      "loss": 0.1682,
+      "step": 10288
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.244760972381258e-06,
+      "loss": 0.1505,
+      "step": 10289
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.242911109099656e-06,
+      "loss": 0.1993,
+      "step": 10290
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.2410614561949155e-06,
+      "loss": 0.166,
+      "step": 10291
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.239212013748825e-06,
+      "loss": 0.1563,
+      "step": 10292
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.237362781843175e-06,
+      "loss": 0.1373,
+      "step": 10293
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.235513760559747e-06,
+      "loss": 0.1852,
+      "step": 10294
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.233664949980314e-06,
+      "loss": 0.1691,
+      "step": 10295
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.231816350186633e-06,
+      "loss": 0.143,
+      "step": 10296
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.229967961260463e-06,
+      "loss": 0.1451,
+      "step": 10297
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.2281197832835395e-06,
+      "loss": 0.181,
+      "step": 10298
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.2262718163376e-06,
+      "loss": 0.1599,
+      "step": 10299
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.224424060504369e-06,
+      "loss": 0.1597,
+      "step": 10300
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.22257651586556e-06,
+      "loss": 0.1629,
+      "step": 10301
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.220729182502879e-06,
+      "loss": 0.1539,
+      "step": 10302
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.218882060498023e-06,
+      "loss": 0.1866,
+      "step": 10303
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.217035149932679e-06,
+      "loss": 0.1807,
+      "step": 10304
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.215188450888522e-06,
+      "loss": 0.1831,
+      "step": 10305
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.213341963447223e-06,
+      "loss": 0.1614,
+      "step": 10306
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.211495687690439e-06,
+      "loss": 0.1866,
+      "step": 10307
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.20964962369982e-06,
+      "loss": 0.1703,
+      "step": 10308
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.20780377155701e-06,
+      "loss": 0.1572,
+      "step": 10309
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.205958131343629e-06,
+      "loss": 0.1891,
+      "step": 10310
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.204112703141309e-06,
+      "loss": 0.1843,
+      "step": 10311
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.2022674870316626e-06,
+      "loss": 0.1881,
+      "step": 10312
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.200422483096279e-06,
+      "loss": 0.174,
+      "step": 10313
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.19857769141677e-06,
+      "loss": 0.1728,
+      "step": 10314
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.1967331120747045e-06,
+      "loss": 0.1427,
+      "step": 10315
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.194888745151658e-06,
+      "loss": 0.1455,
+      "step": 10316
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.193044590729209e-06,
+      "loss": 0.1639,
+      "step": 10317
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.1912006488889e-06,
+      "loss": 0.1487,
+      "step": 10318
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.189356919712281e-06,
+      "loss": 0.1624,
+      "step": 10319
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.187513403280888e-06,
+      "loss": 0.1451,
+      "step": 10320
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.185670099676247e-06,
+      "loss": 0.1754,
+      "step": 10321
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.183827008979885e-06,
+      "loss": 0.1548,
+      "step": 10322
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.181984131273301e-06,
+      "loss": 0.1728,
+      "step": 10323
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.180141466637998e-06,
+      "loss": 0.1728,
+      "step": 10324
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.178299015155463e-06,
+      "loss": 0.1891,
+      "step": 10325
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.176456776907179e-06,
+      "loss": 0.1767,
+      "step": 10326
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.174614751974617e-06,
+      "loss": 0.1771,
+      "step": 10327
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.172772940439237e-06,
+      "loss": 0.1627,
+      "step": 10328
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.1709313423824924e-06,
+      "loss": 0.1677,
+      "step": 10329
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.169089957885825e-06,
+      "loss": 0.1808,
+      "step": 10330
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.167248787030668e-06,
+      "loss": 0.1533,
+      "step": 10331
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.165407829898448e-06,
+      "loss": 0.1832,
+      "step": 10332
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.163567086570569e-06,
+      "loss": 0.1656,
+      "step": 10333
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.1617265571284455e-06,
+      "loss": 0.1692,
+      "step": 10334
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.1598862416534765e-06,
+      "loss": 0.1702,
+      "step": 10335
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.158046140227033e-06,
+      "loss": 0.1661,
+      "step": 10336
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.156206252930509e-06,
+      "loss": 0.1726,
+      "step": 10337
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.1543665798452535e-06,
+      "loss": 0.1673,
+      "step": 10338
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.152527121052639e-06,
+      "loss": 0.169,
+      "step": 10339
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.150687876634009e-06,
+      "loss": 0.1757,
+      "step": 10340
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.148848846670695e-06,
+      "loss": 0.1649,
+      "step": 10341
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.1470100312440375e-06,
+      "loss": 0.1319,
+      "step": 10342
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.145171430435347e-06,
+      "loss": 0.1706,
+      "step": 10343
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.143333044325934e-06,
+      "loss": 0.1675,
+      "step": 10344
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.141494872997107e-06,
+      "loss": 0.1568,
+      "step": 10345
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.139656916530149e-06,
+      "loss": 0.1938,
+      "step": 10346
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.137819175006344e-06,
+      "loss": 0.1635,
+      "step": 10347
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.1359816485069644e-06,
+      "loss": 0.1704,
+      "step": 10348
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.134144337113274e-06,
+      "loss": 0.1632,
+      "step": 10349
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.132307240906523e-06,
+      "loss": 0.1529,
+      "step": 10350
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.130470359967956e-06,
+      "loss": 0.182,
+      "step": 10351
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.128633694378806e-06,
+      "loss": 0.2073,
+      "step": 10352
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.126797244220301e-06,
+      "loss": 0.1862,
+      "step": 10353
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.1249610095736535e-06,
+      "loss": 0.1357,
+      "step": 10354
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.123124990520068e-06,
+      "loss": 0.1835,
+      "step": 10355
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.121289187140742e-06,
+      "loss": 0.1691,
+      "step": 10356
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.119453599516862e-06,
+      "loss": 0.1898,
+      "step": 10357
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.117618227729601e-06,
+      "loss": 0.1717,
+      "step": 10358
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.115783071860132e-06,
+      "loss": 0.1769,
+      "step": 10359
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.113948131989614e-06,
+      "loss": 0.1822,
+      "step": 10360
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.1121134081991816e-06,
+      "loss": 0.1526,
+      "step": 10361
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.110278900569987e-06,
+      "loss": 0.1789,
+      "step": 10362
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.108444609183161e-06,
+      "loss": 0.2155,
+      "step": 10363
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.106610534119807e-06,
+      "loss": 0.1859,
+      "step": 10364
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.104776675461053e-06,
+      "loss": 0.1871,
+      "step": 10365
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.102943033287988e-06,
+      "loss": 0.1413,
+      "step": 10366
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.101109607681702e-06,
+      "loss": 0.1923,
+      "step": 10367
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.099276398723287e-06,
+      "loss": 0.172,
+      "step": 10368
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.097443406493805e-06,
+      "loss": 0.1704,
+      "step": 10369
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.095610631074321e-06,
+      "loss": 0.1847,
+      "step": 10370
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.093778072545886e-06,
+      "loss": 0.1705,
+      "step": 10371
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.091945730989541e-06,
+      "loss": 0.191,
+      "step": 10372
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.090113606486331e-06,
+      "loss": 0.1665,
+      "step": 10373
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.088281699117266e-06,
+      "loss": 0.1723,
+      "step": 10374
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.086450008963364e-06,
+      "loss": 0.1726,
+      "step": 10375
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.08461853610563e-06,
+      "loss": 0.1478,
+      "step": 10376
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.0827872806250615e-06,
+      "loss": 0.1505,
+      "step": 10377
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.080956242602639e-06,
+      "loss": 0.1918,
+      "step": 10378
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.079125422119342e-06,
+      "loss": 0.1719,
+      "step": 10379
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.077294819256133e-06,
+      "loss": 0.143,
+      "step": 10380
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.075464434093972e-06,
+      "loss": 0.1743,
+      "step": 10381
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.073634266713803e-06,
+      "loss": 0.193,
+      "step": 10382
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.071804317196568e-06,
+      "loss": 0.1393,
+      "step": 10383
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.069974585623183e-06,
+      "loss": 0.1808,
+      "step": 10384
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.068145072074576e-06,
+      "loss": 0.1603,
+      "step": 10385
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.066315776631658e-06,
+      "loss": 0.1355,
+      "step": 10386
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.064486699375313e-06,
+      "loss": 0.1713,
+      "step": 10387
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.0626578403864454e-06,
+      "loss": 0.1492,
+      "step": 10388
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.060829199745921e-06,
+      "loss": 0.167,
+      "step": 10389
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.059000777534619e-06,
+      "loss": 0.1493,
+      "step": 10390
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.0571725738334025e-06,
+      "loss": 0.1649,
+      "step": 10391
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.055344588723105e-06,
+      "loss": 0.164,
+      "step": 10392
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.053516822284589e-06,
+      "loss": 0.1812,
+      "step": 10393
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.051689274598668e-06,
+      "loss": 0.2182,
+      "step": 10394
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.049861945746166e-06,
+      "loss": 0.1538,
+      "step": 10395
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.048034835807904e-06,
+      "loss": 0.1811,
+      "step": 10396
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.046207944864676e-06,
+      "loss": 0.1716,
+      "step": 10397
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.044381272997274e-06,
+      "loss": 0.1777,
+      "step": 10398
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.042554820286484e-06,
+      "loss": 0.1674,
+      "step": 10399
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.040728586813077e-06,
+      "loss": 0.1692,
+      "step": 10400
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.038902572657816e-06,
+      "loss": 0.1628,
+      "step": 10401
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.037076777901454e-06,
+      "loss": 0.1753,
+      "step": 10402
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.035251202624736e-06,
+      "loss": 0.1786,
+      "step": 10403
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.033425846908395e-06,
+      "loss": 0.1649,
+      "step": 10404
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.031600710833156e-06,
+      "loss": 0.1871,
+      "step": 10405
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.029775794479732e-06,
+      "loss": 0.1655,
+      "step": 10406
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.027951097928832e-06,
+      "loss": 0.1759,
+      "step": 10407
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.0261266212611444e-06,
+      "loss": 0.172,
+      "step": 10408
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.02430236455736e-06,
+      "loss": 0.1618,
+      "step": 10409
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.022478327898155e-06,
+      "loss": 0.184,
+      "step": 10410
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.020654511364191e-06,
+      "loss": 0.172,
+      "step": 10411
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.018830915036126e-06,
+      "loss": 0.1902,
+      "step": 10412
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.0170075389946086e-06,
+      "loss": 0.144,
+      "step": 10413
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.015184383320277e-06,
+      "loss": 0.1551,
+      "step": 10414
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.013361448093748e-06,
+      "loss": 0.1712,
+      "step": 10415
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.0115387333956525e-06,
+      "loss": 0.1636,
+      "step": 10416
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.009716239306588e-06,
+      "loss": 0.1546,
+      "step": 10417
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.007893965907151e-06,
+      "loss": 0.1745,
+      "step": 10418
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.006071913277943e-06,
+      "loss": 0.1947,
+      "step": 10419
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.004250081499524e-06,
+      "loss": 0.1753,
+      "step": 10420
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.00242847065248e-06,
+      "loss": 0.1781,
+      "step": 10421
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.000607080817359e-06,
+      "loss": 0.1722,
+      "step": 10422
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.998785912074708e-06,
+      "loss": 0.1677,
+      "step": 10423
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.996964964505077e-06,
+      "loss": 0.1948,
+      "step": 10424
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.995144238188986e-06,
+      "loss": 0.1394,
+      "step": 10425
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.993323733206957e-06,
+      "loss": 0.1622,
+      "step": 10426
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.9915034496395e-06,
+      "loss": 0.1654,
+      "step": 10427
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.9896833875671155e-06,
+      "loss": 0.157,
+      "step": 10428
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.987863547070293e-06,
+      "loss": 0.1994,
+      "step": 10429
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.986043928229512e-06,
+      "loss": 0.1685,
+      "step": 10430
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.9842245311252455e-06,
+      "loss": 0.1688,
+      "step": 10431
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.982405355837953e-06,
+      "loss": 0.15,
+      "step": 10432
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.9805864024480845e-06,
+      "loss": 0.1784,
+      "step": 10433
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.978767671036083e-06,
+      "loss": 0.1714,
+      "step": 10434
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.976949161682379e-06,
+      "loss": 0.1795,
+      "step": 10435
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.975130874467393e-06,
+      "loss": 0.163,
+      "step": 10436
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.973312809471538e-06,
+      "loss": 0.1577,
+      "step": 10437
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.971494966775216e-06,
+      "loss": 0.1869,
+      "step": 10438
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.969677346458816e-06,
+      "loss": 0.16,
+      "step": 10439
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.967859948602726e-06,
+      "loss": 0.1654,
+      "step": 10440
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.966042773287313e-06,
+      "loss": 0.1578,
+      "step": 10441
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.964225820592946e-06,
+      "loss": 0.1946,
+      "step": 10442
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.962409090599965e-06,
+      "loss": 0.1715,
+      "step": 10443
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.960592583388727e-06,
+      "loss": 0.1598,
+      "step": 10444
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.95877629903956e-06,
+      "loss": 0.1686,
+      "step": 10445
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.956960237632781e-06,
+      "loss": 0.1812,
+      "step": 10446
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.955144399248714e-06,
+      "loss": 0.1534,
+      "step": 10447
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.953328783967653e-06,
+      "loss": 0.1917,
+      "step": 10448
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.9515133918698934e-06,
+      "loss": 0.1569,
+      "step": 10449
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.949698223035727e-06,
+      "loss": 0.1786,
+      "step": 10450
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.947883277545418e-06,
+      "loss": 0.1869,
+      "step": 10451
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.946068555479233e-06,
+      "loss": 0.2122,
+      "step": 10452
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.944254056917427e-06,
+      "loss": 0.1704,
+      "step": 10453
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.94243978194024e-06,
+      "loss": 0.1971,
+      "step": 10454
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.940625730627918e-06,
+      "loss": 0.1838,
+      "step": 10455
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.938811903060673e-06,
+      "loss": 0.1583,
+      "step": 10456
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.936998299318725e-06,
+      "loss": 0.1762,
+      "step": 10457
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.935184919482276e-06,
+      "loss": 0.1816,
+      "step": 10458
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.933371763631523e-06,
+      "loss": 0.1676,
+      "step": 10459
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.931558831846649e-06,
+      "loss": 0.153,
+      "step": 10460
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.92974612420783e-06,
+      "loss": 0.1581,
+      "step": 10461
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.927933640795231e-06,
+      "loss": 0.1706,
+      "step": 10462
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.926121381689006e-06,
+      "loss": 0.1635,
+      "step": 10463
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.9243093469693016e-06,
+      "loss": 0.1561,
+      "step": 10464
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.9224975367162556e-06,
+      "loss": 0.1631,
+      "step": 10465
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.920685951009983e-06,
+      "loss": 0.1824,
+      "step": 10466
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.918874589930608e-06,
+      "loss": 0.1882,
+      "step": 10467
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.917063453558238e-06,
+      "loss": 0.163,
+      "step": 10468
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.915252541972958e-06,
+      "loss": 0.1741,
+      "step": 10469
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.9134418552548664e-06,
+      "loss": 0.1757,
+      "step": 10470
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.9116313934840255e-06,
+      "loss": 0.1841,
+      "step": 10471
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.909821156740511e-06,
+      "loss": 0.153,
+      "step": 10472
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.9080111451043785e-06,
+      "loss": 0.1731,
+      "step": 10473
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.906201358655663e-06,
+      "loss": 0.1788,
+      "step": 10474
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.904391797474416e-06,
+      "loss": 0.1578,
+      "step": 10475
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.902582461640652e-06,
+      "loss": 0.1886,
+      "step": 10476
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.9007733512343835e-06,
+      "loss": 0.1595,
+      "step": 10477
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.898964466335631e-06,
+      "loss": 0.1619,
+      "step": 10478
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.897155807024378e-06,
+      "loss": 0.178,
+      "step": 10479
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.895347373380615e-06,
+      "loss": 0.1938,
+      "step": 10480
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.893539165484318e-06,
+      "loss": 0.1717,
+      "step": 10481
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.891731183415451e-06,
+      "loss": 0.1582,
+      "step": 10482
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.8899234272539734e-06,
+      "loss": 0.1801,
+      "step": 10483
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.888115897079827e-06,
+      "loss": 0.1768,
+      "step": 10484
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.8863085929729494e-06,
+      "loss": 0.1856,
+      "step": 10485
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.884501515013268e-06,
+      "loss": 0.1483,
+      "step": 10486
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.882694663280697e-06,
+      "loss": 0.18,
+      "step": 10487
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.880888037855144e-06,
+      "loss": 0.1873,
+      "step": 10488
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.879081638816503e-06,
+      "loss": 0.1495,
+      "step": 10489
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.877275466244662e-06,
+      "loss": 0.168,
+      "step": 10490
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.875469520219495e-06,
+      "loss": 0.1819,
+      "step": 10491
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.873663800820869e-06,
+      "loss": 0.183,
+      "step": 10492
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.871858308128644e-06,
+      "loss": 0.1592,
+      "step": 10493
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.8700530422226535e-06,
+      "loss": 0.1708,
+      "step": 10494
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.868248003182745e-06,
+      "loss": 0.1762,
+      "step": 10495
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.866443191088745e-06,
+      "loss": 0.188,
+      "step": 10496
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.864638606020456e-06,
+      "loss": 0.1659,
+      "step": 10497
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.862834248057703e-06,
+      "loss": 0.1607,
+      "step": 10498
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.861030117280265e-06,
+      "loss": 0.1604,
+      "step": 10499
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.859226213767931e-06,
+      "loss": 0.1792,
+      "step": 10500
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.857422537600488e-06,
+      "loss": 0.1766,
+      "step": 10501
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.855619088857689e-06,
+      "loss": 0.1656,
+      "step": 10502
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.853815867619294e-06,
+      "loss": 0.1644,
+      "step": 10503
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.8520128739650465e-06,
+      "loss": 0.1636,
+      "step": 10504
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.85021010797468e-06,
+      "loss": 0.1572,
+      "step": 10505
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.848407569727932e-06,
+      "loss": 0.1703,
+      "step": 10506
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.846605259304503e-06,
+      "loss": 0.1648,
+      "step": 10507
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.844803176784106e-06,
+      "loss": 0.1727,
+      "step": 10508
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.843001322246434e-06,
+      "loss": 0.1418,
+      "step": 10509
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.841199695771171e-06,
+      "loss": 0.161,
+      "step": 10510
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.839398297437993e-06,
+      "loss": 0.1701,
+      "step": 10511
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.8375971273265646e-06,
+      "loss": 0.1662,
+      "step": 10512
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.8357961855165406e-06,
+      "loss": 0.1561,
+      "step": 10513
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.833995472087565e-06,
+      "loss": 0.1534,
+      "step": 10514
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.832194987119274e-06,
+      "loss": 0.1574,
+      "step": 10515
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.830394730691295e-06,
+      "loss": 0.1534,
+      "step": 10516
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.8285947028832295e-06,
+      "loss": 0.1742,
+      "step": 10517
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.8267949037746946e-06,
+      "loss": 0.1746,
+      "step": 10518
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.824995333445284e-06,
+      "loss": 0.1672,
+      "step": 10519
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.82319599197457e-06,
+      "loss": 0.1617,
+      "step": 10520
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.821396879442141e-06,
+      "loss": 0.1727,
+      "step": 10521
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.819597995927548e-06,
+      "loss": 0.1675,
+      "step": 10522
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.817799341510353e-06,
+      "loss": 0.1469,
+      "step": 10523
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.8160009162701015e-06,
+      "loss": 0.1659,
+      "step": 10524
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.814202720286315e-06,
+      "loss": 0.1637,
+      "step": 10525
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.812404753638531e-06,
+      "loss": 0.1477,
+      "step": 10526
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.810607016406252e-06,
+      "loss": 0.1829,
+      "step": 10527
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.8088095086689805e-06,
+      "loss": 0.1613,
+      "step": 10528
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.8070122305062195e-06,
+      "loss": 0.1515,
+      "step": 10529
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.805215181997442e-06,
+      "loss": 0.1416,
+      "step": 10530
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.803418363222125e-06,
+      "loss": 0.1958,
+      "step": 10531
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.801621774259729e-06,
+      "loss": 0.1763,
+      "step": 10532
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.799825415189706e-06,
+      "loss": 0.1593,
+      "step": 10533
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.7980292860914975e-06,
+      "loss": 0.1694,
+      "step": 10534
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.796233387044538e-06,
+      "loss": 0.1553,
+      "step": 10535
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.794437718128247e-06,
+      "loss": 0.1926,
+      "step": 10536
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.792642279422035e-06,
+      "loss": 0.1502,
+      "step": 10537
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.790847071005306e-06,
+      "loss": 0.1847,
+      "step": 10538
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.789052092957448e-06,
+      "loss": 0.1727,
+      "step": 10539
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.787257345357844e-06,
+      "loss": 0.1616,
+      "step": 10540
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.785462828285865e-06,
+      "loss": 0.1682,
+      "step": 10541
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.7836685418208695e-06,
+      "loss": 0.1653,
+      "step": 10542
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.781874486042209e-06,
+      "loss": 0.157,
+      "step": 10543
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.780080661029227e-06,
+      "loss": 0.1937,
+      "step": 10544
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.7782870668612436e-06,
+      "loss": 0.1754,
+      "step": 10545
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.7764937036175874e-06,
+      "loss": 0.1446,
+      "step": 10546
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.774700571377568e-06,
+      "loss": 0.1861,
+      "step": 10547
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.772907670220476e-06,
+      "loss": 0.1552,
+      "step": 10548
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.7711150002256125e-06,
+      "loss": 0.1931,
+      "step": 10549
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.769322561472246e-06,
+      "loss": 0.1708,
+      "step": 10550
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.767530354039646e-06,
+      "loss": 0.1633,
+      "step": 10551
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.765738378007081e-06,
+      "loss": 0.1495,
+      "step": 10552
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.763946633453785e-06,
+      "loss": 0.1693,
+      "step": 10553
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.762155120459009e-06,
+      "loss": 0.1861,
+      "step": 10554
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.760363839101971e-06,
+      "loss": 0.1832,
+      "step": 10555
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.7585727894618895e-06,
+      "loss": 0.1696,
+      "step": 10556
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.75678197161798e-06,
+      "loss": 0.1565,
+      "step": 10557
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.75499138564943e-06,
+      "loss": 0.1881,
+      "step": 10558
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.75320103163543e-06,
+      "loss": 0.1764,
+      "step": 10559
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.751410909655155e-06,
+      "loss": 0.1503,
+      "step": 10560
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.749621019787773e-06,
+      "loss": 0.1749,
+      "step": 10561
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.747831362112438e-06,
+      "loss": 0.19,
+      "step": 10562
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.746041936708298e-06,
+      "loss": 0.1681,
+      "step": 10563
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.744252743654485e-06,
+      "loss": 0.135,
+      "step": 10564
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.742463783030126e-06,
+      "loss": 0.1462,
+      "step": 10565
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.740675054914335e-06,
+      "loss": 0.1696,
+      "step": 10566
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.738886559386219e-06,
+      "loss": 0.1697,
+      "step": 10567
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.737098296524868e-06,
+      "loss": 0.1739,
+      "step": 10568
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.7353102664093685e-06,
+      "loss": 0.1626,
+      "step": 10569
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.733522469118794e-06,
+      "loss": 0.1588,
+      "step": 10570
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.731734904732208e-06,
+      "loss": 0.1598,
+      "step": 10571
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.729947573328662e-06,
+      "loss": 0.1669,
+      "step": 10572
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.7281604749872e-06,
+      "loss": 0.1812,
+      "step": 10573
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.726373609786854e-06,
+      "loss": 0.1633,
+      "step": 10574
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.72458697780665e-06,
+      "loss": 0.1623,
+      "step": 10575
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.722800579125588e-06,
+      "loss": 0.1779,
+      "step": 10576
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.721014413822682e-06,
+      "loss": 0.155,
+      "step": 10577
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.719228481976922e-06,
+      "loss": 0.1933,
+      "step": 10578
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.717442783667278e-06,
+      "loss": 0.1541,
+      "step": 10579
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.715657318972734e-06,
+      "loss": 0.1737,
+      "step": 10580
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.713872087972241e-06,
+      "loss": 0.1825,
+      "step": 10581
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.712087090744748e-06,
+      "loss": 0.171,
+      "step": 10582
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.7103023273692065e-06,
+      "loss": 0.1653,
+      "step": 10583
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.708517797924532e-06,
+      "loss": 0.209,
+      "step": 10584
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.7067335024896485e-06,
+      "loss": 0.155,
+      "step": 10585
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.704949441143466e-06,
+      "loss": 0.1772,
+      "step": 10586
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.703165613964877e-06,
+      "loss": 0.1548,
+      "step": 10587
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.701382021032781e-06,
+      "loss": 0.15,
+      "step": 10588
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.6995986624260435e-06,
+      "loss": 0.1693,
+      "step": 10589
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.697815538223536e-06,
+      "loss": 0.1743,
+      "step": 10590
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.696032648504115e-06,
+      "loss": 0.1553,
+      "step": 10591
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.6942499933466265e-06,
+      "loss": 0.1625,
+      "step": 10592
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.692467572829908e-06,
+      "loss": 0.1958,
+      "step": 10593
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.6906853870327805e-06,
+      "loss": 0.163,
+      "step": 10594
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.688903436034065e-06,
+      "loss": 0.1626,
+      "step": 10595
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.687121719912562e-06,
+      "loss": 0.1821,
+      "step": 10596
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.685340238747068e-06,
+      "loss": 0.1726,
+      "step": 10597
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.6835589926163705e-06,
+      "loss": 0.2,
+      "step": 10598
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.68177798159923e-06,
+      "loss": 0.1701,
+      "step": 10599
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.6799972057744244e-06,
+      "loss": 0.1392,
+      "step": 10600
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.678216665220704e-06,
+      "loss": 0.1764,
+      "step": 10601
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.6764363600168e-06,
+      "loss": 0.1557,
+      "step": 10602
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.674656290241459e-06,
+      "loss": 0.1931,
+      "step": 10603
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.672876455973389e-06,
+      "loss": 0.1724,
+      "step": 10604
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.6710968572913116e-06,
+      "loss": 0.1521,
+      "step": 10605
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.669317494273928e-06,
+      "loss": 0.1671,
+      "step": 10606
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.6675383669999176e-06,
+      "loss": 0.1662,
+      "step": 10607
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.665759475547973e-06,
+      "loss": 0.1639,
+      "step": 10608
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.663980819996755e-06,
+      "loss": 0.1772,
+      "step": 10609
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.662202400424922e-06,
+      "loss": 0.1834,
+      "step": 10610
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.660424216911132e-06,
+      "loss": 0.1683,
+      "step": 10611
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.658646269534015e-06,
+      "loss": 0.1845,
+      "step": 10612
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.656868558372201e-06,
+      "loss": 0.1728,
+      "step": 10613
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.655091083504306e-06,
+      "loss": 0.1731,
+      "step": 10614
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.653313845008938e-06,
+      "loss": 0.1702,
+      "step": 10615
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.651536842964693e-06,
+      "loss": 0.1857,
+      "step": 10616
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.649760077450158e-06,
+      "loss": 0.1631,
+      "step": 10617
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.647983548543907e-06,
+      "loss": 0.1588,
+      "step": 10618
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.6462072563245065e-06,
+      "loss": 0.1681,
+      "step": 10619
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.64443120087051e-06,
+      "loss": 0.1738,
+      "step": 10620
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.642655382260461e-06,
+      "loss": 0.1676,
+      "step": 10621
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.640879800572893e-06,
+      "loss": 0.1508,
+      "step": 10622
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.639104455886331e-06,
+      "loss": 0.1731,
+      "step": 10623
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.637329348279286e-06,
+      "loss": 0.169,
+      "step": 10624
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.63555447783026e-06,
+      "loss": 0.1778,
+      "step": 10625
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.6337798446177485e-06,
+      "loss": 0.1644,
+      "step": 10626
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.632005448720223e-06,
+      "loss": 0.1363,
+      "step": 10627
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.630231290216163e-06,
+      "loss": 0.1856,
+      "step": 10628
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.62845736918403e-06,
+      "loss": 0.1794,
+      "step": 10629
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.626683685702262e-06,
+      "loss": 0.1442,
+      "step": 10630
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.624910239849315e-06,
+      "loss": 0.135,
+      "step": 10631
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.623137031703603e-06,
+      "loss": 0.1747,
+      "step": 10632
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.621364061343546e-06,
+      "loss": 0.185,
+      "step": 10633
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.619591328847564e-06,
+      "loss": 0.1936,
+      "step": 10634
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.61781883429404e-06,
+      "loss": 0.1479,
+      "step": 10635
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.616046577761366e-06,
+      "loss": 0.1556,
+      "step": 10636
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.614274559327918e-06,
+      "loss": 0.176,
+      "step": 10637
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.612502779072057e-06,
+      "loss": 0.1892,
+      "step": 10638
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.610731237072151e-06,
+      "loss": 0.1671,
+      "step": 10639
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.608959933406531e-06,
+      "loss": 0.1762,
+      "step": 10640
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.607188868153536e-06,
+      "loss": 0.1529,
+      "step": 10641
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.60541804139149e-06,
+      "loss": 0.154,
+      "step": 10642
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.603647453198703e-06,
+      "loss": 0.1688,
+      "step": 10643
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.60187710365348e-06,
+      "loss": 0.1524,
+      "step": 10644
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.600106992834112e-06,
+      "loss": 0.1661,
+      "step": 10645
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.598337120818879e-06,
+      "loss": 0.1529,
+      "step": 10646
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.596567487686052e-06,
+      "loss": 0.1636,
+      "step": 10647
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.5947980935138926e-06,
+      "loss": 0.1683,
+      "step": 10648
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.593028938380653e-06,
+      "loss": 0.162,
+      "step": 10649
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.59126002236456e-06,
+      "loss": 0.158,
+      "step": 10650
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.589491345543855e-06,
+      "loss": 0.1686,
+      "step": 10651
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.587722907996753e-06,
+      "loss": 0.1455,
+      "step": 10652
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.585954709801452e-06,
+      "loss": 0.1667,
+      "step": 10653
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.584186751036164e-06,
+      "loss": 0.1664,
+      "step": 10654
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.582419031779057e-06,
+      "loss": 0.1719,
+      "step": 10655
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.580651552108322e-06,
+      "loss": 0.1523,
+      "step": 10656
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.57888431210212e-06,
+      "loss": 0.1931,
+      "step": 10657
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.577117311838594e-06,
+      "loss": 0.1741,
+      "step": 10658
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.575350551395906e-06,
+      "loss": 0.1583,
+      "step": 10659
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.573584030852175e-06,
+      "loss": 0.177,
+      "step": 10660
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.571817750285524e-06,
+      "loss": 0.1615,
+      "step": 10661
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.570051709774076e-06,
+      "loss": 0.1793,
+      "step": 10662
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.5682859093959205e-06,
+      "loss": 0.1884,
+      "step": 10663
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.566520349229153e-06,
+      "loss": 0.1862,
+      "step": 10664
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.5647550293518525e-06,
+      "loss": 0.167,
+      "step": 10665
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.562989949842088e-06,
+      "loss": 0.1578,
+      "step": 10666
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.561225110777918e-06,
+      "loss": 0.1686,
+      "step": 10667
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.559460512237391e-06,
+      "loss": 0.1426,
+      "step": 10668
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.557696154298545e-06,
+      "loss": 0.1465,
+      "step": 10669
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.555932037039406e-06,
+      "loss": 0.1787,
+      "step": 10670
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.55416816053799e-06,
+      "loss": 0.161,
+      "step": 10671
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.552404524872301e-06,
+      "loss": 0.173,
+      "step": 10672
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.550641130120338e-06,
+      "loss": 0.1506,
+      "step": 10673
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.548877976360081e-06,
+      "loss": 0.1575,
+      "step": 10674
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.547115063669505e-06,
+      "loss": 0.1566,
+      "step": 10675
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.545352392126573e-06,
+      "loss": 0.1863,
+      "step": 10676
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.54358996180924e-06,
+      "loss": 0.1628,
+      "step": 10677
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.5418277727954374e-06,
+      "loss": 0.1512,
+      "step": 10678
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.540065825163107e-06,
+      "loss": 0.1624,
+      "step": 10679
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.538304118990168e-06,
+      "loss": 0.1761,
+      "step": 10680
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.53654265435452e-06,
+      "loss": 0.1687,
+      "step": 10681
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.534781431334075e-06,
+      "loss": 0.1577,
+      "step": 10682
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.533020450006712e-06,
+      "loss": 0.1509,
+      "step": 10683
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.531259710450307e-06,
+      "loss": 0.1621,
+      "step": 10684
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.529499212742738e-06,
+      "loss": 0.164,
+      "step": 10685
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.527738956961847e-06,
+      "loss": 0.1738,
+      "step": 10686
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.5259789431854925e-06,
+      "loss": 0.1606,
+      "step": 10687
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.524219171491499e-06,
+      "loss": 0.1598,
+      "step": 10688
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.522459641957692e-06,
+      "loss": 0.1667,
+      "step": 10689
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.520700354661892e-06,
+      "loss": 0.1581,
+      "step": 10690
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.5189413096818944e-06,
+      "loss": 0.162,
+      "step": 10691
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.517182507095492e-06,
+      "loss": 0.157,
+      "step": 10692
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.515423946980466e-06,
+      "loss": 0.2044,
+      "step": 10693
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.513665629414589e-06,
+      "loss": 0.1851,
+      "step": 10694
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.5119075544756175e-06,
+      "loss": 0.1543,
+      "step": 10695
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.510149722241303e-06,
+      "loss": 0.1754,
+      "step": 10696
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.508392132789382e-06,
+      "loss": 0.1751,
+      "step": 10697
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.506634786197582e-06,
+      "loss": 0.1571,
+      "step": 10698
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.504877682543621e-06,
+      "loss": 0.147,
+      "step": 10699
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.503120821905202e-06,
+      "loss": 0.1405,
+      "step": 10700
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.5013642043600245e-06,
+      "loss": 0.1729,
+      "step": 10701
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.499607829985769e-06,
+      "loss": 0.1899,
+      "step": 10702
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.497851698860112e-06,
+      "loss": 0.1638,
+      "step": 10703
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.496095811060715e-06,
+      "loss": 0.1404,
+      "step": 10704
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.494340166665229e-06,
+      "loss": 0.1543,
+      "step": 10705
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4925847657512976e-06,
+      "loss": 0.1689,
+      "step": 10706
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.49082960839655e-06,
+      "loss": 0.1748,
+      "step": 10707
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4890746946786104e-06,
+      "loss": 0.1692,
+      "step": 10708
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.487320024675076e-06,
+      "loss": 0.1677,
+      "step": 10709
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.485565598463558e-06,
+      "loss": 0.1634,
+      "step": 10710
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.483811416121642e-06,
+      "loss": 0.1704,
+      "step": 10711
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4820574777268945e-06,
+      "loss": 0.1605,
+      "step": 10712
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.480303783356896e-06,
+      "loss": 0.1625,
+      "step": 10713
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.47855033308919e-06,
+      "loss": 0.1866,
+      "step": 10714
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.476797127001321e-06,
+      "loss": 0.1715,
+      "step": 10715
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.475044165170835e-06,
+      "loss": 0.1768,
+      "step": 10716
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.473291447675242e-06,
+      "loss": 0.1574,
+      "step": 10717
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.471538974592059e-06,
+      "loss": 0.1696,
+      "step": 10718
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.469786745998786e-06,
+      "loss": 0.167,
+      "step": 10719
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.468034761972912e-06,
+      "loss": 0.1864,
+      "step": 10720
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.466283022591923e-06,
+      "loss": 0.1969,
+      "step": 10721
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.46453152793328e-06,
+      "loss": 0.1821,
+      "step": 10722
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4627802780744446e-06,
+      "loss": 0.1801,
+      "step": 10723
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.461029273092863e-06,
+      "loss": 0.1635,
+      "step": 10724
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.459278513065971e-06,
+      "loss": 0.1994,
+      "step": 10725
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.457527998071194e-06,
+      "loss": 0.146,
+      "step": 10726
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.455777728185949e-06,
+      "loss": 0.1741,
+      "step": 10727
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4540277034876354e-06,
+      "loss": 0.1834,
+      "step": 10728
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.45227792405365e-06,
+      "loss": 0.1728,
+      "step": 10729
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.450528389961372e-06,
+      "loss": 0.1605,
+      "step": 10730
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4487791012881775e-06,
+      "loss": 0.1783,
+      "step": 10731
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.447030058111416e-06,
+      "loss": 0.1576,
+      "step": 10732
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4452812605084474e-06,
+      "loss": 0.1615,
+      "step": 10733
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.44353270855661e-06,
+      "loss": 0.1747,
+      "step": 10734
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.441784402333222e-06,
+      "loss": 0.1584,
+      "step": 10735
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.440036341915614e-06,
+      "loss": 0.165,
+      "step": 10736
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4382885273810775e-06,
+      "loss": 0.171,
+      "step": 10737
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.436540958806918e-06,
+      "loss": 0.1633,
+      "step": 10738
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.434793636270419e-06,
+      "loss": 0.1376,
+      "step": 10739
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.433046559848846e-06,
+      "loss": 0.1896,
+      "step": 10740
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.431299729619473e-06,
+      "loss": 0.1725,
+      "step": 10741
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.429553145659542e-06,
+      "loss": 0.1614,
+      "step": 10742
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.427806808046293e-06,
+      "loss": 0.1849,
+      "step": 10743
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.426060716856968e-06,
+      "loss": 0.1488,
+      "step": 10744
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.424314872168774e-06,
+      "loss": 0.1761,
+      "step": 10745
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.422569274058922e-06,
+      "loss": 0.1758,
+      "step": 10746
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.42082392260461e-06,
+      "loss": 0.1899,
+      "step": 10747
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4190788178830245e-06,
+      "loss": 0.1784,
+      "step": 10748
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.417333959971339e-06,
+      "loss": 0.1517,
+      "step": 10749
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.415589348946721e-06,
+      "loss": 0.1652,
+      "step": 10750
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.413844984886322e-06,
+      "loss": 0.1802,
+      "step": 10751
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.412100867867284e-06,
+      "loss": 0.1797,
+      "step": 10752
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.410356997966739e-06,
+      "loss": 0.1809,
+      "step": 10753
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.408613375261807e-06,
+      "loss": 0.1695,
+      "step": 10754
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4068699998296e-06,
+      "loss": 0.1626,
+      "step": 10755
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.405126871747216e-06,
+      "loss": 0.1669,
+      "step": 10756
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.403383991091742e-06,
+      "loss": 0.1625,
+      "step": 10757
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.401641357940255e-06,
+      "loss": 0.1728,
+      "step": 10758
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.399898972369825e-06,
+      "loss": 0.1673,
+      "step": 10759
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.398156834457496e-06,
+      "loss": 0.1639,
+      "step": 10760
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.396414944280321e-06,
+      "loss": 0.1896,
+      "step": 10761
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.394673301915336e-06,
+      "loss": 0.1833,
+      "step": 10762
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.392931907439553e-06,
+      "loss": 0.1707,
+      "step": 10763
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.391190760929994e-06,
+      "loss": 0.1902,
+      "step": 10764
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.389449862463652e-06,
+      "loss": 0.1563,
+      "step": 10765
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.3877092121175134e-06,
+      "loss": 0.156,
+      "step": 10766
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.385968809968569e-06,
+      "loss": 0.1574,
+      "step": 10767
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.384228656093775e-06,
+      "loss": 0.1778,
+      "step": 10768
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.3824887505700905e-06,
+      "loss": 0.1886,
+      "step": 10769
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.3807490934744625e-06,
+      "loss": 0.1642,
+      "step": 10770
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.3790096848838195e-06,
+      "loss": 0.1745,
+      "step": 10771
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.377270524875098e-06,
+      "loss": 0.1819,
+      "step": 10772
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.375531613525198e-06,
+      "loss": 0.1601,
+      "step": 10773
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.373792950911024e-06,
+      "loss": 0.1834,
+      "step": 10774
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.3720545371094695e-06,
+      "loss": 0.1693,
+      "step": 10775
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.37031637219741e-06,
+      "loss": 0.1279,
+      "step": 10776
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.368578456251717e-06,
+      "loss": 0.1564,
+      "step": 10777
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.366840789349246e-06,
+      "loss": 0.1839,
+      "step": 10778
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.365103371566843e-06,
+      "loss": 0.1673,
+      "step": 10779
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.363366202981345e-06,
+      "loss": 0.1595,
+      "step": 10780
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.361629283669575e-06,
+      "loss": 0.1477,
+      "step": 10781
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.35989261370835e-06,
+      "loss": 0.1776,
+      "step": 10782
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.358156193174462e-06,
+      "loss": 0.1874,
+      "step": 10783
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.356420022144714e-06,
+      "loss": 0.1694,
+      "step": 10784
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.354684100695884e-06,
+      "loss": 0.1637,
+      "step": 10785
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.3529484289047305e-06,
+      "loss": 0.1652,
+      "step": 10786
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.351213006848028e-06,
+      "loss": 0.159,
+      "step": 10787
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.3494778346025075e-06,
+      "loss": 0.1672,
+      "step": 10788
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.347742912244918e-06,
+      "loss": 0.1612,
+      "step": 10789
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.346008239851981e-06,
+      "loss": 0.1693,
+      "step": 10790
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.344273817500402e-06,
+      "loss": 0.167,
+      "step": 10791
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.342539645266898e-06,
+      "loss": 0.1595,
+      "step": 10792
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.340805723228149e-06,
+      "loss": 0.1616,
+      "step": 10793
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.339072051460837e-06,
+      "loss": 0.1707,
+      "step": 10794
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.337338630041642e-06,
+      "loss": 0.1641,
+      "step": 10795
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.335605459047212e-06,
+      "loss": 0.1836,
+      "step": 10796
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.333872538554197e-06,
+      "loss": 0.1511,
+      "step": 10797
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.332139868639234e-06,
+      "loss": 0.1519,
+      "step": 10798
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.330407449378949e-06,
+      "loss": 0.1577,
+      "step": 10799
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.328675280849957e-06,
+      "loss": 0.1586,
+      "step": 10800
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.326943363128858e-06,
+      "loss": 0.1836,
+      "step": 10801
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.325211696292248e-06,
+      "loss": 0.1624,
+      "step": 10802
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.323480280416705e-06,
+      "loss": 0.1784,
+      "step": 10803
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.321749115578801e-06,
+      "loss": 0.1474,
+      "step": 10804
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.320018201855094e-06,
+      "loss": 0.1798,
+      "step": 10805
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.318287539322131e-06,
+      "loss": 0.176,
+      "step": 10806
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.316557128056449e-06,
+      "loss": 0.1665,
+      "step": 10807
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.314826968134573e-06,
+      "loss": 0.1735,
+      "step": 10808
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.31309705963302e-06,
+      "loss": 0.1601,
+      "step": 10809
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.3113674026282935e-06,
+      "loss": 0.1842,
+      "step": 10810
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.309637997196876e-06,
+      "loss": 0.1804,
+      "step": 10811
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.3079088434152595e-06,
+      "loss": 0.1599,
+      "step": 10812
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.306179941359915e-06,
+      "loss": 0.1675,
+      "step": 10813
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.304451291107287e-06,
+      "loss": 0.1708,
+      "step": 10814
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.302722892733841e-06,
+      "loss": 0.1401,
+      "step": 10815
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.300994746316002e-06,
+      "loss": 0.1624,
+      "step": 10816
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.299266851930194e-06,
+      "loss": 0.1704,
+      "step": 10817
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.297539209652842e-06,
+      "loss": 0.208,
+      "step": 10818
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.295811819560337e-06,
+      "loss": 0.1538,
+      "step": 10819
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.294084681729083e-06,
+      "loss": 0.1776,
+      "step": 10820
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.292357796235449e-06,
+      "loss": 0.1384,
+      "step": 10821
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.290631163155807e-06,
+      "loss": 0.1708,
+      "step": 10822
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.288904782566526e-06,
+      "loss": 0.1592,
+      "step": 10823
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.28717865454394e-06,
+      "loss": 0.1595,
+      "step": 10824
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.285452779164391e-06,
+      "loss": 0.1631,
+      "step": 10825
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.283727156504202e-06,
+      "loss": 0.171,
+      "step": 10826
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.2820017866396905e-06,
+      "loss": 0.1687,
+      "step": 10827
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.280276669647155e-06,
+      "loss": 0.1713,
+      "step": 10828
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.278551805602887e-06,
+      "loss": 0.1706,
+      "step": 10829
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.27682719458317e-06,
+      "loss": 0.1761,
+      "step": 10830
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.275102836664269e-06,
+      "loss": 0.1627,
+      "step": 10831
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.273378731922444e-06,
+      "loss": 0.1811,
+      "step": 10832
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.271654880433942e-06,
+      "loss": 0.1817,
+      "step": 10833
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.269931282274996e-06,
+      "loss": 0.1885,
+      "step": 10834
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.268207937521834e-06,
+      "loss": 0.1735,
+      "step": 10835
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.266484846250665e-06,
+      "loss": 0.1712,
+      "step": 10836
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.264762008537693e-06,
+      "loss": 0.1628,
+      "step": 10837
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.263039424459107e-06,
+      "loss": 0.1488,
+      "step": 10838
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.261317094091089e-06,
+      "loss": 0.1616,
+      "step": 10839
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.259595017509804e-06,
+      "loss": 0.1879,
+      "step": 10840
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.2578731947914145e-06,
+      "loss": 0.1432,
+      "step": 10841
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.256151626012054e-06,
+      "loss": 0.1522,
+      "step": 10842
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.254430311247869e-06,
+      "loss": 0.1617,
+      "step": 10843
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.252709250574982e-06,
+      "loss": 0.1375,
+      "step": 10844
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.2509884440694944e-06,
+      "loss": 0.1707,
+      "step": 10845
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.249267891807521e-06,
+      "loss": 0.1885,
+      "step": 10846
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.247547593865141e-06,
+      "loss": 0.1857,
+      "step": 10847
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.245827550318433e-06,
+      "loss": 0.1744,
+      "step": 10848
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.244107761243473e-06,
+      "loss": 0.1785,
+      "step": 10849
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.2423882267163075e-06,
+      "loss": 0.1594,
+      "step": 10850
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.240668946812984e-06,
+      "loss": 0.1835,
+      "step": 10851
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.238949921609535e-06,
+      "loss": 0.1704,
+      "step": 10852
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.237231151181981e-06,
+      "loss": 0.1597,
+      "step": 10853
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.2355126356063415e-06,
+      "loss": 0.1716,
+      "step": 10854
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.233794374958605e-06,
+      "loss": 0.1815,
+      "step": 10855
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.232076369314765e-06,
+      "loss": 0.1695,
+      "step": 10856
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.230358618750796e-06,
+      "loss": 0.1544,
+      "step": 10857
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.228641123342665e-06,
+      "loss": 0.1732,
+      "step": 10858
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.226923883166327e-06,
+      "loss": 0.1783,
+      "step": 10859
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.225206898297724e-06,
+      "loss": 0.1944,
+      "step": 10860
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.223490168812787e-06,
+      "loss": 0.1908,
+      "step": 10861
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.221773694787438e-06,
+      "loss": 0.1732,
+      "step": 10862
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.220057476297585e-06,
+      "loss": 0.1888,
+      "step": 10863
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.21834151341913e-06,
+      "loss": 0.1649,
+      "step": 10864
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.216625806227948e-06,
+      "loss": 0.1556,
+      "step": 10865
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.2149103547999245e-06,
+      "loss": 0.172,
+      "step": 10866
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.213195159210927e-06,
+      "loss": 0.1624,
+      "step": 10867
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.2114802195367925e-06,
+      "loss": 0.1655,
+      "step": 10868
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.20976553585338e-06,
+      "loss": 0.1786,
+      "step": 10869
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.2080511082365014e-06,
+      "loss": 0.1625,
+      "step": 10870
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.206336936761991e-06,
+      "loss": 0.1785,
+      "step": 10871
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.204623021505652e-06,
+      "loss": 0.1879,
+      "step": 10872
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.2029093625432724e-06,
+      "loss": 0.1662,
+      "step": 10873
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.2011959599506504e-06,
+      "loss": 0.1788,
+      "step": 10874
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.199482813803547e-06,
+      "loss": 0.1742,
+      "step": 10875
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.197769924177726e-06,
+      "loss": 0.1549,
+      "step": 10876
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.196057291148947e-06,
+      "loss": 0.1689,
+      "step": 10877
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.1943449147929405e-06,
+      "loss": 0.1608,
+      "step": 10878
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.192632795185436e-06,
+      "loss": 0.1896,
+      "step": 10879
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.190920932402152e-06,
+      "loss": 0.1454,
+      "step": 10880
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.189209326518793e-06,
+      "loss": 0.1703,
+      "step": 10881
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.187497977611051e-06,
+      "loss": 0.1628,
+      "step": 10882
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.185786885754611e-06,
+      "loss": 0.1784,
+      "step": 10883
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.184076051025143e-06,
+      "loss": 0.1529,
+      "step": 10884
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.1823654734983065e-06,
+      "loss": 0.1503,
+      "step": 10885
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.18065515324975e-06,
+      "loss": 0.1881,
+      "step": 10886
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.178945090355111e-06,
+      "loss": 0.1604,
+      "step": 10887
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.177235284890014e-06,
+      "loss": 0.1657,
+      "step": 10888
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.175525736930075e-06,
+      "loss": 0.1517,
+      "step": 10889
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.173816446550895e-06,
+      "loss": 0.1926,
+      "step": 10890
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.172107413828067e-06,
+      "loss": 0.1806,
+      "step": 10891
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.170398638837173e-06,
+      "loss": 0.186,
+      "step": 10892
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.168690121653773e-06,
+      "loss": 0.1688,
+      "step": 10893
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.166981862353432e-06,
+      "loss": 0.1699,
+      "step": 10894
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.165273861011698e-06,
+      "loss": 0.1691,
+      "step": 10895
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.1635661177040945e-06,
+      "loss": 0.1787,
+      "step": 10896
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.161858632506159e-06,
+      "loss": 0.1718,
+      "step": 10897
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.16015140549339e-06,
+      "loss": 0.1685,
+      "step": 10898
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.15844443674129e-06,
+      "loss": 0.182,
+      "step": 10899
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.156737726325359e-06,
+      "loss": 0.1742,
+      "step": 10900
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.155031274321062e-06,
+      "loss": 0.1644,
+      "step": 10901
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.153325080803869e-06,
+      "loss": 0.1566,
+      "step": 10902
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.151619145849235e-06,
+      "loss": 0.1629,
+      "step": 10903
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.1499134695325986e-06,
+      "loss": 0.171,
+      "step": 10904
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.1482080519294035e-06,
+      "loss": 0.1606,
+      "step": 10905
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.146502893115056e-06,
+      "loss": 0.1648,
+      "step": 10906
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.144797993164973e-06,
+      "loss": 0.1633,
+      "step": 10907
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.143093352154548e-06,
+      "loss": 0.1619,
+      "step": 10908
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.141388970159168e-06,
+      "loss": 0.1613,
+      "step": 10909
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.1396848472542074e-06,
+      "loss": 0.151,
+      "step": 10910
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.13798098351503e-06,
+      "loss": 0.1788,
+      "step": 10911
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.136277379016986e-06,
+      "loss": 0.1723,
+      "step": 10912
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.134574033835415e-06,
+      "loss": 0.168,
+      "step": 10913
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.132870948045647e-06,
+      "loss": 0.1752,
+      "step": 10914
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.131168121723003e-06,
+      "loss": 0.1809,
+      "step": 10915
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.129465554942775e-06,
+      "loss": 0.1879,
+      "step": 10916
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.1277632477802705e-06,
+      "loss": 0.1814,
+      "step": 10917
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.126061200310771e-06,
+      "loss": 0.1571,
+      "step": 10918
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.124359412609538e-06,
+      "loss": 0.1806,
+      "step": 10919
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.122657884751843e-06,
+      "loss": 0.1522,
+      "step": 10920
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.1209566168129214e-06,
+      "loss": 0.1538,
+      "step": 10921
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.119255608868021e-06,
+      "loss": 0.1868,
+      "step": 10922
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.117554860992366e-06,
+      "loss": 0.1617,
+      "step": 10923
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.11585437326116e-06,
+      "loss": 0.1683,
+      "step": 10924
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.1141541457496194e-06,
+      "loss": 0.1484,
+      "step": 10925
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.112454178532924e-06,
+      "loss": 0.1528,
+      "step": 10926
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.110754471686253e-06,
+      "loss": 0.1861,
+      "step": 10927
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.109055025284783e-06,
+      "loss": 0.1728,
+      "step": 10928
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.1073558394036614e-06,
+      "loss": 0.1742,
+      "step": 10929
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.105656914118037e-06,
+      "loss": 0.156,
+      "step": 10930
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.1039582495030415e-06,
+      "loss": 0.1649,
+      "step": 10931
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.102259845633796e-06,
+      "loss": 0.1629,
+      "step": 10932
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.100561702585412e-06,
+      "loss": 0.1595,
+      "step": 10933
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.098863820432986e-06,
+      "loss": 0.1597,
+      "step": 10934
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.0971661992516054e-06,
+      "loss": 0.1542,
+      "step": 10935
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.095468839116347e-06,
+      "loss": 0.152,
+      "step": 10936
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.093771740102273e-06,
+      "loss": 0.1615,
+      "step": 10937
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.092074902284438e-06,
+      "loss": 0.1735,
+      "step": 10938
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.090378325737881e-06,
+      "loss": 0.1523,
+      "step": 10939
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.088682010537629e-06,
+      "loss": 0.1724,
+      "step": 10940
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.086985956758703e-06,
+      "loss": 0.1656,
+      "step": 10941
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.085290164476109e-06,
+      "loss": 0.1452,
+      "step": 10942
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.083594633764843e-06,
+      "loss": 0.1803,
+      "step": 10943
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.081899364699879e-06,
+      "loss": 0.1913,
+      "step": 10944
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.080204357356198e-06,
+      "loss": 0.1547,
+      "step": 10945
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.07850961180876e-06,
+      "loss": 0.189,
+      "step": 10946
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.076815128132503e-06,
+      "loss": 0.1649,
+      "step": 10947
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.075120906402377e-06,
+      "loss": 0.1459,
+      "step": 10948
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.073426946693296e-06,
+      "loss": 0.1636,
+      "step": 10949
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.071733249080175e-06,
+      "loss": 0.1533,
+      "step": 10950
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.070039813637926e-06,
+      "loss": 0.1782,
+      "step": 10951
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.068346640441425e-06,
+      "loss": 0.1497,
+      "step": 10952
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.066653729565564e-06,
+      "loss": 0.1791,
+      "step": 10953
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.0649610810852e-06,
+      "loss": 0.1656,
+      "step": 10954
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.063268695075189e-06,
+      "loss": 0.1589,
+      "step": 10955
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.061576571610385e-06,
+      "loss": 0.1772,
+      "step": 10956
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.059884710765609e-06,
+      "loss": 0.1445,
+      "step": 10957
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.058193112615687e-06,
+      "loss": 0.1566,
+      "step": 10958
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.056501777235425e-06,
+      "loss": 0.1677,
+      "step": 10959
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.054810704699624e-06,
+      "loss": 0.1645,
+      "step": 10960
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.053119895083068e-06,
+      "loss": 0.1598,
+      "step": 10961
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.051429348460529e-06,
+      "loss": 0.2125,
+      "step": 10962
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.049739064906774e-06,
+      "loss": 0.1388,
+      "step": 10963
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.048049044496549e-06,
+      "loss": 0.168,
+      "step": 10964
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.0463592873045974e-06,
+      "loss": 0.1704,
+      "step": 10965
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.0446697934056445e-06,
+      "loss": 0.1647,
+      "step": 10966
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.042980562874408e-06,
+      "loss": 0.1563,
+      "step": 10967
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.041291595785589e-06,
+      "loss": 0.1527,
+      "step": 10968
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.0396028922138844e-06,
+      "loss": 0.1588,
+      "step": 10969
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.037914452233971e-06,
+      "loss": 0.168,
+      "step": 10970
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.036226275920521e-06,
+      "loss": 0.1651,
+      "step": 10971
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.034538363348192e-06,
+      "loss": 0.1283,
+      "step": 10972
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.032850714591628e-06,
+      "loss": 0.1526,
+      "step": 10973
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.031163329725468e-06,
+      "loss": 0.1742,
+      "step": 10974
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.029476208824326e-06,
+      "loss": 0.1484,
+      "step": 10975
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.02778935196282e-06,
+      "loss": 0.1629,
+      "step": 10976
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.026102759215551e-06,
+      "loss": 0.1434,
+      "step": 10977
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.0244164306570965e-06,
+      "loss": 0.162,
+      "step": 10978
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.022730366362046e-06,
+      "loss": 0.1806,
+      "step": 10979
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.021044566404955e-06,
+      "loss": 0.1738,
+      "step": 10980
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.0193590308603735e-06,
+      "loss": 0.1661,
+      "step": 10981
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.017673759802853e-06,
+      "loss": 0.1847,
+      "step": 10982
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.015988753306915e-06,
+      "loss": 0.1854,
+      "step": 10983
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.014304011447078e-06,
+      "loss": 0.1585,
+      "step": 10984
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.012619534297848e-06,
+      "loss": 0.1829,
+      "step": 10985
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.010935321933716e-06,
+      "loss": 0.184,
+      "step": 10986
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.009251374429176e-06,
+      "loss": 0.154,
+      "step": 10987
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.007567691858687e-06,
+      "loss": 0.1742,
+      "step": 10988
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.005884274296711e-06,
+      "loss": 0.1766,
+      "step": 10989
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.004201121817696e-06,
+      "loss": 0.183,
+      "step": 10990
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.002518234496077e-06,
+      "loss": 0.1723,
+      "step": 10991
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.0008356124062775e-06,
+      "loss": 0.1771,
+      "step": 10992
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.999153255622711e-06,
+      "loss": 0.1482,
+      "step": 10993
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.9974711642197765e-06,
+      "loss": 0.1506,
+      "step": 10994
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.995789338271863e-06,
+      "loss": 0.1811,
+      "step": 10995
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.994107777853347e-06,
+      "loss": 0.1792,
+      "step": 10996
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.992426483038598e-06,
+      "loss": 0.1821,
+      "step": 10997
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.990745453901958e-06,
+      "loss": 0.1667,
+      "step": 10998
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.98906469051778e-06,
+      "loss": 0.1731,
+      "step": 10999
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.987384192960393e-06,
+      "loss": 0.1695,
+      "step": 11000
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.985703961304105e-06,
+      "loss": 0.1576,
+      "step": 11001
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.984023995623235e-06,
+      "loss": 0.1707,
+      "step": 11002
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.9823442959920656e-06,
+      "loss": 0.1778,
+      "step": 11003
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.98066486248489e-06,
+      "loss": 0.1781,
+      "step": 11004
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.978985695175978e-06,
+      "loss": 0.1556,
+      "step": 11005
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.977306794139578e-06,
+      "loss": 0.1577,
+      "step": 11006
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.975628159449953e-06,
+      "loss": 0.15,
+      "step": 11007
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.973949791181329e-06,
+      "loss": 0.1722,
+      "step": 11008
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.972271689407927e-06,
+      "loss": 0.1585,
+      "step": 11009
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.970593854203971e-06,
+      "loss": 0.1635,
+      "step": 11010
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.968916285643652e-06,
+      "loss": 0.1704,
+      "step": 11011
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.9672389838011606e-06,
+      "loss": 0.1504,
+      "step": 11012
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.965561948750674e-06,
+      "loss": 0.1835,
+      "step": 11013
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.963885180566357e-06,
+      "loss": 0.1551,
+      "step": 11014
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.962208679322363e-06,
+      "loss": 0.1433,
+      "step": 11015
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.960532445092834e-06,
+      "loss": 0.1793,
+      "step": 11016
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.958856477951898e-06,
+      "loss": 0.1724,
+      "step": 11017
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.957180777973674e-06,
+      "loss": 0.173,
+      "step": 11018
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.955505345232268e-06,
+      "loss": 0.162,
+      "step": 11019
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.953830179801773e-06,
+      "loss": 0.1476,
+      "step": 11020
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.9521552817562724e-06,
+      "loss": 0.1676,
+      "step": 11021
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.950480651169837e-06,
+      "loss": 0.1699,
+      "step": 11022
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.948806288116523e-06,
+      "loss": 0.1456,
+      "step": 11023
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.947132192670381e-06,
+      "loss": 0.1731,
+      "step": 11024
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.945458364905446e-06,
+      "loss": 0.1769,
+      "step": 11025
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.943784804895732e-06,
+      "loss": 0.1681,
+      "step": 11026
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.942111512715261e-06,
+      "loss": 0.1569,
+      "step": 11027
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.940438488438032e-06,
+      "loss": 0.1537,
+      "step": 11028
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.938765732138022e-06,
+      "loss": 0.1787,
+      "step": 11029
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.937093243889222e-06,
+      "loss": 0.1912,
+      "step": 11030
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.935421023765584e-06,
+      "loss": 0.18,
+      "step": 11031
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.933749071841061e-06,
+      "loss": 0.1605,
+      "step": 11032
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.932077388189602e-06,
+      "loss": 0.1776,
+      "step": 11033
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.930405972885127e-06,
+      "loss": 0.1818,
+      "step": 11034
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.928734826001555e-06,
+      "loss": 0.1569,
+      "step": 11035
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.92706394761279e-06,
+      "loss": 0.1664,
+      "step": 11036
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.9253933377927225e-06,
+      "loss": 0.1615,
+      "step": 11037
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.923722996615243e-06,
+      "loss": 0.1739,
+      "step": 11038
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.922052924154208e-06,
+      "loss": 0.1924,
+      "step": 11039
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.920383120483482e-06,
+      "loss": 0.1407,
+      "step": 11040
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.918713585676908e-06,
+      "loss": 0.1713,
+      "step": 11041
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.917044319808319e-06,
+      "loss": 0.1667,
+      "step": 11042
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.915375322951536e-06,
+      "loss": 0.1939,
+      "step": 11043
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.913706595180371e-06,
+      "loss": 0.1441,
+      "step": 11044
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.9120381365686186e-06,
+      "loss": 0.2014,
+      "step": 11045
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.910369947190067e-06,
+      "loss": 0.1801,
+      "step": 11046
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.908702027118488e-06,
+      "loss": 0.1392,
+      "step": 11047
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.9070343764276485e-06,
+      "loss": 0.1574,
+      "step": 11048
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.905366995191288e-06,
+      "loss": 0.1706,
+      "step": 11049
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.903699883483152e-06,
+      "loss": 0.1784,
+      "step": 11050
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.90203304137697e-06,
+      "loss": 0.1744,
+      "step": 11051
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.900366468946445e-06,
+      "loss": 0.1802,
+      "step": 11052
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.898700166265292e-06,
+      "loss": 0.1642,
+      "step": 11053
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.897034133407189e-06,
+      "loss": 0.1646,
+      "step": 11054
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.895368370445824e-06,
+      "loss": 0.1615,
+      "step": 11055
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.893702877454863e-06,
+      "loss": 0.1538,
+      "step": 11056
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.892037654507952e-06,
+      "loss": 0.1502,
+      "step": 11057
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.890372701678745e-06,
+      "loss": 0.1805,
+      "step": 11058
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.888708019040864e-06,
+      "loss": 0.1713,
+      "step": 11059
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.887043606667927e-06,
+      "loss": 0.1586,
+      "step": 11060
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.885379464633552e-06,
+      "loss": 0.1949,
+      "step": 11061
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.883715593011321e-06,
+      "loss": 0.1674,
+      "step": 11062
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.882051991874824e-06,
+      "loss": 0.1763,
+      "step": 11063
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.880388661297628e-06,
+      "loss": 0.1543,
+      "step": 11064
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.878725601353294e-06,
+      "loss": 0.1748,
+      "step": 11065
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.87706281211537e-06,
+      "loss": 0.1427,
+      "step": 11066
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.875400293657388e-06,
+      "loss": 0.1872,
+      "step": 11067
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.873738046052873e-06,
+      "loss": 0.1475,
+      "step": 11068
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8720760693753365e-06,
+      "loss": 0.1747,
+      "step": 11069
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.870414363698276e-06,
+      "loss": 0.1598,
+      "step": 11070
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.868752929095179e-06,
+      "loss": 0.171,
+      "step": 11071
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.867091765639521e-06,
+      "loss": 0.17,
+      "step": 11072
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.865430873404765e-06,
+      "loss": 0.1604,
+      "step": 11073
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.863770252464362e-06,
+      "loss": 0.1727,
+      "step": 11074
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.862109902891752e-06,
+      "loss": 0.189,
+      "step": 11075
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.860449824760363e-06,
+      "loss": 0.1719,
+      "step": 11076
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.858790018143601e-06,
+      "loss": 0.1707,
+      "step": 11077
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.85713048311488e-06,
+      "loss": 0.1631,
+      "step": 11078
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.855471219747592e-06,
+      "loss": 0.1731,
+      "step": 11079
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8538122281151035e-06,
+      "loss": 0.1547,
+      "step": 11080
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.852153508290795e-06,
+      "loss": 0.1738,
+      "step": 11081
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8504950603480154e-06,
+      "loss": 0.1882,
+      "step": 11082
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.848836884360102e-06,
+      "loss": 0.1653,
+      "step": 11083
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.847178980400399e-06,
+      "loss": 0.2097,
+      "step": 11084
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.845521348542213e-06,
+      "loss": 0.1797,
+      "step": 11085
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.843863988858862e-06,
+      "loss": 0.1434,
+      "step": 11086
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.84220690142363e-06,
+      "loss": 0.1493,
+      "step": 11087
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.840550086309802e-06,
+      "loss": 0.1747,
+      "step": 11088
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8388935435906596e-06,
+      "loss": 0.184,
+      "step": 11089
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.83723727333945e-06,
+      "loss": 0.1513,
+      "step": 11090
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.835581275629421e-06,
+      "loss": 0.1658,
+      "step": 11091
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8339255505338116e-06,
+      "loss": 0.1844,
+      "step": 11092
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.832270098125841e-06,
+      "loss": 0.1664,
+      "step": 11093
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.830614918478721e-06,
+      "loss": 0.1704,
+      "step": 11094
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.828960011665651e-06,
+      "loss": 0.1488,
+      "step": 11095
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.827305377759816e-06,
+      "loss": 0.1439,
+      "step": 11096
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.825651016834391e-06,
+      "loss": 0.1905,
+      "step": 11097
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.823996928962537e-06,
+      "loss": 0.1377,
+      "step": 11098
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.822343114217405e-06,
+      "loss": 0.1957,
+      "step": 11099
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.820689572672134e-06,
+      "loss": 0.1785,
+      "step": 11100
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.819036304399848e-06,
+      "loss": 0.1851,
+      "step": 11101
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.817383309473664e-06,
+      "loss": 0.1934,
+      "step": 11102
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8157305879666796e-06,
+      "loss": 0.1766,
+      "step": 11103
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8140781399519876e-06,
+      "loss": 0.1776,
+      "step": 11104
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.812425965502665e-06,
+      "loss": 0.1447,
+      "step": 11105
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.810774064691778e-06,
+      "loss": 0.1512,
+      "step": 11106
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.809122437592382e-06,
+      "loss": 0.1948,
+      "step": 11107
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.807471084277509e-06,
+      "loss": 0.1487,
+      "step": 11108
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8058200048201974e-06,
+      "loss": 0.1604,
+      "step": 11109
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.804169199293465e-06,
+      "loss": 0.1601,
+      "step": 11110
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8025186677703075e-06,
+      "loss": 0.1759,
+      "step": 11111
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8008684103237304e-06,
+      "loss": 0.1486,
+      "step": 11112
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.799218427026704e-06,
+      "loss": 0.1629,
+      "step": 11113
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.7975687179521982e-06,
+      "loss": 0.1619,
+      "step": 11114
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.7959192831731773e-06,
+      "loss": 0.1555,
+      "step": 11115
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7942701227625788e-06,
+      "loss": 0.1626,
+      "step": 11116
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7926212367933356e-06,
+      "loss": 0.168,
+      "step": 11117
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7909726253383692e-06,
+      "loss": 0.1904,
+      "step": 11118
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.789324288470584e-06,
+      "loss": 0.1733,
+      "step": 11119
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7876762262628853e-06,
+      "loss": 0.1844,
+      "step": 11120
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7860284387881475e-06,
+      "loss": 0.1672,
+      "step": 11121
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7843809261192442e-06,
+      "loss": 0.1564,
+      "step": 11122
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7827336883290368e-06,
+      "loss": 0.1753,
+      "step": 11123
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7810867254903704e-06,
+      "loss": 0.1808,
+      "step": 11124
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.779440037676082e-06,
+      "loss": 0.1414,
+      "step": 11125
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.777793624958993e-06,
+      "loss": 0.1848,
+      "step": 11126
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7761474874119152e-06,
+      "loss": 0.187,
+      "step": 11127
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.774501625107646e-06,
+      "loss": 0.1561,
+      "step": 11128
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7728560381189737e-06,
+      "loss": 0.1616,
+      "step": 11129
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7712107265186736e-06,
+      "loss": 0.1631,
+      "step": 11130
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7695656903794984e-06,
+      "loss": 0.1393,
+      "step": 11131
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7679209297742093e-06,
+      "loss": 0.1599,
+      "step": 11132
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.766276444775543e-06,
+      "loss": 0.1775,
+      "step": 11133
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7646322354562137e-06,
+      "loss": 0.1574,
+      "step": 11134
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.76298830188895e-06,
+      "loss": 0.1811,
+      "step": 11135
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7613446441464398e-06,
+      "loss": 0.1641,
+      "step": 11136
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7597012623013797e-06,
+      "loss": 0.1568,
+      "step": 11137
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7580581564264484e-06,
+      "loss": 0.1549,
+      "step": 11138
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7564153265943004e-06,
+      "loss": 0.1698,
+      "step": 11139
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7547727728775998e-06,
+      "loss": 0.1449,
+      "step": 11140
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.753130495348978e-06,
+      "loss": 0.1566,
+      "step": 11141
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7514884940810627e-06,
+      "loss": 0.1666,
+      "step": 11142
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.749846769146479e-06,
+      "loss": 0.156,
+      "step": 11143
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7482053206178214e-06,
+      "loss": 0.1527,
+      "step": 11144
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.746564148567683e-06,
+      "loss": 0.1577,
+      "step": 11145
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.744923253068643e-06,
+      "loss": 0.1745,
+      "step": 11146
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7432826341932693e-06,
+      "loss": 0.1644,
+      "step": 11147
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.741642292014115e-06,
+      "loss": 0.1852,
+      "step": 11148
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.740002226603724e-06,
+      "loss": 0.1708,
+      "step": 11149
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.738362438034625e-06,
+      "loss": 0.1967,
+      "step": 11150
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7367229263793358e-06,
+      "loss": 0.1452,
+      "step": 11151
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7350836917103628e-06,
+      "loss": 0.178,
+      "step": 11152
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7334447341001976e-06,
+      "loss": 0.1568,
+      "step": 11153
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.731806053621324e-06,
+      "loss": 0.1717,
+      "step": 11154
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7301676503462083e-06,
+      "loss": 0.1555,
+      "step": 11155
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7285295243473084e-06,
+      "loss": 0.1595,
+      "step": 11156
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7268916756970685e-06,
+      "loss": 0.1506,
+      "step": 11157
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7252541044679223e-06,
+      "loss": 0.1846,
+      "step": 11158
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7236168107322824e-06,
+      "loss": 0.1681,
+      "step": 11159
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7219797945625636e-06,
+      "loss": 0.1764,
+      "step": 11160
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7203430560311616e-06,
+      "loss": 0.1371,
+      "step": 11161
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.718706595210451e-06,
+      "loss": 0.1699,
+      "step": 11162
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.717070412172814e-06,
+      "loss": 0.182,
+      "step": 11163
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7154345069905996e-06,
+      "loss": 0.1423,
+      "step": 11164
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7137988797361536e-06,
+      "loss": 0.1631,
+      "step": 11165
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7121635304818193e-06,
+      "loss": 0.1741,
+      "step": 11166
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.7105284592999093e-06,
+      "loss": 0.2258,
+      "step": 11167
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.7088936662627364e-06,
+      "loss": 0.1581,
+      "step": 11168
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.707259151442596e-06,
+      "loss": 0.1559,
+      "step": 11169
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.7056249149117697e-06,
+      "loss": 0.1666,
+      "step": 11170
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.7039909567425403e-06,
+      "loss": 0.1459,
+      "step": 11171
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.702357277007157e-06,
+      "loss": 0.1655,
+      "step": 11172
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.7007238757778717e-06,
+      "loss": 0.1594,
+      "step": 11173
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6990907531269195e-06,
+      "loss": 0.177,
+      "step": 11174
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6974579091265216e-06,
+      "loss": 0.1612,
+      "step": 11175
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.695825343848892e-06,
+      "loss": 0.147,
+      "step": 11176
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6941930573662264e-06,
+      "loss": 0.167,
+      "step": 11177
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6925610497507127e-06,
+      "loss": 0.1726,
+      "step": 11178
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6909293210745224e-06,
+      "loss": 0.1803,
+      "step": 11179
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6892978714098182e-06,
+      "loss": 0.1713,
+      "step": 11180
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6876667008287525e-06,
+      "loss": 0.1574,
+      "step": 11181
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.686035809403451e-06,
+      "loss": 0.17,
+      "step": 11182
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.68440519720605e-06,
+      "loss": 0.1778,
+      "step": 11183
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6827748643086592e-06,
+      "loss": 0.1646,
+      "step": 11184
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.681144810783369e-06,
+      "loss": 0.1523,
+      "step": 11185
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6795150367022814e-06,
+      "loss": 0.1606,
+      "step": 11186
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.677885542137456e-06,
+      "loss": 0.1587,
+      "step": 11187
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6762563271609663e-06,
+      "loss": 0.1619,
+      "step": 11188
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6746273918448627e-06,
+      "loss": 0.1693,
+      "step": 11189
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.672998736261172e-06,
+      "loss": 0.1535,
+      "step": 11190
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.671370360481934e-06,
+      "loss": 0.1754,
+      "step": 11191
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6697422645791514e-06,
+      "loss": 0.1911,
+      "step": 11192
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6681144486248244e-06,
+      "loss": 0.159,
+      "step": 11193
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.666486912690952e-06,
+      "loss": 0.1588,
+      "step": 11194
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.664859656849501e-06,
+      "loss": 0.1583,
+      "step": 11195
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6632326811724362e-06,
+      "loss": 0.165,
+      "step": 11196
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.66160598573171e-06,
+      "loss": 0.1655,
+      "step": 11197
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.65997957059926e-06,
+      "loss": 0.1456,
+      "step": 11198
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.658353435847014e-06,
+      "loss": 0.1473,
+      "step": 11199
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.656727581546886e-06,
+      "loss": 0.1823,
+      "step": 11200
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.655102007770777e-06,
+      "loss": 0.1712,
+      "step": 11201
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.653476714590575e-06,
+      "loss": 0.1892,
+      "step": 11202
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6518517020781583e-06,
+      "loss": 0.1686,
+      "step": 11203
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.650226970305392e-06,
+      "loss": 0.154,
+      "step": 11204
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6486025193441254e-06,
+      "loss": 0.1718,
+      "step": 11205
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6469783492662e-06,
+      "loss": 0.1652,
+      "step": 11206
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.645354460143443e-06,
+      "loss": 0.1636,
+      "step": 11207
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6437308520476677e-06,
+      "loss": 0.1404,
+      "step": 11208
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.64210752505068e-06,
+      "loss": 0.1701,
+      "step": 11209
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.64048447922426e-06,
+      "loss": 0.1568,
+      "step": 11210
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6388617146401962e-06,
+      "loss": 0.1756,
+      "step": 11211
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.637239231370251e-06,
+      "loss": 0.1522,
+      "step": 11212
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.635617029486169e-06,
+      "loss": 0.1802,
+      "step": 11213
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6339951090597025e-06,
+      "loss": 0.1633,
+      "step": 11214
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.63237347016257e-06,
+      "loss": 0.1569,
+      "step": 11215
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.630752112866486e-06,
+      "loss": 0.183,
+      "step": 11216
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.629131037243163e-06,
+      "loss": 0.1972,
+      "step": 11217
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6275102433642818e-06,
+      "loss": 0.1652,
+      "step": 11218
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.625889731301523e-06,
+      "loss": 0.1621,
+      "step": 11219
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6242695011265518e-06,
+      "loss": 0.1601,
+      "step": 11220
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6226495529110183e-06,
+      "loss": 0.1545,
+      "step": 11221
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6210298867265725e-06,
+      "loss": 0.1642,
+      "step": 11222
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6194105026448324e-06,
+      "loss": 0.1764,
+      "step": 11223
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.617791400737416e-06,
+      "loss": 0.165,
+      "step": 11224
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6161725810759286e-06,
+      "loss": 0.1677,
+      "step": 11225
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.614554043731957e-06,
+      "loss": 0.1848,
+      "step": 11226
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6129357887770832e-06,
+      "loss": 0.1668,
+      "step": 11227
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6113178162828687e-06,
+      "loss": 0.1867,
+      "step": 11228
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.60970012632087e-06,
+      "loss": 0.1769,
+      "step": 11229
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6080827189626254e-06,
+      "loss": 0.1763,
+      "step": 11230
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6064655942796645e-06,
+      "loss": 0.1781,
+      "step": 11231
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6048487523435007e-06,
+      "loss": 0.1463,
+      "step": 11232
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.603232193225639e-06,
+      "loss": 0.1872,
+      "step": 11233
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6016159169975686e-06,
+      "loss": 0.1821,
+      "step": 11234
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.599999923730768e-06,
+      "loss": 0.1729,
+      "step": 11235
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.598384213496703e-06,
+      "loss": 0.1627,
+      "step": 11236
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5967687863668268e-06,
+      "loss": 0.1696,
+      "step": 11237
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5951536424125776e-06,
+      "loss": 0.1554,
+      "step": 11238
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5935387817053857e-06,
+      "loss": 0.1619,
+      "step": 11239
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.591924204316669e-06,
+      "loss": 0.1534,
+      "step": 11240
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5903099103178206e-06,
+      "loss": 0.1713,
+      "step": 11241
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.588695899780241e-06,
+      "loss": 0.1913,
+      "step": 11242
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.587082172775307e-06,
+      "loss": 0.1589,
+      "step": 11243
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.585468729374374e-06,
+      "loss": 0.1905,
+      "step": 11244
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5838555696488096e-06,
+      "loss": 0.1702,
+      "step": 11245
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5822426936699416e-06,
+      "loss": 0.1583,
+      "step": 11246
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5806301015090993e-06,
+      "loss": 0.1553,
+      "step": 11247
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.579017793237608e-06,
+      "loss": 0.1436,
+      "step": 11248
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5774057689267582e-06,
+      "loss": 0.1698,
+      "step": 11249
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5757940286478456e-06,
+      "loss": 0.1852,
+      "step": 11250
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.574182572472146e-06,
+      "loss": 0.1696,
+      "step": 11251
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.572571400470921e-06,
+      "loss": 0.1421,
+      "step": 11252
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5709605127154346e-06,
+      "loss": 0.1817,
+      "step": 11253
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5693499092769137e-06,
+      "loss": 0.1651,
+      "step": 11254
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5677395902265898e-06,
+      "loss": 0.1751,
+      "step": 11255
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.566129555635679e-06,
+      "loss": 0.1657,
+      "step": 11256
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5645198055753815e-06,
+      "loss": 0.1302,
+      "step": 11257
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5629103401168864e-06,
+      "loss": 0.1783,
+      "step": 11258
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5613011593313718e-06,
+      "loss": 0.171,
+      "step": 11259
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.559692263290002e-06,
+      "loss": 0.1534,
+      "step": 11260
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.558083652063927e-06,
+      "loss": 0.1915,
+      "step": 11261
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.556475325724288e-06,
+      "loss": 0.1581,
+      "step": 11262
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5548672843422114e-06,
+      "loss": 0.1781,
+      "step": 11263
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.553259527988805e-06,
+      "loss": 0.1566,
+      "step": 11264
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5516520567351775e-06,
+      "loss": 0.1619,
+      "step": 11265
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.550044870652418e-06,
+      "loss": 0.1762,
+      "step": 11266
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5484379698115924e-06,
+      "loss": 0.1567,
+      "step": 11267
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5468313542837773e-06,
+      "loss": 0.1533,
+      "step": 11268
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.545225024140011e-06,
+      "loss": 0.164,
+      "step": 11269
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.543618979451341e-06,
+      "loss": 0.1721,
+      "step": 11270
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5420132202887923e-06,
+      "loss": 0.1447,
+      "step": 11271
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.540407746723369e-06,
+      "loss": 0.1664,
+      "step": 11272
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5388025588260834e-06,
+      "loss": 0.194,
+      "step": 11273
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.537197656667913e-06,
+      "loss": 0.1649,
+      "step": 11274
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.535593040319835e-06,
+      "loss": 0.1725,
+      "step": 11275
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.533988709852819e-06,
+      "loss": 0.1614,
+      "step": 11276
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.532384665337807e-06,
+      "loss": 0.1832,
+      "step": 11277
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5307809068457367e-06,
+      "loss": 0.1607,
+      "step": 11278
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.529177434447535e-06,
+      "loss": 0.186,
+      "step": 11279
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.527574248214113e-06,
+      "loss": 0.1656,
+      "step": 11280
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5259713482163682e-06,
+      "loss": 0.2076,
+      "step": 11281
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5243687345251896e-06,
+      "loss": 0.1526,
+      "step": 11282
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.522766407211449e-06,
+      "loss": 0.1568,
+      "step": 11283
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5211643663460083e-06,
+      "loss": 0.1586,
+      "step": 11284
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5195626119997163e-06,
+      "loss": 0.1679,
+      "step": 11285
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.517961144243408e-06,
+      "loss": 0.1538,
+      "step": 11286
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.516359963147906e-06,
+      "loss": 0.1566,
+      "step": 11287
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5147590687840226e-06,
+      "loss": 0.1671,
+      "step": 11288
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5131584612225546e-06,
+      "loss": 0.1711,
+      "step": 11289
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5115581405342857e-06,
+      "loss": 0.1844,
+      "step": 11290
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5099581067899935e-06,
+      "loss": 0.1636,
+      "step": 11291
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5083583600604276e-06,
+      "loss": 0.1753,
+      "step": 11292
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5067589004163426e-06,
+      "loss": 0.1746,
+      "step": 11293
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5051597279284765e-06,
+      "loss": 0.1624,
+      "step": 11294
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5035608426675385e-06,
+      "loss": 0.1538,
+      "step": 11295
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5019622447042502e-06,
+      "loss": 0.1824,
+      "step": 11296
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.500363934109299e-06,
+      "loss": 0.1552,
+      "step": 11297
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.498765910953369e-06,
+      "loss": 0.1811,
+      "step": 11298
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4971681753071397e-06,
+      "loss": 0.1433,
+      "step": 11299
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.49557072724126e-06,
+      "loss": 0.163,
+      "step": 11300
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4939735668263776e-06,
+      "loss": 0.169,
+      "step": 11301
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.492376694133125e-06,
+      "loss": 0.1768,
+      "step": 11302
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4907801092321203e-06,
+      "loss": 0.1781,
+      "step": 11303
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4891838121939804e-06,
+      "loss": 0.1841,
+      "step": 11304
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4875878030892885e-06,
+      "loss": 0.1744,
+      "step": 11305
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4859920819886295e-06,
+      "loss": 0.1613,
+      "step": 11306
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4843966489625735e-06,
+      "loss": 0.172,
+      "step": 11307
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4828015040816763e-06,
+      "loss": 0.1757,
+      "step": 11308
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4812066474164817e-06,
+      "loss": 0.1494,
+      "step": 11309
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4796120790375197e-06,
+      "loss": 0.164,
+      "step": 11310
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4780177990153087e-06,
+      "loss": 0.1692,
+      "step": 11311
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4764238074203536e-06,
+      "loss": 0.1704,
+      "step": 11312
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.474830104323148e-06,
+      "loss": 0.1856,
+      "step": 11313
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.473236689794174e-06,
+      "loss": 0.1694,
+      "step": 11314
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.471643563903888e-06,
+      "loss": 0.1587,
+      "step": 11315
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4700507267227544e-06,
+      "loss": 0.1615,
+      "step": 11316
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.468458178321217e-06,
+      "loss": 0.1429,
+      "step": 11317
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.466865918769691e-06,
+      "loss": 0.1582,
+      "step": 11318
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.465273948138607e-06,
+      "loss": 0.1764,
+      "step": 11319
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.463682266498355e-06,
+      "loss": 0.1484,
+      "step": 11320
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.462090873919335e-06,
+      "loss": 0.1597,
+      "step": 11321
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4604997704719244e-06,
+      "loss": 0.1745,
+      "step": 11322
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4589089562264787e-06,
+      "loss": 0.1667,
+      "step": 11323
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4573184312533623e-06,
+      "loss": 0.1702,
+      "step": 11324
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.455728195622906e-06,
+      "loss": 0.1543,
+      "step": 11325
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4541382494054343e-06,
+      "loss": 0.1573,
+      "step": 11326
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.452548592671271e-06,
+      "loss": 0.1597,
+      "step": 11327
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4509592254907077e-06,
+      "loss": 0.1826,
+      "step": 11328
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.449370147934035e-06,
+      "loss": 0.1667,
+      "step": 11329
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.447781360071528e-06,
+      "loss": 0.1799,
+      "step": 11330
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4461928619734496e-06,
+      "loss": 0.1346,
+      "step": 11331
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4446046537100498e-06,
+      "loss": 0.1548,
+      "step": 11332
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4430167353515633e-06,
+      "loss": 0.1845,
+      "step": 11333
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.441429106968216e-06,
+      "loss": 0.163,
+      "step": 11334
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4398417686302175e-06,
+      "loss": 0.1509,
+      "step": 11335
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.438254720407768e-06,
+      "loss": 0.1645,
+      "step": 11336
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.436667962371051e-06,
+      "loss": 0.1451,
+      "step": 11337
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.43508149459024e-06,
+      "loss": 0.1712,
+      "step": 11338
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4334953171354936e-06,
+      "loss": 0.1879,
+      "step": 11339
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.431909430076961e-06,
+      "loss": 0.1729,
+      "step": 11340
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4303238334847734e-06,
+      "loss": 0.1734,
+      "step": 11341
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4287385274290573e-06,
+      "loss": 0.176,
+      "step": 11342
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4271535119799104e-06,
+      "loss": 0.1695,
+      "step": 11343
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4255687872074395e-06,
+      "loss": 0.1549,
+      "step": 11344
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.423984353181725e-06,
+      "loss": 0.1572,
+      "step": 11345
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.422400209972828e-06,
+      "loss": 0.1632,
+      "step": 11346
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.420816357650818e-06,
+      "loss": 0.1712,
+      "step": 11347
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4192327962857307e-06,
+      "loss": 0.1641,
+      "step": 11348
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.417649525947595e-06,
+      "loss": 0.1538,
+      "step": 11349
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4160665467064413e-06,
+      "loss": 0.1743,
+      "step": 11350
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4144838586322637e-06,
+      "loss": 0.1736,
+      "step": 11351
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4129014617950595e-06,
+      "loss": 0.2,
+      "step": 11352
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4113193562648072e-06,
+      "loss": 0.1681,
+      "step": 11353
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4097375421114697e-06,
+      "loss": 0.1815,
+      "step": 11354
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4081560194050122e-06,
+      "loss": 0.199,
+      "step": 11355
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4065747882153668e-06,
+      "loss": 0.1589,
+      "step": 11356
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.404993848612462e-06,
+      "loss": 0.1766,
+      "step": 11357
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.403413200666216e-06,
+      "loss": 0.1909,
+      "step": 11358
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4018328444465286e-06,
+      "loss": 0.1872,
+      "step": 11359
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.400252780023291e-06,
+      "loss": 0.1632,
+      "step": 11360
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.3986730074663787e-06,
+      "loss": 0.173,
+      "step": 11361
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.397093526845656e-06,
+      "loss": 0.1566,
+      "step": 11362
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.395514338230974e-06,
+      "loss": 0.1825,
+      "step": 11363
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.39393544169217e-06,
+      "loss": 0.1634,
+      "step": 11364
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.392356837299069e-06,
+      "loss": 0.1722,
+      "step": 11365
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.390778525121482e-06,
+      "loss": 0.1479,
+      "step": 11366
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.3892005052292098e-06,
+      "loss": 0.1888,
+      "step": 11367
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.3876227776920377e-06,
+      "loss": 0.158,
+      "step": 11368
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.3860453425797392e-06,
+      "loss": 0.1602,
+      "step": 11369
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.384468199962073e-06,
+      "loss": 0.171,
+      "step": 11370
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.382891349908789e-06,
+      "loss": 0.1645,
+      "step": 11371
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3813147924896194e-06,
+      "loss": 0.1685,
+      "step": 11372
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.379738527774291e-06,
+      "loss": 0.1842,
+      "step": 11373
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3781625558325003e-06,
+      "loss": 0.1857,
+      "step": 11374
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3765868767339547e-06,
+      "loss": 0.1736,
+      "step": 11375
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3750114905483355e-06,
+      "loss": 0.1689,
+      "step": 11376
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.373436397345303e-06,
+      "loss": 0.2255,
+      "step": 11377
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.371861597194527e-06,
+      "loss": 0.1621,
+      "step": 11378
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3702870901656417e-06,
+      "loss": 0.1474,
+      "step": 11379
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.368712876328277e-06,
+      "loss": 0.1617,
+      "step": 11380
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.367138955752062e-06,
+      "loss": 0.1768,
+      "step": 11381
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3655653285065893e-06,
+      "loss": 0.1679,
+      "step": 11382
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3639919946614564e-06,
+      "loss": 0.1567,
+      "step": 11383
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3624189542862405e-06,
+      "loss": 0.1548,
+      "step": 11384
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3608462074505064e-06,
+      "loss": 0.1627,
+      "step": 11385
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3592737542238154e-06,
+      "loss": 0.1731,
+      "step": 11386
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3577015946756974e-06,
+      "loss": 0.171,
+      "step": 11387
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.356129728875683e-06,
+      "loss": 0.1554,
+      "step": 11388
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.354558156893287e-06,
+      "loss": 0.1761,
+      "step": 11389
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.352986878798009e-06,
+      "loss": 0.1743,
+      "step": 11390
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3514158946593377e-06,
+      "loss": 0.171,
+      "step": 11391
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3498452045467477e-06,
+      "loss": 0.1502,
+      "step": 11392
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3482748085297025e-06,
+      "loss": 0.1806,
+      "step": 11393
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.346704706677648e-06,
+      "loss": 0.1602,
+      "step": 11394
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.345134899060023e-06,
+      "loss": 0.1675,
+      "step": 11395
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3435653857462536e-06,
+      "loss": 0.1481,
+      "step": 11396
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3419961668057378e-06,
+      "loss": 0.1793,
+      "step": 11397
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3404272423078844e-06,
+      "loss": 0.158,
+      "step": 11398
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3388586123220768e-06,
+      "loss": 0.1904,
+      "step": 11399
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.337290276917675e-06,
+      "loss": 0.1694,
+      "step": 11400
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.335722236164052e-06,
+      "loss": 0.1616,
+      "step": 11401
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3341544901305367e-06,
+      "loss": 0.1776,
+      "step": 11402
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.332587038886473e-06,
+      "loss": 0.1899,
+      "step": 11403
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3310198825011795e-06,
+      "loss": 0.2069,
+      "step": 11404
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.32945302104395e-06,
+      "loss": 0.1616,
+      "step": 11405
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3278864545840937e-06,
+      "loss": 0.16,
+      "step": 11406
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3263201831908775e-06,
+      "loss": 0.166,
+      "step": 11407
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3247542069335692e-06,
+      "loss": 0.1773,
+      "step": 11408
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3231885258814322e-06,
+      "loss": 0.1724,
+      "step": 11409
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3216231401036956e-06,
+      "loss": 0.1892,
+      "step": 11410
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3200580496695935e-06,
+      "loss": 0.1638,
+      "step": 11411
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3184932546483363e-06,
+      "loss": 0.1847,
+      "step": 11412
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3169287551091277e-06,
+      "loss": 0.1632,
+      "step": 11413
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3153645511211553e-06,
+      "loss": 0.1701,
+      "step": 11414
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.313800642753594e-06,
+      "loss": 0.1571,
+      "step": 11415
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3122370300756067e-06,
+      "loss": 0.1895,
+      "step": 11416
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3106737131563417e-06,
+      "loss": 0.1525,
+      "step": 11417
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.309110692064934e-06,
+      "loss": 0.1725,
+      "step": 11418
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.307547966870509e-06,
+      "loss": 0.1965,
+      "step": 11419
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3059855376421735e-06,
+      "loss": 0.1576,
+      "step": 11420
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3044234044490264e-06,
+      "loss": 0.1838,
+      "step": 11421
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3028615673601506e-06,
+      "loss": 0.1851,
+      "step": 11422
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.301300026444616e-06,
+      "loss": 0.1587,
+      "step": 11423
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.299738781771483e-06,
+      "loss": 0.1637,
+      "step": 11424
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.298177833409788e-06,
+      "loss": 0.171,
+      "step": 11425
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2966171814285697e-06,
+      "loss": 0.1863,
+      "step": 11426
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2950568258968474e-06,
+      "loss": 0.155,
+      "step": 11427
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.293496766883616e-06,
+      "loss": 0.1825,
+      "step": 11428
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.291937004457881e-06,
+      "loss": 0.1712,
+      "step": 11429
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2903775386886106e-06,
+      "loss": 0.1673,
+      "step": 11430
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2888183696447697e-06,
+      "loss": 0.1498,
+      "step": 11431
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.287259497395322e-06,
+      "loss": 0.1413,
+      "step": 11432
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.285700922009196e-06,
+      "loss": 0.1702,
+      "step": 11433
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2841426435553227e-06,
+      "loss": 0.1813,
+      "step": 11434
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2825846621026123e-06,
+      "loss": 0.1595,
+      "step": 11435
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2810269777199657e-06,
+      "loss": 0.1617,
+      "step": 11436
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2794695904762753e-06,
+      "loss": 0.1509,
+      "step": 11437
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2779125004404076e-06,
+      "loss": 0.1363,
+      "step": 11438
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.276355707681226e-06,
+      "loss": 0.1427,
+      "step": 11439
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2747992122675765e-06,
+      "loss": 0.1643,
+      "step": 11440
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2732430142682945e-06,
+      "loss": 0.1642,
+      "step": 11441
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.271687113752202e-06,
+      "loss": 0.1757,
+      "step": 11442
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.270131510788105e-06,
+      "loss": 0.1859,
+      "step": 11443
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2685762054448e-06,
+      "loss": 0.1629,
+      "step": 11444
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.267021197791067e-06,
+      "loss": 0.1594,
+      "step": 11445
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2654664878956755e-06,
+      "loss": 0.154,
+      "step": 11446
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2639120758273835e-06,
+      "loss": 0.184,
+      "step": 11447
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.262357961654924e-06,
+      "loss": 0.175,
+      "step": 11448
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.260804145447035e-06,
+      "loss": 0.1407,
+      "step": 11449
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2592506272724333e-06,
+      "loss": 0.1589,
+      "step": 11450
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.25769740719981e-06,
+      "loss": 0.1855,
+      "step": 11451
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2561444852978697e-06,
+      "loss": 0.1735,
+      "step": 11452
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2545918616352736e-06,
+      "loss": 0.1483,
+      "step": 11453
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2530395362806956e-06,
+      "loss": 0.1614,
+      "step": 11454
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2514875093027855e-06,
+      "loss": 0.1539,
+      "step": 11455
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2499357807701683e-06,
+      "loss": 0.1817,
+      "step": 11456
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2483843507514833e-06,
+      "loss": 0.1649,
+      "step": 11457
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2468332193153295e-06,
+      "loss": 0.1574,
+      "step": 11458
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.245282386530304e-06,
+      "loss": 0.1541,
+      "step": 11459
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2437318524650006e-06,
+      "loss": 0.1705,
+      "step": 11460
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2421816171879805e-06,
+      "loss": 0.166,
+      "step": 11461
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.240631680767803e-06,
+      "loss": 0.1772,
+      "step": 11462
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.239082043273013e-06,
+      "loss": 0.1473,
+      "step": 11463
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.237532704772143e-06,
+      "loss": 0.142,
+      "step": 11464
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2359836653337085e-06,
+      "loss": 0.1632,
+      "step": 11465
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2344349250262143e-06,
+      "loss": 0.1863,
+      "step": 11466
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.232886483918153e-06,
+      "loss": 0.1949,
+      "step": 11467
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.231338342078002e-06,
+      "loss": 0.1733,
+      "step": 11468
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2297904995742256e-06,
+      "loss": 0.1604,
+      "step": 11469
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.228242956475276e-06,
+      "loss": 0.1485,
+      "step": 11470
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.226695712849591e-06,
+      "loss": 0.1494,
+      "step": 11471
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.225148768765597e-06,
+      "loss": 0.1965,
+      "step": 11472
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2236021242917038e-06,
+      "loss": 0.1466,
+      "step": 11473
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2220557794963116e-06,
+      "loss": 0.1937,
+      "step": 11474
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2205097344478075e-06,
+      "loss": 0.1706,
+      "step": 11475
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.218963989214556e-06,
+      "loss": 0.1579,
+      "step": 11476
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2174185438649243e-06,
+      "loss": 0.1594,
+      "step": 11477
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2158733984672575e-06,
+      "loss": 0.1724,
+      "step": 11478
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2143285530898795e-06,
+      "loss": 0.1839,
+      "step": 11479
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.212784007801121e-06,
+      "loss": 0.1929,
+      "step": 11480
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.211239762669279e-06,
+      "loss": 0.1788,
+      "step": 11481
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.209695817762646e-06,
+      "loss": 0.1779,
+      "step": 11482
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.208152173149509e-06,
+      "loss": 0.1635,
+      "step": 11483
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2066088288981266e-06,
+      "loss": 0.1906,
+      "step": 11484
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.205065785076754e-06,
+      "loss": 0.1576,
+      "step": 11485
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.203523041753631e-06,
+      "loss": 0.1553,
+      "step": 11486
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.20198059899698e-06,
+      "loss": 0.1825,
+      "step": 11487
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2004384568750224e-06,
+      "loss": 0.1434,
+      "step": 11488
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1988966154559486e-06,
+      "loss": 0.154,
+      "step": 11489
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1973550748079487e-06,
+      "loss": 0.1481,
+      "step": 11490
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1958138349991964e-06,
+      "loss": 0.1658,
+      "step": 11491
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.194272896097849e-06,
+      "loss": 0.1748,
+      "step": 11492
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1927322581720534e-06,
+      "loss": 0.1751,
+      "step": 11493
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1911919212899435e-06,
+      "loss": 0.166,
+      "step": 11494
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.189651885519639e-06,
+      "loss": 0.1684,
+      "step": 11495
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1881121509292445e-06,
+      "loss": 0.1648,
+      "step": 11496
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1865727175868555e-06,
+      "loss": 0.1884,
+      "step": 11497
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1850335855605496e-06,
+      "loss": 0.169,
+      "step": 11498
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.183494754918395e-06,
+      "loss": 0.1826,
+      "step": 11499
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.181956225728443e-06,
+      "loss": 0.1542,
+      "step": 11500
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1804179980587336e-06,
+      "loss": 0.1495,
+      "step": 11501
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.178880071977294e-06,
+      "loss": 0.1742,
+      "step": 11502
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1773424475521376e-06,
+      "loss": 0.1492,
+      "step": 11503
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1758051248512623e-06,
+      "loss": 0.1634,
+      "step": 11504
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.174268103942656e-06,
+      "loss": 0.1727,
+      "step": 11505
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.172731384894294e-06,
+      "loss": 0.1655,
+      "step": 11506
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.171194967774127e-06,
+      "loss": 0.1715,
+      "step": 11507
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1696588526501102e-06,
+      "loss": 0.1464,
+      "step": 11508
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1681230395901774e-06,
+      "loss": 0.1713,
+      "step": 11509
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1665875286622383e-06,
+      "loss": 0.1674,
+      "step": 11510
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.165052319934212e-06,
+      "loss": 0.1648,
+      "step": 11511
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.163517413473981e-06,
+      "loss": 0.155,
+      "step": 11512
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1619828093494254e-06,
+      "loss": 0.1309,
+      "step": 11513
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1604485076284207e-06,
+      "loss": 0.1681,
+      "step": 11514
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.158914508378811e-06,
+      "loss": 0.1891,
+      "step": 11515
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1573808116684377e-06,
+      "loss": 0.1728,
+      "step": 11516
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1558474175651267e-06,
+      "loss": 0.1584,
+      "step": 11517
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1543143261366913e-06,
+      "loss": 0.185,
+      "step": 11518
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.152781537450932e-06,
+      "loss": 0.1621,
+      "step": 11519
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.151249051575631e-06,
+      "loss": 0.1714,
+      "step": 11520
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.149716868578564e-06,
+      "loss": 0.1656,
+      "step": 11521
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1481849885274897e-06,
+      "loss": 0.1878,
+      "step": 11522
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1466534114901517e-06,
+      "loss": 0.1596,
+      "step": 11523
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1451221375342846e-06,
+      "loss": 0.1527,
+      "step": 11524
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1435911667276053e-06,
+      "loss": 0.1889,
+      "step": 11525
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.142060499137821e-06,
+      "loss": 0.1875,
+      "step": 11526
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1405301348326225e-06,
+      "loss": 0.1523,
+      "step": 11527
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1390000738796887e-06,
+      "loss": 0.1699,
+      "step": 11528
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1374703163466892e-06,
+      "loss": 0.1977,
+      "step": 11529
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1359408623012643e-06,
+      "loss": 0.1496,
+      "step": 11530
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1344117118110627e-06,
+      "loss": 0.1654,
+      "step": 11531
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1328828649437105e-06,
+      "loss": 0.1846,
+      "step": 11532
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1313543217668075e-06,
+      "loss": 0.1623,
+      "step": 11533
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1298260823479654e-06,
+      "loss": 0.1905,
+      "step": 11534
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1282981467547568e-06,
+      "loss": 0.1621,
+      "step": 11535
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.126770515054762e-06,
+      "loss": 0.1545,
+      "step": 11536
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.12524318731554e-06,
+      "loss": 0.1458,
+      "step": 11537
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1237161636046233e-06,
+      "loss": 0.1644,
+      "step": 11538
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1221894439895563e-06,
+      "loss": 0.1574,
+      "step": 11539
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.120663028537847e-06,
+      "loss": 0.1625,
+      "step": 11540
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1191369173170007e-06,
+      "loss": 0.1829,
+      "step": 11541
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.117611110394515e-06,
+      "loss": 0.1591,
+      "step": 11542
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1160856078378597e-06,
+      "loss": 0.1729,
+      "step": 11543
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.114560409714501e-06,
+      "loss": 0.162,
+      "step": 11544
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.113035516091888e-06,
+      "loss": 0.1596,
+      "step": 11545
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.111510927037459e-06,
+      "loss": 0.1682,
+      "step": 11546
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1099866426186355e-06,
+      "loss": 0.17,
+      "step": 11547
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.108462662902828e-06,
+      "loss": 0.1512,
+      "step": 11548
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.106938987957433e-06,
+      "loss": 0.1475,
+      "step": 11549
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.105415617849834e-06,
+      "loss": 0.1579,
+      "step": 11550
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1038925526473984e-06,
+      "loss": 0.1421,
+      "step": 11551
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1023697924174833e-06,
+      "loss": 0.1395,
+      "step": 11552
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.100847337227432e-06,
+      "loss": 0.1601,
+      "step": 11553
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.099325187144571e-06,
+      "loss": 0.1679,
+      "step": 11554
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0978033422362174e-06,
+      "loss": 0.1973,
+      "step": 11555
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.096281802569673e-06,
+      "loss": 0.1496,
+      "step": 11556
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0947605682122284e-06,
+      "loss": 0.1652,
+      "step": 11557
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.093239639231149e-06,
+      "loss": 0.1647,
+      "step": 11558
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.091719015693706e-06,
+      "loss": 0.1688,
+      "step": 11559
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0901986976671485e-06,
+      "loss": 0.1676,
+      "step": 11560
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.088678685218699e-06,
+      "loss": 0.1486,
+      "step": 11561
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.087158978415592e-06,
+      "loss": 0.1792,
+      "step": 11562
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0856395773250256e-06,
+      "loss": 0.1847,
+      "step": 11563
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0841204820141944e-06,
+      "loss": 0.1455,
+      "step": 11564
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0826016925502865e-06,
+      "loss": 0.1683,
+      "step": 11565
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0810832090004604e-06,
+      "loss": 0.1533,
+      "step": 11566
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0795650314318716e-06,
+      "loss": 0.1649,
+      "step": 11567
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.07804715991166e-06,
+      "loss": 0.1525,
+      "step": 11568
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0765295945069485e-06,
+      "loss": 0.1754,
+      "step": 11569
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0750123352848603e-06,
+      "loss": 0.1727,
+      "step": 11570
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0734953823124834e-06,
+      "loss": 0.1601,
+      "step": 11571
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0719787356569075e-06,
+      "loss": 0.1512,
+      "step": 11572
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.070462395385203e-06,
+      "loss": 0.168,
+      "step": 11573
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.068946361564431e-06,
+      "loss": 0.1774,
+      "step": 11574
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0674306342616345e-06,
+      "loss": 0.1492,
+      "step": 11575
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0659152135438453e-06,
+      "loss": 0.1917,
+      "step": 11576
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.064400099478081e-06,
+      "loss": 0.1503,
+      "step": 11577
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0628852921313455e-06,
+      "loss": 0.153,
+      "step": 11578
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0613707915706305e-06,
+      "loss": 0.1552,
+      "step": 11579
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.059856597862916e-06,
+      "loss": 0.1564,
+      "step": 11580
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.058342711075154e-06,
+      "loss": 0.1746,
+      "step": 11581
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.056829131274307e-06,
+      "loss": 0.179,
+      "step": 11582
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.055315858527309e-06,
+      "loss": 0.1859,
+      "step": 11583
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0538028929010743e-06,
+      "loss": 0.1719,
+      "step": 11584
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.052290234462525e-06,
+      "loss": 0.1684,
+      "step": 11585
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.050777883278543e-06,
+      "loss": 0.1646,
+      "step": 11586
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0492658394160203e-06,
+      "loss": 0.1791,
+      "step": 11587
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0477541029418257e-06,
+      "loss": 0.1697,
+      "step": 11588
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0462426739228045e-06,
+      "loss": 0.1868,
+      "step": 11589
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.04473155242581e-06,
+      "loss": 0.1811,
+      "step": 11590
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0432207385176616e-06,
+      "loss": 0.2074,
+      "step": 11591
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0417102322651716e-06,
+      "loss": 0.1591,
+      "step": 11592
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.040200033735151e-06,
+      "loss": 0.1634,
+      "step": 11593
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.038690142994377e-06,
+      "loss": 0.1612,
+      "step": 11594
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.037180560109625e-06,
+      "loss": 0.1597,
+      "step": 11595
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.035671285147657e-06,
+      "loss": 0.1815,
+      "step": 11596
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0341623181752155e-06,
+      "loss": 0.1654,
+      "step": 11597
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.032653659259036e-06,
+      "loss": 0.1817,
+      "step": 11598
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.031145308465836e-06,
+      "loss": 0.1786,
+      "step": 11599
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0296372658623195e-06,
+      "loss": 0.1357,
+      "step": 11600
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.02812953151518e-06,
+      "loss": 0.1758,
+      "step": 11601
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0266221054910936e-06,
+      "loss": 0.1478,
+      "step": 11602
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0251149878567267e-06,
+      "loss": 0.1358,
+      "step": 11603
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0236081786787275e-06,
+      "loss": 0.1611,
+      "step": 11604
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0221016780237345e-06,
+      "loss": 0.1676,
+      "step": 11605
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.02059548595837e-06,
+      "loss": 0.175,
+      "step": 11606
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0190896025492435e-06,
+      "loss": 0.1874,
+      "step": 11607
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.017584027862954e-06,
+      "loss": 0.1796,
+      "step": 11608
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.016078761966077e-06,
+      "loss": 0.1639,
+      "step": 11609
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.014573804925187e-06,
+      "loss": 0.1923,
+      "step": 11610
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0130691568068416e-06,
+      "loss": 0.1672,
+      "step": 11611
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0115648176775716e-06,
+      "loss": 0.1716,
+      "step": 11612
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0100607876039166e-06,
+      "loss": 0.1562,
+      "step": 11613
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0085570666523833e-06,
+      "loss": 0.1722,
+      "step": 11614
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0070536548894703e-06,
+      "loss": 0.1734,
+      "step": 11615
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.005550552381674e-06,
+      "loss": 0.142,
+      "step": 11616
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0040477591954576e-06,
+      "loss": 0.1546,
+      "step": 11617
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.002545275397285e-06,
+      "loss": 0.1623,
+      "step": 11618
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.001043101053601e-06,
+      "loss": 0.1652,
+      "step": 11619
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.999541236230834e-06,
+      "loss": 0.2077,
+      "step": 11620
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.9980396809954114e-06,
+      "loss": 0.1742,
+      "step": 11621
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.9965384354137305e-06,
+      "loss": 0.1534,
+      "step": 11622
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.9950374995521824e-06,
+      "loss": 0.2011,
+      "step": 11623
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.9935368734771477e-06,
+      "loss": 0.162,
+      "step": 11624
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.992036557254986e-06,
+      "loss": 0.1835,
+      "step": 11625
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.990536550952049e-06,
+      "loss": 0.1916,
+      "step": 11626
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.989036854634674e-06,
+      "loss": 0.1571,
+      "step": 11627
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.9875374683691816e-06,
+      "loss": 0.1661,
+      "step": 11628
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9860383922218805e-06,
+      "loss": 0.177,
+      "step": 11629
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9845396262590667e-06,
+      "loss": 0.1607,
+      "step": 11630
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9830411705470208e-06,
+      "loss": 0.1685,
+      "step": 11631
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9815430251520092e-06,
+      "loss": 0.1483,
+      "step": 11632
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9800451901402882e-06,
+      "loss": 0.1524,
+      "step": 11633
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9785476655780965e-06,
+      "loss": 0.1991,
+      "step": 11634
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.97705045153166e-06,
+      "loss": 0.1801,
+      "step": 11635
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9755535480671917e-06,
+      "loss": 0.1519,
+      "step": 11636
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.97405695525089e-06,
+      "loss": 0.1722,
+      "step": 11637
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.972560673148941e-06,
+      "loss": 0.1631,
+      "step": 11638
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9710647018275186e-06,
+      "loss": 0.1496,
+      "step": 11639
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.969569041352771e-06,
+      "loss": 0.1837,
+      "step": 11640
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9680736917908513e-06,
+      "loss": 0.1876,
+      "step": 11641
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.96657865320789e-06,
+      "loss": 0.1805,
+      "step": 11642
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.965083925669995e-06,
+      "loss": 0.1679,
+      "step": 11643
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.96358950924328e-06,
+      "loss": 0.1632,
+      "step": 11644
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9620954039938245e-06,
+      "loss": 0.1554,
+      "step": 11645
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9606016099877042e-06,
+      "loss": 0.1498,
+      "step": 11646
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9591081272909894e-06,
+      "loss": 0.1723,
+      "step": 11647
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.957614955969719e-06,
+      "loss": 0.1759,
+      "step": 11648
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9561220960899296e-06,
+      "loss": 0.1664,
+      "step": 11649
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9546295477176412e-06,
+      "loss": 0.1899,
+      "step": 11650
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.95313731091886e-06,
+      "loss": 0.1744,
+      "step": 11651
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9516453857595773e-06,
+      "loss": 0.1911,
+      "step": 11652
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.950153772305774e-06,
+      "loss": 0.1411,
+      "step": 11653
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9486624706234144e-06,
+      "loss": 0.1463,
+      "step": 11654
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.947171480778449e-06,
+      "loss": 0.1427,
+      "step": 11655
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9456808028368144e-06,
+      "loss": 0.1539,
+      "step": 11656
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.944190436864436e-06,
+      "loss": 0.177,
+      "step": 11657
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9427003829272225e-06,
+      "loss": 0.181,
+      "step": 11658
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9412106410910703e-06,
+      "loss": 0.1697,
+      "step": 11659
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9397212114218596e-06,
+      "loss": 0.1597,
+      "step": 11660
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9382320939854614e-06,
+      "loss": 0.1549,
+      "step": 11661
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.936743288847733e-06,
+      "loss": 0.1656,
+      "step": 11662
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9352547960745046e-06,
+      "loss": 0.176,
+      "step": 11663
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9337666157316123e-06,
+      "loss": 0.1698,
+      "step": 11664
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9322787478848703e-06,
+      "loss": 0.1714,
+      "step": 11665
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9307911926000676e-06,
+      "loss": 0.1671,
+      "step": 11666
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9293039499430034e-06,
+      "loss": 0.1471,
+      "step": 11667
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9278170199794354e-06,
+      "loss": 0.1498,
+      "step": 11668
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9263304027751315e-06,
+      "loss": 0.1572,
+      "step": 11669
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.924844098395835e-06,
+      "loss": 0.1429,
+      "step": 11670
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.923358106907267e-06,
+      "loss": 0.1879,
+      "step": 11671
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9218724283751575e-06,
+      "loss": 0.1712,
+      "step": 11672
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9203870628651975e-06,
+      "loss": 0.1857,
+      "step": 11673
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9189020104430766e-06,
+      "loss": 0.1944,
+      "step": 11674
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9174172711744798e-06,
+      "loss": 0.1715,
+      "step": 11675
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9159328451250567e-06,
+      "loss": 0.1949,
+      "step": 11676
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.914448732360459e-06,
+      "loss": 0.1765,
+      "step": 11677
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9129649329463193e-06,
+      "loss": 0.1558,
+      "step": 11678
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.911481446948258e-06,
+      "loss": 0.1619,
+      "step": 11679
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.909998274431878e-06,
+      "loss": 0.1439,
+      "step": 11680
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.908515415462774e-06,
+      "loss": 0.1715,
+      "step": 11681
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.907032870106521e-06,
+      "loss": 0.149,
+      "step": 11682
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9055506384286847e-06,
+      "loss": 0.1762,
+      "step": 11683
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.904068720494815e-06,
+      "loss": 0.1359,
+      "step": 11684
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.902587116370448e-06,
+      "loss": 0.1667,
+      "step": 11685
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.9011058261211043e-06,
+      "loss": 0.1539,
+      "step": 11686
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.899624849812295e-06,
+      "loss": 0.2003,
+      "step": 11687
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.898144187509513e-06,
+      "loss": 0.1572,
+      "step": 11688
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.896663839278239e-06,
+      "loss": 0.1632,
+      "step": 11689
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.895183805183942e-06,
+      "loss": 0.1816,
+      "step": 11690
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8937040852920674e-06,
+      "loss": 0.1413,
+      "step": 11691
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.892224679668062e-06,
+      "loss": 0.1732,
+      "step": 11692
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.890745588377352e-06,
+      "loss": 0.1366,
+      "step": 11693
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.889266811485337e-06,
+      "loss": 0.1865,
+      "step": 11694
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8877883490574298e-06,
+      "loss": 0.1439,
+      "step": 11695
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8863102011590028e-06,
+      "loss": 0.1518,
+      "step": 11696
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8848323678554246e-06,
+      "loss": 0.1643,
+      "step": 11697
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8833548492120613e-06,
+      "loss": 0.1413,
+      "step": 11698
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8818776452942444e-06,
+      "loss": 0.1659,
+      "step": 11699
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8804007561673053e-06,
+      "loss": 0.192,
+      "step": 11700
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8789241818965573e-06,
+      "loss": 0.1548,
+      "step": 11701
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.877447922547296e-06,
+      "loss": 0.1685,
+      "step": 11702
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8759719781848195e-06,
+      "loss": 0.1433,
+      "step": 11703
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8744963488743884e-06,
+      "loss": 0.1477,
+      "step": 11704
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8730210346812636e-06,
+      "loss": 0.1817,
+      "step": 11705
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.87154603567069e-06,
+      "loss": 0.1804,
+      "step": 11706
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8700713519078984e-06,
+      "loss": 0.1747,
+      "step": 11707
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.868596983458104e-06,
+      "loss": 0.1634,
+      "step": 11708
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8671229303865088e-06,
+      "loss": 0.1387,
+      "step": 11709
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.865649192758302e-06,
+      "loss": 0.1591,
+      "step": 11710
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8641757706386565e-06,
+      "loss": 0.171,
+      "step": 11711
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8627026640927345e-06,
+      "loss": 0.1898,
+      "step": 11712
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8612298731856845e-06,
+      "loss": 0.1535,
+      "step": 11713
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.859757397982631e-06,
+      "loss": 0.1672,
+      "step": 11714
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8582852385486993e-06,
+      "loss": 0.1889,
+      "step": 11715
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8568133949489975e-06,
+      "loss": 0.153,
+      "step": 11716
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8553418672486033e-06,
+      "loss": 0.1525,
+      "step": 11717
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8538706555126085e-06,
+      "loss": 0.1576,
+      "step": 11718
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8523997598060616e-06,
+      "loss": 0.1657,
+      "step": 11719
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.850929180194021e-06,
+      "loss": 0.1405,
+      "step": 11720
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.849458916741522e-06,
+      "loss": 0.1551,
+      "step": 11721
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8479889695135766e-06,
+      "loss": 0.1559,
+      "step": 11722
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.846519338575202e-06,
+      "loss": 0.1671,
+      "step": 11723
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8450500239913836e-06,
+      "loss": 0.1506,
+      "step": 11724
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.843581025827098e-06,
+      "loss": 0.1562,
+      "step": 11725
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.842112344147322e-06,
+      "loss": 0.1912,
+      "step": 11726
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.840643979016995e-06,
+      "loss": 0.1835,
+      "step": 11727
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.839175930501057e-06,
+      "loss": 0.1683,
+      "step": 11728
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8377081986644315e-06,
+      "loss": 0.1615,
+      "step": 11729
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8362407835720283e-06,
+      "loss": 0.1769,
+      "step": 11730
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8347736852887397e-06,
+      "loss": 0.1558,
+      "step": 11731
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8333069038794483e-06,
+      "loss": 0.1589,
+      "step": 11732
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8318404394090195e-06,
+      "loss": 0.1665,
+      "step": 11733
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8303742919423073e-06,
+      "loss": 0.17,
+      "step": 11734
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.82890846154415e-06,
+      "loss": 0.1513,
+      "step": 11735
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.827442948279372e-06,
+      "loss": 0.1886,
+      "step": 11736
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.825977752212784e-06,
+      "loss": 0.1325,
+      "step": 11737
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8245128734091832e-06,
+      "loss": 0.1881,
+      "step": 11738
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.823048311933352e-06,
+      "loss": 0.1602,
+      "step": 11739
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8215840678500574e-06,
+      "loss": 0.151,
+      "step": 11740
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8201201412240597e-06,
+      "loss": 0.1676,
+      "step": 11741
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8186565321200878e-06,
+      "loss": 0.1504,
+      "step": 11742
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.817193240602879e-06,
+      "loss": 0.1546,
+      "step": 11743
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8157302667371443e-06,
+      "loss": 0.1805,
+      "step": 11744
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8142676105875744e-06,
+      "loss": 0.1452,
+      "step": 11745
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.812805272218865e-06,
+      "loss": 0.1867,
+      "step": 11746
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8113432516956774e-06,
+      "loss": 0.1617,
+      "step": 11747
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.809881549082667e-06,
+      "loss": 0.1618,
+      "step": 11748
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8084201644444864e-06,
+      "loss": 0.1393,
+      "step": 11749
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.806959097845754e-06,
+      "loss": 0.1577,
+      "step": 11750
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8054983493510857e-06,
+      "loss": 0.1479,
+      "step": 11751
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8040379190250834e-06,
+      "loss": 0.1962,
+      "step": 11752
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8025778069323286e-06,
+      "loss": 0.1618,
+      "step": 11753
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.801118013137403e-06,
+      "loss": 0.1626,
+      "step": 11754
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7996585377048546e-06,
+      "loss": 0.1581,
+      "step": 11755
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7981993806992304e-06,
+      "loss": 0.1537,
+      "step": 11756
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7967405421850603e-06,
+      "loss": 0.1784,
+      "step": 11757
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7952820222268582e-06,
+      "loss": 0.1785,
+      "step": 11758
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7938238208891278e-06,
+      "loss": 0.1898,
+      "step": 11759
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7923659382363544e-06,
+      "loss": 0.1697,
+      "step": 11760
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.790908374333012e-06,
+      "loss": 0.1717,
+      "step": 11761
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.789451129243561e-06,
+      "loss": 0.1478,
+      "step": 11762
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.787994203032445e-06,
+      "loss": 0.1759,
+      "step": 11763
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.786537595764094e-06,
+      "loss": 0.1711,
+      "step": 11764
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.785081307502926e-06,
+      "loss": 0.1865,
+      "step": 11765
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7836253383133447e-06,
+      "loss": 0.1556,
+      "step": 11766
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7821696882597392e-06,
+      "loss": 0.1728,
+      "step": 11767
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7807143574064778e-06,
+      "loss": 0.1504,
+      "step": 11768
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7792593458179272e-06,
+      "loss": 0.1707,
+      "step": 11769
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7778046535584325e-06,
+      "loss": 0.1608,
+      "step": 11770
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7763502806923248e-06,
+      "loss": 0.1834,
+      "step": 11771
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7748962272839253e-06,
+      "loss": 0.1681,
+      "step": 11772
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.773442493397529e-06,
+      "loss": 0.1712,
+      "step": 11773
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.771989079097435e-06,
+      "loss": 0.1609,
+      "step": 11774
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7705359844479175e-06,
+      "loss": 0.1792,
+      "step": 11775
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7690832095132315e-06,
+      "loss": 0.1548,
+      "step": 11776
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.767630754357633e-06,
+      "loss": 0.1536,
+      "step": 11777
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.76617861904535e-06,
+      "loss": 0.1685,
+      "step": 11778
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.764726803640598e-06,
+      "loss": 0.1564,
+      "step": 11779
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7632753082075926e-06,
+      "loss": 0.1854,
+      "step": 11780
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7618241328105155e-06,
+      "loss": 0.1864,
+      "step": 11781
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.760373277513546e-06,
+      "loss": 0.1603,
+      "step": 11782
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7589227423808473e-06,
+      "loss": 0.1566,
+      "step": 11783
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.757472527476567e-06,
+      "loss": 0.17,
+      "step": 11784
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7560226328648387e-06,
+      "loss": 0.1617,
+      "step": 11785
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.754573058609783e-06,
+      "loss": 0.1524,
+      "step": 11786
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7531238047755047e-06,
+      "loss": 0.19,
+      "step": 11787
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7516748714260977e-06,
+      "loss": 0.1687,
+      "step": 11788
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.750226258625637e-06,
+      "loss": 0.1548,
+      "step": 11789
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7487779664381877e-06,
+      "loss": 0.179,
+      "step": 11790
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7473299949277976e-06,
+      "loss": 0.1671,
+      "step": 11791
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.745882344158501e-06,
+      "loss": 0.1767,
+      "step": 11792
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.744435014194321e-06,
+      "loss": 0.1713,
+      "step": 11793
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.742988005099262e-06,
+      "loss": 0.1974,
+      "step": 11794
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7415413169373206e-06,
+      "loss": 0.1683,
+      "step": 11795
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.740094949772465e-06,
+      "loss": 0.1865,
+      "step": 11796
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.738648903668668e-06,
+      "loss": 0.149,
+      "step": 11797
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7372031786898813e-06,
+      "loss": 0.19,
+      "step": 11798
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7357577749000285e-06,
+      "loss": 0.1661,
+      "step": 11799
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.734312692363046e-06,
+      "loss": 0.1748,
+      "step": 11800
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7328679311428262e-06,
+      "loss": 0.1649,
+      "step": 11801
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7314234913032734e-06,
+      "loss": 0.1642,
+      "step": 11802
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.729979372908265e-06,
+      "loss": 0.1609,
+      "step": 11803
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7285355760216558e-06,
+      "loss": 0.1809,
+      "step": 11804
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.727092100707309e-06,
+      "loss": 0.1472,
+      "step": 11805
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7256489470290524e-06,
+      "loss": 0.1476,
+      "step": 11806
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.724206115050706e-06,
+      "loss": 0.1914,
+      "step": 11807
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7227636048360885e-06,
+      "loss": 0.1459,
+      "step": 11808
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.721321416448982e-06,
+      "loss": 0.1806,
+      "step": 11809
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7198795499531707e-06,
+      "loss": 0.1781,
+      "step": 11810
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.718438005412418e-06,
+      "loss": 0.1516,
+      "step": 11811
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7169967828904764e-06,
+      "loss": 0.1502,
+      "step": 11812
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7155558824510808e-06,
+      "loss": 0.1697,
+      "step": 11813
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7141153041579526e-06,
+      "loss": 0.1638,
+      "step": 11814
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.712675048074802e-06,
+      "loss": 0.1779,
+      "step": 11815
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7112351142653213e-06,
+      "loss": 0.1679,
+      "step": 11816
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.70979550279319e-06,
+      "loss": 0.1568,
+      "step": 11817
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.708356213722074e-06,
+      "loss": 0.1343,
+      "step": 11818
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7069172471156237e-06,
+      "loss": 0.1457,
+      "step": 11819
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.705478603037476e-06,
+      "loss": 0.1451,
+      "step": 11820
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.704040281551252e-06,
+      "loss": 0.1577,
+      "step": 11821
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.702602282720561e-06,
+      "loss": 0.1617,
+      "step": 11822
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7011646066090004e-06,
+      "loss": 0.1748,
+      "step": 11823
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.69972725328014e-06,
+      "loss": 0.1587,
+      "step": 11824
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.6982902227975547e-06,
+      "loss": 0.1619,
+      "step": 11825
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.6968535152247945e-06,
+      "loss": 0.1627,
+      "step": 11826
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.6954171306253883e-06,
+      "loss": 0.1626,
+      "step": 11827
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.69398106906287e-06,
+      "loss": 0.1679,
+      "step": 11828
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.6925453306007386e-06,
+      "loss": 0.1722,
+      "step": 11829
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.6911099153024878e-06,
+      "loss": 0.1556,
+      "step": 11830
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.6896748232316073e-06,
+      "loss": 0.1709,
+      "step": 11831
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.688240054451553e-06,
+      "loss": 0.145,
+      "step": 11832
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.6868056090257788e-06,
+      "loss": 0.1652,
+      "step": 11833
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6853714870177205e-06,
+      "loss": 0.1768,
+      "step": 11834
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6839376884907986e-06,
+      "loss": 0.1699,
+      "step": 11835
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.68250421350843e-06,
+      "loss": 0.1616,
+      "step": 11836
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6810710621339997e-06,
+      "loss": 0.1729,
+      "step": 11837
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.67963823443089e-06,
+      "loss": 0.1645,
+      "step": 11838
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6782057304624663e-06,
+      "loss": 0.1452,
+      "step": 11839
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.676773550292079e-06,
+      "loss": 0.186,
+      "step": 11840
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6753416939830647e-06,
+      "loss": 0.1596,
+      "step": 11841
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6739101615987463e-06,
+      "loss": 0.1715,
+      "step": 11842
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.67247895320243e-06,
+      "loss": 0.1511,
+      "step": 11843
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6710480688574114e-06,
+      "loss": 0.1724,
+      "step": 11844
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.669617508626968e-06,
+      "loss": 0.184,
+      "step": 11845
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6681872725743698e-06,
+      "loss": 0.1788,
+      "step": 11846
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.666757360762856e-06,
+      "loss": 0.1582,
+      "step": 11847
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.665327773255674e-06,
+      "loss": 0.1527,
+      "step": 11848
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6638985101160443e-06,
+      "loss": 0.2084,
+      "step": 11849
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6624695714071647e-06,
+      "loss": 0.1616,
+      "step": 11850
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6610409571922424e-06,
+      "loss": 0.1755,
+      "step": 11851
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6596126675344437e-06,
+      "loss": 0.1583,
+      "step": 11852
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.658184702496941e-06,
+      "loss": 0.1612,
+      "step": 11853
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.656757062142885e-06,
+      "loss": 0.1606,
+      "step": 11854
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6553297465354035e-06,
+      "loss": 0.1711,
+      "step": 11855
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6539027557376273e-06,
+      "loss": 0.1662,
+      "step": 11856
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.652476089812658e-06,
+      "loss": 0.1993,
+      "step": 11857
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6510497488235855e-06,
+      "loss": 0.1687,
+      "step": 11858
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6496237328334983e-06,
+      "loss": 0.1511,
+      "step": 11859
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.648198041905451e-06,
+      "loss": 0.1596,
+      "step": 11860
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.646772676102496e-06,
+      "loss": 0.1673,
+      "step": 11861
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6453476354876695e-06,
+      "loss": 0.1819,
+      "step": 11862
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6439229201239903e-06,
+      "loss": 0.1558,
+      "step": 11863
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.642498530074468e-06,
+      "loss": 0.1782,
+      "step": 11864
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6410744654020904e-06,
+      "loss": 0.1575,
+      "step": 11865
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.639650726169839e-06,
+      "loss": 0.1495,
+      "step": 11866
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6382273124406744e-06,
+      "loss": 0.1707,
+      "step": 11867
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6368042242775473e-06,
+      "loss": 0.1954,
+      "step": 11868
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6353814617433915e-06,
+      "loss": 0.1638,
+      "step": 11869
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.633959024901126e-06,
+      "loss": 0.1548,
+      "step": 11870
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.632536913813658e-06,
+      "loss": 0.1459,
+      "step": 11871
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6311151285438785e-06,
+      "loss": 0.1764,
+      "step": 11872
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6296936691546638e-06,
+      "loss": 0.1744,
+      "step": 11873
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.62827253570888e-06,
+      "loss": 0.1649,
+      "step": 11874
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.626851728269365e-06,
+      "loss": 0.177,
+      "step": 11875
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.625431246898962e-06,
+      "loss": 0.167,
+      "step": 11876
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.62401109166049e-06,
+      "loss": 0.1596,
+      "step": 11877
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.622591262616745e-06,
+      "loss": 0.1783,
+      "step": 11878
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.62117175983053e-06,
+      "loss": 0.1759,
+      "step": 11879
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6197525833646108e-06,
+      "loss": 0.1484,
+      "step": 11880
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6183337332817484e-06,
+      "loss": 0.1571,
+      "step": 11881
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6169152096447004e-06,
+      "loss": 0.1865,
+      "step": 11882
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6154970125161903e-06,
+      "loss": 0.1477,
+      "step": 11883
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.614079141958937e-06,
+      "loss": 0.1464,
+      "step": 11884
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.612661598035647e-06,
+      "loss": 0.2074,
+      "step": 11885
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.611244380809006e-06,
+      "loss": 0.1551,
+      "step": 11886
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.6098274903416964e-06,
+      "loss": 0.1569,
+      "step": 11887
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.6084109266963707e-06,
+      "loss": 0.1732,
+      "step": 11888
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.606994689935677e-06,
+      "loss": 0.1455,
+      "step": 11889
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.605578780122248e-06,
+      "loss": 0.1664,
+      "step": 11890
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.6041631973186997e-06,
+      "loss": 0.1882,
+      "step": 11891
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.6027479415876344e-06,
+      "loss": 0.1739,
+      "step": 11892
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.6013330129916413e-06,
+      "loss": 0.1644,
+      "step": 11893
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.599918411593294e-06,
+      "loss": 0.1615,
+      "step": 11894
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.59850413745515e-06,
+      "loss": 0.1509,
+      "step": 11895
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5970901906397562e-06,
+      "loss": 0.1553,
+      "step": 11896
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.59567657120964e-06,
+      "loss": 0.1522,
+      "step": 11897
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5942632792273205e-06,
+      "loss": 0.1662,
+      "step": 11898
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5928503147552965e-06,
+      "loss": 0.1683,
+      "step": 11899
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5914376778560577e-06,
+      "loss": 0.1749,
+      "step": 11900
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5900253685920683e-06,
+      "loss": 0.1644,
+      "step": 11901
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5886133870257956e-06,
+      "loss": 0.1721,
+      "step": 11902
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5872017332196786e-06,
+      "loss": 0.1689,
+      "step": 11903
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.585790407236147e-06,
+      "loss": 0.1537,
+      "step": 11904
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5843794091376174e-06,
+      "loss": 0.1463,
+      "step": 11905
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5829687389864812e-06,
+      "loss": 0.1554,
+      "step": 11906
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.581558396845132e-06,
+      "loss": 0.1807,
+      "step": 11907
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.580148382775941e-06,
+      "loss": 0.1682,
+      "step": 11908
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5787386968412555e-06,
+      "loss": 0.1686,
+      "step": 11909
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.577329339103429e-06,
+      "loss": 0.1692,
+      "step": 11910
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.575920309624781e-06,
+      "loss": 0.1828,
+      "step": 11911
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5745116084676247e-06,
+      "loss": 0.1936,
+      "step": 11912
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5731032356942608e-06,
+      "loss": 0.1709,
+      "step": 11913
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.571695191366972e-06,
+      "loss": 0.1605,
+      "step": 11914
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5702874755480267e-06,
+      "loss": 0.1718,
+      "step": 11915
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.568880088299681e-06,
+      "loss": 0.1618,
+      "step": 11916
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5674730296841734e-06,
+      "loss": 0.1782,
+      "step": 11917
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5660662997637308e-06,
+      "loss": 0.1646,
+      "step": 11918
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5646598986005644e-06,
+      "loss": 0.184,
+      "step": 11919
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5632538262568696e-06,
+      "loss": 0.1744,
+      "step": 11920
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5618480827948277e-06,
+      "loss": 0.1672,
+      "step": 11921
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.560442668276609e-06,
+      "loss": 0.1882,
+      "step": 11922
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.559037582764363e-06,
+      "loss": 0.1849,
+      "step": 11923
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.557632826320231e-06,
+      "loss": 0.1491,
+      "step": 11924
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5562283990063354e-06,
+      "loss": 0.1874,
+      "step": 11925
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.554824300884784e-06,
+      "loss": 0.1475,
+      "step": 11926
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.553420532017674e-06,
+      "loss": 0.159,
+      "step": 11927
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.552017092467086e-06,
+      "loss": 0.1648,
+      "step": 11928
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5506139822950795e-06,
+      "loss": 0.1519,
+      "step": 11929
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5492112015637114e-06,
+      "loss": 0.1669,
+      "step": 11930
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5478087503350203e-06,
+      "loss": 0.163,
+      "step": 11931
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5464066286710176e-06,
+      "loss": 0.1759,
+      "step": 11932
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5450048366337233e-06,
+      "loss": 0.1682,
+      "step": 11933
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5436033742851183e-06,
+      "loss": 0.1701,
+      "step": 11934
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5422022416871885e-06,
+      "loss": 0.1468,
+      "step": 11935
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.540801438901899e-06,
+      "loss": 0.1675,
+      "step": 11936
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.539400965991188e-06,
+      "loss": 0.1587,
+      "step": 11937
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5380008230170027e-06,
+      "loss": 0.1691,
+      "step": 11938
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.536601010041254e-06,
+      "loss": 0.1755,
+      "step": 11939
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5352015271258467e-06,
+      "loss": 0.1795,
+      "step": 11940
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5338023743326812e-06,
+      "loss": 0.1389,
+      "step": 11941
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5324035517236236e-06,
+      "loss": 0.1901,
+      "step": 11942
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5310050593605373e-06,
+      "loss": 0.1798,
+      "step": 11943
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5296068973052713e-06,
+      "loss": 0.1556,
+      "step": 11944
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.528209065619656e-06,
+      "loss": 0.166,
+      "step": 11945
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.526811564365509e-06,
+      "loss": 0.1694,
+      "step": 11946
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.525414393604635e-06,
+      "loss": 0.1798,
+      "step": 11947
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.52401755339882e-06,
+      "loss": 0.1818,
+      "step": 11948
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5226210438098385e-06,
+      "loss": 0.1803,
+      "step": 11949
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5212248648994496e-06,
+      "loss": 0.1665,
+      "step": 11950
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.519829016729398e-06,
+      "loss": 0.1729,
+      "step": 11951
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5184334993614133e-06,
+      "loss": 0.1413,
+      "step": 11952
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5170383128572116e-06,
+      "loss": 0.1735,
+      "step": 11953
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5156434572784914e-06,
+      "loss": 0.1739,
+      "step": 11954
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.51424893268694e-06,
+      "loss": 0.1576,
+      "step": 11955
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5128547391442314e-06,
+      "loss": 0.1883,
+      "step": 11956
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.511460876712013e-06,
+      "loss": 0.1767,
+      "step": 11957
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5100673454519365e-06,
+      "loss": 0.1777,
+      "step": 11958
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.508674145425628e-06,
+      "loss": 0.1496,
+      "step": 11959
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.507281276694693e-06,
+      "loss": 0.1471,
+      "step": 11960
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.50588873932074e-06,
+      "loss": 0.1585,
+      "step": 11961
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.504496533365344e-06,
+      "loss": 0.1682,
+      "step": 11962
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.503104658890073e-06,
+      "loss": 0.1636,
+      "step": 11963
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.501713115956491e-06,
+      "loss": 0.1735,
+      "step": 11964
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5003219046261275e-06,
+      "loss": 0.1988,
+      "step": 11965
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4989310249605114e-06,
+      "loss": 0.1816,
+      "step": 11966
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.497540477021152e-06,
+      "loss": 0.1701,
+      "step": 11967
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.496150260869542e-06,
+      "loss": 0.187,
+      "step": 11968
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4947603765671704e-06,
+      "loss": 0.1492,
+      "step": 11969
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4933708241754952e-06,
+      "loss": 0.1969,
+      "step": 11970
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.491981603755971e-06,
+      "loss": 0.1864,
+      "step": 11971
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4905927153700326e-06,
+      "loss": 0.1502,
+      "step": 11972
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.489204159079105e-06,
+      "loss": 0.1818,
+      "step": 11973
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.487815934944593e-06,
+      "loss": 0.188,
+      "step": 11974
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4864280430278896e-06,
+      "loss": 0.1533,
+      "step": 11975
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.485040483390374e-06,
+      "loss": 0.168,
+      "step": 11976
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4836532560934077e-06,
+      "loss": 0.1522,
+      "step": 11977
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.482266361198341e-06,
+      "loss": 0.1498,
+      "step": 11978
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.48087979876651e-06,
+      "loss": 0.1694,
+      "step": 11979
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.479493568859225e-06,
+      "loss": 0.19,
+      "step": 11980
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4781076715377995e-06,
+      "loss": 0.1849,
+      "step": 11981
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4767221068635216e-06,
+      "loss": 0.1815,
+      "step": 11982
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.47533687489766e-06,
+      "loss": 0.2004,
+      "step": 11983
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.473951975701485e-06,
+      "loss": 0.1792,
+      "step": 11984
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.472567409336232e-06,
+      "loss": 0.167,
+      "step": 11985
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4711831758631386e-06,
+      "loss": 0.1559,
+      "step": 11986
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.469799275343423e-06,
+      "loss": 0.1822,
+      "step": 11987
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4684157078382766e-06,
+      "loss": 0.1519,
+      "step": 11988
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4670324734088978e-06,
+      "loss": 0.1727,
+      "step": 11989
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.465649572116451e-06,
+      "loss": 0.1539,
+      "step": 11990
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4642670040220928e-06,
+      "loss": 0.1636,
+      "step": 11991
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4628847691869727e-06,
+      "loss": 0.1534,
+      "step": 11992
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.461502867672213e-06,
+      "loss": 0.1664,
+      "step": 11993
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4601212995389267e-06,
+      "loss": 0.17,
+      "step": 11994
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.458740064848214e-06,
+      "loss": 0.1692,
+      "step": 11995
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4573591636611572e-06,
+      "loss": 0.1605,
+      "step": 11996
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.455978596038826e-06,
+      "loss": 0.1692,
+      "step": 11997
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4545983620422733e-06,
+      "loss": 0.1612,
+      "step": 11998
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4532184617325395e-06,
+      "loss": 0.1601,
+      "step": 11999
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4518388951706497e-06,
+      "loss": 0.1546,
+      "step": 12000
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4504596624176115e-06,
+      "loss": 0.1783,
+      "step": 12001
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4490807635344216e-06,
+      "loss": 0.1545,
+      "step": 12002
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4477021985820595e-06,
+      "loss": 0.145,
+      "step": 12003
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4463239676214914e-06,
+      "loss": 0.1365,
+      "step": 12004
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4449460707136673e-06,
+      "loss": 0.1643,
+      "step": 12005
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4435685079195224e-06,
+      "loss": 0.1591,
+      "step": 12006
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4421912792999834e-06,
+      "loss": 0.1954,
+      "step": 12007
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.440814384915945e-06,
+      "loss": 0.18,
+      "step": 12008
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.439437824828309e-06,
+      "loss": 0.1994,
+      "step": 12009
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4380615990979527e-06,
+      "loss": 0.1634,
+      "step": 12010
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4366857077857288e-06,
+      "loss": 0.1816,
+      "step": 12011
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4353101509524945e-06,
+      "loss": 0.158,
+      "step": 12012
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4339349286590765e-06,
+      "loss": 0.1744,
+      "step": 12013
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4325600409662906e-06,
+      "loss": 0.1898,
+      "step": 12014
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4311854879349483e-06,
+      "loss": 0.1793,
+      "step": 12015
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4298112696258304e-06,
+      "loss": 0.1526,
+      "step": 12016
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4284373860997112e-06,
+      "loss": 0.1339,
+      "step": 12017
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4270638374173493e-06,
+      "loss": 0.1356,
+      "step": 12018
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4256906236394873e-06,
+      "loss": 0.1591,
+      "step": 12019
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4243177448268607e-06,
+      "loss": 0.1581,
+      "step": 12020
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4229452010401766e-06,
+      "loss": 0.1625,
+      "step": 12021
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4215729923401355e-06,
+      "loss": 0.149,
+      "step": 12022
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.420201118787423e-06,
+      "loss": 0.153,
+      "step": 12023
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4188295804427075e-06,
+      "loss": 0.1635,
+      "step": 12024
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.417458377366644e-06,
+      "loss": 0.1956,
+      "step": 12025
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.416087509619873e-06,
+      "loss": 0.1542,
+      "step": 12026
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.41471697726302e-06,
+      "loss": 0.1693,
+      "step": 12027
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4133467803566933e-06,
+      "loss": 0.2006,
+      "step": 12028
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4119769189614894e-06,
+      "loss": 0.1704,
+      "step": 12029
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4106073931379904e-06,
+      "loss": 0.1836,
+      "step": 12030
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.409238202946759e-06,
+      "loss": 0.2031,
+      "step": 12031
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4078693484483486e-06,
+      "loss": 0.1723,
+      "step": 12032
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4065008297032966e-06,
+      "loss": 0.1552,
+      "step": 12033
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.405132646772117e-06,
+      "loss": 0.1641,
+      "step": 12034
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4037647997153224e-06,
+      "loss": 0.1653,
+      "step": 12035
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4023972885934046e-06,
+      "loss": 0.1696,
+      "step": 12036
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.401030113466837e-06,
+      "loss": 0.1595,
+      "step": 12037
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.3996632743960866e-06,
+      "loss": 0.1702,
+      "step": 12038
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.3982967714415907e-06,
+      "loss": 0.1702,
+      "step": 12039
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3969306046637896e-06,
+      "loss": 0.1729,
+      "step": 12040
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3955647741231026e-06,
+      "loss": 0.1625,
+      "step": 12041
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3941992798799198e-06,
+      "loss": 0.1572,
+      "step": 12042
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.392834121994644e-06,
+      "loss": 0.1483,
+      "step": 12043
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.391469300527636e-06,
+      "loss": 0.155,
+      "step": 12044
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.390104815539257e-06,
+      "loss": 0.1798,
+      "step": 12045
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3887406670898515e-06,
+      "loss": 0.1737,
+      "step": 12046
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3873768552397457e-06,
+      "loss": 0.1707,
+      "step": 12047
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3860133800492536e-06,
+      "loss": 0.1514,
+      "step": 12048
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.384650241578673e-06,
+      "loss": 0.1629,
+      "step": 12049
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3832874398882864e-06,
+      "loss": 0.1663,
+      "step": 12050
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3819249750383635e-06,
+      "loss": 0.1492,
+      "step": 12051
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.380562847089157e-06,
+      "loss": 0.1671,
+      "step": 12052
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.379201056100906e-06,
+      "loss": 0.1915,
+      "step": 12053
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3778396021338335e-06,
+      "loss": 0.1537,
+      "step": 12054
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3764784852481483e-06,
+      "loss": 0.1759,
+      "step": 12055
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.375117705504044e-06,
+      "loss": 0.1742,
+      "step": 12056
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.373757262961701e-06,
+      "loss": 0.1466,
+      "step": 12057
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3723971576812823e-06,
+      "loss": 0.1421,
+      "step": 12058
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3710373897229365e-06,
+      "loss": 0.1637,
+      "step": 12059
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3696779591467977e-06,
+      "loss": 0.1538,
+      "step": 12060
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3683188660129896e-06,
+      "loss": 0.1494,
+      "step": 12061
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3669601103816077e-06,
+      "loss": 0.1794,
+      "step": 12062
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.365601692312748e-06,
+      "loss": 0.1723,
+      "step": 12063
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.364243611866487e-06,
+      "loss": 0.1729,
+      "step": 12064
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3628858691028744e-06,
+      "loss": 0.1461,
+      "step": 12065
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.361528464081968e-06,
+      "loss": 0.1606,
+      "step": 12066
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.360171396863784e-06,
+      "loss": 0.1492,
+      "step": 12067
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.358814667508347e-06,
+      "loss": 0.1656,
+      "step": 12068
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.357458276075657e-06,
+      "loss": 0.1566,
+      "step": 12069
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3561022226256893e-06,
+      "loss": 0.1661,
+      "step": 12070
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.354746507218426e-06,
+      "loss": 0.1949,
+      "step": 12071
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3533911299138123e-06,
+      "loss": 0.1712,
+      "step": 12072
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3520360907717897e-06,
+      "loss": 0.1675,
+      "step": 12073
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3506813898522917e-06,
+      "loss": 0.1602,
+      "step": 12074
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.349327027215219e-06,
+      "loss": 0.1629,
+      "step": 12075
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3479730029204706e-06,
+      "loss": 0.1571,
+      "step": 12076
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.346619317027926e-06,
+      "loss": 0.1676,
+      "step": 12077
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3452659695974512e-06,
+      "loss": 0.1715,
+      "step": 12078
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3439129606888946e-06,
+      "loss": 0.1472,
+      "step": 12079
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.342560290362094e-06,
+      "loss": 0.1745,
+      "step": 12080
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3412079586768666e-06,
+      "loss": 0.1597,
+      "step": 12081
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.339855965693021e-06,
+      "loss": 0.1897,
+      "step": 12082
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3385043114703453e-06,
+      "loss": 0.1696,
+      "step": 12083
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3371529960686156e-06,
+      "loss": 0.1658,
+      "step": 12084
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3358020195475928e-06,
+      "loss": 0.2014,
+      "step": 12085
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3344513819670223e-06,
+      "loss": 0.1644,
+      "step": 12086
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.333101083386633e-06,
+      "loss": 0.172,
+      "step": 12087
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.331751123866143e-06,
+      "loss": 0.1679,
+      "step": 12088
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3304015034652525e-06,
+      "loss": 0.1574,
+      "step": 12089
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3290522222436405e-06,
+      "loss": 0.1436,
+      "step": 12090
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3277032802609846e-06,
+      "loss": 0.1657,
+      "step": 12091
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3263546775769418e-06,
+      "loss": 0.1841,
+      "step": 12092
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.325006414251143e-06,
+      "loss": 0.1935,
+      "step": 12093
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.323658490343225e-06,
+      "loss": 0.1455,
+      "step": 12094
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.32231090591279e-06,
+      "loss": 0.1488,
+      "step": 12095
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.320963661019433e-06,
+      "loss": 0.1844,
+      "step": 12096
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3196167557227432e-06,
+      "loss": 0.1735,
+      "step": 12097
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.318270190082277e-06,
+      "loss": 0.1538,
+      "step": 12098
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.316923964157588e-06,
+      "loss": 0.1767,
+      "step": 12099
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3155780780082114e-06,
+      "loss": 0.1705,
+      "step": 12100
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3142325316936655e-06,
+      "loss": 0.1967,
+      "step": 12101
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.312887325273462e-06,
+      "loss": 0.1578,
+      "step": 12102
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3115424588070846e-06,
+      "loss": 0.1638,
+      "step": 12103
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3101979323540103e-06,
+      "loss": 0.1615,
+      "step": 12104
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3088537459737002e-06,
+      "loss": 0.1579,
+      "step": 12105
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3075098997255984e-06,
+      "loss": 0.1738,
+      "step": 12106
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.306166393669135e-06,
+      "loss": 0.152,
+      "step": 12107
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.304823227863725e-06,
+      "loss": 0.16,
+      "step": 12108
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3034804023687697e-06,
+      "loss": 0.1877,
+      "step": 12109
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.302137917243653e-06,
+      "loss": 0.1765,
+      "step": 12110
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3007957725477436e-06,
+      "loss": 0.177,
+      "step": 12111
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2994539683404014e-06,
+      "loss": 0.1487,
+      "step": 12112
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.298112504680957e-06,
+      "loss": 0.1783,
+      "step": 12113
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.296771381628743e-06,
+      "loss": 0.1652,
+      "step": 12114
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.295430599243069e-06,
+      "loss": 0.1539,
+      "step": 12115
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.294090157583222e-06,
+      "loss": 0.1849,
+      "step": 12116
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2927500567084914e-06,
+      "loss": 0.1746,
+      "step": 12117
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2914102966781327e-06,
+      "loss": 0.1585,
+      "step": 12118
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2900708775514003e-06,
+      "loss": 0.1827,
+      "step": 12119
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.288731799387531e-06,
+      "loss": 0.1611,
+      "step": 12120
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2873930622457365e-06,
+      "loss": 0.1525,
+      "step": 12121
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.286054666185229e-06,
+      "loss": 0.1914,
+      "step": 12122
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.284716611265191e-06,
+      "loss": 0.1508,
+      "step": 12123
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.283378897544797e-06,
+      "loss": 0.1745,
+      "step": 12124
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.282041525083213e-06,
+      "loss": 0.1741,
+      "step": 12125
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.280704493939574e-06,
+      "loss": 0.1716,
+      "step": 12126
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.279367804173014e-06,
+      "loss": 0.1834,
+      "step": 12127
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.278031455842643e-06,
+      "loss": 0.1685,
+      "step": 12128
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2766954490075632e-06,
+      "loss": 0.164,
+      "step": 12129
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.275359783726856e-06,
+      "loss": 0.1545,
+      "step": 12130
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2740244600595883e-06,
+      "loss": 0.1689,
+      "step": 12131
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2726894780648167e-06,
+      "loss": 0.1569,
+      "step": 12132
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.271354837801577e-06,
+      "loss": 0.1478,
+      "step": 12133
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2700205393288922e-06,
+      "loss": 0.1669,
+      "step": 12134
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.26868658270577e-06,
+      "loss": 0.1773,
+      "step": 12135
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2673529679912054e-06,
+      "loss": 0.1557,
+      "step": 12136
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2660196952441725e-06,
+      "loss": 0.1664,
+      "step": 12137
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2646867645236357e-06,
+      "loss": 0.1873,
+      "step": 12138
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2633541758885436e-06,
+      "loss": 0.1565,
+      "step": 12139
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2620219293978284e-06,
+      "loss": 0.1781,
+      "step": 12140
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2606900251104e-06,
+      "loss": 0.1809,
+      "step": 12141
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2593584630851706e-06,
+      "loss": 0.1684,
+      "step": 12142
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2580272433810234e-06,
+      "loss": 0.1685,
+      "step": 12143
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2566963660568254e-06,
+      "loss": 0.1956,
+      "step": 12144
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.255365831171441e-06,
+      "loss": 0.1842,
+      "step": 12145
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.254035638783706e-06,
+      "loss": 0.1618,
+      "step": 12146
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.252705788952444e-06,
+      "loss": 0.1488,
+      "step": 12147
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.251376281736477e-06,
+      "loss": 0.1701,
+      "step": 12148
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2500471171945903e-06,
+      "loss": 0.1772,
+      "step": 12149
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.24871829538557e-06,
+      "loss": 0.1878,
+      "step": 12150
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2473898163681796e-06,
+      "loss": 0.179,
+      "step": 12151
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.246061680201168e-06,
+      "loss": 0.1564,
+      "step": 12152
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2447338869432777e-06,
+      "loss": 0.1649,
+      "step": 12153
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2434064366532217e-06,
+      "loss": 0.1705,
+      "step": 12154
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.242079329389708e-06,
+      "loss": 0.1655,
+      "step": 12155
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.240752565211425e-06,
+      "loss": 0.1667,
+      "step": 12156
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.239426144177047e-06,
+      "loss": 0.1539,
+      "step": 12157
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2381000663452356e-06,
+      "loss": 0.1783,
+      "step": 12158
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.236774331774635e-06,
+      "loss": 0.1705,
+      "step": 12159
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.235448940523871e-06,
+      "loss": 0.1514,
+      "step": 12160
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2341238926515617e-06,
+      "loss": 0.1577,
+      "step": 12161
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2327991882163035e-06,
+      "loss": 0.17,
+      "step": 12162
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2314748272766795e-06,
+      "loss": 0.1675,
+      "step": 12163
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2301508098912604e-06,
+      "loss": 0.1582,
+      "step": 12164
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2288271361185964e-06,
+      "loss": 0.1785,
+      "step": 12165
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2275038060172307e-06,
+      "loss": 0.1558,
+      "step": 12166
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.226180819645677e-06,
+      "loss": 0.1664,
+      "step": 12167
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.224858177062451e-06,
+      "loss": 0.1626,
+      "step": 12168
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.223535878326042e-06,
+      "loss": 0.174,
+      "step": 12169
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2222139234949267e-06,
+      "loss": 0.1905,
+      "step": 12170
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2208923126275716e-06,
+      "loss": 0.187,
+      "step": 12171
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.219571045782414e-06,
+      "loss": 0.1682,
+      "step": 12172
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2182501230178966e-06,
+      "loss": 0.1596,
+      "step": 12173
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2169295443924265e-06,
+      "loss": 0.1705,
+      "step": 12174
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.215609309964407e-06,
+      "loss": 0.1865,
+      "step": 12175
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2142894197922303e-06,
+      "loss": 0.1606,
+      "step": 12176
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2129698739342597e-06,
+      "loss": 0.1714,
+      "step": 12177
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.211650672448853e-06,
+      "loss": 0.1595,
+      "step": 12178
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2103318153943507e-06,
+      "loss": 0.1866,
+      "step": 12179
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2090133028290772e-06,
+      "loss": 0.1595,
+      "step": 12180
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2076951348113428e-06,
+      "loss": 0.1719,
+      "step": 12181
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2063773113994422e-06,
+      "loss": 0.1601,
+      "step": 12182
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2050598326516536e-06,
+      "loss": 0.1732,
+      "step": 12183
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.203742698626241e-06,
+      "loss": 0.155,
+      "step": 12184
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2024259093814547e-06,
+      "loss": 0.1707,
+      "step": 12185
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.201109464975527e-06,
+      "loss": 0.1749,
+      "step": 12186
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.199793365466677e-06,
+      "loss": 0.1571,
+      "step": 12187
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.1984776109131067e-06,
+      "loss": 0.1542,
+      "step": 12188
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.1971622013730044e-06,
+      "loss": 0.1826,
+      "step": 12189
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.195847136904542e-06,
+      "loss": 0.1799,
+      "step": 12190
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.1945324175658765e-06,
+      "loss": 0.1614,
+      "step": 12191
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.193218043415152e-06,
+      "loss": 0.1813,
+      "step": 12192
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.191904014510493e-06,
+      "loss": 0.1572,
+      "step": 12193
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.190590330910015e-06,
+      "loss": 0.1936,
+      "step": 12194
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.189276992671806e-06,
+      "loss": 0.1395,
+      "step": 12195
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1879639998539536e-06,
+      "loss": 0.1468,
+      "step": 12196
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1866513525145237e-06,
+      "loss": 0.1551,
+      "step": 12197
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1853390507115602e-06,
+      "loss": 0.1694,
+      "step": 12198
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.184027094503107e-06,
+      "loss": 0.1827,
+      "step": 12199
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1827154839471753e-06,
+      "loss": 0.1657,
+      "step": 12200
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.181404219101775e-06,
+      "loss": 0.1664,
+      "step": 12201
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.180093300024898e-06,
+      "loss": 0.1659,
+      "step": 12202
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1787827267745075e-06,
+      "loss": 0.218,
+      "step": 12203
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.177472499408575e-06,
+      "loss": 0.1738,
+      "step": 12204
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1761626179850335e-06,
+      "loss": 0.1544,
+      "step": 12205
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.174853082561814e-06,
+      "loss": 0.1558,
+      "step": 12206
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1735438931968355e-06,
+      "loss": 0.1609,
+      "step": 12207
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.172235049947987e-06,
+      "loss": 0.1682,
+      "step": 12208
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1709265528731538e-06,
+      "loss": 0.1785,
+      "step": 12209
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1696184020302026e-06,
+      "loss": 0.1588,
+      "step": 12210
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1683105974769857e-06,
+      "loss": 0.177,
+      "step": 12211
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1670031392713388e-06,
+      "loss": 0.1484,
+      "step": 12212
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1656960274710816e-06,
+      "loss": 0.16,
+      "step": 12213
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.164389262134021e-06,
+      "loss": 0.1652,
+      "step": 12214
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1630828433179463e-06,
+      "loss": 0.167,
+      "step": 12215
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1617767710806336e-06,
+      "loss": 0.1595,
+      "step": 12216
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.160471045479842e-06,
+      "loss": 0.1541,
+      "step": 12217
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.159165666573314e-06,
+      "loss": 0.1572,
+      "step": 12218
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1578606344187814e-06,
+      "loss": 0.1651,
+      "step": 12219
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.156555949073956e-06,
+      "loss": 0.1744,
+      "step": 12220
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.155251610596536e-06,
+      "loss": 0.1438,
+      "step": 12221
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.153947619044208e-06,
+      "loss": 0.1488,
+      "step": 12222
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1526439744746307e-06,
+      "loss": 0.1652,
+      "step": 12223
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1513406769454635e-06,
+      "loss": 0.1642,
+      "step": 12224
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1500377265143455e-06,
+      "loss": 0.1651,
+      "step": 12225
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1487351232388886e-06,
+      "loss": 0.156,
+      "step": 12226
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1474328671767096e-06,
+      "loss": 0.1666,
+      "step": 12227
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.146130958385392e-06,
+      "loss": 0.1865,
+      "step": 12228
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1448293969225096e-06,
+      "loss": 0.1734,
+      "step": 12229
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1435281828456335e-06,
+      "loss": 0.188,
+      "step": 12230
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1422273162122974e-06,
+      "loss": 0.1769,
+      "step": 12231
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1409267970800364e-06,
+      "loss": 0.1884,
+      "step": 12232
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1396266255063603e-06,
+      "loss": 0.1691,
+      "step": 12233
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.138326801548769e-06,
+      "loss": 0.1592,
+      "step": 12234
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1370273252647524e-06,
+      "loss": 0.1983,
+      "step": 12235
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.135728196711768e-06,
+      "loss": 0.16,
+      "step": 12236
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1344294159472754e-06,
+      "loss": 0.1743,
+      "step": 12237
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1331309830287083e-06,
+      "loss": 0.1705,
+      "step": 12238
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1318328980134896e-06,
+      "loss": 0.151,
+      "step": 12239
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1305351609590254e-06,
+      "loss": 0.1679,
+      "step": 12240
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1292377719227075e-06,
+      "loss": 0.1689,
+      "step": 12241
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1279407309619094e-06,
+      "loss": 0.1655,
+      "step": 12242
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.126644038133994e-06,
+      "loss": 0.1446,
+      "step": 12243
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1253476934963036e-06,
+      "loss": 0.1608,
+      "step": 12244
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.124051697106172e-06,
+      "loss": 0.1996,
+      "step": 12245
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1227560490209043e-06,
+      "loss": 0.1731,
+      "step": 12246
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.121460749297807e-06,
+      "loss": 0.1749,
+      "step": 12247
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.120165797994164e-06,
+      "loss": 0.1708,
+      "step": 12248
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1188711951672335e-06,
+      "loss": 0.1683,
+      "step": 12249
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1175769408742808e-06,
+      "loss": 0.1686,
+      "step": 12250
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1162830351725304e-06,
+      "loss": 0.16,
+      "step": 12251
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1149894781192117e-06,
+      "loss": 0.1748,
+      "step": 12252
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1136962697715323e-06,
+      "loss": 0.1633,
+      "step": 12253
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.112403410186673e-06,
+      "loss": 0.1645,
+      "step": 12254
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1111108994218222e-06,
+      "loss": 0.1415,
+      "step": 12255
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1098187375341294e-06,
+      "loss": 0.2008,
+      "step": 12256
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1085269245807394e-06,
+      "loss": 0.1665,
+      "step": 12257
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1072354606187896e-06,
+      "loss": 0.1551,
+      "step": 12258
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1059443457053864e-06,
+      "loss": 0.17,
+      "step": 12259
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1046535798976286e-06,
+      "loss": 0.184,
+      "step": 12260
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.103363163252601e-06,
+      "loss": 0.1791,
+      "step": 12261
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1020730958273683e-06,
+      "loss": 0.1619,
+      "step": 12262
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.100783377678984e-06,
+      "loss": 0.143,
+      "step": 12263
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0994940088644855e-06,
+      "loss": 0.1992,
+      "step": 12264
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0982049894408907e-06,
+      "loss": 0.183,
+      "step": 12265
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0969163194652076e-06,
+      "loss": 0.1764,
+      "step": 12266
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.095627998994425e-06,
+      "loss": 0.1486,
+      "step": 12267
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0943400280855177e-06,
+      "loss": 0.178,
+      "step": 12268
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.093052406795444e-06,
+      "loss": 0.1484,
+      "step": 12269
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0917651351811485e-06,
+      "loss": 0.1624,
+      "step": 12270
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.090478213299558e-06,
+      "loss": 0.1728,
+      "step": 12271
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0891916412075873e-06,
+      "loss": 0.164,
+      "step": 12272
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.087905418962135e-06,
+      "loss": 0.1471,
+      "step": 12273
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0866195466200743e-06,
+      "loss": 0.1391,
+      "step": 12274
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.08533402423828e-06,
+      "loss": 0.1794,
+      "step": 12275
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0840488518736037e-06,
+      "loss": 0.1757,
+      "step": 12276
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.082764029582871e-06,
+      "loss": 0.1855,
+      "step": 12277
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.081479557422914e-06,
+      "loss": 0.1441,
+      "step": 12278
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.080195435450527e-06,
+      "loss": 0.1806,
+      "step": 12279
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0789116637225015e-06,
+      "loss": 0.1808,
+      "step": 12280
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0776282422956174e-06,
+      "loss": 0.1654,
+      "step": 12281
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0763451712266237e-06,
+      "loss": 0.1387,
+      "step": 12282
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.075062450572267e-06,
+      "loss": 0.16,
+      "step": 12283
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0737800803892727e-06,
+      "loss": 0.1581,
+      "step": 12284
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0724980607343503e-06,
+      "loss": 0.1807,
+      "step": 12285
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0712163916642047e-06,
+      "loss": 0.1755,
+      "step": 12286
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0699350732355073e-06,
+      "loss": 0.1699,
+      "step": 12287
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.068654105504925e-06,
+      "loss": 0.1569,
+      "step": 12288
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.067373488529106e-06,
+      "loss": 0.1453,
+      "step": 12289
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.066093222364688e-06,
+      "loss": 0.1763,
+      "step": 12290
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0648133070682863e-06,
+      "loss": 0.1697,
+      "step": 12291
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.063533742696505e-06,
+      "loss": 0.1793,
+      "step": 12292
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0622545293059304e-06,
+      "loss": 0.1782,
+      "step": 12293
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0609756669531355e-06,
+      "loss": 0.1703,
+      "step": 12294
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0596971556946753e-06,
+      "loss": 0.1601,
+      "step": 12295
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0584189955870913e-06,
+      "loss": 0.1627,
+      "step": 12296
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.057141186686907e-06,
+      "loss": 0.1796,
+      "step": 12297
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0558637290506358e-06,
+      "loss": 0.1782,
+      "step": 12298
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.05458662273477e-06,
+      "loss": 0.1781,
+      "step": 12299
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.053309867795783e-06,
+      "loss": 0.1442,
+      "step": 12300
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0520334642901463e-06,
+      "loss": 0.1589,
+      "step": 12301
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0507574122743023e-06,
+      "loss": 0.1659,
+      "step": 12302
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.049481711804684e-06,
+      "loss": 0.1698,
+      "step": 12303
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0482063629377114e-06,
+      "loss": 0.1815,
+      "step": 12304
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.046931365729776e-06,
+      "loss": 0.174,
+      "step": 12305
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0456567202372746e-06,
+      "loss": 0.1621,
+      "step": 12306
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0443824265165692e-06,
+      "loss": 0.1821,
+      "step": 12307
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0431084846240137e-06,
+      "loss": 0.1644,
+      "step": 12308
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0418348946159537e-06,
+      "loss": 0.185,
+      "step": 12309
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0405616565487065e-06,
+      "loss": 0.1664,
+      "step": 12310
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.039288770478579e-06,
+      "loss": 0.1844,
+      "step": 12311
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.038016236461866e-06,
+      "loss": 0.167,
+      "step": 12312
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0367440545548433e-06,
+      "loss": 0.1736,
+      "step": 12313
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0354722248137715e-06,
+      "loss": 0.1618,
+      "step": 12314
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0342007472948956e-06,
+      "loss": 0.1492,
+      "step": 12315
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.032929622054445e-06,
+      "loss": 0.1829,
+      "step": 12316
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.031658849148633e-06,
+      "loss": 0.1645,
+      "step": 12317
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0303884286336607e-06,
+      "loss": 0.1713,
+      "step": 12318
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0291183605657095e-06,
+      "loss": 0.1745,
+      "step": 12319
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.027848645000946e-06,
+      "loss": 0.169,
+      "step": 12320
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0265792819955232e-06,
+      "loss": 0.1415,
+      "step": 12321
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.025310271605576e-06,
+      "loss": 0.1585,
+      "step": 12322
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0240416138872256e-06,
+      "loss": 0.1629,
+      "step": 12323
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0227733088965773e-06,
+      "loss": 0.1522,
+      "step": 12324
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0215053566897203e-06,
+      "loss": 0.1651,
+      "step": 12325
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0202377573227282e-06,
+      "loss": 0.1721,
+      "step": 12326
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.018970510851661e-06,
+      "loss": 0.1561,
+      "step": 12327
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.017703617332555e-06,
+      "loss": 0.1662,
+      "step": 12328
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0164370768214438e-06,
+      "loss": 0.1879,
+      "step": 12329
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.015170889374338e-06,
+      "loss": 0.1977,
+      "step": 12330
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0139050550472273e-06,
+      "loss": 0.1743,
+      "step": 12331
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0126395738961014e-06,
+      "loss": 0.1667,
+      "step": 12332
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0113744459769147e-06,
+      "loss": 0.1667,
+      "step": 12333
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0101096713456225e-06,
+      "loss": 0.1811,
+      "step": 12334
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.00884525005816e-06,
+      "loss": 0.1427,
+      "step": 12335
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.007581182170435e-06,
+      "loss": 0.1542,
+      "step": 12336
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0063174677383613e-06,
+      "loss": 0.1552,
+      "step": 12337
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0050541068178173e-06,
+      "loss": 0.1595,
+      "step": 12338
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0037910994646736e-06,
+      "loss": 0.1751,
+      "step": 12339
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0025284457347914e-06,
+      "loss": 0.2168,
+      "step": 12340
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0012661456840055e-06,
+      "loss": 0.1588,
+      "step": 12341
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0000041993681396e-06,
+      "loss": 0.1645,
+      "step": 12342
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9987426068430025e-06,
+      "loss": 0.1664,
+      "step": 12343
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.997481368164388e-06,
+      "loss": 0.1601,
+      "step": 12344
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9962204833880714e-06,
+      "loss": 0.1641,
+      "step": 12345
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9949599525698147e-06,
+      "loss": 0.1577,
+      "step": 12346
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.993699775765363e-06,
+      "loss": 0.1359,
+      "step": 12347
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.992439953030447e-06,
+      "loss": 0.1647,
+      "step": 12348
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9911804844207793e-06,
+      "loss": 0.1426,
+      "step": 12349
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9899213699920604e-06,
+      "loss": 0.1503,
+      "step": 12350
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9886626097999717e-06,
+      "loss": 0.1514,
+      "step": 12351
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.987404203900181e-06,
+      "loss": 0.1825,
+      "step": 12352
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9861461523483407e-06,
+      "loss": 0.1349,
+      "step": 12353
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.984888455200085e-06,
+      "loss": 0.1659,
+      "step": 12354
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9836311125110376e-06,
+      "loss": 0.1998,
+      "step": 12355
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9823741243367956e-06,
+      "loss": 0.1553,
+      "step": 12356
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9811174907329555e-06,
+      "loss": 0.1634,
+      "step": 12357
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.979861211755091e-06,
+      "loss": 0.1595,
+      "step": 12358
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.978605287458749e-06,
+      "loss": 0.1522,
+      "step": 12359
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9773497178994862e-06,
+      "loss": 0.1444,
+      "step": 12360
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.976094503132817e-06,
+      "loss": 0.163,
+      "step": 12361
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9748396432142545e-06,
+      "loss": 0.1599,
+      "step": 12362
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9735851381992998e-06,
+      "loss": 0.1743,
+      "step": 12363
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.972330988143425e-06,
+      "loss": 0.184,
+      "step": 12364
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9710771931020957e-06,
+      "loss": 0.1678,
+      "step": 12365
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9698237531307584e-06,
+      "loss": 0.1706,
+      "step": 12366
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.968570668284845e-06,
+      "loss": 0.1577,
+      "step": 12367
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9673179386197772e-06,
+      "loss": 0.1867,
+      "step": 12368
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.966065564190949e-06,
+      "loss": 0.1777,
+      "step": 12369
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9648135450537478e-06,
+      "loss": 0.1804,
+      "step": 12370
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9635618812635414e-06,
+      "loss": 0.1601,
+      "step": 12371
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.962310572875685e-06,
+      "loss": 0.156,
+      "step": 12372
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9610596199455157e-06,
+      "loss": 0.1685,
+      "step": 12373
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.959809022528355e-06,
+      "loss": 0.174,
+      "step": 12374
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9585587806795103e-06,
+      "loss": 0.1656,
+      "step": 12375
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9573088944542707e-06,
+      "loss": 0.1904,
+      "step": 12376
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9560593639079108e-06,
+      "loss": 0.1844,
+      "step": 12377
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.954810189095694e-06,
+      "loss": 0.1648,
+      "step": 12378
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.953561370072855e-06,
+      "loss": 0.1771,
+      "step": 12379
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9523129068946277e-06,
+      "loss": 0.1986,
+      "step": 12380
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9510647996162268e-06,
+      "loss": 0.1801,
+      "step": 12381
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.949817048292839e-06,
+      "loss": 0.1493,
+      "step": 12382
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.948569652979655e-06,
+      "loss": 0.1488,
+      "step": 12383
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.947322613731829e-06,
+      "loss": 0.1535,
+      "step": 12384
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9460759306045185e-06,
+      "loss": 0.1682,
+      "step": 12385
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9448296036528556e-06,
+      "loss": 0.1686,
+      "step": 12386
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9435836329319512e-06,
+      "loss": 0.1644,
+      "step": 12387
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9423380184969166e-06,
+      "loss": 0.1593,
+      "step": 12388
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9410927604028297e-06,
+      "loss": 0.1431,
+      "step": 12389
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9398478587047596e-06,
+      "loss": 0.1884,
+      "step": 12390
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.938603313457771e-06,
+      "loss": 0.1597,
+      "step": 12391
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9373591247168934e-06,
+      "loss": 0.1578,
+      "step": 12392
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.936115292537152e-06,
+      "loss": 0.1609,
+      "step": 12393
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9348718169735536e-06,
+      "loss": 0.1624,
+      "step": 12394
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9336286980810904e-06,
+      "loss": 0.1529,
+      "step": 12395
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.932385935914738e-06,
+      "loss": 0.1705,
+      "step": 12396
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9311435305294546e-06,
+      "loss": 0.1635,
+      "step": 12397
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9299014819801854e-06,
+      "loss": 0.1662,
+      "step": 12398
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9286597903218584e-06,
+      "loss": 0.1688,
+      "step": 12399
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9274184556093865e-06,
+      "loss": 0.1559,
+      "step": 12400
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.926177477897665e-06,
+      "loss": 0.1573,
+      "step": 12401
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.924936857241576e-06,
+      "loss": 0.1654,
+      "step": 12402
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.923696593695984e-06,
+      "loss": 0.1566,
+      "step": 12403
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.922456687315738e-06,
+      "loss": 0.1598,
+      "step": 12404
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.921217138155672e-06,
+      "loss": 0.1402,
+      "step": 12405
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9199779462706068e-06,
+      "loss": 0.1814,
+      "step": 12406
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9187391117153355e-06,
+      "loss": 0.1767,
+      "step": 12407
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9175006345446513e-06,
+      "loss": 0.1528,
+      "step": 12408
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9162625148133252e-06,
+      "loss": 0.1868,
+      "step": 12409
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9150247525761047e-06,
+      "loss": 0.1599,
+      "step": 12410
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9137873478877377e-06,
+      "loss": 0.1487,
+      "step": 12411
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9125503008029388e-06,
+      "loss": 0.1843,
+      "step": 12412
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9113136113764174e-06,
+      "loss": 0.1623,
+      "step": 12413
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9100772796628698e-06,
+      "loss": 0.1811,
+      "step": 12414
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9088413057169643e-06,
+      "loss": 0.1672,
+      "step": 12415
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9076056895933635e-06,
+      "loss": 0.1744,
+      "step": 12416
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9063704313467112e-06,
+      "loss": 0.1691,
+      "step": 12417
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9051355310316333e-06,
+      "loss": 0.1377,
+      "step": 12418
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9039009887027482e-06,
+      "loss": 0.1856,
+      "step": 12419
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9026668044146445e-06,
+      "loss": 0.1931,
+      "step": 12420
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9014329782219054e-06,
+      "loss": 0.1475,
+      "step": 12421
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.900199510179097e-06,
+      "loss": 0.1769,
+      "step": 12422
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.898966400340766e-06,
+      "loss": 0.1724,
+      "step": 12423
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8977336487614461e-06,
+      "loss": 0.1664,
+      "step": 12424
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8965012554956542e-06,
+      "loss": 0.1764,
+      "step": 12425
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8952692205978918e-06,
+      "loss": 0.1761,
+      "step": 12426
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8940375441226432e-06,
+      "loss": 0.1994,
+      "step": 12427
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8928062261243796e-06,
+      "loss": 0.1622,
+      "step": 12428
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8915752666575537e-06,
+      "loss": 0.1703,
+      "step": 12429
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8903446657766033e-06,
+      "loss": 0.1651,
+      "step": 12430
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8891144235359505e-06,
+      "loss": 0.18,
+      "step": 12431
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8878845399900037e-06,
+      "loss": 0.1673,
+      "step": 12432
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8866550151931452e-06,
+      "loss": 0.1931,
+      "step": 12433
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8854258491997569e-06,
+      "loss": 0.1666,
+      "step": 12434
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8841970420641953e-06,
+      "loss": 0.1513,
+      "step": 12435
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8829685938408038e-06,
+      "loss": 0.1376,
+      "step": 12436
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8817405045839099e-06,
+      "loss": 0.1688,
+      "step": 12437
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.880512774347818e-06,
+      "loss": 0.1896,
+      "step": 12438
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8792854031868324e-06,
+      "loss": 0.1509,
+      "step": 12439
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8780583911552254e-06,
+      "loss": 0.1626,
+      "step": 12440
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8768317383072598e-06,
+      "loss": 0.1725,
+      "step": 12441
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8756054446971904e-06,
+      "loss": 0.1739,
+      "step": 12442
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8743795103792406e-06,
+      "loss": 0.1868,
+      "step": 12443
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.87315393540763e-06,
+      "loss": 0.1632,
+      "step": 12444
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8719287198365576e-06,
+      "loss": 0.1665,
+      "step": 12445
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8707038637202068e-06,
+      "loss": 0.1629,
+      "step": 12446
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8694793671127454e-06,
+      "loss": 0.1681,
+      "step": 12447
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8682552300683255e-06,
+      "loss": 0.1682,
+      "step": 12448
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8670314526410837e-06,
+      "loss": 0.1613,
+      "step": 12449
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8658080348851394e-06,
+      "loss": 0.1566,
+      "step": 12450
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8645849768545966e-06,
+      "loss": 0.1434,
+      "step": 12451
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8633622786035455e-06,
+      "loss": 0.1621,
+      "step": 12452
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8621399401860563e-06,
+      "loss": 0.1799,
+      "step": 12453
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8609179616561868e-06,
+      "loss": 0.144,
+      "step": 12454
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.859696343067977e-06,
+      "loss": 0.172,
+      "step": 12455
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8584750844754528e-06,
+      "loss": 0.1714,
+      "step": 12456
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8572541859326209e-06,
+      "loss": 0.1689,
+      "step": 12457
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.856033647493476e-06,
+      "loss": 0.1742,
+      "step": 12458
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8548134692119946e-06,
+      "loss": 0.1579,
+      "step": 12459
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8535936511421405e-06,
+      "loss": 0.1775,
+      "step": 12460
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8523741933378492e-06,
+      "loss": 0.1666,
+      "step": 12461
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.85115509585306e-06,
+      "loss": 0.1945,
+      "step": 12462
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.849936358741684e-06,
+      "loss": 0.1761,
+      "step": 12463
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8487179820576128e-06,
+      "loss": 0.1701,
+      "step": 12464
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8474999658547366e-06,
+      "loss": 0.1739,
+      "step": 12465
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.846282310186911e-06,
+      "loss": 0.1734,
+      "step": 12466
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.845065015107993e-06,
+      "loss": 0.1396,
+      "step": 12467
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8438480806718163e-06,
+      "loss": 0.1983,
+      "step": 12468
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.842631506932191e-06,
+      "loss": 0.1766,
+      "step": 12469
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8414152939429274e-06,
+      "loss": 0.204,
+      "step": 12470
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8401994417578062e-06,
+      "loss": 0.1571,
+      "step": 12471
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.838983950430594e-06,
+      "loss": 0.2012,
+      "step": 12472
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8377688200150545e-06,
+      "loss": 0.1636,
+      "step": 12473
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.836554050564917e-06,
+      "loss": 0.1634,
+      "step": 12474
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8353396421339054e-06,
+      "loss": 0.1585,
+      "step": 12475
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8341255947757253e-06,
+      "loss": 0.1469,
+      "step": 12476
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8329119085440684e-06,
+      "loss": 0.1577,
+      "step": 12477
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.831698583492607e-06,
+      "loss": 0.1661,
+      "step": 12478
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8304856196749998e-06,
+      "loss": 0.1513,
+      "step": 12479
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.829273017144888e-06,
+      "loss": 0.1423,
+      "step": 12480
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8280607759558977e-06,
+      "loss": 0.1594,
+      "step": 12481
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8268488961616392e-06,
+      "loss": 0.1675,
+      "step": 12482
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8256373778157066e-06,
+      "loss": 0.1496,
+      "step": 12483
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.824426220971678e-06,
+      "loss": 0.1657,
+      "step": 12484
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8232154256831159e-06,
+      "loss": 0.1816,
+      "step": 12485
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8220049920035643e-06,
+      "loss": 0.173,
+      "step": 12486
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8207949199865548e-06,
+      "loss": 0.1591,
+      "step": 12487
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.819585209685605e-06,
+      "loss": 0.1822,
+      "step": 12488
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8183758611542024e-06,
+      "loss": 0.1722,
+      "step": 12489
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8171668744458403e-06,
+      "loss": 0.1765,
+      "step": 12490
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8159582496139817e-06,
+      "loss": 0.187,
+      "step": 12491
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.81474998671207e-06,
+      "loss": 0.202,
+      "step": 12492
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8135420857935504e-06,
+      "loss": 0.1444,
+      "step": 12493
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8123345469118326e-06,
+      "loss": 0.191,
+      "step": 12494
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.811127370120318e-06,
+      "loss": 0.161,
+      "step": 12495
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.809920555472402e-06,
+      "loss": 0.1702,
+      "step": 12496
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.808714103021445e-06,
+      "loss": 0.1653,
+      "step": 12497
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8075080128208045e-06,
+      "loss": 0.1637,
+      "step": 12498
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8063022849238188e-06,
+      "loss": 0.1687,
+      "step": 12499
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8050969193838063e-06,
+      "loss": 0.166,
+      "step": 12500
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8038919162540814e-06,
+      "loss": 0.1659,
+      "step": 12501
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.8026872755879254e-06,
+      "loss": 0.177,
+      "step": 12502
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.8014829974386162e-06,
+      "loss": 0.162,
+      "step": 12503
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.8002790818594106e-06,
+      "loss": 0.1773,
+      "step": 12504
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.79907552890355e-06,
+      "loss": 0.1688,
+      "step": 12505
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7978723386242614e-06,
+      "loss": 0.1642,
+      "step": 12506
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7966695110747546e-06,
+      "loss": 0.1549,
+      "step": 12507
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7954670463082214e-06,
+      "loss": 0.1756,
+      "step": 12508
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7942649443778415e-06,
+      "loss": 0.1428,
+      "step": 12509
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7930632053367747e-06,
+      "loss": 0.1534,
+      "step": 12510
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7918618292381707e-06,
+      "loss": 0.1481,
+      "step": 12511
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7906608161351503e-06,
+      "loss": 0.1494,
+      "step": 12512
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7894601660808352e-06,
+      "loss": 0.1766,
+      "step": 12513
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7882598791283222e-06,
+      "loss": 0.1564,
+      "step": 12514
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7870599553306856e-06,
+      "loss": 0.1872,
+      "step": 12515
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7858603947410003e-06,
+      "loss": 0.1539,
+      "step": 12516
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7846611974123062e-06,
+      "loss": 0.1694,
+      "step": 12517
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7834623633976433e-06,
+      "loss": 0.1634,
+      "step": 12518
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.782263892750028e-06,
+      "loss": 0.1745,
+      "step": 12519
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7810657855224555e-06,
+      "loss": 0.189,
+      "step": 12520
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7798680417679193e-06,
+      "loss": 0.185,
+      "step": 12521
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7786706615393813e-06,
+      "loss": 0.1878,
+      "step": 12522
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7774736448897944e-06,
+      "loss": 0.1724,
+      "step": 12523
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7762769918721035e-06,
+      "loss": 0.1917,
+      "step": 12524
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.775080702539219e-06,
+      "loss": 0.1711,
+      "step": 12525
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7738847769440514e-06,
+      "loss": 0.1608,
+      "step": 12526
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.772689215139487e-06,
+      "loss": 0.1681,
+      "step": 12527
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.771494017178399e-06,
+      "loss": 0.1459,
+      "step": 12528
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.770299183113644e-06,
+      "loss": 0.1972,
+      "step": 12529
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7691047129980598e-06,
+      "loss": 0.1597,
+      "step": 12530
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7679106068844743e-06,
+      "loss": 0.1571,
+      "step": 12531
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7667168648256917e-06,
+      "loss": 0.1596,
+      "step": 12532
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7655234868745064e-06,
+      "loss": 0.1472,
+      "step": 12533
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7643304730836942e-06,
+      "loss": 0.1736,
+      "step": 12534
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7631378235060126e-06,
+      "loss": 0.1746,
+      "step": 12535
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7619455381942075e-06,
+      "loss": 0.151,
+      "step": 12536
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7607536172010053e-06,
+      "loss": 0.184,
+      "step": 12537
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7595620605791176e-06,
+      "loss": 0.1416,
+      "step": 12538
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7583708683812417e-06,
+      "loss": 0.1628,
+      "step": 12539
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7571800406600503e-06,
+      "loss": 0.1513,
+      "step": 12540
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7559895774682123e-06,
+      "loss": 0.1589,
+      "step": 12541
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7547994788583755e-06,
+      "loss": 0.1547,
+      "step": 12542
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7536097448831636e-06,
+      "loss": 0.1495,
+      "step": 12543
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7524203755952008e-06,
+      "loss": 0.1631,
+      "step": 12544
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7512313710470775e-06,
+      "loss": 0.1502,
+      "step": 12545
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.750042731291377e-06,
+      "loss": 0.1752,
+      "step": 12546
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7488544563806732e-06,
+      "loss": 0.1632,
+      "step": 12547
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7476665463675068e-06,
+      "loss": 0.1677,
+      "step": 12548
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7464790013044163e-06,
+      "loss": 0.167,
+      "step": 12549
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7452918212439196e-06,
+      "loss": 0.1623,
+      "step": 12550
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7441050062385144e-06,
+      "loss": 0.1831,
+      "step": 12551
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7429185563406947e-06,
+      "loss": 0.1465,
+      "step": 12552
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7417324716029226e-06,
+      "loss": 0.1707,
+      "step": 12553
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.740546752077653e-06,
+      "loss": 0.1423,
+      "step": 12554
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.739361397817324e-06,
+      "loss": 0.162,
+      "step": 12555
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7381764088743547e-06,
+      "loss": 0.1657,
+      "step": 12556
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.736991785301153e-06,
+      "loss": 0.1771,
+      "step": 12557
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.735807527150104e-06,
+      "loss": 0.1645,
+      "step": 12558
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7346236344735834e-06,
+      "loss": 0.1616,
+      "step": 12559
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.733440107323945e-06,
+      "loss": 0.1705,
+      "step": 12560
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.73225694575353e-06,
+      "loss": 0.1689,
+      "step": 12561
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7310741498146633e-06,
+      "loss": 0.1751,
+      "step": 12562
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.729891719559651e-06,
+      "loss": 0.1647,
+      "step": 12563
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7287096550407855e-06,
+      "loss": 0.1771,
+      "step": 12564
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7275279563103454e-06,
+      "loss": 0.1886,
+      "step": 12565
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7263466234205806e-06,
+      "loss": 0.1985,
+      "step": 12566
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7251656564237462e-06,
+      "loss": 0.1408,
+      "step": 12567
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7239850553720582e-06,
+      "loss": 0.183,
+      "step": 12568
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.722804820317735e-06,
+      "loss": 0.1881,
+      "step": 12569
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7216249513129712e-06,
+      "loss": 0.1614,
+      "step": 12570
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7204454484099375e-06,
+      "loss": 0.1856,
+      "step": 12571
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7192663116608067e-06,
+      "loss": 0.1629,
+      "step": 12572
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7180875411177157e-06,
+      "loss": 0.1475,
+      "step": 12573
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7169091368327962e-06,
+      "loss": 0.163,
+      "step": 12574
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7157310988581688e-06,
+      "loss": 0.1459,
+      "step": 12575
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7145534272459218e-06,
+      "loss": 0.1769,
+      "step": 12576
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7133761220481415e-06,
+      "loss": 0.1618,
+      "step": 12577
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7121991833168905e-06,
+      "loss": 0.1691,
+      "step": 12578
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7110226111042193e-06,
+      "loss": 0.1698,
+      "step": 12579
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.70984640546216e-06,
+      "loss": 0.1573,
+      "step": 12580
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7086705664427272e-06,
+      "loss": 0.16,
+      "step": 12581
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.707495094097924e-06,
+      "loss": 0.1593,
+      "step": 12582
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.706319988479732e-06,
+      "loss": 0.1681,
+      "step": 12583
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7051452496401211e-06,
+      "loss": 0.1488,
+      "step": 12584
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7039708776310404e-06,
+      "loss": 0.1652,
+      "step": 12585
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7027968725044264e-06,
+      "loss": 0.2121,
+      "step": 12586
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7016232343121985e-06,
+      "loss": 0.1795,
+      "step": 12587
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7004499631062587e-06,
+      "loss": 0.1807,
+      "step": 12588
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6992770589384943e-06,
+      "loss": 0.1457,
+      "step": 12589
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.698104521860775e-06,
+      "loss": 0.16,
+      "step": 12590
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6969323519249548e-06,
+      "loss": 0.1512,
+      "step": 12591
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6957605491828732e-06,
+      "loss": 0.1652,
+      "step": 12592
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6945891136863513e-06,
+      "loss": 0.1831,
+      "step": 12593
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6934180454871908e-06,
+      "loss": 0.1766,
+      "step": 12594
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6922473446371846e-06,
+      "loss": 0.1496,
+      "step": 12595
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6910770111881082e-06,
+      "loss": 0.1487,
+      "step": 12596
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.68990704519171e-06,
+      "loss": 0.1784,
+      "step": 12597
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6887374466997408e-06,
+      "loss": 0.1611,
+      "step": 12598
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.687568215763914e-06,
+      "loss": 0.1985,
+      "step": 12599
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6863993524359457e-06,
+      "loss": 0.1556,
+      "step": 12600
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.685230856767527e-06,
+      "loss": 0.145,
+      "step": 12601
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.684062728810325e-06,
+      "loss": 0.1716,
+      "step": 12602
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6828949686160112e-06,
+      "loss": 0.1611,
+      "step": 12603
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6817275762362195e-06,
+      "loss": 0.1717,
+      "step": 12604
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6805605517225765e-06,
+      "loss": 0.1623,
+      "step": 12605
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6793938951266998e-06,
+      "loss": 0.1632,
+      "step": 12606
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6782276065001758e-06,
+      "loss": 0.1865,
+      "step": 12607
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6770616858945865e-06,
+      "loss": 0.1539,
+      "step": 12608
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.675896133361491e-06,
+      "loss": 0.1665,
+      "step": 12609
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6747309489524365e-06,
+      "loss": 0.1995,
+      "step": 12610
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.673566132718951e-06,
+      "loss": 0.1713,
+      "step": 12611
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6724016847125479e-06,
+      "loss": 0.1945,
+      "step": 12612
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6712376049847235e-06,
+      "loss": 0.173,
+      "step": 12613
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6700738935869565e-06,
+      "loss": 0.1745,
+      "step": 12614
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6689105505707115e-06,
+      "loss": 0.1765,
+      "step": 12615
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6677475759874374e-06,
+      "loss": 0.1435,
+      "step": 12616
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6665849698885628e-06,
+      "loss": 0.1726,
+      "step": 12617
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6654227323255047e-06,
+      "loss": 0.1801,
+      "step": 12618
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6642608633496615e-06,
+      "loss": 0.1643,
+      "step": 12619
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6630993630124149e-06,
+      "loss": 0.1754,
+      "step": 12620
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6619382313651333e-06,
+      "loss": 0.1579,
+      "step": 12621
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6607774684591605e-06,
+      "loss": 0.158,
+      "step": 12622
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6596170743458351e-06,
+      "loss": 0.1707,
+      "step": 12623
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.658457049076476e-06,
+      "loss": 0.1551,
+      "step": 12624
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.657297392702375e-06,
+      "loss": 0.148,
+      "step": 12625
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6561381052748283e-06,
+      "loss": 0.1425,
+      "step": 12626
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.654979186845096e-06,
+      "loss": 0.197,
+      "step": 12627
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6538206374644283e-06,
+      "loss": 0.179,
+      "step": 12628
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.652662457184071e-06,
+      "loss": 0.1794,
+      "step": 12629
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.651504646055233e-06,
+      "loss": 0.1718,
+      "step": 12630
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6503472041291224e-06,
+      "loss": 0.1843,
+      "step": 12631
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6491901314569236e-06,
+      "loss": 0.1543,
+      "step": 12632
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6480334280898059e-06,
+      "loss": 0.1727,
+      "step": 12633
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.646877094078929e-06,
+      "loss": 0.1496,
+      "step": 12634
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.645721129475425e-06,
+      "loss": 0.1528,
+      "step": 12635
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6445655343304157e-06,
+      "loss": 0.1691,
+      "step": 12636
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.643410308695007e-06,
+      "loss": 0.1697,
+      "step": 12637
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6422554526202883e-06,
+      "loss": 0.1645,
+      "step": 12638
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6411009661573295e-06,
+      "loss": 0.1811,
+      "step": 12639
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.639946849357188e-06,
+      "loss": 0.1598,
+      "step": 12640
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6387931022709036e-06,
+      "loss": 0.1636,
+      "step": 12641
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.637639724949498e-06,
+      "loss": 0.1533,
+      "step": 12642
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6364867174439792e-06,
+      "loss": 0.1824,
+      "step": 12643
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6353340798053395e-06,
+      "loss": 0.1731,
+      "step": 12644
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6341818120845464e-06,
+      "loss": 0.169,
+      "step": 12645
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6330299143325646e-06,
+      "loss": 0.1491,
+      "step": 12646
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6318783866003352e-06,
+      "loss": 0.1699,
+      "step": 12647
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6307272289387755e-06,
+      "loss": 0.18,
+      "step": 12648
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6295764413988058e-06,
+      "loss": 0.1665,
+      "step": 12649
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6284260240313054e-06,
+      "loss": 0.1986,
+      "step": 12650
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6272759768871605e-06,
+      "loss": 0.1836,
+      "step": 12651
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.62612630001723e-06,
+      "loss": 0.1361,
+      "step": 12652
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6249769934723481e-06,
+      "loss": 0.1455,
+      "step": 12653
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6238280573033526e-06,
+      "loss": 0.1872,
+      "step": 12654
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6226794915610477e-06,
+      "loss": 0.159,
+      "step": 12655
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6215312962962249e-06,
+      "loss": 0.1803,
+      "step": 12656
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6203834715596711e-06,
+      "loss": 0.152,
+      "step": 12657
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6192360174021393e-06,
+      "loss": 0.1604,
+      "step": 12658
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6180889338743767e-06,
+      "loss": 0.163,
+      "step": 12659
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6169422210271123e-06,
+      "loss": 0.1553,
+      "step": 12660
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6157958789110584e-06,
+      "loss": 0.1782,
+      "step": 12661
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6146499075769095e-06,
+      "loss": 0.1913,
+      "step": 12662
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6135043070753465e-06,
+      "loss": 0.1439,
+      "step": 12663
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6123590774570307e-06,
+      "loss": 0.171,
+      "step": 12664
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.61121421877261e-06,
+      "loss": 0.1762,
+      "step": 12665
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6100697310727132e-06,
+      "loss": 0.1608,
+      "step": 12666
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6089256144079545e-06,
+      "loss": 0.175,
+      "step": 12667
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6077818688289316e-06,
+      "loss": 0.1665,
+      "step": 12668
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6066384943862245e-06,
+      "loss": 0.1696,
+      "step": 12669
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6054954911303988e-06,
+      "loss": 0.1668,
+      "step": 12670
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6043528591120007e-06,
+      "loss": 0.1695,
+      "step": 12671
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.603210598381566e-06,
+      "loss": 0.1804,
+      "step": 12672
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6020687089896026e-06,
+      "loss": 0.1595,
+      "step": 12673
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6009271909866152e-06,
+      "loss": 0.1531,
+      "step": 12674
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.599786044423086e-06,
+      "loss": 0.1692,
+      "step": 12675
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5986452693494759e-06,
+      "loss": 0.1439,
+      "step": 12676
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5975048658162417e-06,
+      "loss": 0.1824,
+      "step": 12677
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5963648338738114e-06,
+      "loss": 0.1594,
+      "step": 12678
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5952251735725989e-06,
+      "loss": 0.1523,
+      "step": 12679
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5940858849630137e-06,
+      "loss": 0.1862,
+      "step": 12680
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5929469680954313e-06,
+      "loss": 0.1468,
+      "step": 12681
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5918084230202224e-06,
+      "loss": 0.1744,
+      "step": 12682
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5906702497877369e-06,
+      "loss": 0.1711,
+      "step": 12683
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5895324484483076e-06,
+      "loss": 0.1554,
+      "step": 12684
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.588395019052259e-06,
+      "loss": 0.1655,
+      "step": 12685
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.587257961649885e-06,
+      "loss": 0.1671,
+      "step": 12686
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5861212762914747e-06,
+      "loss": 0.1555,
+      "step": 12687
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.584984963027295e-06,
+      "loss": 0.1859,
+      "step": 12688
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5838490219075996e-06,
+      "loss": 0.1876,
+      "step": 12689
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5827134529826226e-06,
+      "loss": 0.141,
+      "step": 12690
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.581578256302585e-06,
+      "loss": 0.1705,
+      "step": 12691
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5804434319176876e-06,
+      "loss": 0.1568,
+      "step": 12692
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5793089798781181e-06,
+      "loss": 0.1879,
+      "step": 12693
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5781749002340462e-06,
+      "loss": 0.1855,
+      "step": 12694
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5770411930356255e-06,
+      "loss": 0.1601,
+      "step": 12695
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5759078583329923e-06,
+      "loss": 0.1566,
+      "step": 12696
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5747748961762677e-06,
+      "loss": 0.1519,
+      "step": 12697
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.573642306615557e-06,
+      "loss": 0.1689,
+      "step": 12698
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5725100897009416e-06,
+      "loss": 0.1658,
+      "step": 12699
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5713782454825012e-06,
+      "loss": 0.1594,
+      "step": 12700
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5702467740102811e-06,
+      "loss": 0.1904,
+      "step": 12701
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5691156753343262e-06,
+      "loss": 0.1749,
+      "step": 12702
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5679849495046595e-06,
+      "loss": 0.1649,
+      "step": 12703
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5668545965712767e-06,
+      "loss": 0.1898,
+      "step": 12704
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5657246165841777e-06,
+      "loss": 0.1665,
+      "step": 12705
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5645950095933271e-06,
+      "loss": 0.165,
+      "step": 12706
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5634657756486792e-06,
+      "loss": 0.1606,
+      "step": 12707
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5623369148001822e-06,
+      "loss": 0.136,
+      "step": 12708
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.56120842709775e-06,
+      "loss": 0.1533,
+      "step": 12709
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5600803125912922e-06,
+      "loss": 0.1721,
+      "step": 12710
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5589525713306974e-06,
+      "loss": 0.1477,
+      "step": 12711
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5578252033658392e-06,
+      "loss": 0.1751,
+      "step": 12712
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5566982087465743e-06,
+      "loss": 0.1381,
+      "step": 12713
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5555715875227417e-06,
+      "loss": 0.1554,
+      "step": 12714
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.554445339744166e-06,
+      "loss": 0.1912,
+      "step": 12715
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5533194654606553e-06,
+      "loss": 0.1497,
+      "step": 12716
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5521939647219974e-06,
+      "loss": 0.1679,
+      "step": 12717
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.551068837577968e-06,
+      "loss": 0.159,
+      "step": 12718
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.549944084078324e-06,
+      "loss": 0.1768,
+      "step": 12719
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.548819704272807e-06,
+      "loss": 0.1664,
+      "step": 12720
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5476956982111413e-06,
+      "loss": 0.1458,
+      "step": 12721
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.546572065943034e-06,
+      "loss": 0.1561,
+      "step": 12722
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5454488075181772e-06,
+      "loss": 0.1403,
+      "step": 12723
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5443259229862451e-06,
+      "loss": 0.172,
+      "step": 12724
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5432034123968963e-06,
+      "loss": 0.1591,
+      "step": 12725
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.542081275799776e-06,
+      "loss": 0.1706,
+      "step": 12726
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.540959513244501e-06,
+      "loss": 0.1753,
+      "step": 12727
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.539838124780687e-06,
+      "loss": 0.1758,
+      "step": 12728
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5387171104579268e-06,
+      "loss": 0.1583,
+      "step": 12729
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5375964703257884e-06,
+      "loss": 0.1611,
+      "step": 12730
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.536476204433841e-06,
+      "loss": 0.1682,
+      "step": 12731
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5353563128316162e-06,
+      "loss": 0.1788,
+      "step": 12732
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5342367955686488e-06,
+      "loss": 0.1493,
+      "step": 12733
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5331176526944468e-06,
+      "loss": 0.1829,
+      "step": 12734
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5319988842584976e-06,
+      "loss": 0.1611,
+      "step": 12735
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5308804903102847e-06,
+      "loss": 0.1755,
+      "step": 12736
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.529762470899262e-06,
+      "loss": 0.1332,
+      "step": 12737
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.528644826074873e-06,
+      "loss": 0.1588,
+      "step": 12738
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5275275558865499e-06,
+      "loss": 0.1696,
+      "step": 12739
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5264106603836981e-06,
+      "loss": 0.1545,
+      "step": 12740
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5252941396157106e-06,
+      "loss": 0.1373,
+      "step": 12741
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5241779936319656e-06,
+      "loss": 0.1884,
+      "step": 12742
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5230622224818237e-06,
+      "loss": 0.1544,
+      "step": 12743
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5219468262146276e-06,
+      "loss": 0.1548,
+      "step": 12744
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5208318048797054e-06,
+      "loss": 0.1772,
+      "step": 12745
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.519717158526367e-06,
+      "loss": 0.1655,
+      "step": 12746
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5186028872039072e-06,
+      "loss": 0.1569,
+      "step": 12747
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.517488990961603e-06,
+      "loss": 0.1797,
+      "step": 12748
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5163754698487144e-06,
+      "loss": 0.1586,
+      "step": 12749
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.515262323914487e-06,
+      "loss": 0.1651,
+      "step": 12750
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5141495532081462e-06,
+      "loss": 0.1757,
+      "step": 12751
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.513037157778906e-06,
+      "loss": 0.1552,
+      "step": 12752
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.511925137675959e-06,
+      "loss": 0.2002,
+      "step": 12753
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5108134929484853e-06,
+      "loss": 0.1579,
+      "step": 12754
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5097022236456393e-06,
+      "loss": 0.185,
+      "step": 12755
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5085913298165733e-06,
+      "loss": 0.1671,
+      "step": 12756
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5074808115104146e-06,
+      "loss": 0.1587,
+      "step": 12757
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.506370668776268e-06,
+      "loss": 0.1655,
+      "step": 12758
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.505260901663237e-06,
+      "loss": 0.1677,
+      "step": 12759
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5041515102203942e-06,
+      "loss": 0.1828,
+      "step": 12760
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5030424944967993e-06,
+      "loss": 0.1604,
+      "step": 12761
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.5019338545415052e-06,
+      "loss": 0.155,
+      "step": 12762
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.500825590403534e-06,
+      "loss": 0.1616,
+      "step": 12763
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4997177021318988e-06,
+      "loss": 0.1752,
+      "step": 12764
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4986101897755946e-06,
+      "loss": 0.1878,
+      "step": 12765
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4975030533835977e-06,
+      "loss": 0.169,
+      "step": 12766
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.496396293004878e-06,
+      "loss": 0.1796,
+      "step": 12767
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4952899086883722e-06,
+      "loss": 0.1488,
+      "step": 12768
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4941839004830117e-06,
+      "loss": 0.1975,
+      "step": 12769
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4930782684377088e-06,
+      "loss": 0.1797,
+      "step": 12770
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4919730126013587e-06,
+      "loss": 0.1546,
+      "step": 12771
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.490868133022839e-06,
+      "loss": 0.1976,
+      "step": 12772
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4897636297510144e-06,
+      "loss": 0.1541,
+      "step": 12773
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4886595028347273e-06,
+      "loss": 0.1691,
+      "step": 12774
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.487555752322809e-06,
+      "loss": 0.1476,
+      "step": 12775
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4864523782640694e-06,
+      "loss": 0.1822,
+      "step": 12776
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4853493807073083e-06,
+      "loss": 0.1585,
+      "step": 12777
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4842467597012954e-06,
+      "loss": 0.1757,
+      "step": 12778
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.483144515294802e-06,
+      "loss": 0.1587,
+      "step": 12779
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4820426475365723e-06,
+      "loss": 0.1562,
+      "step": 12780
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4809411564753285e-06,
+      "loss": 0.1741,
+      "step": 12781
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4798400421597915e-06,
+      "loss": 0.1427,
+      "step": 12782
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4787393046386478e-06,
+      "loss": 0.1786,
+      "step": 12783
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4776389439605842e-06,
+      "loss": 0.1506,
+      "step": 12784
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.476538960174263e-06,
+      "loss": 0.1936,
+      "step": 12785
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4754393533283206e-06,
+      "loss": 0.1485,
+      "step": 12786
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4743401234713973e-06,
+      "loss": 0.165,
+      "step": 12787
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4732412706520982e-06,
+      "loss": 0.1479,
+      "step": 12788
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.472142794919017e-06,
+      "loss": 0.1723,
+      "step": 12789
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4710446963207414e-06,
+      "loss": 0.141,
+      "step": 12790
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.469946974905826e-06,
+      "loss": 0.1626,
+      "step": 12791
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4688496307228173e-06,
+      "loss": 0.163,
+      "step": 12792
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4677526638202467e-06,
+      "loss": 0.1572,
+      "step": 12793
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.466656074246624e-06,
+      "loss": 0.1581,
+      "step": 12794
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4655598620504464e-06,
+      "loss": 0.1762,
+      "step": 12795
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4644640272801913e-06,
+      "loss": 0.1729,
+      "step": 12796
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4633685699843214e-06,
+      "loss": 0.1658,
+      "step": 12797
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4622734902112823e-06,
+      "loss": 0.1546,
+      "step": 12798
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4611787880095018e-06,
+      "loss": 0.1649,
+      "step": 12799
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4600844634273913e-06,
+      "loss": 0.1868,
+      "step": 12800
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4589905165133488e-06,
+      "loss": 0.195,
+      "step": 12801
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.45789694731575e-06,
+      "loss": 0.1644,
+      "step": 12802
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4568037558829585e-06,
+      "loss": 0.1677,
+      "step": 12803
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4557109422633186e-06,
+      "loss": 0.1632,
+      "step": 12804
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.454618506505161e-06,
+      "loss": 0.154,
+      "step": 12805
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4535264486567912e-06,
+      "loss": 0.1836,
+      "step": 12806
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4524347687665108e-06,
+      "loss": 0.1839,
+      "step": 12807
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4513434668825976e-06,
+      "loss": 0.1764,
+      "step": 12808
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4502525430533076e-06,
+      "loss": 0.1892,
+      "step": 12809
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4491619973268932e-06,
+      "loss": 0.1707,
+      "step": 12810
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4480718297515771e-06,
+      "loss": 0.1735,
+      "step": 12811
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4469820403755697e-06,
+      "loss": 0.1651,
+      "step": 12812
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4458926292470721e-06,
+      "loss": 0.1663,
+      "step": 12813
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.444803596414257e-06,
+      "loss": 0.1787,
+      "step": 12814
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.443714941925287e-06,
+      "loss": 0.1594,
+      "step": 12815
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4426266658283072e-06,
+      "loss": 0.1505,
+      "step": 12816
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4415387681714432e-06,
+      "loss": 0.1643,
+      "step": 12817
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4404512490028112e-06,
+      "loss": 0.1636,
+      "step": 12818
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4393641083705e-06,
+      "loss": 0.1924,
+      "step": 12819
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4382773463225896e-06,
+      "loss": 0.1618,
+      "step": 12820
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4371909629071401e-06,
+      "loss": 0.1671,
+      "step": 12821
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4361049581721964e-06,
+      "loss": 0.1701,
+      "step": 12822
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4350193321657845e-06,
+      "loss": 0.202,
+      "step": 12823
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4339340849359173e-06,
+      "loss": 0.1586,
+      "step": 12824
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4328492165305862e-06,
+      "loss": 0.1562,
+      "step": 12825
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4317647269977685e-06,
+      "loss": 0.1566,
+      "step": 12826
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.430680616385427e-06,
+      "loss": 0.1923,
+      "step": 12827
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4295968847415043e-06,
+      "loss": 0.1584,
+      "step": 12828
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.428513532113923e-06,
+      "loss": 0.1784,
+      "step": 12829
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4274305585505977e-06,
+      "loss": 0.1808,
+      "step": 12830
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4263479640994238e-06,
+      "loss": 0.1592,
+      "step": 12831
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.42526574880827e-06,
+      "loss": 0.1645,
+      "step": 12832
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4241839127250047e-06,
+      "loss": 0.176,
+      "step": 12833
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4231024558974616e-06,
+      "loss": 0.1589,
+      "step": 12834
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.422021378373476e-06,
+      "loss": 0.1678,
+      "step": 12835
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4209406802008542e-06,
+      "loss": 0.1642,
+      "step": 12836
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4198603614273843e-06,
+      "loss": 0.1637,
+      "step": 12837
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4187804221008494e-06,
+      "loss": 0.148,
+      "step": 12838
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4177008622690036e-06,
+      "loss": 0.149,
+      "step": 12839
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4166216819795886e-06,
+      "loss": 0.1638,
+      "step": 12840
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4155428812803362e-06,
+      "loss": 0.1444,
+      "step": 12841
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4144644602189496e-06,
+      "loss": 0.1548,
+      "step": 12842
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4133864188431213e-06,
+      "loss": 0.1708,
+      "step": 12843
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4123087572005267e-06,
+      "loss": 0.158,
+      "step": 12844
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4112314753388257e-06,
+      "loss": 0.1637,
+      "step": 12845
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4101545733056588e-06,
+      "loss": 0.1923,
+      "step": 12846
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4090780511486501e-06,
+      "loss": 0.1762,
+      "step": 12847
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4080019089154085e-06,
+      "loss": 0.1512,
+      "step": 12848
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4069261466535255e-06,
+      "loss": 0.1494,
+      "step": 12849
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4058507644105756e-06,
+      "loss": 0.1781,
+      "step": 12850
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.404775762234114e-06,
+      "loss": 0.1591,
+      "step": 12851
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.403701140171685e-06,
+      "loss": 0.1409,
+      "step": 12852
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.402626898270809e-06,
+      "loss": 0.1815,
+      "step": 12853
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4015530365789965e-06,
+      "loss": 0.1626,
+      "step": 12854
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4004795551437345e-06,
+      "loss": 0.1507,
+      "step": 12855
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.3994064540124997e-06,
+      "loss": 0.1566,
+      "step": 12856
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.3983337332327463e-06,
+      "loss": 0.1569,
+      "step": 12857
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.3972613928519151e-06,
+      "loss": 0.1629,
+      "step": 12858
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.3961894329174318e-06,
+      "loss": 0.1852,
+      "step": 12859
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.3951178534766952e-06,
+      "loss": 0.1798,
+      "step": 12860
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3940466545771015e-06,
+      "loss": 0.1616,
+      "step": 12861
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3929758362660239e-06,
+      "loss": 0.1677,
+      "step": 12862
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3919053985908104e-06,
+      "loss": 0.1391,
+      "step": 12863
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3908353415988096e-06,
+      "loss": 0.1537,
+      "step": 12864
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.389765665337337e-06,
+      "loss": 0.1622,
+      "step": 12865
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.388696369853697e-06,
+      "loss": 0.1663,
+      "step": 12866
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3876274551951852e-06,
+      "loss": 0.1734,
+      "step": 12867
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3865589214090648e-06,
+      "loss": 0.1639,
+      "step": 12868
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3854907685425977e-06,
+      "loss": 0.171,
+      "step": 12869
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3844229966430166e-06,
+      "loss": 0.1498,
+      "step": 12870
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3833556057575415e-06,
+      "loss": 0.1767,
+      "step": 12871
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3822885959333843e-06,
+      "loss": 0.1808,
+      "step": 12872
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.381221967217724e-06,
+      "loss": 0.1816,
+      "step": 12873
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3801557196577354e-06,
+      "loss": 0.1564,
+      "step": 12874
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3790898533005715e-06,
+      "loss": 0.193,
+      "step": 12875
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3780243681933669e-06,
+      "loss": 0.1805,
+      "step": 12876
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.376959264383244e-06,
+      "loss": 0.179,
+      "step": 12877
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3758945419173054e-06,
+      "loss": 0.1709,
+      "step": 12878
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3748302008426362e-06,
+      "loss": 0.1508,
+      "step": 12879
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3737662412063058e-06,
+      "loss": 0.1393,
+      "step": 12880
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3727026630553674e-06,
+      "loss": 0.1643,
+      "step": 12881
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.371639466436856e-06,
+      "loss": 0.143,
+      "step": 12882
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3705766513977914e-06,
+      "loss": 0.1693,
+      "step": 12883
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3695142179851728e-06,
+      "loss": 0.1587,
+      "step": 12884
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3684521662459881e-06,
+      "loss": 0.1515,
+      "step": 12885
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3673904962272033e-06,
+      "loss": 0.1374,
+      "step": 12886
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3663292079757728e-06,
+      "loss": 0.1538,
+      "step": 12887
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3652683015386236e-06,
+      "loss": 0.1575,
+      "step": 12888
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3642077769626795e-06,
+      "loss": 0.1562,
+      "step": 12889
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3631476342948423e-06,
+      "loss": 0.1468,
+      "step": 12890
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3620878735819877e-06,
+      "loss": 0.1735,
+      "step": 12891
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3610284948709917e-06,
+      "loss": 0.1751,
+      "step": 12892
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3599694982086975e-06,
+      "loss": 0.17,
+      "step": 12893
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3589108836419385e-06,
+      "loss": 0.1615,
+      "step": 12894
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.357852651217536e-06,
+      "loss": 0.1635,
+      "step": 12895
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.356794800982284e-06,
+      "loss": 0.1504,
+      "step": 12896
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.355737332982966e-06,
+      "loss": 0.1826,
+      "step": 12897
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3546802472663478e-06,
+      "loss": 0.1833,
+      "step": 12898
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3536235438791755e-06,
+      "loss": 0.1604,
+      "step": 12899
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3525672228681875e-06,
+      "loss": 0.1521,
+      "step": 12900
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3515112842800905e-06,
+      "loss": 0.1678,
+      "step": 12901
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.350455728161586e-06,
+      "loss": 0.1646,
+      "step": 12902
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3494005545593535e-06,
+      "loss": 0.1615,
+      "step": 12903
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.348345763520058e-06,
+      "loss": 0.1528,
+      "step": 12904
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3472913550903455e-06,
+      "loss": 0.1384,
+      "step": 12905
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3462373293168463e-06,
+      "loss": 0.1748,
+      "step": 12906
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3451836862461742e-06,
+      "loss": 0.1775,
+      "step": 12907
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3441304259249243e-06,
+      "loss": 0.1569,
+      "step": 12908
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.343077548399676e-06,
+      "loss": 0.1694,
+      "step": 12909
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3420250537169953e-06,
+      "loss": 0.1506,
+      "step": 12910
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3409729419234185e-06,
+      "loss": 0.1502,
+      "step": 12911
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3399212130654827e-06,
+      "loss": 0.159,
+      "step": 12912
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3388698671896982e-06,
+      "loss": 0.1622,
+      "step": 12913
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3378189043425538e-06,
+      "loss": 0.1715,
+      "step": 12914
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3367683245705355e-06,
+      "loss": 0.1569,
+      "step": 12915
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3357181279200959e-06,
+      "loss": 0.1667,
+      "step": 12916
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3346683144376837e-06,
+      "loss": 0.1456,
+      "step": 12917
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3336188841697273e-06,
+      "loss": 0.1787,
+      "step": 12918
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3325698371626294e-06,
+      "loss": 0.1485,
+      "step": 12919
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.331521173462792e-06,
+      "loss": 0.1543,
+      "step": 12920
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3304728931165833e-06,
+      "loss": 0.1347,
+      "step": 12921
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.329424996170363e-06,
+      "loss": 0.149,
+      "step": 12922
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.328377482670481e-06,
+      "loss": 0.1582,
+      "step": 12923
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.327330352663254e-06,
+      "loss": 0.1884,
+      "step": 12924
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3262836061949935e-06,
+      "loss": 0.1859,
+      "step": 12925
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3252372433119897e-06,
+      "loss": 0.1517,
+      "step": 12926
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3241912640605182e-06,
+      "loss": 0.1652,
+      "step": 12927
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3231456684868349e-06,
+      "loss": 0.143,
+      "step": 12928
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.322100456637181e-06,
+      "loss": 0.161,
+      "step": 12929
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3210556285577792e-06,
+      "loss": 0.2042,
+      "step": 12930
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3200111842948348e-06,
+      "loss": 0.1608,
+      "step": 12931
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3189671238945401e-06,
+      "loss": 0.1651,
+      "step": 12932
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3179234474030644e-06,
+      "loss": 0.1824,
+      "step": 12933
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3168801548665656e-06,
+      "loss": 0.167,
+      "step": 12934
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3158372463311797e-06,
+      "loss": 0.1529,
+      "step": 12935
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3147947218430302e-06,
+      "loss": 0.1969,
+      "step": 12936
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.313752581448221e-06,
+      "loss": 0.176,
+      "step": 12937
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3127108251928422e-06,
+      "loss": 0.1908,
+      "step": 12938
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3116694531229556e-06,
+      "loss": 0.1875,
+      "step": 12939
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3106284652846234e-06,
+      "loss": 0.1573,
+      "step": 12940
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3095878617238812e-06,
+      "loss": 0.1585,
+      "step": 12941
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3085476424867427e-06,
+      "loss": 0.178,
+      "step": 12942
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3075078076192182e-06,
+      "loss": 0.1466,
+      "step": 12943
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3064683571672875e-06,
+      "loss": 0.1672,
+      "step": 12944
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3054292911769184e-06,
+      "loss": 0.1676,
+      "step": 12945
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3043906096940695e-06,
+      "loss": 0.1589,
+      "step": 12946
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3033523127646686e-06,
+      "loss": 0.1863,
+      "step": 12947
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3023144004346356e-06,
+      "loss": 0.1687,
+      "step": 12948
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3012768727498703e-06,
+      "loss": 0.1919,
+      "step": 12949
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3002397297562552e-06,
+      "loss": 0.1383,
+      "step": 12950
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2992029714996612e-06,
+      "loss": 0.1435,
+      "step": 12951
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.298166598025934e-06,
+      "loss": 0.1557,
+      "step": 12952
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2971306093809054e-06,
+      "loss": 0.1592,
+      "step": 12953
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2960950056103937e-06,
+      "loss": 0.1435,
+      "step": 12954
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2950597867601944e-06,
+      "loss": 0.1766,
+      "step": 12955
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2940249528760918e-06,
+      "loss": 0.193,
+      "step": 12956
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.292990504003847e-06,
+      "loss": 0.1789,
+      "step": 12957
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2919564401892105e-06,
+      "loss": 0.1666,
+      "step": 12958
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2909227614779107e-06,
+      "loss": 0.1581,
+      "step": 12959
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2898894679156614e-06,
+      "loss": 0.1473,
+      "step": 12960
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2888565595481607e-06,
+      "loss": 0.1761,
+      "step": 12961
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2878240364210814e-06,
+      "loss": 0.1824,
+      "step": 12962
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2867918985800932e-06,
+      "loss": 0.1648,
+      "step": 12963
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2857601460708392e-06,
+      "loss": 0.1997,
+      "step": 12964
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.284728778938943e-06,
+      "loss": 0.1594,
+      "step": 12965
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2836977972300224e-06,
+      "loss": 0.151,
+      "step": 12966
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2826672009896647e-06,
+      "loss": 0.1487,
+      "step": 12967
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2816369902634519e-06,
+      "loss": 0.1556,
+      "step": 12968
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2806071650969443e-06,
+      "loss": 0.1611,
+      "step": 12969
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2795777255356789e-06,
+      "loss": 0.1719,
+      "step": 12970
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2785486716251893e-06,
+      "loss": 0.178,
+      "step": 12971
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2775200034109781e-06,
+      "loss": 0.17,
+      "step": 12972
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2764917209385364e-06,
+      "loss": 0.1551,
+      "step": 12973
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2754638242533469e-06,
+      "loss": 0.1544,
+      "step": 12974
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2744363134008587e-06,
+      "loss": 0.1313,
+      "step": 12975
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2734091884265154e-06,
+      "loss": 0.1656,
+      "step": 12976
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.272382449375741e-06,
+      "loss": 0.1534,
+      "step": 12977
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2713560962939408e-06,
+      "loss": 0.1757,
+      "step": 12978
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2703301292265047e-06,
+      "loss": 0.1402,
+      "step": 12979
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.269304548218805e-06,
+      "loss": 0.1858,
+      "step": 12980
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2682793533161963e-06,
+      "loss": 0.1658,
+      "step": 12981
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2672545445640172e-06,
+      "loss": 0.191,
+      "step": 12982
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.266230122007589e-06,
+      "loss": 0.1549,
+      "step": 12983
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2652060856922156e-06,
+      "loss": 0.1529,
+      "step": 12984
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2641824356631815e-06,
+      "loss": 0.1824,
+      "step": 12985
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.26315917196576e-06,
+      "loss": 0.157,
+      "step": 12986
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.262136294645202e-06,
+      "loss": 0.1568,
+      "step": 12987
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.261113803746743e-06,
+      "loss": 0.1655,
+      "step": 12988
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.260091699315601e-06,
+      "loss": 0.1742,
+      "step": 12989
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2590699813969798e-06,
+      "loss": 0.1338,
+      "step": 12990
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.258048650036061e-06,
+      "loss": 0.171,
+      "step": 12991
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2570277052780144e-06,
+      "loss": 0.164,
+      "step": 12992
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2560071471679858e-06,
+      "loss": 0.1558,
+      "step": 12993
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2549869757511123e-06,
+      "loss": 0.1648,
+      "step": 12994
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2539671910725103e-06,
+      "loss": 0.1511,
+      "step": 12995
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2529477931772738e-06,
+      "loss": 0.1586,
+      "step": 12996
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2519287821104908e-06,
+      "loss": 0.1593,
+      "step": 12997
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2509101579172211e-06,
+      "loss": 0.1477,
+      "step": 12998
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.249891920642511e-06,
+      "loss": 0.1693,
+      "step": 12999
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2488740703313984e-06,
+      "loss": 0.144,
+      "step": 13000
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2478566070288878e-06,
+      "loss": 0.1596,
+      "step": 13001
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.246839530779983e-06,
+      "loss": 0.1463,
+      "step": 13002
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.245822841629658e-06,
+      "loss": 0.1641,
+      "step": 13003
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2448065396228738e-06,
+      "loss": 0.1642,
+      "step": 13004
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2437906248045828e-06,
+      "loss": 0.16,
+      "step": 13005
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.242775097219704e-06,
+      "loss": 0.1771,
+      "step": 13006
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2417599569131523e-06,
+      "loss": 0.189,
+      "step": 13007
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2407452039298196e-06,
+      "loss": 0.1623,
+      "step": 13008
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2397308383145823e-06,
+      "loss": 0.1926,
+      "step": 13009
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2387168601123012e-06,
+      "loss": 0.1616,
+      "step": 13010
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.237703269367817e-06,
+      "loss": 0.1619,
+      "step": 13011
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.236690066125954e-06,
+      "loss": 0.1378,
+      "step": 13012
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2356772504315217e-06,
+      "loss": 0.1579,
+      "step": 13013
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.23466482232931e-06,
+      "loss": 0.167,
+      "step": 13014
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.233652781864092e-06,
+      "loss": 0.1684,
+      "step": 13015
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2326411290806238e-06,
+      "loss": 0.1919,
+      "step": 13016
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2316298640236457e-06,
+      "loss": 0.1744,
+      "step": 13017
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2306189867378781e-06,
+      "loss": 0.1482,
+      "step": 13018
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2296084972680278e-06,
+      "loss": 0.1773,
+      "step": 13019
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2285983956587832e-06,
+      "loss": 0.1598,
+      "step": 13020
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2275886819548099e-06,
+      "loss": 0.1835,
+      "step": 13021
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2265793562007656e-06,
+      "loss": 0.1621,
+      "step": 13022
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2255704184412887e-06,
+      "loss": 0.1881,
+      "step": 13023
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2245618687209915e-06,
+      "loss": 0.167,
+      "step": 13024
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2235537070844838e-06,
+      "loss": 0.1854,
+      "step": 13025
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2225459335763445e-06,
+      "loss": 0.1866,
+      "step": 13026
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.22153854824114e-06,
+      "loss": 0.1798,
+      "step": 13027
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2205315511234294e-06,
+      "loss": 0.1596,
+      "step": 13028
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.219524942267738e-06,
+      "loss": 0.173,
+      "step": 13029
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2185187217185845e-06,
+      "loss": 0.1508,
+      "step": 13030
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.217512889520468e-06,
+      "loss": 0.1771,
+      "step": 13031
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2165074457178682e-06,
+      "loss": 0.1607,
+      "step": 13032
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2155023903552554e-06,
+      "loss": 0.1348,
+      "step": 13033
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2144977234770717e-06,
+      "loss": 0.1362,
+      "step": 13034
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.213493445127749e-06,
+      "loss": 0.1737,
+      "step": 13035
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2124895553517003e-06,
+      "loss": 0.1822,
+      "step": 13036
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2114860541933204e-06,
+      "loss": 0.1623,
+      "step": 13037
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.21048294169699e-06,
+      "loss": 0.166,
+      "step": 13038
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2094802179070697e-06,
+      "loss": 0.1691,
+      "step": 13039
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2084778828679044e-06,
+      "loss": 0.1458,
+      "step": 13040
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2074759366238197e-06,
+      "loss": 0.1851,
+      "step": 13041
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.206474379219127e-06,
+      "loss": 0.1608,
+      "step": 13042
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2054732106981205e-06,
+      "loss": 0.1682,
+      "step": 13043
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2044724311050693e-06,
+      "loss": 0.1719,
+      "step": 13044
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2034720404842392e-06,
+      "loss": 0.1356,
+      "step": 13045
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2024720388798694e-06,
+      "loss": 0.1721,
+      "step": 13046
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.20147242633618e-06,
+      "loss": 0.1574,
+      "step": 13047
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.200473202897383e-06,
+      "loss": 0.1737,
+      "step": 13048
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1994743686076626e-06,
+      "loss": 0.16,
+      "step": 13049
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1984759235111955e-06,
+      "loss": 0.1449,
+      "step": 13050
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.197477867652137e-06,
+      "loss": 0.1915,
+      "step": 13051
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1964802010746202e-06,
+      "loss": 0.1798,
+      "step": 13052
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1954829238227716e-06,
+      "loss": 0.1669,
+      "step": 13053
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1944860359406907e-06,
+      "loss": 0.1635,
+      "step": 13054
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.193489537472463e-06,
+      "loss": 0.1472,
+      "step": 13055
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1924934284621625e-06,
+      "loss": 0.1746,
+      "step": 13056
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1914977089538371e-06,
+      "loss": 0.1477,
+      "step": 13057
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1905023789915215e-06,
+      "loss": 0.1934,
+      "step": 13058
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1895074386192341e-06,
+      "loss": 0.1672,
+      "step": 13059
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1885128878809738e-06,
+      "loss": 0.1518,
+      "step": 13060
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1875187268207245e-06,
+      "loss": 0.1723,
+      "step": 13061
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.186524955482452e-06,
+      "loss": 0.1379,
+      "step": 13062
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1855315739101047e-06,
+      "loss": 0.1704,
+      "step": 13063
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1845385821476129e-06,
+      "loss": 0.1932,
+      "step": 13064
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1835459802388905e-06,
+      "loss": 0.1777,
+      "step": 13065
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1825537682278355e-06,
+      "loss": 0.169,
+      "step": 13066
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1815619461583273e-06,
+      "loss": 0.1561,
+      "step": 13067
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1805705140742263e-06,
+      "loss": 0.2063,
+      "step": 13068
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1795794720193788e-06,
+      "loss": 0.153,
+      "step": 13069
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1785888200376116e-06,
+      "loss": 0.1851,
+      "step": 13070
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1775985581727388e-06,
+      "loss": 0.1438,
+      "step": 13071
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1766086864685467e-06,
+      "loss": 0.1591,
+      "step": 13072
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1756192049688164e-06,
+      "loss": 0.1641,
+      "step": 13073
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1746301137173088e-06,
+      "loss": 0.181,
+      "step": 13074
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1736414127577567e-06,
+      "loss": 0.1537,
+      "step": 13075
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1726531021338938e-06,
+      "loss": 0.1581,
+      "step": 13076
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.171665181889421e-06,
+      "loss": 0.1804,
+      "step": 13077
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1706776520680264e-06,
+      "loss": 0.1641,
+      "step": 13078
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1696905127133907e-06,
+      "loss": 0.1608,
+      "step": 13079
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1687037638691622e-06,
+      "loss": 0.1746,
+      "step": 13080
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1677174055789798e-06,
+      "loss": 0.141,
+      "step": 13081
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1667314378864657e-06,
+      "loss": 0.1741,
+      "step": 13082
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.165745860835219e-06,
+      "loss": 0.1585,
+      "step": 13083
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1647606744688346e-06,
+      "loss": 0.1652,
+      "step": 13084
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1637758788308728e-06,
+      "loss": 0.1501,
+      "step": 13085
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1627914739648872e-06,
+      "loss": 0.1559,
+      "step": 13086
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1618074599144137e-06,
+      "loss": 0.1695,
+      "step": 13087
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1608238367229685e-06,
+      "loss": 0.173,
+      "step": 13088
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1598406044340504e-06,
+      "loss": 0.1508,
+      "step": 13089
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1588577630911425e-06,
+      "loss": 0.1642,
+      "step": 13090
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1578753127377085e-06,
+      "loss": 0.174,
+      "step": 13091
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1568932534171983e-06,
+      "loss": 0.1577,
+      "step": 13092
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1559115851730419e-06,
+      "loss": 0.164,
+      "step": 13093
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.154930308048653e-06,
+      "loss": 0.18,
+      "step": 13094
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1539494220874225e-06,
+      "loss": 0.1753,
+      "step": 13095
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.152968927332735e-06,
+      "loss": 0.1801,
+      "step": 13096
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.151988823827951e-06,
+      "loss": 0.1667,
+      "step": 13097
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.151009111616409e-06,
+      "loss": 0.1809,
+      "step": 13098
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1500297907414449e-06,
+      "loss": 0.1467,
+      "step": 13099
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.149050861246358e-06,
+      "loss": 0.1904,
+      "step": 13100
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1480723231744472e-06,
+      "loss": 0.1664,
+      "step": 13101
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.147094176568987e-06,
+      "loss": 0.1661,
+      "step": 13102
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1461164214732302e-06,
+      "loss": 0.1425,
+      "step": 13103
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1451390579304222e-06,
+      "loss": 0.138,
+      "step": 13104
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1441620859837832e-06,
+      "loss": 0.1439,
+      "step": 13105
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.143185505676515e-06,
+      "loss": 0.1671,
+      "step": 13106
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1422093170518146e-06,
+      "loss": 0.1474,
+      "step": 13107
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1412335201528456e-06,
+      "loss": 0.1742,
+      "step": 13108
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1402581150227643e-06,
+      "loss": 0.165,
+      "step": 13109
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1392831017047056e-06,
+      "loss": 0.1615,
+      "step": 13110
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1383084802417898e-06,
+      "loss": 0.1544,
+      "step": 13111
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1373342506771167e-06,
+      "loss": 0.1491,
+      "step": 13112
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1363604130537719e-06,
+      "loss": 0.1653,
+      "step": 13113
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1353869674148221e-06,
+      "loss": 0.2078,
+      "step": 13114
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1344139138033172e-06,
+      "loss": 0.1581,
+      "step": 13115
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1334412522622873e-06,
+      "loss": 0.1738,
+      "step": 13116
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1324689828347502e-06,
+      "loss": 0.1487,
+      "step": 13117
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1314971055637013e-06,
+      "loss": 0.1606,
+      "step": 13118
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.130525620492121e-06,
+      "loss": 0.1478,
+      "step": 13119
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1295545276629728e-06,
+      "loss": 0.1586,
+      "step": 13120
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.128583827119203e-06,
+      "loss": 0.1739,
+      "step": 13121
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1276135189037373e-06,
+      "loss": 0.1448,
+      "step": 13122
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1266436030594884e-06,
+      "loss": 0.1918,
+      "step": 13123
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1256740796293486e-06,
+      "loss": 0.1839,
+      "step": 13124
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1247049486561968e-06,
+      "loss": 0.1824,
+      "step": 13125
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1237362101828863e-06,
+      "loss": 0.1356,
+      "step": 13126
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.122767864252262e-06,
+      "loss": 0.1797,
+      "step": 13127
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.121799910907151e-06,
+      "loss": 0.1517,
+      "step": 13128
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1208323501903518e-06,
+      "loss": 0.1902,
+      "step": 13129
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1198651821446626e-06,
+      "loss": 0.1467,
+      "step": 13130
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1188984068128483e-06,
+      "loss": 0.1912,
+      "step": 13131
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1179320242376646e-06,
+      "loss": 0.1714,
+      "step": 13132
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.116966034461855e-06,
+      "loss": 0.172,
+      "step": 13133
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1160004375281297e-06,
+      "loss": 0.1514,
+      "step": 13134
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1150352334792003e-06,
+      "loss": 0.195,
+      "step": 13135
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1140704223577447e-06,
+      "loss": 0.1698,
+      "step": 13136
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1131060042064323e-06,
+      "loss": 0.1539,
+      "step": 13137
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1121419790679177e-06,
+      "loss": 0.1348,
+      "step": 13138
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1111783469848292e-06,
+      "loss": 0.1732,
+      "step": 13139
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1102151079997835e-06,
+      "loss": 0.1723,
+      "step": 13140
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1092522621553793e-06,
+      "loss": 0.1675,
+      "step": 13141
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1082898094941974e-06,
+      "loss": 0.1732,
+      "step": 13142
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1073277500587998e-06,
+      "loss": 0.154,
+      "step": 13143
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1063660838917345e-06,
+      "loss": 0.1577,
+      "step": 13144
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1054048110355286e-06,
+      "loss": 0.1816,
+      "step": 13145
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1044439315326937e-06,
+      "loss": 0.1534,
+      "step": 13146
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1034834454257237e-06,
+      "loss": 0.1634,
+      "step": 13147
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1025233527570955e-06,
+      "loss": 0.1833,
+      "step": 13148
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1015636535692675e-06,
+      "loss": 0.1465,
+      "step": 13149
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1006043479046813e-06,
+      "loss": 0.138,
+      "step": 13150
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0996454358057606e-06,
+      "loss": 0.2065,
+      "step": 13151
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.098686917314914e-06,
+      "loss": 0.1835,
+      "step": 13152
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0977287924745317e-06,
+      "loss": 0.1507,
+      "step": 13153
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0967710613269788e-06,
+      "loss": 0.1832,
+      "step": 13154
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0958137239146171e-06,
+      "loss": 0.162,
+      "step": 13155
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0948567802797839e-06,
+      "loss": 0.1612,
+      "step": 13156
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0939002304647918e-06,
+      "loss": 0.1547,
+      "step": 13157
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0929440745119524e-06,
+      "loss": 0.1848,
+      "step": 13158
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0919883124635443e-06,
+      "loss": 0.1514,
+      "step": 13159
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.091032944361834e-06,
+      "loss": 0.1495,
+      "step": 13160
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0900779702490782e-06,
+      "loss": 0.1625,
+      "step": 13161
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0891233901675046e-06,
+      "loss": 0.1528,
+      "step": 13162
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0881692041593283e-06,
+      "loss": 0.1773,
+      "step": 13163
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0872154122667488e-06,
+      "loss": 0.172,
+      "step": 13164
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0862620145319447e-06,
+      "loss": 0.192,
+      "step": 13165
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0853090109970832e-06,
+      "loss": 0.1504,
+      "step": 13166
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0843564017043052e-06,
+      "loss": 0.1914,
+      "step": 13167
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.083404186695739e-06,
+      "loss": 0.1728,
+      "step": 13168
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0824523660134977e-06,
+      "loss": 0.1405,
+      "step": 13169
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0815009396996734e-06,
+      "loss": 0.1667,
+      "step": 13170
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0805499077963422e-06,
+      "loss": 0.1633,
+      "step": 13171
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0795992703455605e-06,
+      "loss": 0.1603,
+      "step": 13172
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0786490273893724e-06,
+      "loss": 0.1721,
+      "step": 13173
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0776991789697977e-06,
+      "loss": 0.1747,
+      "step": 13174
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.076749725128846e-06,
+      "loss": 0.1681,
+      "step": 13175
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0758006659085052e-06,
+      "loss": 0.1573,
+      "step": 13176
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0748520013507412e-06,
+      "loss": 0.1743,
+      "step": 13177
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0739037314975131e-06,
+      "loss": 0.1467,
+      "step": 13178
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0729558563907582e-06,
+      "loss": 0.1714,
+      "step": 13179
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0720083760723887e-06,
+      "loss": 0.1694,
+      "step": 13180
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0710612905843142e-06,
+      "loss": 0.1903,
+      "step": 13181
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0701145999684092e-06,
+      "loss": 0.1603,
+      "step": 13182
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0691683042665467e-06,
+      "loss": 0.1857,
+      "step": 13183
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0682224035205757e-06,
+      "loss": 0.1589,
+      "step": 13184
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0672768977723224e-06,
+      "loss": 0.1693,
+      "step": 13185
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.066331787063607e-06,
+      "loss": 0.1849,
+      "step": 13186
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0653870714362213e-06,
+      "loss": 0.1827,
+      "step": 13187
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0644427509319433e-06,
+      "loss": 0.1757,
+      "step": 13188
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0634988255925416e-06,
+      "loss": 0.1715,
+      "step": 13189
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.062555295459754e-06,
+      "loss": 0.1742,
+      "step": 13190
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.061612160575307e-06,
+      "loss": 0.1734,
+      "step": 13191
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0606694209809132e-06,
+      "loss": 0.1706,
+      "step": 13192
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.059727076718261e-06,
+      "loss": 0.1828,
+      "step": 13193
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0587851278290272e-06,
+      "loss": 0.1641,
+      "step": 13194
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0578435743548655e-06,
+      "loss": 0.1739,
+      "step": 13195
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0569024163374176e-06,
+      "loss": 0.1579,
+      "step": 13196
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0559616538183038e-06,
+      "loss": 0.1532,
+      "step": 13197
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0550212868391284e-06,
+      "loss": 0.1817,
+      "step": 13198
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0540813154414785e-06,
+      "loss": 0.1552,
+      "step": 13199
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.053141739666923e-06,
+      "loss": 0.1768,
+      "step": 13200
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.052202559557014e-06,
+      "loss": 0.1815,
+      "step": 13201
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.051263775153285e-06,
+      "loss": 0.1461,
+      "step": 13202
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.050325386497253e-06,
+      "loss": 0.165,
+      "step": 13203
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0493873936304189e-06,
+      "loss": 0.1641,
+      "step": 13204
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0484497965942586e-06,
+      "loss": 0.1555,
+      "step": 13205
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0475125954302423e-06,
+      "loss": 0.1787,
+      "step": 13206
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0465757901798168e-06,
+      "loss": 0.1581,
+      "step": 13207
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0456393808844046e-06,
+      "loss": 0.1715,
+      "step": 13208
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0447033675854257e-06,
+      "loss": 0.1723,
+      "step": 13209
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0437677503242694e-06,
+      "loss": 0.1553,
+      "step": 13210
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0428325291423102e-06,
+      "loss": 0.1591,
+      "step": 13211
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0418977040809142e-06,
+      "loss": 0.1341,
+      "step": 13212
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.040963275181417e-06,
+      "loss": 0.15,
+      "step": 13213
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0400292424851455e-06,
+      "loss": 0.1582,
+      "step": 13214
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0390956060334045e-06,
+      "loss": 0.1684,
+      "step": 13215
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0381623658674823e-06,
+      "loss": 0.1434,
+      "step": 13216
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0372295220286565e-06,
+      "loss": 0.147,
+      "step": 13217
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0362970745581734e-06,
+      "loss": 0.1535,
+      "step": 13218
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0353650234972735e-06,
+      "loss": 0.1582,
+      "step": 13219
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0344333688871744e-06,
+      "loss": 0.1643,
+      "step": 13220
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.033502110769078e-06,
+      "loss": 0.1856,
+      "step": 13221
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0325712491841677e-06,
+      "loss": 0.1762,
+      "step": 13222
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0316407841736097e-06,
+      "loss": 0.1616,
+      "step": 13223
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0307107157785545e-06,
+      "loss": 0.1464,
+      "step": 13224
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0297810440401302e-06,
+      "loss": 0.1562,
+      "step": 13225
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0288517689994537e-06,
+      "loss": 0.1553,
+      "step": 13226
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0279228906976214e-06,
+      "loss": 0.1697,
+      "step": 13227
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0269944091757056e-06,
+      "loss": 0.1776,
+      "step": 13228
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0260663244747736e-06,
+      "loss": 0.1878,
+      "step": 13229
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0251386366358694e-06,
+      "loss": 0.1637,
+      "step": 13230
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0242113457000135e-06,
+      "loss": 0.1746,
+      "step": 13231
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0232844517082208e-06,
+      "loss": 0.1651,
+      "step": 13232
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.022357954701475e-06,
+      "loss": 0.1562,
+      "step": 13233
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0214318547207558e-06,
+      "loss": 0.2101,
+      "step": 13234
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.020506151807018e-06,
+      "loss": 0.2013,
+      "step": 13235
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0195808460011936e-06,
+      "loss": 0.1803,
+      "step": 13236
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0186559373442118e-06,
+      "loss": 0.1638,
+      "step": 13237
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.01773142587697e-06,
+      "loss": 0.1478,
+      "step": 13238
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0168073116403532e-06,
+      "loss": 0.1546,
+      "step": 13239
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0158835946752355e-06,
+      "loss": 0.1904,
+      "step": 13240
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0149602750224608e-06,
+      "loss": 0.1709,
+      "step": 13241
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0140373527228631e-06,
+      "loss": 0.1555,
+      "step": 13242
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0131148278172597e-06,
+      "loss": 0.1671,
+      "step": 13243
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.012192700346447e-06,
+      "loss": 0.1749,
+      "step": 13244
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0112709703512047e-06,
+      "loss": 0.1518,
+      "step": 13245
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0103496378722966e-06,
+      "loss": 0.154,
+      "step": 13246
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0094287029504658e-06,
+      "loss": 0.1713,
+      "step": 13247
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0085081656264406e-06,
+      "loss": 0.1541,
+      "step": 13248
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.007588025940931e-06,
+      "loss": 0.1618,
+      "step": 13249
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0066682839346287e-06,
+      "loss": 0.1676,
+      "step": 13250
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.005748939648209e-06,
+      "loss": 0.1628,
+      "step": 13251
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0048299931223272e-06,
+      "loss": 0.188,
+      "step": 13252
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.003911444397624e-06,
+      "loss": 0.1788,
+      "step": 13253
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0029932935147202e-06,
+      "loss": 0.1503,
+      "step": 13254
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0020755405142203e-06,
+      "loss": 0.1512,
+      "step": 13255
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0011581854367125e-06,
+      "loss": 0.1602,
+      "step": 13256
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0002412283227636e-06,
+      "loss": 0.144,
+      "step": 13257
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.993246692129277e-07,
+      "loss": 0.1546,
+      "step": 13258
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.984085081477335e-07,
+      "loss": 0.1671,
+      "step": 13259
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.97492745167702e-07,
+      "loss": 0.1708,
+      "step": 13260
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.965773803133317e-07,
+      "loss": 0.1541,
+      "step": 13261
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.95662413625098e-07,
+      "loss": 0.1413,
+      "step": 13262
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.947478451434734e-07,
+      "loss": 0.1668,
+      "step": 13263
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.93833674908895e-07,
+      "loss": 0.1642,
+      "step": 13264
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.92919902961794e-07,
+      "loss": 0.1583,
+      "step": 13265
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.92006529342585e-07,
+      "loss": 0.1653,
+      "step": 13266
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.910935540916532e-07,
+      "loss": 0.1906,
+      "step": 13267
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.90180977249382e-07,
+      "loss": 0.1693,
+      "step": 13268
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.892687988561223e-07,
+      "loss": 0.1654,
+      "step": 13269
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.883570189522163e-07,
+      "loss": 0.1784,
+      "step": 13270
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.874456375779906e-07,
+      "loss": 0.1553,
+      "step": 13271
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.865346547737442e-07,
+      "loss": 0.1574,
+      "step": 13272
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.856240705797672e-07,
+      "loss": 0.1849,
+      "step": 13273
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.847138850363281e-07,
+      "loss": 0.165,
+      "step": 13274
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.838040981836794e-07,
+      "loss": 0.2167,
+      "step": 13275
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.828947100620545e-07,
+      "loss": 0.1635,
+      "step": 13276
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.819857207116724e-07,
+      "loss": 0.1356,
+      "step": 13277
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.810771301727296e-07,
+      "loss": 0.1702,
+      "step": 13278
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.801689384854074e-07,
+      "loss": 0.1432,
+      "step": 13279
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.792611456898726e-07,
+      "loss": 0.1548,
+      "step": 13280
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.783537518262676e-07,
+      "loss": 0.1572,
+      "step": 13281
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.774467569347235e-07,
+      "loss": 0.1582,
+      "step": 13282
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.765401610553494e-07,
+      "loss": 0.1526,
+      "step": 13283
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.756339642282398e-07,
+      "loss": 0.1724,
+      "step": 13284
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.747281664934694e-07,
+      "loss": 0.1744,
+      "step": 13285
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.738227678910984e-07,
+      "loss": 0.1551,
+      "step": 13286
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.729177684611613e-07,
+      "loss": 0.1647,
+      "step": 13287
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.720131682436851e-07,
+      "loss": 0.1702,
+      "step": 13288
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.711089672786767e-07,
+      "loss": 0.1706,
+      "step": 13289
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.702051656061173e-07,
+      "loss": 0.1724,
+      "step": 13290
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.693017632659829e-07,
+      "loss": 0.1585,
+      "step": 13291
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.683987602982203e-07,
+      "loss": 0.1675,
+      "step": 13292
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.674961567427643e-07,
+      "loss": 0.1653,
+      "step": 13293
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.665939526395362e-07,
+      "loss": 0.1818,
+      "step": 13294
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.65692148028431e-07,
+      "loss": 0.1627,
+      "step": 13295
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.647907429493297e-07,
+      "loss": 0.1518,
+      "step": 13296
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.638897374420976e-07,
+      "loss": 0.1713,
+      "step": 13297
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.629891315465779e-07,
+      "loss": 0.177,
+      "step": 13298
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.620889253026055e-07,
+      "loss": 0.1451,
+      "step": 13299
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.611891187499834e-07,
+      "loss": 0.1684,
+      "step": 13300
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.602897119285082e-07,
+      "loss": 0.1565,
+      "step": 13301
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.59390704877955e-07,
+      "loss": 0.1651,
+      "step": 13302
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.584920976380808e-07,
+      "loss": 0.1746,
+      "step": 13303
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.575938902486247e-07,
+      "loss": 0.1327,
+      "step": 13304
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.56696082749311e-07,
+      "loss": 0.1755,
+      "step": 13305
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.55798675179842e-07,
+      "loss": 0.1567,
+      "step": 13306
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.54901667579906e-07,
+      "loss": 0.165,
+      "step": 13307
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.540050599891726e-07,
+      "loss": 0.1383,
+      "step": 13308
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.531088524472932e-07,
+      "loss": 0.1744,
+      "step": 13309
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.522130449938971e-07,
+      "loss": 0.1857,
+      "step": 13310
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.513176376686062e-07,
+      "loss": 0.1519,
+      "step": 13311
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.504226305110187e-07,
+      "loss": 0.1758,
+      "step": 13312
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.495280235607085e-07,
+      "loss": 0.1537,
+      "step": 13313
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.486338168572484e-07,
+      "loss": 0.1941,
+      "step": 13314
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.477400104401735e-07,
+      "loss": 0.1841,
+      "step": 13315
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.468466043490187e-07,
+      "loss": 0.1735,
+      "step": 13316
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.459535986232949e-07,
+      "loss": 0.1728,
+      "step": 13317
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.450609933024857e-07,
+      "loss": 0.1726,
+      "step": 13318
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.441687884260753e-07,
+      "loss": 0.1635,
+      "step": 13319
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.432769840335132e-07,
+      "loss": 0.1774,
+      "step": 13320
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.423855801642401e-07,
+      "loss": 0.1755,
+      "step": 13321
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.414945768576822e-07,
+      "loss": 0.1475,
+      "step": 13322
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.40603974153238e-07,
+      "loss": 0.1871,
+      "step": 13323
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.397137720902938e-07,
+      "loss": 0.1908,
+      "step": 13324
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.388239707082192e-07,
+      "loss": 0.1805,
+      "step": 13325
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.379345700463649e-07,
+      "loss": 0.1523,
+      "step": 13326
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.370455701440617e-07,
+      "loss": 0.1737,
+      "step": 13327
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.361569710406271e-07,
+      "loss": 0.1541,
+      "step": 13328
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.352687727753563e-07,
+      "loss": 0.162,
+      "step": 13329
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.343809753875299e-07,
+      "loss": 0.1799,
+      "step": 13330
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.33493578916409e-07,
+      "loss": 0.1527,
+      "step": 13331
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.326065834012388e-07,
+      "loss": 0.1934,
+      "step": 13332
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.317199888812445e-07,
+      "loss": 0.1696,
+      "step": 13333
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.308337953956359e-07,
+      "loss": 0.1622,
+      "step": 13334
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.299480029836039e-07,
+      "loss": 0.1767,
+      "step": 13335
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.290626116843204e-07,
+      "loss": 0.174,
+      "step": 13336
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.281776215369432e-07,
+      "loss": 0.1401,
+      "step": 13337
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.27293032580604e-07,
+      "loss": 0.1507,
+      "step": 13338
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.264088448544307e-07,
+      "loss": 0.1584,
+      "step": 13339
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.255250583975217e-07,
+      "loss": 0.1708,
+      "step": 13340
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.246416732489594e-07,
+      "loss": 0.1722,
+      "step": 13341
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.237586894478156e-07,
+      "loss": 0.1637,
+      "step": 13342
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.228761070331349e-07,
+      "loss": 0.1661,
+      "step": 13343
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.219939260439481e-07,
+      "loss": 0.1571,
+      "step": 13344
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.21112146519274e-07,
+      "loss": 0.1883,
+      "step": 13345
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.202307684981027e-07,
+      "loss": 0.1997,
+      "step": 13346
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.19349792019415e-07,
+      "loss": 0.1374,
+      "step": 13347
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.184692171221699e-07,
+      "loss": 0.1527,
+      "step": 13348
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.175890438453084e-07,
+      "loss": 0.1671,
+      "step": 13349
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.167092722277604e-07,
+      "loss": 0.1559,
+      "step": 13350
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.158299023084282e-07,
+      "loss": 0.1736,
+      "step": 13351
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.14950934126202e-07,
+      "loss": 0.1541,
+      "step": 13352
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.140723677199536e-07,
+      "loss": 0.1534,
+      "step": 13353
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.131942031285357e-07,
+      "loss": 0.1691,
+      "step": 13354
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.123164403907847e-07,
+      "loss": 0.1775,
+      "step": 13355
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.114390795455197e-07,
+      "loss": 0.1554,
+      "step": 13356
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.105621206315395e-07,
+      "loss": 0.1611,
+      "step": 13357
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.096855636876267e-07,
+      "loss": 0.133,
+      "step": 13358
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.088094087525468e-07,
+      "loss": 0.1793,
+      "step": 13359
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.079336558650476e-07,
+      "loss": 0.1734,
+      "step": 13360
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.070583050638549e-07,
+      "loss": 0.1849,
+      "step": 13361
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.061833563876831e-07,
+      "loss": 0.1497,
+      "step": 13362
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.053088098752272e-07,
+      "loss": 0.1543,
+      "step": 13363
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.04434665565157e-07,
+      "loss": 0.1645,
+      "step": 13364
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.035609234961373e-07,
+      "loss": 0.1639,
+      "step": 13365
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.026875837068027e-07,
+      "loss": 0.158,
+      "step": 13366
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.018146462357802e-07,
+      "loss": 0.148,
+      "step": 13367
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.009421111216743e-07,
+      "loss": 0.1493,
+      "step": 13368
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.000699784030665e-07,
+      "loss": 0.1793,
+      "step": 13369
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 8.991982481185324e-07,
+      "loss": 0.1692,
+      "step": 13370
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 8.983269203066203e-07,
+      "loss": 0.138,
+      "step": 13371
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 8.974559950058615e-07,
+      "loss": 0.1611,
+      "step": 13372
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 8.965854722547773e-07,
+      "loss": 0.1686,
+      "step": 13373
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 8.957153520918605e-07,
+      "loss": 0.1511,
+      "step": 13374
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.948456345555934e-07,
+      "loss": 0.1814,
+      "step": 13375
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.939763196844376e-07,
+      "loss": 0.1491,
+      "step": 13376
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.931074075168377e-07,
+      "loss": 0.157,
+      "step": 13377
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.9223889809122e-07,
+      "loss": 0.1725,
+      "step": 13378
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.913707914459946e-07,
+      "loss": 0.1524,
+      "step": 13379
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.905030876195519e-07,
+      "loss": 0.1688,
+      "step": 13380
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.896357866502647e-07,
+      "loss": 0.1759,
+      "step": 13381
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.887688885764889e-07,
+      "loss": 0.1624,
+      "step": 13382
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.879023934365616e-07,
+      "loss": 0.1852,
+      "step": 13383
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.870363012688021e-07,
+      "loss": 0.1385,
+      "step": 13384
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.861706121115143e-07,
+      "loss": 0.1527,
+      "step": 13385
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.853053260029809e-07,
+      "loss": 0.1593,
+      "step": 13386
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.844404429814679e-07,
+      "loss": 0.1886,
+      "step": 13387
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.835759630852248e-07,
+      "loss": 0.168,
+      "step": 13388
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.827118863524809e-07,
+      "loss": 0.1764,
+      "step": 13389
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.818482128214501e-07,
+      "loss": 0.1479,
+      "step": 13390
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.809849425303296e-07,
+      "loss": 0.1746,
+      "step": 13391
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.801220755172901e-07,
+      "loss": 0.1524,
+      "step": 13392
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.792596118204966e-07,
+      "loss": 0.1554,
+      "step": 13393
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.783975514780896e-07,
+      "loss": 0.1742,
+      "step": 13394
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.775358945281898e-07,
+      "loss": 0.1737,
+      "step": 13395
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.766746410089077e-07,
+      "loss": 0.1682,
+      "step": 13396
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.758137909583276e-07,
+      "loss": 0.1512,
+      "step": 13397
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.749533444145186e-07,
+      "loss": 0.1687,
+      "step": 13398
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.740933014155395e-07,
+      "loss": 0.1621,
+      "step": 13399
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.732336619994164e-07,
+      "loss": 0.1968,
+      "step": 13400
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.723744262041734e-07,
+      "loss": 0.1625,
+      "step": 13401
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.715155940678033e-07,
+      "loss": 0.1559,
+      "step": 13402
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.706571656282881e-07,
+      "loss": 0.157,
+      "step": 13403
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.697991409235951e-07,
+      "loss": 0.1455,
+      "step": 13404
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.689415199916651e-07,
+      "loss": 0.1493,
+      "step": 13405
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.680843028704256e-07,
+      "loss": 0.16,
+      "step": 13406
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.672274895977883e-07,
+      "loss": 0.1603,
+      "step": 13407
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.663710802116431e-07,
+      "loss": 0.1712,
+      "step": 13408
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.655150747498653e-07,
+      "loss": 0.1797,
+      "step": 13409
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.646594732503088e-07,
+      "loss": 0.1766,
+      "step": 13410
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.638042757508125e-07,
+      "loss": 0.1789,
+      "step": 13411
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.62949482289197e-07,
+      "loss": 0.1508,
+      "step": 13412
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.620950929032646e-07,
+      "loss": 0.1673,
+      "step": 13413
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.612411076308003e-07,
+      "loss": 0.1844,
+      "step": 13414
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.603875265095674e-07,
+      "loss": 0.1662,
+      "step": 13415
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.595343495773168e-07,
+      "loss": 0.1778,
+      "step": 13416
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.586815768717804e-07,
+      "loss": 0.1731,
+      "step": 13417
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.578292084306694e-07,
+      "loss": 0.1548,
+      "step": 13418
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.569772442916812e-07,
+      "loss": 0.1732,
+      "step": 13419
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.561256844924881e-07,
+      "loss": 0.1521,
+      "step": 13420
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.552745290707531e-07,
+      "loss": 0.177,
+      "step": 13421
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.544237780641195e-07,
+      "loss": 0.1637,
+      "step": 13422
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.535734315102052e-07,
+      "loss": 0.1705,
+      "step": 13423
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.527234894466208e-07,
+      "loss": 0.1899,
+      "step": 13424
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.518739519109509e-07,
+      "loss": 0.1767,
+      "step": 13425
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.510248189407655e-07,
+      "loss": 0.1969,
+      "step": 13426
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.5017609057362e-07,
+      "loss": 0.1707,
+      "step": 13427
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.493277668470446e-07,
+      "loss": 0.1654,
+      "step": 13428
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.484798477985568e-07,
+      "loss": 0.1566,
+      "step": 13429
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.476323334656545e-07,
+      "loss": 0.1712,
+      "step": 13430
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.467852238858165e-07,
+      "loss": 0.1453,
+      "step": 13431
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.459385190965108e-07,
+      "loss": 0.1904,
+      "step": 13432
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.450922191351763e-07,
+      "loss": 0.186,
+      "step": 13433
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.442463240392429e-07,
+      "loss": 0.162,
+      "step": 13434
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.434008338461175e-07,
+      "loss": 0.1932,
+      "step": 13435
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.425557485931912e-07,
+      "loss": 0.1353,
+      "step": 13436
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.417110683178375e-07,
+      "loss": 0.1467,
+      "step": 13437
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.40866793057411e-07,
+      "loss": 0.1471,
+      "step": 13438
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.400229228492485e-07,
+      "loss": 0.1774,
+      "step": 13439
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.391794577306711e-07,
+      "loss": 0.186,
+      "step": 13440
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.38336397738978e-07,
+      "loss": 0.1439,
+      "step": 13441
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.374937429114549e-07,
+      "loss": 0.1591,
+      "step": 13442
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.36651493285362e-07,
+      "loss": 0.1838,
+      "step": 13443
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.358096488979528e-07,
+      "loss": 0.1646,
+      "step": 13444
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.349682097864565e-07,
+      "loss": 0.1628,
+      "step": 13445
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.341271759880786e-07,
+      "loss": 0.1824,
+      "step": 13446
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.33286547540022e-07,
+      "loss": 0.1615,
+      "step": 13447
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.324463244794523e-07,
+      "loss": 0.1588,
+      "step": 13448
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.316065068435364e-07,
+      "loss": 0.1648,
+      "step": 13449
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.307670946694113e-07,
+      "loss": 0.1656,
+      "step": 13450
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.299280879941951e-07,
+      "loss": 0.1814,
+      "step": 13451
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.290894868549992e-07,
+      "loss": 0.1728,
+      "step": 13452
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.282512912889029e-07,
+      "loss": 0.1858,
+      "step": 13453
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.274135013329765e-07,
+      "loss": 0.1868,
+      "step": 13454
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.265761170242747e-07,
+      "loss": 0.1832,
+      "step": 13455
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.257391383998247e-07,
+      "loss": 0.1801,
+      "step": 13456
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.249025654966425e-07,
+      "loss": 0.1542,
+      "step": 13457
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.24066398351725e-07,
+      "loss": 0.1506,
+      "step": 13458
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.232306370020493e-07,
+      "loss": 0.1508,
+      "step": 13459
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.223952814845781e-07,
+      "loss": 0.1646,
+      "step": 13460
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.215603318362531e-07,
+      "loss": 0.1463,
+      "step": 13461
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.207257880939979e-07,
+      "loss": 0.1485,
+      "step": 13462
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.198916502947218e-07,
+      "loss": 0.1935,
+      "step": 13463
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.190579184753111e-07,
+      "loss": 0.1641,
+      "step": 13464
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.182245926726373e-07,
+      "loss": 0.1788,
+      "step": 13465
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.173916729235543e-07,
+      "loss": 0.1797,
+      "step": 13466
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.16559159264897e-07,
+      "loss": 0.1694,
+      "step": 13467
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.157270517334803e-07,
+      "loss": 0.1703,
+      "step": 13468
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.148953503661062e-07,
+      "loss": 0.1666,
+      "step": 13469
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.140640551995549e-07,
+      "loss": 0.1688,
+      "step": 13470
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.13233166270585e-07,
+      "loss": 0.1504,
+      "step": 13471
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.124026836159482e-07,
+      "loss": 0.187,
+      "step": 13472
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.115726072723695e-07,
+      "loss": 0.1527,
+      "step": 13473
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.107429372765551e-07,
+      "loss": 0.1748,
+      "step": 13474
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.099136736652014e-07,
+      "loss": 0.1669,
+      "step": 13475
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.090848164749765e-07,
+      "loss": 0.1639,
+      "step": 13476
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.082563657425369e-07,
+      "loss": 0.1535,
+      "step": 13477
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.074283215045242e-07,
+      "loss": 0.1797,
+      "step": 13478
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.066006837975526e-07,
+      "loss": 0.1733,
+      "step": 13479
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.057734526582262e-07,
+      "loss": 0.1735,
+      "step": 13480
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.049466281231266e-07,
+      "loss": 0.1622,
+      "step": 13481
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.041202102288181e-07,
+      "loss": 0.1623,
+      "step": 13482
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.032941990118548e-07,
+      "loss": 0.1462,
+      "step": 13483
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.024685945087584e-07,
+      "loss": 0.1494,
+      "step": 13484
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.016433967560434e-07,
+      "loss": 0.1687,
+      "step": 13485
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.008186057902035e-07,
+      "loss": 0.1662,
+      "step": 13486
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.99994221647713e-07,
+      "loss": 0.1713,
+      "step": 13487
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.991702443650306e-07,
+      "loss": 0.152,
+      "step": 13488
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.983466739785961e-07,
+      "loss": 0.1786,
+      "step": 13489
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.975235105248292e-07,
+      "loss": 0.1763,
+      "step": 13490
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.967007540401339e-07,
+      "loss": 0.1649,
+      "step": 13491
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.958784045608969e-07,
+      "loss": 0.1517,
+      "step": 13492
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.950564621234857e-07,
+      "loss": 0.1681,
+      "step": 13493
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.942349267642468e-07,
+      "loss": 0.1664,
+      "step": 13494
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.934137985195145e-07,
+      "loss": 0.1718,
+      "step": 13495
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.92593077425603e-07,
+      "loss": 0.1803,
+      "step": 13496
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.917727635188033e-07,
+      "loss": 0.1614,
+      "step": 13497
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.909528568353997e-07,
+      "loss": 0.1434,
+      "step": 13498
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.901333574116432e-07,
+      "loss": 0.1886,
+      "step": 13499
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.893142652837826e-07,
+      "loss": 0.1643,
+      "step": 13500
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.88495580488039e-07,
+      "loss": 0.1757,
+      "step": 13501
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.876773030606155e-07,
+      "loss": 0.1651,
+      "step": 13502
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.868594330377033e-07,
+      "loss": 0.1461,
+      "step": 13503
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.86041970455469e-07,
+      "loss": 0.1687,
+      "step": 13504
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.852249153500636e-07,
+      "loss": 0.1514,
+      "step": 13505
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.844082677576237e-07,
+      "loss": 0.1608,
+      "step": 13506
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.835920277142617e-07,
+      "loss": 0.1606,
+      "step": 13507
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.827761952560753e-07,
+      "loss": 0.197,
+      "step": 13508
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.819607704191434e-07,
+      "loss": 0.1418,
+      "step": 13509
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.811457532395295e-07,
+      "loss": 0.1499,
+      "step": 13510
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.803311437532745e-07,
+      "loss": 0.1807,
+      "step": 13511
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.795169419964044e-07,
+      "loss": 0.1726,
+      "step": 13512
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.787031480049256e-07,
+      "loss": 0.1686,
+      "step": 13513
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.778897618148295e-07,
+      "loss": 0.1526,
+      "step": 13514
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.77076783462084e-07,
+      "loss": 0.1737,
+      "step": 13515
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.762642129826448e-07,
+      "loss": 0.1304,
+      "step": 13516
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.754520504124463e-07,
+      "loss": 0.1625,
+      "step": 13517
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.746402957874044e-07,
+      "loss": 0.1727,
+      "step": 13518
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.738289491434181e-07,
+      "loss": 0.1787,
+      "step": 13519
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.730180105163699e-07,
+      "loss": 0.1427,
+      "step": 13520
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.722074799421209e-07,
+      "loss": 0.1736,
+      "step": 13521
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.71397357456517e-07,
+      "loss": 0.165,
+      "step": 13522
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.70587643095384e-07,
+      "loss": 0.1777,
+      "step": 13523
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.697783368945333e-07,
+      "loss": 0.1797,
+      "step": 13524
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.689694388897495e-07,
+      "loss": 0.1663,
+      "step": 13525
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.681609491168118e-07,
+      "loss": 0.1595,
+      "step": 13526
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.673528676114727e-07,
+      "loss": 0.1595,
+      "step": 13527
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.665451944094649e-07,
+      "loss": 0.1808,
+      "step": 13528
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.657379295465139e-07,
+      "loss": 0.1554,
+      "step": 13529
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.649310730583137e-07,
+      "loss": 0.1817,
+      "step": 13530
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.641246249805479e-07,
+      "loss": 0.1739,
+      "step": 13531
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.633185853488844e-07,
+      "loss": 0.1759,
+      "step": 13532
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.62512954198964e-07,
+      "loss": 0.1615,
+      "step": 13533
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.617077315664212e-07,
+      "loss": 0.1565,
+      "step": 13534
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.60902917486861e-07,
+      "loss": 0.1635,
+      "step": 13535
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.600985119958748e-07,
+      "loss": 0.1709,
+      "step": 13536
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.592945151290421e-07,
+      "loss": 0.1744,
+      "step": 13537
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.584909269219131e-07,
+      "loss": 0.185,
+      "step": 13538
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.576877474100286e-07,
+      "loss": 0.1463,
+      "step": 13539
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.568849766289065e-07,
+      "loss": 0.1403,
+      "step": 13540
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.56082614614051e-07,
+      "loss": 0.1587,
+      "step": 13541
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.552806614009422e-07,
+      "loss": 0.1879,
+      "step": 13542
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.544791170250488e-07,
+      "loss": 0.1539,
+      "step": 13543
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.536779815218165e-07,
+      "loss": 0.174,
+      "step": 13544
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.52877254926675e-07,
+      "loss": 0.1549,
+      "step": 13545
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.520769372750347e-07,
+      "loss": 0.1711,
+      "step": 13546
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.512770286022919e-07,
+      "loss": 0.1492,
+      "step": 13547
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.504775289438149e-07,
+      "loss": 0.1773,
+      "step": 13548
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.496784383349676e-07,
+      "loss": 0.1833,
+      "step": 13549
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.488797568110861e-07,
+      "loss": 0.1447,
+      "step": 13550
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.480814844074913e-07,
+      "loss": 0.1514,
+      "step": 13551
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.472836211594881e-07,
+      "loss": 0.2066,
+      "step": 13552
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.464861671023549e-07,
+      "loss": 0.1735,
+      "step": 13553
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.456891222713636e-07,
+      "loss": 0.1615,
+      "step": 13554
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.448924867017637e-07,
+      "loss": 0.1799,
+      "step": 13555
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.440962604287794e-07,
+      "loss": 0.1778,
+      "step": 13556
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.4330044348763e-07,
+      "loss": 0.1738,
+      "step": 13557
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.425050359135044e-07,
+      "loss": 0.1879,
+      "step": 13558
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.417100377415786e-07,
+      "loss": 0.1844,
+      "step": 13559
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.409154490070159e-07,
+      "loss": 0.151,
+      "step": 13560
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.4012126974495e-07,
+      "loss": 0.1782,
+      "step": 13561
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.393274999905054e-07,
+      "loss": 0.1789,
+      "step": 13562
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.38534139778786e-07,
+      "loss": 0.1639,
+      "step": 13563
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.377411891448738e-07,
+      "loss": 0.1701,
+      "step": 13564
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.369486481238419e-07,
+      "loss": 0.1858,
+      "step": 13565
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.361565167507356e-07,
+      "loss": 0.1652,
+      "step": 13566
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.353647950605869e-07,
+      "loss": 0.1687,
+      "step": 13567
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.345734830884088e-07,
+      "loss": 0.1717,
+      "step": 13568
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.337825808691957e-07,
+      "loss": 0.1765,
+      "step": 13569
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.32992088437926e-07,
+      "loss": 0.1935,
+      "step": 13570
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.322020058295554e-07,
+      "loss": 0.1869,
+      "step": 13571
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.314123330790279e-07,
+      "loss": 0.155,
+      "step": 13572
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.306230702212625e-07,
+      "loss": 0.1636,
+      "step": 13573
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.298342172911665e-07,
+      "loss": 0.1673,
+      "step": 13574
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.290457743236256e-07,
+      "loss": 0.1672,
+      "step": 13575
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.282577413535031e-07,
+      "loss": 0.1573,
+      "step": 13576
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.27470118415654e-07,
+      "loss": 0.196,
+      "step": 13577
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.266829055449109e-07,
+      "loss": 0.212,
+      "step": 13578
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.258961027760825e-07,
+      "loss": 0.1503,
+      "step": 13579
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.251097101439697e-07,
+      "loss": 0.1696,
+      "step": 13580
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.243237276833425e-07,
+      "loss": 0.1634,
+      "step": 13581
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.235381554289677e-07,
+      "loss": 0.167,
+      "step": 13582
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.227529934155842e-07,
+      "loss": 0.1465,
+      "step": 13583
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.219682416779106e-07,
+      "loss": 0.1681,
+      "step": 13584
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.211839002506582e-07,
+      "loss": 0.1528,
+      "step": 13585
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.203999691685082e-07,
+      "loss": 0.1843,
+      "step": 13586
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.196164484661305e-07,
+      "loss": 0.1568,
+      "step": 13587
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.188333381781787e-07,
+      "loss": 0.1625,
+      "step": 13588
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.180506383392815e-07,
+      "loss": 0.1797,
+      "step": 13589
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.172683489840527e-07,
+      "loss": 0.1622,
+      "step": 13590
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.164864701470909e-07,
+      "loss": 0.156,
+      "step": 13591
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.157050018629707e-07,
+      "loss": 0.1648,
+      "step": 13592
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.149239441662537e-07,
+      "loss": 0.1774,
+      "step": 13593
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.141432970914807e-07,
+      "loss": 0.201,
+      "step": 13594
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.133630606731745e-07,
+      "loss": 0.1658,
+      "step": 13595
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.125832349458406e-07,
+      "loss": 0.1636,
+      "step": 13596
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.11803819943967e-07,
+      "loss": 0.1602,
+      "step": 13597
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.110248157020206e-07,
+      "loss": 0.1861,
+      "step": 13598
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.102462222544526e-07,
+      "loss": 0.155,
+      "step": 13599
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.094680396356956e-07,
+      "loss": 0.1842,
+      "step": 13600
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.086902678801644e-07,
+      "loss": 0.1998,
+      "step": 13601
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.079129070222534e-07,
+      "loss": 0.1611,
+      "step": 13602
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.071359570963432e-07,
+      "loss": 0.1397,
+      "step": 13603
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.063594181367895e-07,
+      "loss": 0.1696,
+      "step": 13604
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.05583290177938e-07,
+      "loss": 0.184,
+      "step": 13605
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.048075732541115e-07,
+      "loss": 0.1858,
+      "step": 13606
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.040322673996102e-07,
+      "loss": 0.1631,
+      "step": 13607
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.03257372648729e-07,
+      "loss": 0.1621,
+      "step": 13608
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.024828890357305e-07,
+      "loss": 0.1761,
+      "step": 13609
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.01708816594866e-07,
+      "loss": 0.1444,
+      "step": 13610
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.009351553603738e-07,
+      "loss": 0.1675,
+      "step": 13611
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.00161905366461e-07,
+      "loss": 0.181,
+      "step": 13612
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.993890666473269e-07,
+      "loss": 0.1988,
+      "step": 13613
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.986166392371496e-07,
+      "loss": 0.1665,
+      "step": 13614
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.978446231700864e-07,
+      "loss": 0.2002,
+      "step": 13615
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.970730184802832e-07,
+      "loss": 0.1885,
+      "step": 13616
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.963018252018605e-07,
+      "loss": 0.1804,
+      "step": 13617
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.955310433689233e-07,
+      "loss": 0.1601,
+      "step": 13618
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.947606730155599e-07,
+      "loss": 0.1588,
+      "step": 13619
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.939907141758373e-07,
+      "loss": 0.1661,
+      "step": 13620
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.932211668838073e-07,
+      "loss": 0.1819,
+      "step": 13621
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.924520311735017e-07,
+      "loss": 0.136,
+      "step": 13622
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.916833070789353e-07,
+      "loss": 0.1554,
+      "step": 13623
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.909149946341043e-07,
+      "loss": 0.1776,
+      "step": 13624
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.901470938729849e-07,
+      "loss": 0.1682,
+      "step": 13625
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.893796048295399e-07,
+      "loss": 0.1473,
+      "step": 13626
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.886125275377042e-07,
+      "loss": 0.1699,
+      "step": 13627
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.878458620314077e-07,
+      "loss": 0.1761,
+      "step": 13628
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.87079608344553e-07,
+      "loss": 0.1706,
+      "step": 13629
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.863137665110243e-07,
+      "loss": 0.1642,
+      "step": 13630
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.855483365646943e-07,
+      "loss": 0.1621,
+      "step": 13631
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.847833185394093e-07,
+      "loss": 0.145,
+      "step": 13632
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.840187124690035e-07,
+      "loss": 0.1721,
+      "step": 13633
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.83254518387293e-07,
+      "loss": 0.1621,
+      "step": 13634
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.824907363280675e-07,
+      "loss": 0.1581,
+      "step": 13635
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.817273663251112e-07,
+      "loss": 0.1743,
+      "step": 13636
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.809644084121781e-07,
+      "loss": 0.1818,
+      "step": 13637
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.802018626230089e-07,
+      "loss": 0.1783,
+      "step": 13638
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.794397289913323e-07,
+      "loss": 0.1586,
+      "step": 13639
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.78678007550847e-07,
+      "loss": 0.1552,
+      "step": 13640
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.779166983352403e-07,
+      "loss": 0.1649,
+      "step": 13641
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.77155801378182e-07,
+      "loss": 0.1492,
+      "step": 13642
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.763953167133208e-07,
+      "loss": 0.1964,
+      "step": 13643
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.756352443742875e-07,
+      "loss": 0.1794,
+      "step": 13644
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.748755843946975e-07,
+      "loss": 0.157,
+      "step": 13645
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.741163368081438e-07,
+      "loss": 0.1634,
+      "step": 13646
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.733575016482041e-07,
+      "loss": 0.1843,
+      "step": 13647
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.725990789484382e-07,
+      "loss": 0.1635,
+      "step": 13648
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.718410687423849e-07,
+      "loss": 0.1506,
+      "step": 13649
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.710834710635661e-07,
+      "loss": 0.1647,
+      "step": 13650
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.703262859454873e-07,
+      "loss": 0.1612,
+      "step": 13651
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.69569513421633e-07,
+      "loss": 0.1806,
+      "step": 13652
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.688131535254716e-07,
+      "loss": 0.1673,
+      "step": 13653
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.6805720629045e-07,
+      "loss": 0.1661,
+      "step": 13654
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.673016717500024e-07,
+      "loss": 0.1685,
+      "step": 13655
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.66546549937539e-07,
+      "loss": 0.1741,
+      "step": 13656
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.657918408864561e-07,
+      "loss": 0.1293,
+      "step": 13657
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.65037544630126e-07,
+      "loss": 0.1644,
+      "step": 13658
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.6428366120191e-07,
+      "loss": 0.1782,
+      "step": 13659
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.635301906351499e-07,
+      "loss": 0.1592,
+      "step": 13660
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.627771329631605e-07,
+      "loss": 0.1701,
+      "step": 13661
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.620244882192517e-07,
+      "loss": 0.1641,
+      "step": 13662
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.612722564367036e-07,
+      "loss": 0.1781,
+      "step": 13663
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.605204376487829e-07,
+      "loss": 0.175,
+      "step": 13664
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.597690318887417e-07,
+      "loss": 0.1581,
+      "step": 13665
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.59018039189806e-07,
+      "loss": 0.1578,
+      "step": 13666
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.582674595851923e-07,
+      "loss": 0.1887,
+      "step": 13667
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.575172931080898e-07,
+      "loss": 0.1662,
+      "step": 13668
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.567675397916739e-07,
+      "loss": 0.1787,
+      "step": 13669
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.560181996691051e-07,
+      "loss": 0.165,
+      "step": 13670
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.552692727735199e-07,
+      "loss": 0.1525,
+      "step": 13671
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.545207591380376e-07,
+      "loss": 0.1889,
+      "step": 13672
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.537726587957615e-07,
+      "loss": 0.1646,
+      "step": 13673
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.530249717797765e-07,
+      "loss": 0.1927,
+      "step": 13674
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.522776981231471e-07,
+      "loss": 0.1863,
+      "step": 13675
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.515308378589203e-07,
+      "loss": 0.1312,
+      "step": 13676
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.507843910201251e-07,
+      "loss": 0.1571,
+      "step": 13677
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.50038357639774e-07,
+      "loss": 0.1624,
+      "step": 13678
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.492927377508573e-07,
+      "loss": 0.1661,
+      "step": 13679
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.485475313863521e-07,
+      "loss": 0.1563,
+      "step": 13680
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.478027385792096e-07,
+      "loss": 0.1681,
+      "step": 13681
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.470583593623713e-07,
+      "loss": 0.1719,
+      "step": 13682
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.463143937687555e-07,
+      "loss": 0.1857,
+      "step": 13683
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.455708418312634e-07,
+      "loss": 0.1652,
+      "step": 13684
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.448277035827787e-07,
+      "loss": 0.1676,
+      "step": 13685
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.44084979056162e-07,
+      "loss": 0.1392,
+      "step": 13686
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.433426682842637e-07,
+      "loss": 0.1375,
+      "step": 13687
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.426007712999116e-07,
+      "loss": 0.1643,
+      "step": 13688
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.418592881359098e-07,
+      "loss": 0.1672,
+      "step": 13689
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.411182188250565e-07,
+      "loss": 0.1878,
+      "step": 13690
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.403775634001208e-07,
+      "loss": 0.1523,
+      "step": 13691
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.396373218938568e-07,
+      "loss": 0.1599,
+      "step": 13692
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.388974943390048e-07,
+      "loss": 0.1849,
+      "step": 13693
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.381580807682785e-07,
+      "loss": 0.1831,
+      "step": 13694
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.374190812143787e-07,
+      "loss": 0.1981,
+      "step": 13695
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.366804957099892e-07,
+      "loss": 0.1569,
+      "step": 13696
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.359423242877682e-07,
+      "loss": 0.1748,
+      "step": 13697
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.352045669803675e-07,
+      "loss": 0.1555,
+      "step": 13698
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.344672238204086e-07,
+      "loss": 0.2045,
+      "step": 13699
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.337302948404999e-07,
+      "loss": 0.1634,
+      "step": 13700
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.329937800732323e-07,
+      "loss": 0.162,
+      "step": 13701
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.32257679551177e-07,
+      "loss": 0.1484,
+      "step": 13702
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.315219933068884e-07,
+      "loss": 0.1821,
+      "step": 13703
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.307867213728991e-07,
+      "loss": 0.1922,
+      "step": 13704
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.300518637817276e-07,
+      "loss": 0.1569,
+      "step": 13705
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.293174205658714e-07,
+      "loss": 0.1653,
+      "step": 13706
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.28583391757811e-07,
+      "loss": 0.1673,
+      "step": 13707
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.278497773900083e-07,
+      "loss": 0.1663,
+      "step": 13708
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.271165774949028e-07,
+      "loss": 0.152,
+      "step": 13709
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.263837921049243e-07,
+      "loss": 0.1737,
+      "step": 13710
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.256514212524789e-07,
+      "loss": 0.1796,
+      "step": 13711
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.249194649699497e-07,
+      "loss": 0.1844,
+      "step": 13712
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.241879232897141e-07,
+      "loss": 0.1689,
+      "step": 13713
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.234567962441163e-07,
+      "loss": 0.1734,
+      "step": 13714
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.227260838654958e-07,
+      "loss": 0.1567,
+      "step": 13715
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.219957861861659e-07,
+      "loss": 0.1793,
+      "step": 13716
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.212659032384194e-07,
+      "loss": 0.1791,
+      "step": 13717
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.205364350545396e-07,
+      "loss": 0.1666,
+      "step": 13718
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.198073816667838e-07,
+      "loss": 0.1514,
+      "step": 13719
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.190787431073919e-07,
+      "loss": 0.1551,
+      "step": 13720
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.183505194085926e-07,
+      "loss": 0.166,
+      "step": 13721
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.176227106025856e-07,
+      "loss": 0.1522,
+      "step": 13722
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.168953167215586e-07,
+      "loss": 0.1487,
+      "step": 13723
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.161683377976813e-07,
+      "loss": 0.1696,
+      "step": 13724
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.154417738631014e-07,
+      "loss": 0.1853,
+      "step": 13725
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.147156249499519e-07,
+      "loss": 0.16,
+      "step": 13726
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.139898910903464e-07,
+      "loss": 0.1389,
+      "step": 13727
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.132645723163778e-07,
+      "loss": 0.1514,
+      "step": 13728
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.125396686601237e-07,
+      "loss": 0.1686,
+      "step": 13729
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.118151801536421e-07,
+      "loss": 0.1953,
+      "step": 13730
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.110911068289726e-07,
+      "loss": 0.1522,
+      "step": 13731
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.103674487181365e-07,
+      "loss": 0.1436,
+      "step": 13732
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.09644205853136e-07,
+      "loss": 0.1666,
+      "step": 13733
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.089213782659575e-07,
+      "loss": 0.1786,
+      "step": 13734
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.081989659885645e-07,
+      "loss": 0.1737,
+      "step": 13735
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.074769690529092e-07,
+      "loss": 0.1611,
+      "step": 13736
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.067553874909149e-07,
+      "loss": 0.1701,
+      "step": 13737
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.060342213344983e-07,
+      "loss": 0.1718,
+      "step": 13738
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.053134706155506e-07,
+      "loss": 0.162,
+      "step": 13739
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.045931353659429e-07,
+      "loss": 0.1601,
+      "step": 13740
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.038732156175375e-07,
+      "loss": 0.1438,
+      "step": 13741
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.031537114021668e-07,
+      "loss": 0.1814,
+      "step": 13742
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.024346227516498e-07,
+      "loss": 0.1618,
+      "step": 13743
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.01715949697792e-07,
+      "loss": 0.1431,
+      "step": 13744
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.009976922723726e-07,
+      "loss": 0.1821,
+      "step": 13745
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.002798505071561e-07,
+      "loss": 0.187,
+      "step": 13746
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.995624244338882e-07,
+      "loss": 0.1655,
+      "step": 13747
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.988454140842948e-07,
+      "loss": 0.1522,
+      "step": 13748
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.981288194900892e-07,
+      "loss": 0.1586,
+      "step": 13749
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.974126406829583e-07,
+      "loss": 0.172,
+      "step": 13750
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.966968776945758e-07,
+      "loss": 0.1553,
+      "step": 13751
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.959815305565952e-07,
+      "loss": 0.159,
+      "step": 13752
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.952665993006512e-07,
+      "loss": 0.2136,
+      "step": 13753
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.945520839583607e-07,
+      "loss": 0.1367,
+      "step": 13754
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.93837984561324e-07,
+      "loss": 0.1939,
+      "step": 13755
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.931243011411203e-07,
+      "loss": 0.1645,
+      "step": 13756
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.92411033729312e-07,
+      "loss": 0.1963,
+      "step": 13757
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.916981823574408e-07,
+      "loss": 0.1534,
+      "step": 13758
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.909857470570357e-07,
+      "loss": 0.1483,
+      "step": 13759
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.90273727859596e-07,
+      "loss": 0.1494,
+      "step": 13760
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.895621247966177e-07,
+      "loss": 0.156,
+      "step": 13761
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.888509378995677e-07,
+      "loss": 0.1565,
+      "step": 13762
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.881401671998954e-07,
+      "loss": 0.1622,
+      "step": 13763
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.874298127290379e-07,
+      "loss": 0.1928,
+      "step": 13764
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.867198745184055e-07,
+      "loss": 0.1562,
+      "step": 13765
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.860103525993966e-07,
+      "loss": 0.1804,
+      "step": 13766
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.853012470033914e-07,
+      "loss": 0.2012,
+      "step": 13767
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.845925577617429e-07,
+      "loss": 0.1693,
+      "step": 13768
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.838842849058002e-07,
+      "loss": 0.1484,
+      "step": 13769
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.831764284668795e-07,
+      "loss": 0.1655,
+      "step": 13770
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.824689884762858e-07,
+      "loss": 0.1479,
+      "step": 13771
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.817619649653084e-07,
+      "loss": 0.163,
+      "step": 13772
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.810553579652112e-07,
+      "loss": 0.1625,
+      "step": 13773
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.803491675072437e-07,
+      "loss": 0.1599,
+      "step": 13774
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.796433936226375e-07,
+      "loss": 0.1374,
+      "step": 13775
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.789380363426045e-07,
+      "loss": 0.135,
+      "step": 13776
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.782330956983373e-07,
+      "loss": 0.1595,
+      "step": 13777
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.77528571721011e-07,
+      "loss": 0.1677,
+      "step": 13778
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.768244644417842e-07,
+      "loss": 0.1725,
+      "step": 13779
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.761207738917929e-07,
+      "loss": 0.1532,
+      "step": 13780
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.754175001021589e-07,
+      "loss": 0.1584,
+      "step": 13781
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.74714643103983e-07,
+      "loss": 0.1577,
+      "step": 13782
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.740122029283479e-07,
+      "loss": 0.138,
+      "step": 13783
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.733101796063189e-07,
+      "loss": 0.2253,
+      "step": 13784
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.726085731689413e-07,
+      "loss": 0.163,
+      "step": 13785
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.719073836472433e-07,
+      "loss": 0.1869,
+      "step": 13786
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.712066110722347e-07,
+      "loss": 0.1938,
+      "step": 13787
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.705062554749064e-07,
+      "loss": 0.1757,
+      "step": 13788
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.6980631688623e-07,
+      "loss": 0.169,
+      "step": 13789
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.69106795337161e-07,
+      "loss": 0.1629,
+      "step": 13790
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.684076908586311e-07,
+      "loss": 0.1861,
+      "step": 13791
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.677090034815613e-07,
+      "loss": 0.1757,
+      "step": 13792
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.670107332368513e-07,
+      "loss": 0.1689,
+      "step": 13793
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.663128801553752e-07,
+      "loss": 0.1694,
+      "step": 13794
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.656154442680017e-07,
+      "loss": 0.1602,
+      "step": 13795
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.649184256055707e-07,
+      "loss": 0.1611,
+      "step": 13796
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.64221824198905e-07,
+      "loss": 0.1709,
+      "step": 13797
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.635256400788158e-07,
+      "loss": 0.1483,
+      "step": 13798
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.628298732760861e-07,
+      "loss": 0.1739,
+      "step": 13799
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.621345238214915e-07,
+      "loss": 0.1646,
+      "step": 13800
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.614395917457782e-07,
+      "loss": 0.1621,
+      "step": 13801
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.607450770796797e-07,
+      "loss": 0.1622,
+      "step": 13802
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.6005097985391e-07,
+      "loss": 0.185,
+      "step": 13803
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.593573000991659e-07,
+      "loss": 0.165,
+      "step": 13804
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.586640378461239e-07,
+      "loss": 0.1775,
+      "step": 13805
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.579711931254428e-07,
+      "loss": 0.1466,
+      "step": 13806
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.572787659677626e-07,
+      "loss": 0.1687,
+      "step": 13807
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.565867564037053e-07,
+      "loss": 0.1647,
+      "step": 13808
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.558951644638744e-07,
+      "loss": 0.1946,
+      "step": 13809
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.552039901788553e-07,
+      "loss": 0.1852,
+      "step": 13810
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.545132335792125e-07,
+      "loss": 0.1863,
+      "step": 13811
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.53822894695496e-07,
+      "loss": 0.154,
+      "step": 13812
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.531329735582358e-07,
+      "loss": 0.1919,
+      "step": 13813
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.524434701979398e-07,
+      "loss": 0.1573,
+      "step": 13814
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.517543846451024e-07,
+      "loss": 0.1675,
+      "step": 13815
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.510657169301969e-07,
+      "loss": 0.1832,
+      "step": 13816
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.503774670836803e-07,
+      "loss": 0.1919,
+      "step": 13817
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.496896351359904e-07,
+      "loss": 0.142,
+      "step": 13818
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.490022211175417e-07,
+      "loss": 0.1655,
+      "step": 13819
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.483152250587376e-07,
+      "loss": 0.1579,
+      "step": 13820
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.476286469899595e-07,
+      "loss": 0.1657,
+      "step": 13821
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.469424869415685e-07,
+      "loss": 0.1634,
+      "step": 13822
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.462567449439127e-07,
+      "loss": 0.1631,
+      "step": 13823
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.455714210273155e-07,
+      "loss": 0.1544,
+      "step": 13824
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.448865152220828e-07,
+      "loss": 0.1483,
+      "step": 13825
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.442020275585103e-07,
+      "loss": 0.1703,
+      "step": 13826
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.435179580668625e-07,
+      "loss": 0.165,
+      "step": 13827
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.428343067773945e-07,
+      "loss": 0.1511,
+      "step": 13828
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.421510737203384e-07,
+      "loss": 0.1703,
+      "step": 13829
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.41468258925909e-07,
+      "loss": 0.1731,
+      "step": 13830
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.407858624243079e-07,
+      "loss": 0.1701,
+      "step": 13831
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.401038842457074e-07,
+      "loss": 0.1583,
+      "step": 13832
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.394223244202701e-07,
+      "loss": 0.1536,
+      "step": 13833
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.387411829781364e-07,
+      "loss": 0.1591,
+      "step": 13834
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.380604599494288e-07,
+      "loss": 0.1751,
+      "step": 13835
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.373801553642522e-07,
+      "loss": 0.1583,
+      "step": 13836
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.367002692526924e-07,
+      "loss": 0.1896,
+      "step": 13837
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.360208016448165e-07,
+      "loss": 0.1562,
+      "step": 13838
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.353417525706728e-07,
+      "loss": 0.1617,
+      "step": 13839
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.346631220602904e-07,
+      "loss": 0.1599,
+      "step": 13840
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.339849101436845e-07,
+      "loss": 0.1516,
+      "step": 13841
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.33307116850843e-07,
+      "loss": 0.1729,
+      "step": 13842
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.326297422117454e-07,
+      "loss": 0.1537,
+      "step": 13843
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.319527862563467e-07,
+      "loss": 0.163,
+      "step": 13844
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.312762490145806e-07,
+      "loss": 0.1925,
+      "step": 13845
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.306001305163721e-07,
+      "loss": 0.1481,
+      "step": 13846
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.299244307916162e-07,
+      "loss": 0.1727,
+      "step": 13847
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.292491498702002e-07,
+      "loss": 0.1539,
+      "step": 13848
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.285742877819855e-07,
+      "loss": 0.1595,
+      "step": 13849
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.27899844556814e-07,
+      "loss": 0.1565,
+      "step": 13850
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.272258202245173e-07,
+      "loss": 0.1839,
+      "step": 13851
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.265522148149005e-07,
+      "loss": 0.1662,
+      "step": 13852
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.258790283577509e-07,
+      "loss": 0.1833,
+      "step": 13853
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.252062608828456e-07,
+      "loss": 0.1428,
+      "step": 13854
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.245339124199323e-07,
+      "loss": 0.1667,
+      "step": 13855
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.238619829987457e-07,
+      "loss": 0.1752,
+      "step": 13856
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.231904726490012e-07,
+      "loss": 0.1465,
+      "step": 13857
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.225193814003959e-07,
+      "loss": 0.1478,
+      "step": 13858
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.218487092826075e-07,
+      "loss": 0.1498,
+      "step": 13859
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.211784563252964e-07,
+      "loss": 0.1651,
+      "step": 13860
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.205086225581035e-07,
+      "loss": 0.1538,
+      "step": 13861
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.198392080106507e-07,
+      "loss": 0.1624,
+      "step": 13862
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.191702127125442e-07,
+      "loss": 0.1762,
+      "step": 13863
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.185016366933671e-07,
+      "loss": 0.1914,
+      "step": 13864
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.178334799826879e-07,
+      "loss": 0.1695,
+      "step": 13865
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.17165742610054e-07,
+      "loss": 0.1843,
+      "step": 13866
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.164984246049976e-07,
+      "loss": 0.1727,
+      "step": 13867
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.15831525997027e-07,
+      "loss": 0.1871,
+      "step": 13868
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.151650468156377e-07,
+      "loss": 0.188,
+      "step": 13869
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.144989870903017e-07,
+      "loss": 0.161,
+      "step": 13870
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.138333468504753e-07,
+      "loss": 0.1657,
+      "step": 13871
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.131681261255983e-07,
+      "loss": 0.1621,
+      "step": 13872
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.12503324945085e-07,
+      "loss": 0.1743,
+      "step": 13873
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.118389433383397e-07,
+      "loss": 0.1716,
+      "step": 13874
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.111749813347411e-07,
+      "loss": 0.1709,
+      "step": 13875
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.105114389636512e-07,
+      "loss": 0.1783,
+      "step": 13876
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.098483162544199e-07,
+      "loss": 0.1749,
+      "step": 13877
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.09185613236367e-07,
+      "loss": 0.1695,
+      "step": 13878
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.085233299388026e-07,
+      "loss": 0.1945,
+      "step": 13879
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.078614663910141e-07,
+      "loss": 0.1687,
+      "step": 13880
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.072000226222718e-07,
+      "loss": 0.1982,
+      "step": 13881
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.065389986618307e-07,
+      "loss": 0.1604,
+      "step": 13882
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.0587839453892e-07,
+      "loss": 0.1596,
+      "step": 13883
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.052182102827552e-07,
+      "loss": 0.1807,
+      "step": 13884
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.045584459225327e-07,
+      "loss": 0.1614,
+      "step": 13885
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.038991014874295e-07,
+      "loss": 0.1493,
+      "step": 13886
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.032401770066031e-07,
+      "loss": 0.1486,
+      "step": 13887
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.025816725091958e-07,
+      "loss": 0.1746,
+      "step": 13888
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.019235880243289e-07,
+      "loss": 0.1507,
+      "step": 13889
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.012659235811035e-07,
+      "loss": 0.1426,
+      "step": 13890
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.006086792086063e-07,
+      "loss": 0.1595,
+      "step": 13891
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.999518549359039e-07,
+      "loss": 0.1736,
+      "step": 13892
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.992954507920388e-07,
+      "loss": 0.164,
+      "step": 13893
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.986394668060446e-07,
+      "loss": 0.1651,
+      "step": 13894
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.979839030069312e-07,
+      "loss": 0.1534,
+      "step": 13895
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.973287594236853e-07,
+      "loss": 0.1656,
+      "step": 13896
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.966740360852861e-07,
+      "loss": 0.1562,
+      "step": 13897
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.960197330206839e-07,
+      "loss": 0.1771,
+      "step": 13898
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.953658502588154e-07,
+      "loss": 0.1966,
+      "step": 13899
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.947123878286009e-07,
+      "loss": 0.1393,
+      "step": 13900
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.940593457589338e-07,
+      "loss": 0.1828,
+      "step": 13901
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.934067240786988e-07,
+      "loss": 0.1981,
+      "step": 13902
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.92754522816754e-07,
+      "loss": 0.1799,
+      "step": 13903
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.921027420019419e-07,
+      "loss": 0.1547,
+      "step": 13904
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.914513816630917e-07,
+      "loss": 0.1545,
+      "step": 13905
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.908004418290046e-07,
+      "loss": 0.1729,
+      "step": 13906
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.901499225284678e-07,
+      "loss": 0.1793,
+      "step": 13907
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.894998237902515e-07,
+      "loss": 0.1832,
+      "step": 13908
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.88850145643105e-07,
+      "loss": 0.169,
+      "step": 13909
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.882008881157596e-07,
+      "loss": 0.1692,
+      "step": 13910
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.875520512369281e-07,
+      "loss": 0.1807,
+      "step": 13911
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.86903635035304e-07,
+      "loss": 0.1745,
+      "step": 13912
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.862556395395634e-07,
+      "loss": 0.1596,
+      "step": 13913
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.856080647783623e-07,
+      "loss": 0.1506,
+      "step": 13914
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.849609107803399e-07,
+      "loss": 0.1571,
+      "step": 13915
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.843141775741156e-07,
+      "loss": 0.1569,
+      "step": 13916
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.836678651882898e-07,
+      "loss": 0.1553,
+      "step": 13917
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.830219736514463e-07,
+      "loss": 0.1575,
+      "step": 13918
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.823765029921479e-07,
+      "loss": 0.1643,
+      "step": 13919
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.817314532389395e-07,
+      "loss": 0.152,
+      "step": 13920
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.810868244203481e-07,
+      "loss": 0.1725,
+      "step": 13921
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.804426165648823e-07,
+      "loss": 0.1684,
+      "step": 13922
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.797988297010326e-07,
+      "loss": 0.1461,
+      "step": 13923
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.791554638572649e-07,
+      "loss": 0.1643,
+      "step": 13924
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.785125190620355e-07,
+      "loss": 0.1614,
+      "step": 13925
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.778699953437782e-07,
+      "loss": 0.1409,
+      "step": 13926
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.772278927309049e-07,
+      "loss": 0.156,
+      "step": 13927
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.7658621125181604e-07,
+      "loss": 0.1663,
+      "step": 13928
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.759449509348846e-07,
+      "loss": 0.1877,
+      "step": 13929
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.7530411180847115e-07,
+      "loss": 0.1895,
+      "step": 13930
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.746636939009186e-07,
+      "loss": 0.145,
+      "step": 13931
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.740236972405443e-07,
+      "loss": 0.1539,
+      "step": 13932
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.7338412185565674e-07,
+      "loss": 0.1366,
+      "step": 13933
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.727449677745355e-07,
+      "loss": 0.1606,
+      "step": 13934
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.721062350254502e-07,
+      "loss": 0.1665,
+      "step": 13935
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.7146792363664485e-07,
+      "loss": 0.2056,
+      "step": 13936
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.708300336363492e-07,
+      "loss": 0.1537,
+      "step": 13937
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.70192565052775e-07,
+      "loss": 0.1777,
+      "step": 13938
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.6955551791411094e-07,
+      "loss": 0.1776,
+      "step": 13939
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.6891889224853106e-07,
+      "loss": 0.1578,
+      "step": 13940
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.6828268808418953e-07,
+      "loss": 0.1468,
+      "step": 13941
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.676469054492216e-07,
+      "loss": 0.1859,
+      "step": 13942
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.670115443717438e-07,
+      "loss": 0.156,
+      "step": 13943
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.663766048798546e-07,
+      "loss": 0.1388,
+      "step": 13944
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.657420870016327e-07,
+      "loss": 0.1531,
+      "step": 13945
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.6510799076514125e-07,
+      "loss": 0.1717,
+      "step": 13946
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.6447431619841664e-07,
+      "loss": 0.1915,
+      "step": 13947
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.6384106332948877e-07,
+      "loss": 0.1562,
+      "step": 13948
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.6320823218636066e-07,
+      "loss": 0.1817,
+      "step": 13949
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.6257582279701786e-07,
+      "loss": 0.1602,
+      "step": 13950
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.6194383518942897e-07,
+      "loss": 0.1623,
+      "step": 13951
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.613122693915395e-07,
+      "loss": 0.1588,
+      "step": 13952
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.6068112543128374e-07,
+      "loss": 0.1687,
+      "step": 13953
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.600504033365738e-07,
+      "loss": 0.1596,
+      "step": 13954
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.594201031352974e-07,
+      "loss": 0.1753,
+      "step": 13955
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.587902248553344e-07,
+      "loss": 0.1675,
+      "step": 13956
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.581607685245382e-07,
+      "loss": 0.1769,
+      "step": 13957
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5753173417074415e-07,
+      "loss": 0.1719,
+      "step": 13958
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5690312182177454e-07,
+      "loss": 0.174,
+      "step": 13959
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5627493150542603e-07,
+      "loss": 0.1677,
+      "step": 13960
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5564716324947964e-07,
+      "loss": 0.1457,
+      "step": 13961
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5501981708169885e-07,
+      "loss": 0.1627,
+      "step": 13962
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.543928930298258e-07,
+      "loss": 0.1922,
+      "step": 13963
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5376639112158835e-07,
+      "loss": 0.1636,
+      "step": 13964
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5314031138469103e-07,
+      "loss": 0.1828,
+      "step": 13965
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5251465384682056e-07,
+      "loss": 0.1629,
+      "step": 13966
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.518894185356471e-07,
+      "loss": 0.1666,
+      "step": 13967
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.512646054788217e-07,
+      "loss": 0.1857,
+      "step": 13968
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.506402147039746e-07,
+      "loss": 0.1731,
+      "step": 13969
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.500162462387192e-07,
+      "loss": 0.1516,
+      "step": 13970
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4939270011065017e-07,
+      "loss": 0.1669,
+      "step": 13971
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4876957634734207e-07,
+      "loss": 0.1655,
+      "step": 13972
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.481468749763529e-07,
+      "loss": 0.1794,
+      "step": 13973
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.475245960252228e-07,
+      "loss": 0.1757,
+      "step": 13974
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4690273952146646e-07,
+      "loss": 0.1859,
+      "step": 13975
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.462813054925874e-07,
+      "loss": 0.1994,
+      "step": 13976
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4566029396607034e-07,
+      "loss": 0.1575,
+      "step": 13977
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.450397049693744e-07,
+      "loss": 0.1527,
+      "step": 13978
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4441953852994767e-07,
+      "loss": 0.1649,
+      "step": 13979
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4379979467521373e-07,
+      "loss": 0.1713,
+      "step": 13980
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4318047343258174e-07,
+      "loss": 0.1606,
+      "step": 13981
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.425615748294421e-07,
+      "loss": 0.1803,
+      "step": 13982
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.419430988931606e-07,
+      "loss": 0.165,
+      "step": 13983
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.413250456510931e-07,
+      "loss": 0.1697,
+      "step": 13984
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4070741513056904e-07,
+      "loss": 0.1779,
+      "step": 13985
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4009020735890197e-07,
+      "loss": 0.1992,
+      "step": 13986
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.394734223633912e-07,
+      "loss": 0.1665,
+      "step": 13987
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.388570601713105e-07,
+      "loss": 0.1737,
+      "step": 13988
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.382411208099169e-07,
+      "loss": 0.1568,
+      "step": 13989
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.3762560430645194e-07,
+      "loss": 0.1591,
+      "step": 13990
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.3701051068813504e-07,
+      "loss": 0.1759,
+      "step": 13991
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.3639583998216886e-07,
+      "loss": 0.1762,
+      "step": 13992
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.357815922157349e-07,
+      "loss": 0.1648,
+      "step": 13993
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.351677674159982e-07,
+      "loss": 0.1428,
+      "step": 13994
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.345543656101048e-07,
+      "loss": 0.165,
+      "step": 13995
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.3394138682518186e-07,
+      "loss": 0.1724,
+      "step": 13996
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.3332883108833767e-07,
+      "loss": 0.148,
+      "step": 13997
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.3271669842666173e-07,
+      "loss": 0.1635,
+      "step": 13998
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.321049888672235e-07,
+      "loss": 0.1862,
+      "step": 13999
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.31493702437078e-07,
+      "loss": 0.1589,
+      "step": 14000
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.308828391632558e-07,
+      "loss": 0.1485,
+      "step": 14001
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.302723990727753e-07,
+      "loss": 0.1668,
+      "step": 14002
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.29662382192626e-07,
+      "loss": 0.1816,
+      "step": 14003
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.2905278854979193e-07,
+      "loss": 0.1569,
+      "step": 14004
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.2844361817123034e-07,
+      "loss": 0.1572,
+      "step": 14005
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.2783487108387646e-07,
+      "loss": 0.1678,
+      "step": 14006
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.272265473146575e-07,
+      "loss": 0.1543,
+      "step": 14007
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.2661864689047314e-07,
+      "loss": 0.1858,
+      "step": 14008
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.260111698382041e-07,
+      "loss": 0.1973,
+      "step": 14009
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.254041161847222e-07,
+      "loss": 0.1637,
+      "step": 14010
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.247974859568682e-07,
+      "loss": 0.1968,
+      "step": 14011
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.2419127918147175e-07,
+      "loss": 0.1805,
+      "step": 14012
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.2358549588534024e-07,
+      "loss": 0.1538,
+      "step": 14013
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.2298013609526454e-07,
+      "loss": 0.1418,
+      "step": 14014
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.2237519983801877e-07,
+      "loss": 0.1723,
+      "step": 14015
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.217706871403504e-07,
+      "loss": 0.1678,
+      "step": 14016
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.2116659802899695e-07,
+      "loss": 0.1744,
+      "step": 14017
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.205629325306726e-07,
+      "loss": 0.1689,
+      "step": 14018
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.1995969067207374e-07,
+      "loss": 0.1682,
+      "step": 14019
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.193568724798769e-07,
+      "loss": 0.1729,
+      "step": 14020
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.18754477980744e-07,
+      "loss": 0.1476,
+      "step": 14021
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.181525072013115e-07,
+      "loss": 0.1674,
+      "step": 14022
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.175509601682037e-07,
+      "loss": 0.1579,
+      "step": 14023
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.1694983690802273e-07,
+      "loss": 0.1931,
+      "step": 14024
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.163491374473527e-07,
+      "loss": 0.1835,
+      "step": 14025
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.1574886181275696e-07,
+      "loss": 0.163,
+      "step": 14026
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.151490100307842e-07,
+      "loss": 0.152,
+      "step": 14027
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.1454958212796215e-07,
+      "loss": 0.1573,
+      "step": 14028
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.139505781307973e-07,
+      "loss": 0.1591,
+      "step": 14029
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.133519980657841e-07,
+      "loss": 0.1746,
+      "step": 14030
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.1275384195938906e-07,
+      "loss": 0.1843,
+      "step": 14031
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.1215610983807e-07,
+      "loss": 0.1603,
+      "step": 14032
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.115588017282579e-07,
+      "loss": 0.1685,
+      "step": 14033
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.109619176563673e-07,
+      "loss": 0.1416,
+      "step": 14034
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.103654576487992e-07,
+      "loss": 0.1583,
+      "step": 14035
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.097694217319259e-07,
+      "loss": 0.1748,
+      "step": 14036
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.091738099321085e-07,
+      "loss": 0.1464,
+      "step": 14037
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.085786222756893e-07,
+      "loss": 0.1555,
+      "step": 14038
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.0798385878898726e-07,
+      "loss": 0.1819,
+      "step": 14039
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.073895194983057e-07,
+      "loss": 0.1807,
+      "step": 14040
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.0679560442992926e-07,
+      "loss": 0.1614,
+      "step": 14041
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.062021136101224e-07,
+      "loss": 0.1606,
+      "step": 14042
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.0560904706513194e-07,
+      "loss": 0.1718,
+      "step": 14043
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.050164048211858e-07,
+      "loss": 0.1496,
+      "step": 14044
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.044241869044918e-07,
+      "loss": 0.1587,
+      "step": 14045
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.0383239334124134e-07,
+      "loss": 0.2039,
+      "step": 14046
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.032410241576057e-07,
+      "loss": 0.1601,
+      "step": 14047
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.026500793797372e-07,
+      "loss": 0.1552,
+      "step": 14048
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.020595590337695e-07,
+      "loss": 0.164,
+      "step": 14049
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.014694631458183e-07,
+      "loss": 0.163,
+      "step": 14050
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.008797917419793e-07,
+      "loss": 0.1651,
+      "step": 14051
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.0029054484833077e-07,
+      "loss": 0.1877,
+      "step": 14052
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.997017224909305e-07,
+      "loss": 0.1718,
+      "step": 14053
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.99113324695819e-07,
+      "loss": 0.1704,
+      "step": 14054
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.985253514890175e-07,
+      "loss": 0.1612,
+      "step": 14055
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.9793780289652974e-07,
+      "loss": 0.1677,
+      "step": 14056
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.97350678944336e-07,
+      "loss": 0.1573,
+      "step": 14057
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.967639796584033e-07,
+      "loss": 0.1654,
+      "step": 14058
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.961777050646798e-07,
+      "loss": 0.1566,
+      "step": 14059
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.9559185518908916e-07,
+      "loss": 0.1741,
+      "step": 14060
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.9500643005754293e-07,
+      "loss": 0.1447,
+      "step": 14061
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.944214296959281e-07,
+      "loss": 0.1713,
+      "step": 14062
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.938368541301163e-07,
+      "loss": 0.156,
+      "step": 14063
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.9325270338596233e-07,
+      "loss": 0.1994,
+      "step": 14064
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.9266897748929555e-07,
+      "loss": 0.1571,
+      "step": 14065
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.920856764659353e-07,
+      "loss": 0.1428,
+      "step": 14066
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.915028003416732e-07,
+      "loss": 0.145,
+      "step": 14067
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.909203491422886e-07,
+      "loss": 0.1464,
+      "step": 14068
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.903383228935387e-07,
+      "loss": 0.1478,
+      "step": 14069
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.8975672162116286e-07,
+      "loss": 0.1563,
+      "step": 14070
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.8917554535088274e-07,
+      "loss": 0.1976,
+      "step": 14071
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.88594794108399e-07,
+      "loss": 0.1673,
+      "step": 14072
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.8801446791939534e-07,
+      "loss": 0.1488,
+      "step": 14073
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.8743456680953693e-07,
+      "loss": 0.1688,
+      "step": 14074
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.8685509080446773e-07,
+      "loss": 0.172,
+      "step": 14075
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.8627603992981487e-07,
+      "loss": 0.1605,
+      "step": 14076
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.8569741421118687e-07,
+      "loss": 0.1947,
+      "step": 14077
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.8511921367417105e-07,
+      "loss": 0.2036,
+      "step": 14078
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.8454143834434134e-07,
+      "loss": 0.1458,
+      "step": 14079
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.8396408824724395e-07,
+      "loss": 0.1689,
+      "step": 14080
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.8338716340841517e-07,
+      "loss": 0.1701,
+      "step": 14081
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.8281066385336907e-07,
+      "loss": 0.1638,
+      "step": 14082
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.822345896075985e-07,
+      "loss": 0.1926,
+      "step": 14083
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.816589406965832e-07,
+      "loss": 0.1434,
+      "step": 14084
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.8108371714577485e-07,
+      "loss": 0.1655,
+      "step": 14085
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.805089189806166e-07,
+      "loss": 0.175,
+      "step": 14086
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.79934546226528e-07,
+      "loss": 0.144,
+      "step": 14087
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.7936059890890773e-07,
+      "loss": 0.151,
+      "step": 14088
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.78787077053141e-07,
+      "loss": 0.1399,
+      "step": 14089
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.782139806845886e-07,
+      "loss": 0.1686,
+      "step": 14090
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.776413098285947e-07,
+      "loss": 0.1592,
+      "step": 14091
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.770690645104891e-07,
+      "loss": 0.1627,
+      "step": 14092
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.764972447555737e-07,
+      "loss": 0.1566,
+      "step": 14093
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.7592585058913946e-07,
+      "loss": 0.1838,
+      "step": 14094
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.75354882036455e-07,
+      "loss": 0.1675,
+      "step": 14095
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.747843391227701e-07,
+      "loss": 0.1511,
+      "step": 14096
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.742142218733169e-07,
+      "loss": 0.15,
+      "step": 14097
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.736445303133085e-07,
+      "loss": 0.1548,
+      "step": 14098
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.730752644679392e-07,
+      "loss": 0.1647,
+      "step": 14099
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.725064243623833e-07,
+      "loss": 0.1665,
+      "step": 14100
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.719380100217973e-07,
+      "loss": 0.1776,
+      "step": 14101
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.713700214713178e-07,
+      "loss": 0.1488,
+      "step": 14102
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.708024587360659e-07,
+      "loss": 0.1699,
+      "step": 14103
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.702353218411381e-07,
+      "loss": 0.1611,
+      "step": 14104
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6966861081161875e-07,
+      "loss": 0.1823,
+      "step": 14105
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.691023256725679e-07,
+      "loss": 0.2078,
+      "step": 14106
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6853646644903094e-07,
+      "loss": 0.1656,
+      "step": 14107
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.67971033166028e-07,
+      "loss": 0.1474,
+      "step": 14108
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.674060258485679e-07,
+      "loss": 0.1747,
+      "step": 14109
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.668414445216395e-07,
+      "loss": 0.1767,
+      "step": 14110
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6627728921020624e-07,
+      "loss": 0.1559,
+      "step": 14111
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.657135599392203e-07,
+      "loss": 0.2244,
+      "step": 14112
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.651502567336096e-07,
+      "loss": 0.1669,
+      "step": 14113
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.645873796182886e-07,
+      "loss": 0.1741,
+      "step": 14114
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6402492861814853e-07,
+      "loss": 0.1624,
+      "step": 14115
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.634629037580617e-07,
+      "loss": 0.2179,
+      "step": 14116
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.629013050628849e-07,
+      "loss": 0.1721,
+      "step": 14117
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.623401325574538e-07,
+      "loss": 0.1752,
+      "step": 14118
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6177938626658414e-07,
+      "loss": 0.1462,
+      "step": 14119
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.612190662150772e-07,
+      "loss": 0.1381,
+      "step": 14120
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.606591724277098e-07,
+      "loss": 0.1543,
+      "step": 14121
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.600997049292443e-07,
+      "loss": 0.1629,
+      "step": 14122
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5954066374442097e-07,
+      "loss": 0.1635,
+      "step": 14123
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.589820488979634e-07,
+      "loss": 0.1795,
+      "step": 14124
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5842386041457734e-07,
+      "loss": 0.1685,
+      "step": 14125
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.578660983189453e-07,
+      "loss": 0.1584,
+      "step": 14126
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.573087626357352e-07,
+      "loss": 0.1611,
+      "step": 14127
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.567518533895953e-07,
+      "loss": 0.1449,
+      "step": 14128
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5619537060515243e-07,
+      "loss": 0.1618,
+      "step": 14129
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.556393143070169e-07,
+      "loss": 0.1745,
+      "step": 14130
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.550836845197814e-07,
+      "loss": 0.1723,
+      "step": 14131
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.545284812680161e-07,
+      "loss": 0.15,
+      "step": 14132
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5397370457627476e-07,
+      "loss": 0.1625,
+      "step": 14133
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.534193544690923e-07,
+      "loss": 0.1672,
+      "step": 14134
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5286543097098447e-07,
+      "loss": 0.1782,
+      "step": 14135
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5231193410644513e-07,
+      "loss": 0.1617,
+      "step": 14136
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5175886389995474e-07,
+      "loss": 0.157,
+      "step": 14137
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.512062203759736e-07,
+      "loss": 0.1634,
+      "step": 14138
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5065400355893676e-07,
+      "loss": 0.1864,
+      "step": 14139
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.501022134732712e-07,
+      "loss": 0.1652,
+      "step": 14140
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.495508501433753e-07,
+      "loss": 0.1617,
+      "step": 14141
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.4899991359363173e-07,
+      "loss": 0.1512,
+      "step": 14142
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.484494038484099e-07,
+      "loss": 0.175,
+      "step": 14143
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.4789932093205023e-07,
+      "loss": 0.1504,
+      "step": 14144
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.473496648688823e-07,
+      "loss": 0.1673,
+      "step": 14145
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.4680043568321444e-07,
+      "loss": 0.1374,
+      "step": 14146
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.4625163339933264e-07,
+      "loss": 0.1492,
+      "step": 14147
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.4570325804151096e-07,
+      "loss": 0.1626,
+      "step": 14148
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.451553096339977e-07,
+      "loss": 0.1735,
+      "step": 14149
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.446077882010268e-07,
+      "loss": 0.1569,
+      "step": 14150
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.440606937668123e-07,
+      "loss": 0.1756,
+      "step": 14151
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.43514026355547e-07,
+      "loss": 0.1489,
+      "step": 14152
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.4296778599140714e-07,
+      "loss": 0.1625,
+      "step": 14153
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.4242197269855006e-07,
+      "loss": 0.154,
+      "step": 14154
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.418765865011142e-07,
+      "loss": 0.1492,
+      "step": 14155
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.41331627423217e-07,
+      "loss": 0.1648,
+      "step": 14156
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.407870954889614e-07,
+      "loss": 0.1781,
+      "step": 14157
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.4024299072242696e-07,
+      "loss": 0.1746,
+      "step": 14158
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.396993131476745e-07,
+      "loss": 0.1683,
+      "step": 14159
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3915606278875023e-07,
+      "loss": 0.1699,
+      "step": 14160
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.386132396696784e-07,
+      "loss": 0.21,
+      "step": 14161
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.38070843814462e-07,
+      "loss": 0.1681,
+      "step": 14162
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.375288752470929e-07,
+      "loss": 0.1448,
+      "step": 14163
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.36987333991533e-07,
+      "loss": 0.1552,
+      "step": 14164
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.364462200717367e-07,
+      "loss": 0.1707,
+      "step": 14165
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3590553351163235e-07,
+      "loss": 0.1667,
+      "step": 14166
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3536527433512877e-07,
+      "loss": 0.1462,
+      "step": 14167
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.348254425661224e-07,
+      "loss": 0.1598,
+      "step": 14168
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3428603822848294e-07,
+      "loss": 0.1778,
+      "step": 14169
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.337470613460658e-07,
+      "loss": 0.1968,
+      "step": 14170
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.332085119427109e-07,
+      "loss": 0.1758,
+      "step": 14171
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.32670390042229e-07,
+      "loss": 0.145,
+      "step": 14172
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.321326956684212e-07,
+      "loss": 0.1621,
+      "step": 14173
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3159542884506515e-07,
+      "loss": 0.1561,
+      "step": 14174
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.310585895959206e-07,
+      "loss": 0.1687,
+      "step": 14175
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.30522177944731e-07,
+      "loss": 0.1482,
+      "step": 14176
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.2998619391521604e-07,
+      "loss": 0.1565,
+      "step": 14177
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.294506375310802e-07,
+      "loss": 0.1712,
+      "step": 14178
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.2891550881600784e-07,
+      "loss": 0.1589,
+      "step": 14179
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.2838080779366346e-07,
+      "loss": 0.1527,
+      "step": 14180
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.2784653448769467e-07,
+      "loss": 0.1682,
+      "step": 14181
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.273126889217293e-07,
+      "loss": 0.1438,
+      "step": 14182
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.2677927111937513e-07,
+      "loss": 0.1585,
+      "step": 14183
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.2624628110422217e-07,
+      "loss": 0.183,
+      "step": 14184
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.257137188998416e-07,
+      "loss": 0.1645,
+      "step": 14185
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.251815845297845e-07,
+      "loss": 0.1461,
+      "step": 14186
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.2464987801758554e-07,
+      "loss": 0.1412,
+      "step": 14187
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.2411859938675796e-07,
+      "loss": 0.1806,
+      "step": 14188
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.2358774866079744e-07,
+      "loss": 0.1795,
+      "step": 14189
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.2305732586317863e-07,
+      "loss": 0.1905,
+      "step": 14190
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.2252733101736044e-07,
+      "loss": 0.1459,
+      "step": 14191
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.2199776414678196e-07,
+      "loss": 0.1499,
+      "step": 14192
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.214686252748589e-07,
+      "loss": 0.142,
+      "step": 14193
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.2093991442499695e-07,
+      "loss": 0.1811,
+      "step": 14194
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.204116316205741e-07,
+      "loss": 0.1665,
+      "step": 14195
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.19883776884955e-07,
+      "loss": 0.1807,
+      "step": 14196
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.19356350241482e-07,
+      "loss": 0.1411,
+      "step": 14197
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.1882935171347997e-07,
+      "loss": 0.1609,
+      "step": 14198
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.183027813242567e-07,
+      "loss": 0.1527,
+      "step": 14199
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.177766390970982e-07,
+      "loss": 0.1388,
+      "step": 14200
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.172509250552713e-07,
+      "loss": 0.1639,
+      "step": 14201
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.1672563922202635e-07,
+      "loss": 0.1736,
+      "step": 14202
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.162007816205925e-07,
+      "loss": 0.1541,
+      "step": 14203
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.156763522741824e-07,
+      "loss": 0.175,
+      "step": 14204
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.151523512059884e-07,
+      "loss": 0.1802,
+      "step": 14205
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.1462877843918216e-07,
+      "loss": 0.1642,
+      "step": 14206
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.141056339969195e-07,
+      "loss": 0.1656,
+      "step": 14207
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.135829179023342e-07,
+      "loss": 0.1775,
+      "step": 14208
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.130606301785455e-07,
+      "loss": 0.1529,
+      "step": 14209
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.1253877084864935e-07,
+      "loss": 0.185,
+      "step": 14210
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.12017339935724e-07,
+      "loss": 0.1694,
+      "step": 14211
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.1149633746283104e-07,
+      "loss": 0.1717,
+      "step": 14212
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.109757634530075e-07,
+      "loss": 0.1627,
+      "step": 14213
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.104556179292784e-07,
+      "loss": 0.1806,
+      "step": 14214
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.099359009146452e-07,
+      "loss": 0.1902,
+      "step": 14215
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0941661243209297e-07,
+      "loss": 0.1652,
+      "step": 14216
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0889775250458776e-07,
+      "loss": 0.1628,
+      "step": 14217
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.083793211550712e-07,
+      "loss": 0.1514,
+      "step": 14218
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.078613184064727e-07,
+      "loss": 0.1742,
+      "step": 14219
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.073437442817029e-07,
+      "loss": 0.1682,
+      "step": 14220
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0682659880364675e-07,
+      "loss": 0.1583,
+      "step": 14221
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.063098819951771e-07,
+      "loss": 0.1436,
+      "step": 14222
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0579359387914455e-07,
+      "loss": 0.1496,
+      "step": 14223
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.052777344783786e-07,
+      "loss": 0.1657,
+      "step": 14224
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0476230381569884e-07,
+      "loss": 0.1385,
+      "step": 14225
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0424730191389364e-07,
+      "loss": 0.1572,
+      "step": 14226
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.037327287957414e-07,
+      "loss": 0.1764,
+      "step": 14227
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.032185844839963e-07,
+      "loss": 0.1826,
+      "step": 14228
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0270486900139894e-07,
+      "loss": 0.1561,
+      "step": 14229
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.021915823706645e-07,
+      "loss": 0.1365,
+      "step": 14230
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.016787246144948e-07,
+      "loss": 0.1402,
+      "step": 14231
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.011662957555694e-07,
+      "loss": 0.1484,
+      "step": 14232
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0065429581655035e-07,
+      "loss": 0.1669,
+      "step": 14233
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0014272482008054e-07,
+      "loss": 0.1586,
+      "step": 14234
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.996315827887819e-07,
+      "loss": 0.1662,
+      "step": 14235
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.991208697452608e-07,
+      "loss": 0.1875,
+      "step": 14236
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.9861058571210357e-07,
+      "loss": 0.1417,
+      "step": 14237
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.981007307118744e-07,
+      "loss": 0.1903,
+      "step": 14238
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.9759130476712304e-07,
+      "loss": 0.1767,
+      "step": 14239
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.970823079003782e-07,
+      "loss": 0.1396,
+      "step": 14240
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.965737401341484e-07,
+      "loss": 0.1604,
+      "step": 14241
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.960656014909247e-07,
+      "loss": 0.1489,
+      "step": 14242
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.955578919931823e-07,
+      "loss": 0.1749,
+      "step": 14243
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.950506116633678e-07,
+      "loss": 0.1423,
+      "step": 14244
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.94543760523921e-07,
+      "loss": 0.1754,
+      "step": 14245
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.940373385972528e-07,
+      "loss": 0.1747,
+      "step": 14246
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.9353134590576205e-07,
+      "loss": 0.1566,
+      "step": 14247
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.930257824718241e-07,
+      "loss": 0.1826,
+      "step": 14248
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.925206483177967e-07,
+      "loss": 0.178,
+      "step": 14249
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.920159434660208e-07,
+      "loss": 0.1679,
+      "step": 14250
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.9151166793881305e-07,
+      "loss": 0.1754,
+      "step": 14251
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.9100782175847665e-07,
+      "loss": 0.1675,
+      "step": 14252
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.905044049472949e-07,
+      "loss": 0.1596,
+      "step": 14253
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.9000141752752787e-07,
+      "loss": 0.1667,
+      "step": 14254
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.894988595214221e-07,
+      "loss": 0.1583,
+      "step": 14255
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8899673095120096e-07,
+      "loss": 0.1499,
+      "step": 14256
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8849503183907115e-07,
+      "loss": 0.1407,
+      "step": 14257
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.879937622072193e-07,
+      "loss": 0.1408,
+      "step": 14258
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8749292207781443e-07,
+      "loss": 0.1987,
+      "step": 14259
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8699251147300433e-07,
+      "loss": 0.1728,
+      "step": 14260
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.864925304149202e-07,
+      "loss": 0.1679,
+      "step": 14261
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.859929789256732e-07,
+      "loss": 0.1801,
+      "step": 14262
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.854938570273547e-07,
+      "loss": 0.176,
+      "step": 14263
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8499516474203904e-07,
+      "loss": 0.1852,
+      "step": 14264
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.844969020917787e-07,
+      "loss": 0.1479,
+      "step": 14265
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.839990690986094e-07,
+      "loss": 0.1616,
+      "step": 14266
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8350166578454687e-07,
+      "loss": 0.1697,
+      "step": 14267
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.830046921715901e-07,
+      "loss": 0.1755,
+      "step": 14268
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.825081482817138e-07,
+      "loss": 0.1642,
+      "step": 14269
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.820120341368804e-07,
+      "loss": 0.1652,
+      "step": 14270
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8151634975902896e-07,
+      "loss": 0.1648,
+      "step": 14271
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8102109517007867e-07,
+      "loss": 0.1465,
+      "step": 14272
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.805262703919343e-07,
+      "loss": 0.1534,
+      "step": 14273
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8003187544647813e-07,
+      "loss": 0.1694,
+      "step": 14274
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7953791035557174e-07,
+      "loss": 0.1614,
+      "step": 14275
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.790443751410643e-07,
+      "loss": 0.1896,
+      "step": 14276
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7855126982477945e-07,
+      "loss": 0.1743,
+      "step": 14277
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.780585944285241e-07,
+      "loss": 0.1884,
+      "step": 14278
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.775663489740865e-07,
+      "loss": 0.1749,
+      "step": 14279
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7707453348323474e-07,
+      "loss": 0.1835,
+      "step": 14280
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7658314797772255e-07,
+      "loss": 0.1497,
+      "step": 14281
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.76092192479277e-07,
+      "loss": 0.1632,
+      "step": 14282
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.756016670096118e-07,
+      "loss": 0.1489,
+      "step": 14283
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7511157159041845e-07,
+      "loss": 0.1557,
+      "step": 14284
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7462190624337193e-07,
+      "loss": 0.1651,
+      "step": 14285
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7413267099012707e-07,
+      "loss": 0.1561,
+      "step": 14286
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.736438658523211e-07,
+      "loss": 0.1706,
+      "step": 14287
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.731554908515677e-07,
+      "loss": 0.1633,
+      "step": 14288
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7266754600946746e-07,
+      "loss": 0.166,
+      "step": 14289
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.721800313475975e-07,
+      "loss": 0.1801,
+      "step": 14290
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7169294688751955e-07,
+      "loss": 0.1921,
+      "step": 14291
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.712062926507708e-07,
+      "loss": 0.1772,
+      "step": 14292
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7072006865887735e-07,
+      "loss": 0.2056,
+      "step": 14293
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.702342749333398e-07,
+      "loss": 0.1626,
+      "step": 14294
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.6974891149563975e-07,
+      "loss": 0.1947,
+      "step": 14295
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.6926397836724685e-07,
+      "loss": 0.1829,
+      "step": 14296
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.687794755696016e-07,
+      "loss": 0.1445,
+      "step": 14297
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.682954031241336e-07,
+      "loss": 0.162,
+      "step": 14298
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.678117610522502e-07,
+      "loss": 0.1787,
+      "step": 14299
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.673285493753375e-07,
+      "loss": 0.1789,
+      "step": 14300
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.668457681147696e-07,
+      "loss": 0.1499,
+      "step": 14301
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.6636341729189276e-07,
+      "loss": 0.1795,
+      "step": 14302
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.6588149692803875e-07,
+      "loss": 0.1706,
+      "step": 14303
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.6540000704452286e-07,
+      "loss": 0.1654,
+      "step": 14304
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.649189476626357e-07,
+      "loss": 0.1752,
+      "step": 14305
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.6443831880365257e-07,
+      "loss": 0.1604,
+      "step": 14306
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.6395812048882864e-07,
+      "loss": 0.16,
+      "step": 14307
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.6347835273940024e-07,
+      "loss": 0.1669,
+      "step": 14308
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.629990155765838e-07,
+      "loss": 0.1664,
+      "step": 14309
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.62520109021579e-07,
+      "loss": 0.1811,
+      "step": 14310
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.620416330955633e-07,
+      "loss": 0.1623,
+      "step": 14311
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.615635878196976e-07,
+      "loss": 0.1736,
+      "step": 14312
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.610859732151227e-07,
+      "loss": 0.1896,
+      "step": 14313
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.6060878930296077e-07,
+      "loss": 0.1587,
+      "step": 14314
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.6013203610431583e-07,
+      "loss": 0.181,
+      "step": 14315
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5965571364026887e-07,
+      "loss": 0.1899,
+      "step": 14316
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.591798219318875e-07,
+      "loss": 0.157,
+      "step": 14317
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.58704361000216e-07,
+      "loss": 0.1785,
+      "step": 14318
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.582293308662809e-07,
+      "loss": 0.1602,
+      "step": 14319
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5775473155109086e-07,
+      "loss": 0.2003,
+      "step": 14320
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5728056307563476e-07,
+      "loss": 0.1641,
+      "step": 14321
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.568068254608813e-07,
+      "loss": 0.1595,
+      "step": 14322
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5633351872777936e-07,
+      "loss": 0.1774,
+      "step": 14323
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.558606428972643e-07,
+      "loss": 0.1524,
+      "step": 14324
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.553881979902462e-07,
+      "loss": 0.1688,
+      "step": 14325
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5491618402761707e-07,
+      "loss": 0.1905,
+      "step": 14326
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.544446010302548e-07,
+      "loss": 0.1749,
+      "step": 14327
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5397344901901156e-07,
+      "loss": 0.1534,
+      "step": 14328
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5350272801472396e-07,
+      "loss": 0.1909,
+      "step": 14329
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5303243803820985e-07,
+      "loss": 0.1969,
+      "step": 14330
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5256257911026703e-07,
+      "loss": 0.1773,
+      "step": 14331
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5209315125167554e-07,
+      "loss": 0.1373,
+      "step": 14332
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5162415448319323e-07,
+      "loss": 0.1619,
+      "step": 14333
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5115558882556237e-07,
+      "loss": 0.1615,
+      "step": 14334
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5068745429950413e-07,
+      "loss": 0.195,
+      "step": 14335
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5021975092572095e-07,
+      "loss": 0.1657,
+      "step": 14336
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.4975247872489727e-07,
+      "loss": 0.1563,
+      "step": 14337
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.4928563771769777e-07,
+      "loss": 0.1531,
+      "step": 14338
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.488192279247681e-07,
+      "loss": 0.1637,
+      "step": 14339
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.4835324936673287e-07,
+      "loss": 0.1939,
+      "step": 14340
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.478877020642012e-07,
+      "loss": 0.1603,
+      "step": 14341
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.4742258603776214e-07,
+      "loss": 0.1455,
+      "step": 14342
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.469579013079826e-07,
+      "loss": 0.1701,
+      "step": 14343
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.4649364789541385e-07,
+      "loss": 0.1529,
+      "step": 14344
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.460298258205884e-07,
+      "loss": 0.1281,
+      "step": 14345
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.4556643510401436e-07,
+      "loss": 0.1694,
+      "step": 14346
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.451034757661885e-07,
+      "loss": 0.1658,
+      "step": 14347
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.446409478275824e-07,
+      "loss": 0.153,
+      "step": 14348
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.4417885130865295e-07,
+      "loss": 0.1696,
+      "step": 14349
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.4371718622983484e-07,
+      "loss": 0.1512,
+      "step": 14350
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.43255952611543e-07,
+      "loss": 0.1503,
+      "step": 14351
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.427951504741777e-07,
+      "loss": 0.1593,
+      "step": 14352
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.423347798381159e-07,
+      "loss": 0.1521,
+      "step": 14353
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.418748407237159e-07,
+      "loss": 0.1476,
+      "step": 14354
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.414153331513214e-07,
+      "loss": 0.1719,
+      "step": 14355
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.4095625714125046e-07,
+      "loss": 0.167,
+      "step": 14356
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.4049761271380587e-07,
+      "loss": 0.1788,
+      "step": 14357
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.400393998892725e-07,
+      "loss": 0.1531,
+      "step": 14358
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3958161868791185e-07,
+      "loss": 0.1684,
+      "step": 14359
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.391242691299689e-07,
+      "loss": 0.1556,
+      "step": 14360
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3866735123567075e-07,
+      "loss": 0.1545,
+      "step": 14361
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3821086502522462e-07,
+      "loss": 0.1533,
+      "step": 14362
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3775481051881544e-07,
+      "loss": 0.1499,
+      "step": 14363
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3729918773661376e-07,
+      "loss": 0.1649,
+      "step": 14364
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.36843996698769e-07,
+      "loss": 0.1719,
+      "step": 14365
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3638923742541065e-07,
+      "loss": 0.1506,
+      "step": 14366
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3593490993665035e-07,
+      "loss": 0.1369,
+      "step": 14367
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3548101425257987e-07,
+      "loss": 0.1766,
+      "step": 14368
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3502755039327307e-07,
+      "loss": 0.1457,
+      "step": 14369
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3457451837878286e-07,
+      "loss": 0.1538,
+      "step": 14370
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3412191822914542e-07,
+      "loss": 0.1799,
+      "step": 14371
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3366974996437474e-07,
+      "loss": 0.1583,
+      "step": 14372
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3321801360446927e-07,
+      "loss": 0.1779,
+      "step": 14373
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3276670916940303e-07,
+      "loss": 0.152,
+      "step": 14374
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3231583667913893e-07,
+      "loss": 0.1685,
+      "step": 14375
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3186539615361437e-07,
+      "loss": 0.1745,
+      "step": 14376
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3141538761274785e-07,
+      "loss": 0.1425,
+      "step": 14377
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3096581107644455e-07,
+      "loss": 0.1753,
+      "step": 14378
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3051666656458193e-07,
+      "loss": 0.1751,
+      "step": 14379
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.300679540970263e-07,
+      "loss": 0.1863,
+      "step": 14380
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2961967369362071e-07,
+      "loss": 0.1801,
+      "step": 14381
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2917182537418703e-07,
+      "loss": 0.1681,
+      "step": 14382
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2872440915853612e-07,
+      "loss": 0.1583,
+      "step": 14383
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.282774250664499e-07,
+      "loss": 0.1687,
+      "step": 14384
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2783087311769593e-07,
+      "loss": 0.1601,
+      "step": 14385
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2738475333202614e-07,
+      "loss": 0.1871,
+      "step": 14386
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2693906572916703e-07,
+      "loss": 0.1669,
+      "step": 14387
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2649381032882834e-07,
+      "loss": 0.1889,
+      "step": 14388
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.260489871507021e-07,
+      "loss": 0.1835,
+      "step": 14389
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.256045962144593e-07,
+      "loss": 0.172,
+      "step": 14390
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2516063753975303e-07,
+      "loss": 0.1406,
+      "step": 14391
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2471711114621763e-07,
+      "loss": 0.1624,
+      "step": 14392
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.242740170534652e-07,
+      "loss": 0.1571,
+      "step": 14393
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2383135528109334e-07,
+      "loss": 0.1593,
+      "step": 14394
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2338912584867756e-07,
+      "loss": 0.1411,
+      "step": 14395
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2294732877577552e-07,
+      "loss": 0.1384,
+      "step": 14396
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2250596408192383e-07,
+      "loss": 0.1759,
+      "step": 14397
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2206503178664241e-07,
+      "loss": 0.182,
+      "step": 14398
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2162453190943013e-07,
+      "loss": 0.1466,
+      "step": 14399
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2118446446976915e-07,
+      "loss": 0.1946,
+      "step": 14400
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.2074482948711951e-07,
+      "loss": 0.1527,
+      "step": 14401
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.203056269809234e-07,
+      "loss": 0.195,
+      "step": 14402
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1986685697060529e-07,
+      "loss": 0.1802,
+      "step": 14403
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1942851947556853e-07,
+      "loss": 0.1593,
+      "step": 14404
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1899061451519766e-07,
+      "loss": 0.1437,
+      "step": 14405
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1855314210886046e-07,
+      "loss": 0.1638,
+      "step": 14406
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1811610227590152e-07,
+      "loss": 0.1866,
+      "step": 14407
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1767949503564867e-07,
+      "loss": 0.1596,
+      "step": 14408
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1724332040741203e-07,
+      "loss": 0.1604,
+      "step": 14409
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1680757841047838e-07,
+      "loss": 0.1714,
+      "step": 14410
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1637226906412013e-07,
+      "loss": 0.1571,
+      "step": 14411
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1593739238758738e-07,
+      "loss": 0.1695,
+      "step": 14412
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1550294840011144e-07,
+      "loss": 0.1539,
+      "step": 14413
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1506893712090804e-07,
+      "loss": 0.1671,
+      "step": 14414
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.146353585691674e-07,
+      "loss": 0.1681,
+      "step": 14415
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1420221276406528e-07,
+      "loss": 0.1453,
+      "step": 14416
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1376949972475635e-07,
+      "loss": 0.1647,
+      "step": 14417
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1333721947037868e-07,
+      "loss": 0.1496,
+      "step": 14418
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.12905372020048e-07,
+      "loss": 0.1738,
+      "step": 14419
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1247395739286246e-07,
+      "loss": 0.1507,
+      "step": 14420
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.120429756079001e-07,
+      "loss": 0.1883,
+      "step": 14421
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1161242668422232e-07,
+      "loss": 0.1505,
+      "step": 14422
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1118231064086947e-07,
+      "loss": 0.1512,
+      "step": 14423
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1075262749686188e-07,
+      "loss": 0.1639,
+      "step": 14424
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.10323377271201e-07,
+      "loss": 0.1659,
+      "step": 14425
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.098945599828728e-07,
+      "loss": 0.1791,
+      "step": 14426
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0946617565083982e-07,
+      "loss": 0.1581,
+      "step": 14427
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0903822429404475e-07,
+      "loss": 0.1696,
+      "step": 14428
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0861070593141686e-07,
+      "loss": 0.1826,
+      "step": 14429
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0818362058185992e-07,
+      "loss": 0.1689,
+      "step": 14430
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0775696826426217e-07,
+      "loss": 0.1742,
+      "step": 14431
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0733074899749295e-07,
+      "loss": 0.1506,
+      "step": 14432
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.069049628003994e-07,
+      "loss": 0.1639,
+      "step": 14433
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0647960969181314e-07,
+      "loss": 0.1603,
+      "step": 14434
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0605468969054354e-07,
+      "loss": 0.1568,
+      "step": 14435
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0563020281538225e-07,
+      "loss": 0.1498,
+      "step": 14436
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.052061490851043e-07,
+      "loss": 0.1664,
+      "step": 14437
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0478252851846016e-07,
+      "loss": 0.1808,
+      "step": 14438
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0435934113418376e-07,
+      "loss": 0.1506,
+      "step": 14439
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0393658695099239e-07,
+      "loss": 0.1648,
+      "step": 14440
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0351426598757995e-07,
+      "loss": 0.1731,
+      "step": 14441
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0309237826262373e-07,
+      "loss": 0.1656,
+      "step": 14442
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0267092379478103e-07,
+      "loss": 0.1656,
+      "step": 14443
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.022499026026914e-07,
+      "loss": 0.1669,
+      "step": 14444
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0182931470497213e-07,
+      "loss": 0.1806,
+      "step": 14445
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0140916012022394e-07,
+      "loss": 0.1631,
+      "step": 14446
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.009894388670286e-07,
+      "loss": 0.1698,
+      "step": 14447
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0057015096394683e-07,
+      "loss": 0.1654,
+      "step": 14448
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0015129642952158e-07,
+      "loss": 0.1456,
+      "step": 14449
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.997328752822758e-07,
+      "loss": 0.1764,
+      "step": 14450
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.993148875407136e-07,
+      "loss": 0.1584,
+      "step": 14451
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.9889733322332127e-07,
+      "loss": 0.1652,
+      "step": 14452
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9848021234856408e-07,
+      "loss": 0.1612,
+      "step": 14453
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9806352493488722e-07,
+      "loss": 0.1553,
+      "step": 14454
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9764727100072045e-07,
+      "loss": 0.1846,
+      "step": 14455
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9723145056447013e-07,
+      "loss": 0.1746,
+      "step": 14456
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.968160636445282e-07,
+      "loss": 0.1714,
+      "step": 14457
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9640111025926113e-07,
+      "loss": 0.1494,
+      "step": 14458
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9598659042702196e-07,
+      "loss": 0.1758,
+      "step": 14459
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.955725041661427e-07,
+      "loss": 0.1567,
+      "step": 14460
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.951588514949343e-07,
+      "loss": 0.161,
+      "step": 14461
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9474563243169098e-07,
+      "loss": 0.1463,
+      "step": 14462
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.94332846994687e-07,
+      "loss": 0.1841,
+      "step": 14463
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.939204952021756e-07,
+      "loss": 0.1638,
+      "step": 14464
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9350857707239656e-07,
+      "loss": 0.1866,
+      "step": 14465
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9309709262356203e-07,
+      "loss": 0.1645,
+      "step": 14466
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9268604187387185e-07,
+      "loss": 0.1698,
+      "step": 14467
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9227542484150486e-07,
+      "loss": 0.1774,
+      "step": 14468
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.918652415446176e-07,
+      "loss": 0.1603,
+      "step": 14469
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9145549200135226e-07,
+      "loss": 0.1829,
+      "step": 14470
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.910461762298288e-07,
+      "loss": 0.1569,
+      "step": 14471
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9063729424814825e-07,
+      "loss": 0.1671,
+      "step": 14472
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9022884607439395e-07,
+      "loss": 0.1508,
+      "step": 14473
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8982083172662813e-07,
+      "loss": 0.1912,
+      "step": 14474
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8941325122289523e-07,
+      "loss": 0.1521,
+      "step": 14475
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8900610458121971e-07,
+      "loss": 0.1754,
+      "step": 14476
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.885993918196083e-07,
+      "loss": 0.1744,
+      "step": 14477
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.881931129560466e-07,
+      "loss": 0.1643,
+      "step": 14478
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8778726800850022e-07,
+      "loss": 0.1504,
+      "step": 14479
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.873818569949193e-07,
+      "loss": 0.1594,
+      "step": 14480
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8697687993323166e-07,
+      "loss": 0.1687,
+      "step": 14481
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8657233684134857e-07,
+      "loss": 0.1469,
+      "step": 14482
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8616822773715904e-07,
+      "loss": 0.1768,
+      "step": 14483
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.857645526385332e-07,
+      "loss": 0.1611,
+      "step": 14484
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.853613115633257e-07,
+      "loss": 0.1641,
+      "step": 14485
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8495850452936892e-07,
+      "loss": 0.1614,
+      "step": 14486
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8455613155447417e-07,
+      "loss": 0.1512,
+      "step": 14487
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8415419265643942e-07,
+      "loss": 0.1783,
+      "step": 14488
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8375268785303713e-07,
+      "loss": 0.1777,
+      "step": 14489
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.833516171620231e-07,
+      "loss": 0.1678,
+      "step": 14490
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.829509806011387e-07,
+      "loss": 0.1979,
+      "step": 14491
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8255077818809642e-07,
+      "loss": 0.1656,
+      "step": 14492
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8215100994059653e-07,
+      "loss": 0.1722,
+      "step": 14493
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8175167587631936e-07,
+      "loss": 0.1505,
+      "step": 14494
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.813527760129241e-07,
+      "loss": 0.1754,
+      "step": 14495
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8095431036805223e-07,
+      "loss": 0.1501,
+      "step": 14496
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8055627895932516e-07,
+      "loss": 0.1806,
+      "step": 14497
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8015868180434437e-07,
+      "loss": 0.1733,
+      "step": 14498
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.797615189206936e-07,
+      "loss": 0.1726,
+      "step": 14499
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.7936479032593878e-07,
+      "loss": 0.1805,
+      "step": 14500
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.7896849603762257e-07,
+      "loss": 0.1889,
+      "step": 14501
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.7857263607327091e-07,
+      "loss": 0.1707,
+      "step": 14502
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.7817721045039205e-07,
+      "loss": 0.1632,
+      "step": 14503
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7778221918647086e-07,
+      "loss": 0.1539,
+      "step": 14504
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7738766229897676e-07,
+      "loss": 0.1555,
+      "step": 14505
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7699353980535906e-07,
+      "loss": 0.1794,
+      "step": 14506
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7659985172304495e-07,
+      "loss": 0.1678,
+      "step": 14507
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7620659806944718e-07,
+      "loss": 0.1387,
+      "step": 14508
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7581377886195738e-07,
+      "loss": 0.1625,
+      "step": 14509
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7542139411794502e-07,
+      "loss": 0.1723,
+      "step": 14510
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7502944385476505e-07,
+      "loss": 0.1526,
+      "step": 14511
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.746379280897492e-07,
+      "loss": 0.1763,
+      "step": 14512
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7424684684021365e-07,
+      "loss": 0.1756,
+      "step": 14513
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.738562001234534e-07,
+      "loss": 0.1753,
+      "step": 14514
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.734659879567424e-07,
+      "loss": 0.1723,
+      "step": 14515
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7307621035733911e-07,
+      "loss": 0.179,
+      "step": 14516
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7268686734248087e-07,
+      "loss": 0.1349,
+      "step": 14517
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7229795892938382e-07,
+      "loss": 0.1469,
+      "step": 14518
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7190948513525096e-07,
+      "loss": 0.1615,
+      "step": 14519
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7152144597725962e-07,
+      "loss": 0.1554,
+      "step": 14520
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.711338414725694e-07,
+      "loss": 0.1661,
+      "step": 14521
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7074667163832436e-07,
+      "loss": 0.1735,
+      "step": 14522
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7035993649164418e-07,
+      "loss": 0.2053,
+      "step": 14523
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6997363604963291e-07,
+      "loss": 0.1634,
+      "step": 14524
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6958777032937356e-07,
+      "loss": 0.1936,
+      "step": 14525
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6920233934793252e-07,
+      "loss": 0.1638,
+      "step": 14526
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6881734312235277e-07,
+      "loss": 0.1585,
+      "step": 14527
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6843278166966182e-07,
+      "loss": 0.1352,
+      "step": 14528
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6804865500686607e-07,
+      "loss": 0.1532,
+      "step": 14529
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6766496315095192e-07,
+      "loss": 0.1852,
+      "step": 14530
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6728170611888917e-07,
+      "loss": 0.1849,
+      "step": 14531
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6689888392762643e-07,
+      "loss": 0.1955,
+      "step": 14532
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6651649659409462e-07,
+      "loss": 0.1603,
+      "step": 14533
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6613454413520357e-07,
+      "loss": 0.1827,
+      "step": 14534
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6575302656784197e-07,
+      "loss": 0.1833,
+      "step": 14535
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6537194390888632e-07,
+      "loss": 0.1752,
+      "step": 14536
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6499129617518871e-07,
+      "loss": 0.1701,
+      "step": 14537
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6461108338358011e-07,
+      "loss": 0.2023,
+      "step": 14538
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6423130555087818e-07,
+      "loss": 0.1591,
+      "step": 14539
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6385196269387615e-07,
+      "loss": 0.1616,
+      "step": 14540
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.634730548293495e-07,
+      "loss": 0.166,
+      "step": 14541
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6309458197405814e-07,
+      "loss": 0.1794,
+      "step": 14542
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6271654414473646e-07,
+      "loss": 0.1836,
+      "step": 14543
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6233894135810447e-07,
+      "loss": 0.1888,
+      "step": 14544
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6196177363086096e-07,
+      "loss": 0.1689,
+      "step": 14545
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6158504097968375e-07,
+      "loss": 0.1437,
+      "step": 14546
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6120874342123726e-07,
+      "loss": 0.16,
+      "step": 14547
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6083288097216042e-07,
+      "loss": 0.1595,
+      "step": 14548
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6045745364907438e-07,
+      "loss": 0.1758,
+      "step": 14549
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6008246146858365e-07,
+      "loss": 0.1724,
+      "step": 14550
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.5970790444727158e-07,
+      "loss": 0.1643,
+      "step": 14551
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.5933378260170274e-07,
+      "loss": 0.1451,
+      "step": 14552
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.5896009594842167e-07,
+      "loss": 0.1715,
+      "step": 14553
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.58586844503954e-07,
+      "loss": 0.1523,
+      "step": 14554
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5821402828480658e-07,
+      "loss": 0.1383,
+      "step": 14555
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5784164730746842e-07,
+      "loss": 0.1581,
+      "step": 14556
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5746970158840525e-07,
+      "loss": 0.1496,
+      "step": 14557
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5709819114406612e-07,
+      "loss": 0.1711,
+      "step": 14558
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.567271159908823e-07,
+      "loss": 0.142,
+      "step": 14559
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5635647614526407e-07,
+      "loss": 0.1713,
+      "step": 14560
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5598627162359935e-07,
+      "loss": 0.1633,
+      "step": 14561
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5561650244226513e-07,
+      "loss": 0.1616,
+      "step": 14562
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5524716861760935e-07,
+      "loss": 0.1769,
+      "step": 14563
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.548782701659679e-07,
+      "loss": 0.1494,
+      "step": 14564
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5450980710365548e-07,
+      "loss": 0.1732,
+      "step": 14565
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5414177944696462e-07,
+      "loss": 0.1723,
+      "step": 14566
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.537741872121723e-07,
+      "loss": 0.1765,
+      "step": 14567
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.534070304155355e-07,
+      "loss": 0.1326,
+      "step": 14568
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5304030907328903e-07,
+      "loss": 0.1449,
+      "step": 14569
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5267402320165436e-07,
+      "loss": 0.1323,
+      "step": 14570
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.523081728168263e-07,
+      "loss": 0.1816,
+      "step": 14571
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5194275793498635e-07,
+      "loss": 0.1521,
+      "step": 14572
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5157777857229384e-07,
+      "loss": 0.1636,
+      "step": 14573
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5121323474488913e-07,
+      "loss": 0.1512,
+      "step": 14574
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5084912646889604e-07,
+      "loss": 0.1822,
+      "step": 14575
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.504854537604139e-07,
+      "loss": 0.1644,
+      "step": 14576
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.501222166355265e-07,
+      "loss": 0.1557,
+      "step": 14577
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.497594151102999e-07,
+      "loss": 0.1751,
+      "step": 14578
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4939704920077568e-07,
+      "loss": 0.1677,
+      "step": 14579
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4903511892297995e-07,
+      "loss": 0.1859,
+      "step": 14580
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4867362429291877e-07,
+      "loss": 0.1724,
+      "step": 14581
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4831256532657935e-07,
+      "loss": 0.1633,
+      "step": 14582
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.479519420399289e-07,
+      "loss": 0.1675,
+      "step": 14583
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4759175444891473e-07,
+      "loss": 0.177,
+      "step": 14584
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4723200256946624e-07,
+      "loss": 0.1435,
+      "step": 14585
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.468726864174941e-07,
+      "loss": 0.1888,
+      "step": 14586
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4651380600888775e-07,
+      "loss": 0.1535,
+      "step": 14587
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.461553613595179e-07,
+      "loss": 0.1683,
+      "step": 14588
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.457973524852363e-07,
+      "loss": 0.1749,
+      "step": 14589
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4543977940187693e-07,
+      "loss": 0.1691,
+      "step": 14590
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4508264212525047e-07,
+      "loss": 0.1559,
+      "step": 14591
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.447259406711532e-07,
+      "loss": 0.1501,
+      "step": 14592
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4436967505535914e-07,
+      "loss": 0.164,
+      "step": 14593
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.440138452936246e-07,
+      "loss": 0.1617,
+      "step": 14594
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4365845140168367e-07,
+      "loss": 0.1432,
+      "step": 14595
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4330349339525483e-07,
+      "loss": 0.1933,
+      "step": 14596
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4294897129003448e-07,
+      "loss": 0.1461,
+      "step": 14597
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.425948851017023e-07,
+      "loss": 0.1738,
+      "step": 14598
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4224123484591678e-07,
+      "loss": 0.1514,
+      "step": 14599
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4188802053831773e-07,
+      "loss": 0.1576,
+      "step": 14600
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4153524219452597e-07,
+      "loss": 0.1865,
+      "step": 14601
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4118289983014232e-07,
+      "loss": 0.1719,
+      "step": 14602
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4083099346074879e-07,
+      "loss": 0.1813,
+      "step": 14603
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4047952310190737e-07,
+      "loss": 0.1387,
+      "step": 14604
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.401284887691634e-07,
+      "loss": 0.1563,
+      "step": 14605
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.397778904780378e-07,
+      "loss": 0.1562,
+      "step": 14606
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3942772824403926e-07,
+      "loss": 0.1534,
+      "step": 14607
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3907800208264988e-07,
+      "loss": 0.158,
+      "step": 14608
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3872871200933834e-07,
+      "loss": 0.1894,
+      "step": 14609
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3837985803955013e-07,
+      "loss": 0.1795,
+      "step": 14610
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3803144018871395e-07,
+      "loss": 0.1456,
+      "step": 14611
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.376834584722364e-07,
+      "loss": 0.1704,
+      "step": 14612
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.373359129055085e-07,
+      "loss": 0.1533,
+      "step": 14613
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.369888035038991e-07,
+      "loss": 0.1662,
+      "step": 14614
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.366421302827592e-07,
+      "loss": 0.1798,
+      "step": 14615
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3629589325741988e-07,
+      "loss": 0.1541,
+      "step": 14616
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3595009244319224e-07,
+      "loss": 0.1686,
+      "step": 14617
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.356047278553707e-07,
+      "loss": 0.1751,
+      "step": 14618
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3525979950922752e-07,
+      "loss": 0.1858,
+      "step": 14619
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3491530742001602e-07,
+      "loss": 0.1661,
+      "step": 14620
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.345712516029729e-07,
+      "loss": 0.1522,
+      "step": 14621
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3422763207331157e-07,
+      "loss": 0.1688,
+      "step": 14622
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3388444884622876e-07,
+      "loss": 0.2029,
+      "step": 14623
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.335417019369034e-07,
+      "loss": 0.1752,
+      "step": 14624
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.331993913604912e-07,
+      "loss": 0.1972,
+      "step": 14625
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3285751713213003e-07,
+      "loss": 0.1727,
+      "step": 14626
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3251607926693888e-07,
+      "loss": 0.1538,
+      "step": 14627
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3217507778001903e-07,
+      "loss": 0.1782,
+      "step": 14628
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3183451268645063e-07,
+      "loss": 0.1619,
+      "step": 14629
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3149438400129277e-07,
+      "loss": 0.1834,
+      "step": 14630
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3115469173959005e-07,
+      "loss": 0.1652,
+      "step": 14631
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3081543591636269e-07,
+      "loss": 0.1776,
+      "step": 14632
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3047661654661426e-07,
+      "loss": 0.1665,
+      "step": 14633
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3013823364533052e-07,
+      "loss": 0.1741,
+      "step": 14634
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2980028722747395e-07,
+      "loss": 0.1569,
+      "step": 14635
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.294627773079904e-07,
+      "loss": 0.1757,
+      "step": 14636
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2912570390180678e-07,
+      "loss": 0.146,
+      "step": 14637
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2878906702382786e-07,
+      "loss": 0.1529,
+      "step": 14638
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2845286668894397e-07,
+      "loss": 0.1607,
+      "step": 14639
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2811710291201984e-07,
+      "loss": 0.1778,
+      "step": 14640
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2778177570790694e-07,
+      "loss": 0.186,
+      "step": 14641
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2744688509143344e-07,
+      "loss": 0.1728,
+      "step": 14642
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2711243107740855e-07,
+      "loss": 0.2041,
+      "step": 14643
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2677841368062603e-07,
+      "loss": 0.1335,
+      "step": 14644
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2644483291585519e-07,
+      "loss": 0.1379,
+      "step": 14645
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.261116887978475e-07,
+      "loss": 0.1571,
+      "step": 14646
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2577898134133905e-07,
+      "loss": 0.1757,
+      "step": 14647
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.254467105610402e-07,
+      "loss": 0.172,
+      "step": 14648
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.251148764716481e-07,
+      "loss": 0.1569,
+      "step": 14649
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2478347908783438e-07,
+      "loss": 0.162,
+      "step": 14650
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.244525184242562e-07,
+      "loss": 0.1761,
+      "step": 14651
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2412199449555296e-07,
+      "loss": 0.1522,
+      "step": 14652
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2379190731633627e-07,
+      "loss": 0.1807,
+      "step": 14653
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2346225690120782e-07,
+      "loss": 0.1636,
+      "step": 14654
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2313304326474485e-07,
+      "loss": 0.1586,
+      "step": 14655
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2280426642150679e-07,
+      "loss": 0.1713,
+      "step": 14656
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.22475926386032e-07,
+      "loss": 0.1648,
+      "step": 14657
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.2214802317284225e-07,
+      "loss": 0.1605,
+      "step": 14658
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.2182055679643924e-07,
+      "loss": 0.1417,
+      "step": 14659
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.2149352727130248e-07,
+      "loss": 0.1563,
+      "step": 14660
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.211669346118971e-07,
+      "loss": 0.1813,
+      "step": 14661
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.2084077883266486e-07,
+      "loss": 0.1738,
+      "step": 14662
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.2051505994802981e-07,
+      "loss": 0.1586,
+      "step": 14663
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.2018977797239706e-07,
+      "loss": 0.1693,
+      "step": 14664
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1986493292015178e-07,
+      "loss": 0.1717,
+      "step": 14665
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1954052480565803e-07,
+      "loss": 0.1676,
+      "step": 14666
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1921655364326546e-07,
+      "loss": 0.1823,
+      "step": 14667
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1889301944729814e-07,
+      "loss": 0.1584,
+      "step": 14668
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1856992223206687e-07,
+      "loss": 0.1869,
+      "step": 14669
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.182472620118591e-07,
+      "loss": 0.156,
+      "step": 14670
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1792503880094231e-07,
+      "loss": 0.1921,
+      "step": 14671
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1760325261357064e-07,
+      "loss": 0.1699,
+      "step": 14672
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.172819034639705e-07,
+      "loss": 0.1482,
+      "step": 14673
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1696099136635498e-07,
+      "loss": 0.1812,
+      "step": 14674
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1664051633491602e-07,
+      "loss": 0.1577,
+      "step": 14675
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1632047838382676e-07,
+      "loss": 0.1517,
+      "step": 14676
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1600087752724031e-07,
+      "loss": 0.1754,
+      "step": 14677
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1568171377928872e-07,
+      "loss": 0.1887,
+      "step": 14678
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1536298715408844e-07,
+      "loss": 0.1576,
+      "step": 14679
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1504469766573601e-07,
+      "loss": 0.1439,
+      "step": 14680
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.147268453283057e-07,
+      "loss": 0.1691,
+      "step": 14681
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1440943015585404e-07,
+      "loss": 0.2035,
+      "step": 14682
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1409245216241871e-07,
+      "loss": 0.1892,
+      "step": 14683
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1377591136201849e-07,
+      "loss": 0.1517,
+      "step": 14684
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1345980776864996e-07,
+      "loss": 0.1866,
+      "step": 14685
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1314414139629526e-07,
+      "loss": 0.1493,
+      "step": 14686
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1282891225891212e-07,
+      "loss": 0.1756,
+      "step": 14687
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1251412037044162e-07,
+      "loss": 0.1554,
+      "step": 14688
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1219976574480706e-07,
+      "loss": 0.1591,
+      "step": 14689
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1188584839590843e-07,
+      "loss": 0.1643,
+      "step": 14690
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1157236833762686e-07,
+      "loss": 0.1614,
+      "step": 14691
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1125932558382902e-07,
+      "loss": 0.1596,
+      "step": 14692
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1094672014835717e-07,
+      "loss": 0.1683,
+      "step": 14693
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1063455204503582e-07,
+      "loss": 0.1564,
+      "step": 14694
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1032282128767169e-07,
+      "loss": 0.1795,
+      "step": 14695
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.100115278900482e-07,
+      "loss": 0.1663,
+      "step": 14696
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0970067186593325e-07,
+      "loss": 0.1717,
+      "step": 14697
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0939025322907582e-07,
+      "loss": 0.16,
+      "step": 14698
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0908027199320048e-07,
+      "loss": 0.1565,
+      "step": 14699
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0877072817201961e-07,
+      "loss": 0.1643,
+      "step": 14700
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0846162177921893e-07,
+      "loss": 0.1421,
+      "step": 14701
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.081529528284686e-07,
+      "loss": 0.1695,
+      "step": 14702
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0784472133342216e-07,
+      "loss": 0.1538,
+      "step": 14703
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.075369273077087e-07,
+      "loss": 0.1481,
+      "step": 14704
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0722957076493956e-07,
+      "loss": 0.1724,
+      "step": 14705
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.069226517187083e-07,
+      "loss": 0.1725,
+      "step": 14706
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0661617018258851e-07,
+      "loss": 0.1727,
+      "step": 14707
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.063101261701316e-07,
+      "loss": 0.1502,
+      "step": 14708
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0600451969487447e-07,
+      "loss": 0.1678,
+      "step": 14709
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0569935077033189e-07,
+      "loss": 0.184,
+      "step": 14710
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0539461940999863e-07,
+      "loss": 0.1659,
+      "step": 14711
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0509032562735167e-07,
+      "loss": 0.1471,
+      "step": 14712
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0478646943584692e-07,
+      "loss": 0.1609,
+      "step": 14713
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0448305084892474e-07,
+      "loss": 0.1365,
+      "step": 14714
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0418006987999996e-07,
+      "loss": 0.1699,
+      "step": 14715
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0387752654247518e-07,
+      "loss": 0.1593,
+      "step": 14716
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0357542084972749e-07,
+      "loss": 0.1602,
+      "step": 14717
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0327375281511842e-07,
+      "loss": 0.1667,
+      "step": 14718
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.029725224519873e-07,
+      "loss": 0.1577,
+      "step": 14719
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0267172977365791e-07,
+      "loss": 0.1789,
+      "step": 14720
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0237137479343185e-07,
+      "loss": 0.1627,
+      "step": 14721
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0207145752458958e-07,
+      "loss": 0.1719,
+      "step": 14722
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.017719779803994e-07,
+      "loss": 0.1569,
+      "step": 14723
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.014729361741007e-07,
+      "loss": 0.1456,
+      "step": 14724
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0117433211891958e-07,
+      "loss": 0.1624,
+      "step": 14725
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0087616582806326e-07,
+      "loss": 0.1501,
+      "step": 14726
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0057843731471673e-07,
+      "loss": 0.1676,
+      "step": 14727
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0028114659204613e-07,
+      "loss": 0.1515,
+      "step": 14728
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.998429367319873e-08,
+      "loss": 0.1672,
+      "step": 14729
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.968787857130291e-08,
+      "loss": 0.1514,
+      "step": 14730
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.93919012994693e-08,
+      "loss": 0.1529,
+      "step": 14731
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.909636187078297e-08,
+      "loss": 0.1448,
+      "step": 14732
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.880126029831794e-08,
+      "loss": 0.1594,
+      "step": 14733
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.850659659512152e-08,
+      "loss": 0.1768,
+      "step": 14734
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.821237077422663e-08,
+      "loss": 0.1631,
+      "step": 14735
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.791858284864509e-08,
+      "loss": 0.1711,
+      "step": 14736
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.762523283136871e-08,
+      "loss": 0.1459,
+      "step": 14737
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.733232073537047e-08,
+      "loss": 0.1415,
+      "step": 14738
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.703984657360555e-08,
+      "loss": 0.1559,
+      "step": 14739
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.674781035900582e-08,
+      "loss": 0.1627,
+      "step": 14740
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.645621210448874e-08,
+      "loss": 0.1733,
+      "step": 14741
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.616505182294733e-08,
+      "loss": 0.1607,
+      "step": 14742
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.587432952726017e-08,
+      "loss": 0.1914,
+      "step": 14743
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.558404523028364e-08,
+      "loss": 0.1785,
+      "step": 14744
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.529419894485415e-08,
+      "loss": 0.1778,
+      "step": 14745
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.500479068379031e-08,
+      "loss": 0.1524,
+      "step": 14746
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.471582045989191e-08,
+      "loss": 0.167,
+      "step": 14747
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.44272882859365e-08,
+      "loss": 0.1528,
+      "step": 14748
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.413919417468609e-08,
+      "loss": 0.1577,
+      "step": 14749
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.385153813887827e-08,
+      "loss": 0.1534,
+      "step": 14750
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.35643201912384e-08,
+      "loss": 0.1729,
+      "step": 14751
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.327754034446524e-08,
+      "loss": 0.1718,
+      "step": 14752
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.299119861124196e-08,
+      "loss": 0.1849,
+      "step": 14753
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.270529500423398e-08,
+      "loss": 0.1534,
+      "step": 14754
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.24198295360812e-08,
+      "loss": 0.18,
+      "step": 14755
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.213480221941018e-08,
+      "loss": 0.1771,
+      "step": 14756
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.18502130668264e-08,
+      "loss": 0.167,
+      "step": 14757
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.156606209091423e-08,
+      "loss": 0.1622,
+      "step": 14758
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.128234930424029e-08,
+      "loss": 0.1705,
+      "step": 14759
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.099907471935232e-08,
+      "loss": 0.1538,
+      "step": 14760
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 9.071623834877697e-08,
+      "loss": 0.1772,
+      "step": 14761
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 9.0433840205022e-08,
+      "loss": 0.1453,
+      "step": 14762
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 9.015188030057742e-08,
+      "loss": 0.1633,
+      "step": 14763
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.987035864791105e-08,
+      "loss": 0.1615,
+      "step": 14764
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.958927525947514e-08,
+      "loss": 0.1728,
+      "step": 14765
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.930863014769752e-08,
+      "loss": 0.1539,
+      "step": 14766
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.902842332499051e-08,
+      "loss": 0.1615,
+      "step": 14767
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.87486548037464e-08,
+      "loss": 0.1685,
+      "step": 14768
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.84693245963375e-08,
+      "loss": 0.1567,
+      "step": 14769
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.81904327151173e-08,
+      "loss": 0.1667,
+      "step": 14770
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.791197917241811e-08,
+      "loss": 0.1486,
+      "step": 14771
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.763396398055567e-08,
+      "loss": 0.1835,
+      "step": 14772
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.735638715182349e-08,
+      "loss": 0.164,
+      "step": 14773
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.707924869849837e-08,
+      "loss": 0.1652,
+      "step": 14774
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.68025486328361e-08,
+      "loss": 0.1563,
+      "step": 14775
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.652628696707244e-08,
+      "loss": 0.1607,
+      "step": 14776
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.625046371342649e-08,
+      "loss": 0.1658,
+      "step": 14777
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.597507888409406e-08,
+      "loss": 0.1904,
+      "step": 14778
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.570013249125541e-08,
+      "loss": 0.1754,
+      "step": 14779
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.54256245470697e-08,
+      "loss": 0.1736,
+      "step": 14780
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.515155506367501e-08,
+      "loss": 0.1461,
+      "step": 14781
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.487792405319494e-08,
+      "loss": 0.1751,
+      "step": 14782
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.460473152772764e-08,
+      "loss": 0.181,
+      "step": 14783
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.433197749935562e-08,
+      "loss": 0.1496,
+      "step": 14784
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.405966198014149e-08,
+      "loss": 0.1666,
+      "step": 14785
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.378778498212781e-08,
+      "loss": 0.1875,
+      "step": 14786
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.35163465173372e-08,
+      "loss": 0.1727,
+      "step": 14787
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.324534659777561e-08,
+      "loss": 0.1927,
+      "step": 14788
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.297478523542568e-08,
+      "loss": 0.1662,
+      "step": 14789
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.27046624422545e-08,
+      "loss": 0.1784,
+      "step": 14790
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.243497823020697e-08,
+      "loss": 0.168,
+      "step": 14791
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.216573261120908e-08,
+      "loss": 0.1836,
+      "step": 14792
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.189692559716911e-08,
+      "loss": 0.1759,
+      "step": 14793
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.16285571999742e-08,
+      "loss": 0.1534,
+      "step": 14794
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.136062743149153e-08,
+      "loss": 0.1718,
+      "step": 14795
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.109313630357163e-08,
+      "loss": 0.1664,
+      "step": 14796
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.08260838280439e-08,
+      "loss": 0.1725,
+      "step": 14797
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.055947001671782e-08,
+      "loss": 0.1625,
+      "step": 14798
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.029329488138393e-08,
+      "loss": 0.1751,
+      "step": 14799
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.002755843381394e-08,
+      "loss": 0.181,
+      "step": 14800
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.976226068575954e-08,
+      "loss": 0.1655,
+      "step": 14801
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.949740164895359e-08,
+      "loss": 0.1632,
+      "step": 14802
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.923298133510892e-08,
+      "loss": 0.1634,
+      "step": 14803
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.896899975591732e-08,
+      "loss": 0.1703,
+      "step": 14804
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.87054569230572e-08,
+      "loss": 0.1852,
+      "step": 14805
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.844235284818036e-08,
+      "loss": 0.1696,
+      "step": 14806
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.817968754292304e-08,
+      "loss": 0.1769,
+      "step": 14807
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.791746101890151e-08,
+      "loss": 0.1472,
+      "step": 14808
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.765567328771206e-08,
+      "loss": 0.17,
+      "step": 14809
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.739432436093208e-08,
+      "loss": 0.1745,
+      "step": 14810
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.713341425012011e-08,
+      "loss": 0.1963,
+      "step": 14811
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.687294296681246e-08,
+      "loss": 0.1925,
+      "step": 14812
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.661291052253217e-08,
+      "loss": 0.1767,
+      "step": 14813
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.635331692877446e-08,
+      "loss": 0.1805,
+      "step": 14814
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.609416219702236e-08,
+      "loss": 0.191,
+      "step": 14815
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.583544633873563e-08,
+      "loss": 0.197,
+      "step": 14816
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.55771693653562e-08,
+      "loss": 0.1801,
+      "step": 14817
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.531933128830604e-08,
+      "loss": 0.1532,
+      "step": 14818
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.506193211898716e-08,
+      "loss": 0.1583,
+      "step": 14819
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.480497186878266e-08,
+      "loss": 0.1771,
+      "step": 14820
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.45484505490579e-08,
+      "loss": 0.1521,
+      "step": 14821
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.429236817115493e-08,
+      "loss": 0.181,
+      "step": 14822
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.403672474640134e-08,
+      "loss": 0.1691,
+      "step": 14823
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.378152028609919e-08,
+      "loss": 0.1664,
+      "step": 14824
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.352675480153837e-08,
+      "loss": 0.1627,
+      "step": 14825
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.327242830398318e-08,
+      "loss": 0.1774,
+      "step": 14826
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.30185408046813e-08,
+      "loss": 0.1753,
+      "step": 14827
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.27650923148615e-08,
+      "loss": 0.1497,
+      "step": 14828
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.251208284573152e-08,
+      "loss": 0.1576,
+      "step": 14829
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.225951240848017e-08,
+      "loss": 0.154,
+      "step": 14830
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.200738101427851e-08,
+      "loss": 0.175,
+      "step": 14831
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.17556886742743e-08,
+      "loss": 0.1525,
+      "step": 14832
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.150443539960084e-08,
+      "loss": 0.1716,
+      "step": 14833
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.125362120136815e-08,
+      "loss": 0.1809,
+      "step": 14834
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.100324609066733e-08,
+      "loss": 0.1734,
+      "step": 14835
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.0753310078574e-08,
+      "loss": 0.1939,
+      "step": 14836
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.050381317613819e-08,
+      "loss": 0.1471,
+      "step": 14837
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.025475539439552e-08,
+      "loss": 0.1848,
+      "step": 14838
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.000613674435941e-08,
+      "loss": 0.1828,
+      "step": 14839
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.975795723702549e-08,
+      "loss": 0.2015,
+      "step": 14840
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.951021688336723e-08,
+      "loss": 0.1673,
+      "step": 14841
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.926291569434362e-08,
+      "loss": 0.1827,
+      "step": 14842
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.901605368088926e-08,
+      "loss": 0.1627,
+      "step": 14843
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.876963085392096e-08,
+      "loss": 0.1637,
+      "step": 14844
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.85236472243378e-08,
+      "loss": 0.1772,
+      "step": 14845
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.827810280301884e-08,
+      "loss": 0.185,
+      "step": 14846
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.803299760082094e-08,
+      "loss": 0.1697,
+      "step": 14847
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.778833162858434e-08,
+      "loss": 0.174,
+      "step": 14848
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.754410489712926e-08,
+      "loss": 0.1383,
+      "step": 14849
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.730031741725707e-08,
+      "loss": 0.1727,
+      "step": 14850
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.705696919974802e-08,
+      "loss": 0.1736,
+      "step": 14851
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.681406025536352e-08,
+      "loss": 0.1644,
+      "step": 14852
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.65715905948472e-08,
+      "loss": 0.1787,
+      "step": 14853
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.632956022892156e-08,
+      "loss": 0.1561,
+      "step": 14854
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.608796916828919e-08,
+      "loss": 0.1569,
+      "step": 14855
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.584681742363596e-08,
+      "loss": 0.1588,
+      "step": 14856
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.560610500562559e-08,
+      "loss": 0.171,
+      "step": 14857
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.536583192490176e-08,
+      "loss": 0.1725,
+      "step": 14858
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.512599819209375e-08,
+      "loss": 0.1684,
+      "step": 14859
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.48866038178042e-08,
+      "loss": 0.1771,
+      "step": 14860
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.464764881262354e-08,
+      "loss": 0.141,
+      "step": 14861
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.440913318711662e-08,
+      "loss": 0.1543,
+      "step": 14862
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.417105695183168e-08,
+      "loss": 0.1713,
+      "step": 14863
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.39334201172992e-08,
+      "loss": 0.1793,
+      "step": 14864
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.369622269402742e-08,
+      "loss": 0.1676,
+      "step": 14865
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.345946469250575e-08,
+      "loss": 0.164,
+      "step": 14866
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.322314612320468e-08,
+      "loss": 0.1613,
+      "step": 14867
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.298726699657587e-08,
+      "loss": 0.1481,
+      "step": 14868
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.275182732304986e-08,
+      "loss": 0.1605,
+      "step": 14869
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.25168271130383e-08,
+      "loss": 0.1458,
+      "step": 14870
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.228226637693624e-08,
+      "loss": 0.169,
+      "step": 14871
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.204814512511425e-08,
+      "loss": 0.1786,
+      "step": 14872
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.18144633679263e-08,
+      "loss": 0.1727,
+      "step": 14873
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.158122111570852e-08,
+      "loss": 0.1765,
+      "step": 14874
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.134841837877492e-08,
+      "loss": 0.1833,
+      "step": 14875
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.111605516741948e-08,
+      "loss": 0.1616,
+      "step": 14876
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.088413149192063e-08,
+      "loss": 0.1557,
+      "step": 14877
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.065264736253351e-08,
+      "loss": 0.1832,
+      "step": 14878
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.042160278949549e-08,
+      "loss": 0.1804,
+      "step": 14879
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.019099778302395e-08,
+      "loss": 0.1561,
+      "step": 14880
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.99608323533174e-08,
+      "loss": 0.151,
+      "step": 14881
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.973110651055547e-08,
+      "loss": 0.1518,
+      "step": 14882
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.950182026489559e-08,
+      "loss": 0.1683,
+      "step": 14883
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.927297362647855e-08,
+      "loss": 0.1676,
+      "step": 14884
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.904456660542623e-08,
+      "loss": 0.1843,
+      "step": 14885
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.881659921183835e-08,
+      "loss": 0.1547,
+      "step": 14886
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.858907145579684e-08,
+      "loss": 0.1528,
+      "step": 14887
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.836198334736254e-08,
+      "loss": 0.1509,
+      "step": 14888
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.8135334896579635e-08,
+      "loss": 0.1571,
+      "step": 14889
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.7909126113472325e-08,
+      "loss": 0.1522,
+      "step": 14890
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.7683357008042617e-08,
+      "loss": 0.1681,
+      "step": 14891
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.7458027590275856e-08,
+      "loss": 0.1769,
+      "step": 14892
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.723313787013518e-08,
+      "loss": 0.1487,
+      "step": 14893
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.70086878575693e-08,
+      "loss": 0.1683,
+      "step": 14894
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.6784677562501387e-08,
+      "loss": 0.1627,
+      "step": 14895
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.656110699483908e-08,
+      "loss": 0.1559,
+      "step": 14896
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.633797616447001e-08,
+      "loss": 0.157,
+      "step": 14897
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.611528508126074e-08,
+      "loss": 0.1542,
+      "step": 14898
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.589303375506117e-08,
+      "loss": 0.1671,
+      "step": 14899
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.567122219569898e-08,
+      "loss": 0.1579,
+      "step": 14900
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.5449850412983006e-08,
+      "loss": 0.1751,
+      "step": 14901
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.52289184167043e-08,
+      "loss": 0.1793,
+      "step": 14902
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.500842621663394e-08,
+      "loss": 0.1604,
+      "step": 14903
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.478837382252078e-08,
+      "loss": 0.1616,
+      "step": 14904
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.456876124409816e-08,
+      "loss": 0.1842,
+      "step": 14905
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.4349588491077186e-08,
+      "loss": 0.1606,
+      "step": 14906
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.413085557315123e-08,
+      "loss": 0.1658,
+      "step": 14907
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.391256249999255e-08,
+      "loss": 0.1749,
+      "step": 14908
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.3694709281255644e-08,
+      "loss": 0.1826,
+      "step": 14909
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.347729592657613e-08,
+      "loss": 0.1763,
+      "step": 14910
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.3260322445566335e-08,
+      "loss": 0.2035,
+      "step": 14911
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.3043788847821906e-08,
+      "loss": 0.1673,
+      "step": 14912
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.2827695142921855e-08,
+      "loss": 0.1539,
+      "step": 14913
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.261204134041853e-08,
+      "loss": 0.1639,
+      "step": 14914
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.239682744985208e-08,
+      "loss": 0.1964,
+      "step": 14915
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.218205348073824e-08,
+      "loss": 0.1628,
+      "step": 14916
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.196771944257606e-08,
+      "loss": 0.1758,
+      "step": 14917
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.175382534484463e-08,
+      "loss": 0.1938,
+      "step": 14918
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.154037119700195e-08,
+      "loss": 0.1774,
+      "step": 14919
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.132735700848934e-08,
+      "loss": 0.1354,
+      "step": 14920
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.1114782788724835e-08,
+      "loss": 0.167,
+      "step": 14921
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.0902648547112024e-08,
+      "loss": 0.1496,
+      "step": 14922
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.069095429303006e-08,
+      "loss": 0.1818,
+      "step": 14923
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.047970003584146e-08,
+      "loss": 0.18,
+      "step": 14924
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.0268885784889864e-08,
+      "loss": 0.1811,
+      "step": 14925
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.00585115494967e-08,
+      "loss": 0.1715,
+      "step": 14926
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.9848577338966756e-08,
+      "loss": 0.1855,
+      "step": 14927
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.963908316258259e-08,
+      "loss": 0.171,
+      "step": 14928
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.9430029029611245e-08,
+      "loss": 0.1558,
+      "step": 14929
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.9221414949295335e-08,
+      "loss": 0.1876,
+      "step": 14930
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.901324093086302e-08,
+      "loss": 0.1847,
+      "step": 14931
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.880550698351805e-08,
+      "loss": 0.1681,
+      "step": 14932
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.8598213116449747e-08,
+      "loss": 0.1868,
+      "step": 14933
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.8391359338823e-08,
+      "loss": 0.171,
+      "step": 14934
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.8184945659788264e-08,
+      "loss": 0.1779,
+      "step": 14935
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.797897208847158e-08,
+      "loss": 0.158,
+      "step": 14936
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.777343863398232e-08,
+      "loss": 0.1784,
+      "step": 14937
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.75683453054121e-08,
+      "loss": 0.1565,
+      "step": 14938
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.7363692111828116e-08,
+      "loss": 0.1806,
+      "step": 14939
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.715947906228202e-08,
+      "loss": 0.1739,
+      "step": 14940
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.695570616580658e-08,
+      "loss": 0.1792,
+      "step": 14941
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.6752373431411255e-08,
+      "loss": 0.1724,
+      "step": 14942
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.654948086808775e-08,
+      "loss": 0.1458,
+      "step": 14943
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.6347028484811093e-08,
+      "loss": 0.1698,
+      "step": 14944
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.6145016290531916e-08,
+      "loss": 0.1661,
+      "step": 14945
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.59434442941864e-08,
+      "loss": 0.1687,
+      "step": 14946
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.574231250468852e-08,
+      "loss": 0.1712,
+      "step": 14947
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.554162093093117e-08,
+      "loss": 0.1648,
+      "step": 14948
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.5341369581790586e-08,
+      "loss": 0.1678,
+      "step": 14949
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.5141558466124116e-08,
+      "loss": 0.176,
+      "step": 14950
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.494218759276581e-08,
+      "loss": 0.1673,
+      "step": 14951
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.4743256970534164e-08,
+      "loss": 0.1635,
+      "step": 14952
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.45447666082266e-08,
+      "loss": 0.1629,
+      "step": 14953
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.434671651462053e-08,
+      "loss": 0.1898,
+      "step": 14954
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.4149106698475605e-08,
+      "loss": 0.1567,
+      "step": 14955
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.395193716853041e-08,
+      "loss": 0.1663,
+      "step": 14956
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.3755207933502405e-08,
+      "loss": 0.1634,
+      "step": 14957
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.355891900209464e-08,
+      "loss": 0.1633,
+      "step": 14958
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.336307038298682e-08,
+      "loss": 0.1723,
+      "step": 14959
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.3167662084839804e-08,
+      "loss": 0.1614,
+      "step": 14960
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.2972694116295564e-08,
+      "loss": 0.1576,
+      "step": 14961
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.27781664859761e-08,
+      "loss": 0.1721,
+      "step": 14962
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.258407920248453e-08,
+      "loss": 0.1505,
+      "step": 14963
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.239043227440398e-08,
+      "loss": 0.1721,
+      "step": 14964
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.219722571029761e-08,
+      "loss": 0.1757,
+      "step": 14965
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.2004459518710795e-08,
+      "loss": 0.1459,
+      "step": 14966
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.1812133708168947e-08,
+      "loss": 0.1646,
+      "step": 14967
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.1620248287175256e-08,
+      "loss": 0.1718,
+      "step": 14968
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.142880326421739e-08,
+      "loss": 0.1768,
+      "step": 14969
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.123779864776079e-08,
+      "loss": 0.1736,
+      "step": 14970
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.104723444625314e-08,
+      "loss": 0.1597,
+      "step": 14971
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.085711066812215e-08,
+      "loss": 0.1434,
+      "step": 14972
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.066742732177442e-08,
+      "loss": 0.158,
+      "step": 14973
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.047818441559992e-08,
+      "loss": 0.1625,
+      "step": 14974
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.02893819579675e-08,
+      "loss": 0.1693,
+      "step": 14975
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.0101019957226037e-08,
+      "loss": 0.1708,
+      "step": 14976
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.9913098421706655e-08,
+      "loss": 0.1744,
+      "step": 14977
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.9725617359719356e-08,
+      "loss": 0.1841,
+      "step": 14978
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.953857677955419e-08,
+      "loss": 0.1617,
+      "step": 14979
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.935197668948454e-08,
+      "loss": 0.1745,
+      "step": 14980
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.916581709776268e-08,
+      "loss": 0.1683,
+      "step": 14981
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.898009801261982e-08,
+      "loss": 0.1892,
+      "step": 14982
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.879481944227048e-08,
+      "loss": 0.1705,
+      "step": 14983
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.860998139490701e-08,
+      "loss": 0.1605,
+      "step": 14984
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.8425583878705096e-08,
+      "loss": 0.1602,
+      "step": 14985
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.824162690181821e-08,
+      "loss": 0.1817,
+      "step": 14986
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.805811047238317e-08,
+      "loss": 0.1717,
+      "step": 14987
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.78750345985146e-08,
+      "loss": 0.1777,
+      "step": 14988
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.7692399288309365e-08,
+      "loss": 0.1462,
+      "step": 14989
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.751020454984322e-08,
+      "loss": 0.1675,
+      "step": 14990
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.732845039117527e-08,
+      "loss": 0.1769,
+      "step": 14991
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.7147136820342435e-08,
+      "loss": 0.1837,
+      "step": 14992
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.696626384536273e-08,
+      "loss": 0.153,
+      "step": 14993
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.678583147423642e-08,
+      "loss": 0.179,
+      "step": 14994
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.660583971494158e-08,
+      "loss": 0.1581,
+      "step": 14995
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.6426288575437395e-08,
+      "loss": 0.1578,
+      "step": 14996
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.624717806366751e-08,
+      "loss": 0.1701,
+      "step": 14997
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.6068508187550034e-08,
+      "loss": 0.1781,
+      "step": 14998
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.5890278954987536e-08,
+      "loss": 0.1669,
+      "step": 14999
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.571249037386038e-08,
+      "loss": 0.1555,
+      "step": 15000
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.553514245203338e-08,
+      "loss": 0.1725,
+      "step": 15001
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.535823519734916e-08,
+      "loss": 0.1563,
+      "step": 15002
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.518176861763034e-08,
+      "loss": 0.1835,
+      "step": 15003
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.500574272068069e-08,
+      "loss": 0.1736,
+      "step": 15004
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.48301575142862e-08,
+      "loss": 0.183,
+      "step": 15005
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.465501300621066e-08,
+      "loss": 0.1739,
+      "step": 15006
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.448030920420009e-08,
+      "loss": 0.1683,
+      "step": 15007
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.430604611598054e-08,
+      "loss": 0.1463,
+      "step": 15008
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.413222374925807e-08,
+      "loss": 0.1641,
+      "step": 15009
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.3958842111720955e-08,
+      "loss": 0.1549,
+      "step": 15010
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.378590121103642e-08,
+      "loss": 0.1872,
+      "step": 15011
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.3613401054851667e-08,
+      "loss": 0.1686,
+      "step": 15012
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.3441341650796153e-08,
+      "loss": 0.1599,
+      "step": 15013
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.326972300647824e-08,
+      "loss": 0.1395,
+      "step": 15014
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.309854512948962e-08,
+      "loss": 0.1735,
+      "step": 15015
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.292780802739759e-08,
+      "loss": 0.167,
+      "step": 15016
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.275751170775498e-08,
+      "loss": 0.1666,
+      "step": 15017
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.258765617809134e-08,
+      "loss": 0.1653,
+      "step": 15018
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.241824144591954e-08,
+      "loss": 0.1706,
+      "step": 15019
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.2249267518732475e-08,
+      "loss": 0.1767,
+      "step": 15020
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.2080734403999725e-08,
+      "loss": 0.1852,
+      "step": 15021
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.191264210917755e-08,
+      "loss": 0.1695,
+      "step": 15022
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.1744990641698894e-08,
+      "loss": 0.1735,
+      "step": 15023
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.157778000897671e-08,
+      "loss": 0.1771,
+      "step": 15024
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.141101021840731e-08,
+      "loss": 0.1669,
+      "step": 15025
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.12446812773648e-08,
+      "loss": 0.1592,
+      "step": 15026
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.10787931932055e-08,
+      "loss": 0.1831,
+      "step": 15027
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.0913345973264674e-08,
+      "loss": 0.1691,
+      "step": 15028
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.07483396248609e-08,
+      "loss": 0.1515,
+      "step": 15029
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.058377415528946e-08,
+      "loss": 0.1675,
+      "step": 15030
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.0419649571828966e-08,
+      "loss": 0.1959,
+      "step": 15031
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.025596588173807e-08,
+      "loss": 0.1483,
+      "step": 15032
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.009272309225431e-08,
+      "loss": 0.1486,
+      "step": 15033
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.9929921210598565e-08,
+      "loss": 0.1706,
+      "step": 15034
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.976756024396843e-08,
+      "loss": 0.1575,
+      "step": 15035
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.9605640199545926e-08,
+      "loss": 0.1705,
+      "step": 15036
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.9444161084490885e-08,
+      "loss": 0.1562,
+      "step": 15037
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.928312290594537e-08,
+      "loss": 0.1864,
+      "step": 15038
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.9122525671029245e-08,
+      "loss": 0.1568,
+      "step": 15039
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.8962369386846823e-08,
+      "loss": 0.1788,
+      "step": 15040
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.8802654060480216e-08,
+      "loss": 0.2028,
+      "step": 15041
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.8643379698992667e-08,
+      "loss": 0.1742,
+      "step": 15042
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.848454630942743e-08,
+      "loss": 0.1776,
+      "step": 15043
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.8326153898808883e-08,
+      "loss": 0.1779,
+      "step": 15044
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.8168202474141425e-08,
+      "loss": 0.1792,
+      "step": 15045
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.801069204241169e-08,
+      "loss": 0.1611,
+      "step": 15046
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.7853622610584108e-08,
+      "loss": 0.1502,
+      "step": 15047
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.769699418560534e-08,
+      "loss": 0.1748,
+      "step": 15048
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.7540806774400963e-08,
+      "loss": 0.1629,
+      "step": 15049
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.738506038387989e-08,
+      "loss": 0.1845,
+      "step": 15050
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.722975502092995e-08,
+      "loss": 0.1833,
+      "step": 15051
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.7074890692417866e-08,
+      "loss": 0.154,
+      "step": 15052
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.692046740519261e-08,
+      "loss": 0.175,
+      "step": 15053
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.6766485166084267e-08,
+      "loss": 0.1784,
+      "step": 15054
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.661294398190184e-08,
+      "loss": 0.1732,
+      "step": 15055
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.6459843859436563e-08,
+      "loss": 0.1984,
+      "step": 15056
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.6307184805457464e-08,
+      "loss": 0.1679,
+      "step": 15057
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.6154966826716922e-08,
+      "loss": 0.1744,
+      "step": 15058
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.6003189929945107e-08,
+      "loss": 0.1761,
+      "step": 15059
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.5851854121856647e-08,
+      "loss": 0.16,
+      "step": 15060
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.5700959409141745e-08,
+      "loss": 0.1677,
+      "step": 15061
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.5550505798475066e-08,
+      "loss": 0.1907,
+      "step": 15062
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.5400493296510177e-08,
+      "loss": 0.1691,
+      "step": 15063
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.5250921909880655e-08,
+      "loss": 0.1587,
+      "step": 15064
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.5101791645201214e-08,
+      "loss": 0.1819,
+      "step": 15065
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.4953102509067683e-08,
+      "loss": 0.2024,
+      "step": 15066
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.48048545080537e-08,
+      "loss": 0.1512,
+      "step": 15067
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.4657047648717347e-08,
+      "loss": 0.17,
+      "step": 15068
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.450968193759562e-08,
+      "loss": 0.1862,
+      "step": 15069
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.4362757381203307e-08,
+      "loss": 0.1661,
+      "step": 15070
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.421627398603965e-08,
+      "loss": 0.1482,
+      "step": 15071
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.4070231758582806e-08,
+      "loss": 0.1486,
+      "step": 15072
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.392463070528983e-08,
+      "loss": 0.1809,
+      "step": 15073
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.3779470832602237e-08,
+      "loss": 0.1304,
+      "step": 15074
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.363475214693711e-08,
+      "loss": 0.1724,
+      "step": 15075
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.3490474654694896e-08,
+      "loss": 0.1696,
+      "step": 15076
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.3346638362257146e-08,
+      "loss": 0.1702,
+      "step": 15077
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.3203243275984332e-08,
+      "loss": 0.1739,
+      "step": 15078
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.306028940221805e-08,
+      "loss": 0.1481,
+      "step": 15079
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.291777674728102e-08,
+      "loss": 0.1754,
+      "step": 15080
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.2775705317473752e-08,
+      "loss": 0.1659,
+      "step": 15081
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.2634075119080114e-08,
+      "loss": 0.1459,
+      "step": 15082
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.249288615836509e-08,
+      "loss": 0.1687,
+      "step": 15083
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.235213844157036e-08,
+      "loss": 0.1854,
+      "step": 15084
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.221183197492094e-08,
+      "loss": 0.1659,
+      "step": 15085
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.2071966764621866e-08,
+      "loss": 0.1673,
+      "step": 15086
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.1932542816859303e-08,
+      "loss": 0.1632,
+      "step": 15087
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.179356013779832e-08,
+      "loss": 0.1779,
+      "step": 15088
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.1655018733586218e-08,
+      "loss": 0.1886,
+      "step": 15089
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.1516918610348103e-08,
+      "loss": 0.1706,
+      "step": 15090
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.1379259774192417e-08,
+      "loss": 0.157,
+      "step": 15091
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.1242042231206516e-08,
+      "loss": 0.1869,
+      "step": 15092
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.1105265987458878e-08,
+      "loss": 0.1597,
+      "step": 15093
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.0968931048999108e-08,
+      "loss": 0.1613,
+      "step": 15094
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.083303742185461e-08,
+      "loss": 0.1814,
+      "step": 15095
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.0697585112037234e-08,
+      "loss": 0.159,
+      "step": 15096
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.0562574125535527e-08,
+      "loss": 0.1703,
+      "step": 15097
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.0428004468321384e-08,
+      "loss": 0.1741,
+      "step": 15098
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.029387614634559e-08,
+      "loss": 0.1592,
+      "step": 15099
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.0160189165538968e-08,
+      "loss": 0.1543,
+      "step": 15100
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.002694353181456e-08,
+      "loss": 0.1812,
+      "step": 15101
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9894139251065427e-08,
+      "loss": 0.1929,
+      "step": 15102
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.976177632916354e-08,
+      "loss": 0.1546,
+      "step": 15103
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9629854771963108e-08,
+      "loss": 0.2044,
+      "step": 15104
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.949837458529835e-08,
+      "loss": 0.1448,
+      "step": 15105
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.936733577498462e-08,
+      "loss": 0.1926,
+      "step": 15106
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9236738346815054e-08,
+      "loss": 0.1537,
+      "step": 15107
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9106582306566148e-08,
+      "loss": 0.1751,
+      "step": 15108
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.897686765999329e-08,
+      "loss": 0.1618,
+      "step": 15109
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.8847594412834125e-08,
+      "loss": 0.1573,
+      "step": 15110
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.8718762570805184e-08,
+      "loss": 0.1587,
+      "step": 15111
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.859037213960302e-08,
+      "loss": 0.1588,
+      "step": 15112
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.8462423124907534e-08,
+      "loss": 0.1363,
+      "step": 15113
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.8334915532374208e-08,
+      "loss": 0.1679,
+      "step": 15114
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.8207849367645193e-08,
+      "loss": 0.1589,
+      "step": 15115
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.8081224636337102e-08,
+      "loss": 0.1506,
+      "step": 15116
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.7955041344052128e-08,
+      "loss": 0.1644,
+      "step": 15117
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.7829299496368024e-08,
+      "loss": 0.1603,
+      "step": 15118
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.7703999098848123e-08,
+      "loss": 0.175,
+      "step": 15119
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.7579140157032437e-08,
+      "loss": 0.1574,
+      "step": 15120
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.7454722676442104e-08,
+      "loss": 0.1646,
+      "step": 15121
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.7330746662580498e-08,
+      "loss": 0.1898,
+      "step": 15122
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.7207212120928794e-08,
+      "loss": 0.1574,
+      "step": 15123
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.708411905695262e-08,
+      "loss": 0.1736,
+      "step": 15124
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.6961467476093174e-08,
+      "loss": 0.1989,
+      "step": 15125
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.6839257383776118e-08,
+      "loss": 0.1691,
+      "step": 15126
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.6717488785406023e-08,
+      "loss": 0.1495,
+      "step": 15127
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.6596161686366352e-08,
+      "loss": 0.1578,
+      "step": 15128
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.6475276092023928e-08,
+      "loss": 0.2267,
+      "step": 15129
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.6354832007725585e-08,
+      "loss": 0.171,
+      "step": 15130
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.6234829438797062e-08,
+      "loss": 0.1662,
+      "step": 15131
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.6115268390544113e-08,
+      "loss": 0.1703,
+      "step": 15132
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.5996148868255845e-08,
+      "loss": 0.1488,
+      "step": 15133
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.5877470877199154e-08,
+      "loss": 0.1647,
+      "step": 15134
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.5759234422623172e-08,
+      "loss": 0.1528,
+      "step": 15135
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.564143950975705e-08,
+      "loss": 0.1774,
+      "step": 15136
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.5524086143808848e-08,
+      "loss": 0.1722,
+      "step": 15137
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.5407174329968854e-08,
+      "loss": 0.1577,
+      "step": 15138
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.529070407340849e-08,
+      "loss": 0.1386,
+      "step": 15139
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.517467537927697e-08,
+      "loss": 0.1577,
+      "step": 15140
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.5059088252705746e-08,
+      "loss": 0.1663,
+      "step": 15141
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.49439426988085e-08,
+      "loss": 0.145,
+      "step": 15142
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.48292387226745e-08,
+      "loss": 0.175,
+      "step": 15143
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4714976329378572e-08,
+      "loss": 0.1699,
+      "step": 15144
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4601155523972233e-08,
+      "loss": 0.1574,
+      "step": 15145
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4487776311490343e-08,
+      "loss": 0.1897,
+      "step": 15146
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4374838696946669e-08,
+      "loss": 0.1684,
+      "step": 15147
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4262342685336105e-08,
+      "loss": 0.185,
+      "step": 15148
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4150288281632451e-08,
+      "loss": 0.1611,
+      "step": 15149
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4038675490791743e-08,
+      "loss": 0.1495,
+      "step": 15150
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3927504317751139e-08,
+      "loss": 0.1507,
+      "step": 15151
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3816774767424491e-08,
+      "loss": 0.17,
+      "step": 15152
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3706486844711208e-08,
+      "loss": 0.1963,
+      "step": 15153
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3596640554487395e-08,
+      "loss": 0.1588,
+      "step": 15154
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3487235901611384e-08,
+      "loss": 0.17,
+      "step": 15155
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3378272890921529e-08,
+      "loss": 0.1666,
+      "step": 15156
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3269751527235086e-08,
+      "loss": 0.2071,
+      "step": 15157
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3161671815353771e-08,
+      "loss": 0.179,
+      "step": 15158
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3054033760054874e-08,
+      "loss": 0.1628,
+      "step": 15159
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.294683736610014e-08,
+      "loss": 0.1486,
+      "step": 15160
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.2840082638230223e-08,
+      "loss": 0.1603,
+      "step": 15161
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.2733769581164678e-08,
+      "loss": 0.1693,
+      "step": 15162
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.2627898199607524e-08,
+      "loss": 0.1559,
+      "step": 15163
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.2522468498238349e-08,
+      "loss": 0.1513,
+      "step": 15164
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.24174804817212e-08,
+      "loss": 0.1704,
+      "step": 15165
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.231293415469903e-08,
+      "loss": 0.1554,
+      "step": 15166
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.2208829521793697e-08,
+      "loss": 0.1759,
+      "step": 15167
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.2105166587611516e-08,
+      "loss": 0.1665,
+      "step": 15168
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.2001945356735489e-08,
+      "loss": 0.1742,
+      "step": 15169
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.1899165833729741e-08,
+      "loss": 0.1718,
+      "step": 15170
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.1796828023141749e-08,
+      "loss": 0.1686,
+      "step": 15171
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.1694931929494557e-08,
+      "loss": 0.2031,
+      "step": 15172
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.1593477557295674e-08,
+      "loss": 0.1536,
+      "step": 15173
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.149246491103373e-08,
+      "loss": 0.1513,
+      "step": 15174
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.1391893995172931e-08,
+      "loss": 0.1695,
+      "step": 15175
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.1291764814161943e-08,
+      "loss": 0.16,
+      "step": 15176
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.1192077372429443e-08,
+      "loss": 0.1797,
+      "step": 15177
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.109283167438302e-08,
+      "loss": 0.1624,
+      "step": 15178
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0994027724412493e-08,
+      "loss": 0.1646,
+      "step": 15179
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0895665526887699e-08,
+      "loss": 0.1711,
+      "step": 15180
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0797745086157386e-08,
+      "loss": 0.2023,
+      "step": 15181
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0700266406552528e-08,
+      "loss": 0.148,
+      "step": 15182
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0603229492384126e-08,
+      "loss": 0.1818,
+      "step": 15183
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.05066343479443e-08,
+      "loss": 0.1764,
+      "step": 15184
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0410480977502968e-08,
+      "loss": 0.1667,
+      "step": 15185
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0314769385313394e-08,
+      "loss": 0.1582,
+      "step": 15186
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.021949957560886e-08,
+      "loss": 0.1731,
+      "step": 15187
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0124671552602661e-08,
+      "loss": 0.1678,
+      "step": 15188
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0030285320486998e-08,
+      "loss": 0.1582,
+      "step": 15189
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.936340883437424e-09,
+      "loss": 0.1477,
+      "step": 15190
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.842838245607279e-09,
+      "loss": 0.1518,
+      "step": 15191
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.749777411133254e-09,
+      "loss": 0.1601,
+      "step": 15192
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.65715838412984e-09,
+      "loss": 0.1525,
+      "step": 15193
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.564981168692644e-09,
+      "loss": 0.1832,
+      "step": 15194
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.47324576889841e-09,
+      "loss": 0.1682,
+      "step": 15195
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.381952188805e-09,
+      "loss": 0.1765,
+      "step": 15196
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.291100432448074e-09,
+      "loss": 0.1677,
+      "step": 15197
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.200690503846643e-09,
+      "loss": 0.1557,
+      "step": 15198
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.110722406998619e-09,
+      "loss": 0.167,
+      "step": 15199
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.021196145883038e-09,
+      "loss": 0.1569,
+      "step": 15200
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.932111724457847e-09,
+      "loss": 0.1682,
+      "step": 15201
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.843469146664341e-09,
+      "loss": 0.1802,
+      "step": 15202
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.755268416422712e-09,
+      "loss": 0.1892,
+      "step": 15203
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.667509537632069e-09,
+      "loss": 0.1712,
+      "step": 15204
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.580192514173746e-09,
+      "loss": 0.171,
+      "step": 15205
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.493317349910213e-09,
+      "loss": 0.1584,
+      "step": 15206
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.40688404868284e-09,
+      "loss": 0.1477,
+      "step": 15207
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.320892614315235e-09,
+      "loss": 0.1837,
+      "step": 15208
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.235343050607692e-09,
+      "loss": 0.179,
+      "step": 15209
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.15023536134496e-09,
+      "loss": 0.163,
+      "step": 15210
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.065569550291808e-09,
+      "loss": 0.1805,
+      "step": 15211
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.981345621190795e-09,
+      "loss": 0.1719,
+      "step": 15212
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.89756357776783e-09,
+      "loss": 0.149,
+      "step": 15213
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.814223423726619e-09,
+      "loss": 0.1739,
+      "step": 15214
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.73132516275532e-09,
+      "loss": 0.1658,
+      "step": 15215
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.648868798517672e-09,
+      "loss": 0.1569,
+      "step": 15216
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.566854334660756e-09,
+      "loss": 0.1577,
+      "step": 15217
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.48528177481167e-09,
+      "loss": 0.1778,
+      "step": 15218
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.40415112257864e-09,
+      "loss": 0.1856,
+      "step": 15219
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.323462381548796e-09,
+      "loss": 0.1735,
+      "step": 15220
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.243215555290395e-09,
+      "loss": 0.1795,
+      "step": 15221
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.16341064735393e-09,
+      "loss": 0.1754,
+      "step": 15222
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.08404766126547e-09,
+      "loss": 0.1812,
+      "step": 15223
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.0051266005377596e-09,
+      "loss": 0.1589,
+      "step": 15224
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.92664746865912e-09,
+      "loss": 0.1365,
+      "step": 15225
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.848610269101219e-09,
+      "loss": 0.1617,
+      "step": 15226
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.771015005314629e-09,
+      "loss": 0.1433,
+      "step": 15227
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.693861680731051e-09,
+      "loss": 0.1633,
+      "step": 15228
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.6171502987633085e-09,
+      "loss": 0.166,
+      "step": 15229
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.5408808628020234e-09,
+      "loss": 0.2038,
+      "step": 15230
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.465053376222275e-09,
+      "loss": 0.1801,
+      "step": 15231
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.389667842375824e-09,
+      "loss": 0.1594,
+      "step": 15232
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.314724264596672e-09,
+      "loss": 0.1669,
+      "step": 15233
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.240222646199945e-09,
+      "loss": 0.1849,
+      "step": 15234
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.166162990480784e-09,
+      "loss": 0.1377,
+      "step": 15235
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.0925453007132375e-09,
+      "loss": 0.1585,
+      "step": 15236
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.019369580152479e-09,
+      "loss": 0.1869,
+      "step": 15237
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.9466358320359184e-09,
+      "loss": 0.1816,
+      "step": 15238
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.874344059579873e-09,
+      "loss": 0.1671,
+      "step": 15239
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.802494265981784e-09,
+      "loss": 0.1684,
+      "step": 15240
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.731086454418e-09,
+      "loss": 0.2138,
+      "step": 15241
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.660120628045995e-09,
+      "loss": 0.1563,
+      "step": 15242
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.589596790006591e-09,
+      "loss": 0.1536,
+      "step": 15243
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.5195149434161825e-09,
+      "loss": 0.1734,
+      "step": 15244
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.449875091375623e-09,
+      "loss": 0.1617,
+      "step": 15245
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.380677236963561e-09,
+      "loss": 0.1665,
+      "step": 15246
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.311921383240881e-09,
+      "loss": 0.16,
+      "step": 15247
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.243607533247374e-09,
+      "loss": 0.1844,
+      "step": 15248
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.175735690006178e-09,
+      "loss": 0.1597,
+      "step": 15249
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.108305856517115e-09,
+      "loss": 0.1763,
+      "step": 15250
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.041318035761134e-09,
+      "loss": 0.1703,
+      "step": 15251
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.974772230703639e-09,
+      "loss": 0.1822,
+      "step": 15252
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.9086684442856135e-09,
+      "loss": 0.1832,
+      "step": 15253
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.8430066794313835e-09,
+      "loss": 0.1938,
+      "step": 15254
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.777786939043072e-09,
+      "loss": 0.1406,
+      "step": 15255
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.713009226007259e-09,
+      "loss": 0.1696,
+      "step": 15256
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.6486735431861e-09,
+      "loss": 0.1589,
+      "step": 15257
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.584779893427316e-09,
+      "loss": 0.1638,
+      "step": 15258
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.521328279555315e-09,
+      "loss": 0.1634,
+      "step": 15259
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.458318704375631e-09,
+      "loss": 0.1617,
+      "step": 15260
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.395751170676032e-09,
+      "loss": 0.1787,
+      "step": 15261
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.333625681222087e-09,
+      "loss": 0.1451,
+      "step": 15262
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.271942238762705e-09,
+      "loss": 0.1664,
+      "step": 15263
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.210700846024596e-09,
+      "loss": 0.1851,
+      "step": 15264
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.149901505716702e-09,
+      "loss": 0.1541,
+      "step": 15265
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.089544220527986e-09,
+      "loss": 0.1546,
+      "step": 15266
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.0296289931274215e-09,
+      "loss": 0.145,
+      "step": 15267
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.970155826164002e-09,
+      "loss": 0.1686,
+      "step": 15268
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.9111247222689556e-09,
+      "loss": 0.1729,
+      "step": 15269
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.852535684052417e-09,
+      "loss": 0.1677,
+      "step": 15270
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.794388714104535e-09,
+      "loss": 0.1709,
+      "step": 15271
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.736683814998809e-09,
+      "loss": 0.1939,
+      "step": 15272
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.6794209892854203e-09,
+      "loss": 0.1411,
+      "step": 15273
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.6226002394978978e-09,
+      "loss": 0.1557,
+      "step": 15274
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.5662215681475653e-09,
+      "loss": 0.168,
+      "step": 15275
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.510284977729095e-09,
+      "loss": 0.1701,
+      "step": 15276
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.454790470716063e-09,
+      "loss": 0.1711,
+      "step": 15277
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.3997380495620623e-09,
+      "loss": 0.159,
+      "step": 15278
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.3451277167018127e-09,
+      "loss": 0.174,
+      "step": 15279
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.2909594745511588e-09,
+      "loss": 0.1628,
+      "step": 15280
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.2372333255048517e-09,
+      "loss": 0.1608,
+      "step": 15281
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.183949271939879e-09,
+      "loss": 0.1789,
+      "step": 15282
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.131107316211024e-09,
+      "loss": 0.165,
+      "step": 15283
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.078707460656416e-09,
+      "loss": 0.1634,
+      "step": 15284
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.0267497075919803e-09,
+      "loss": 0.1972,
+      "step": 15285
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.9752340593180994e-09,
+      "loss": 0.1924,
+      "step": 15286
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.92416051811073e-09,
+      "loss": 0.1659,
+      "step": 15287
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.8735290862291764e-09,
+      "loss": 0.1582,
+      "step": 15288
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.823339765911648e-09,
+      "loss": 0.1477,
+      "step": 15289
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.773592559379701e-09,
+      "loss": 0.1821,
+      "step": 15290
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.724287468832687e-09,
+      "loss": 0.1712,
+      "step": 15291
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.675424496449974e-09,
+      "loss": 0.1772,
+      "step": 15292
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.6270036443931668e-09,
+      "loss": 0.1653,
+      "step": 15293
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.5790249148038847e-09,
+      "loss": 0.1518,
+      "step": 15294
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.5314883098037647e-09,
+      "loss": 0.1482,
+      "step": 15295
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.4843938314944584e-09,
+      "loss": 0.1548,
+      "step": 15296
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.437741481959854e-09,
+      "loss": 0.1747,
+      "step": 15297
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.3915312632627475e-09,
+      "loss": 0.1779,
+      "step": 15298
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.3457631774448375e-09,
+      "loss": 0.1808,
+      "step": 15299
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.3004372265333918e-09,
+      "loss": 0.1668,
+      "step": 15300
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.2555534125301427e-09,
+      "loss": 0.1754,
+      "step": 15301
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.211111737422389e-09,
+      "loss": 0.1781,
+      "step": 15302
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.167112203173005e-09,
+      "loss": 0.141,
+      "step": 15303
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.1235548117293227e-09,
+      "loss": 0.139,
+      "step": 15304
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.080439565017578e-09,
+      "loss": 0.1636,
+      "step": 15305
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.0377664649440242e-09,
+      "loss": 0.172,
+      "step": 15306
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.995535513397151e-09,
+      "loss": 0.1449,
+      "step": 15307
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.953746712242133e-09,
+      "loss": 0.1668,
+      "step": 15308
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.9124000633297113e-09,
+      "loss": 0.1559,
+      "step": 15309
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8714955684862036e-09,
+      "loss": 0.1396,
+      "step": 15310
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8310332295212729e-09,
+      "loss": 0.1617,
+      "step": 15311
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.7910130482257093e-09,
+      "loss": 0.1765,
+      "step": 15312
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.7514350263669877e-09,
+      "loss": 0.1923,
+      "step": 15313
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.71229916569704e-09,
+      "loss": 0.178,
+      "step": 15314
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.6736054679455938e-09,
+      "loss": 0.1861,
+      "step": 15315
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.6353539348246129e-09,
+      "loss": 0.1731,
+      "step": 15316
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.5975445680260771e-09,
+      "loss": 0.1513,
+      "step": 15317
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.5601773692208721e-09,
+      "loss": 0.1634,
+      "step": 15318
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.5232523400632305e-09,
+      "loss": 0.1636,
+      "step": 15319
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4867694821840695e-09,
+      "loss": 0.1788,
+      "step": 15320
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4507287971976536e-09,
+      "loss": 0.1864,
+      "step": 15321
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.415130286698263e-09,
+      "loss": 0.1906,
+      "step": 15322
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.3799739522590838e-09,
+      "loss": 0.2046,
+      "step": 15323
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.3452597954366486e-09,
+      "loss": 0.1628,
+      "step": 15324
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.3109878177652858e-09,
+      "loss": 0.1739,
+      "step": 15325
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.2771580207593392e-09,
+      "loss": 0.1632,
+      "step": 15326
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.2437704059176103e-09,
+      "loss": 0.1584,
+      "step": 15327
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.2108249747133649e-09,
+      "loss": 0.1655,
+      "step": 15328
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.1783217286065462e-09,
+      "loss": 0.1698,
+      "step": 15329
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.1462606690326727e-09,
+      "loss": 0.1804,
+      "step": 15330
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.1146417974106095e-09,
+      "loss": 0.1691,
+      "step": 15331
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.083465115138127e-09,
+      "loss": 0.1785,
+      "step": 15332
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.0527306235941225e-09,
+      "loss": 0.1708,
+      "step": 15333
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.0224383241386193e-09,
+      "loss": 0.1509,
+      "step": 15334
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 9.92588218109436e-10,
+      "loss": 0.1549,
+      "step": 15335
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 9.631803068277379e-10,
+      "loss": 0.1835,
+      "step": 15336
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 9.342145915947066e-10,
+      "loss": 0.1654,
+      "step": 15337
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 9.05691073690429e-10,
+      "loss": 0.157,
+      "step": 15338
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.776097543761186e-10,
+      "loss": 0.1762,
+      "step": 15339
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.49970634894115e-10,
+      "loss": 0.1604,
+      "step": 15340
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.22773716467884e-10,
+      "loss": 0.1815,
+      "step": 15341
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.960190002975765e-10,
+      "loss": 0.1774,
+      "step": 15342
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.697064875678006e-10,
+      "loss": 0.1638,
+      "step": 15343
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.438361794420701e-10,
+      "loss": 0.1536,
+      "step": 15344
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.184080770650248e-10,
+      "loss": 0.1352,
+      "step": 15345
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.934221815613206e-10,
+      "loss": 0.1569,
+      "step": 15346
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.688784940345194e-10,
+      "loss": 0.1626,
+      "step": 15347
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.447770155715293e-10,
+      "loss": 0.1697,
+      "step": 15348
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.211177472370544e-10,
+      "loss": 0.176,
+      "step": 15349
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.979006900769246e-10,
+      "loss": 0.1632,
+      "step": 15350
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.751258451203168e-10,
+      "loss": 0.1656,
+      "step": 15351
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.527932133719827e-10,
+      "loss": 0.1667,
+      "step": 15352
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.309027958211311e-10,
+      "loss": 0.192,
+      "step": 15353
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.094545934347661e-10,
+      "loss": 0.2007,
+      "step": 15354
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.884486071632388e-10,
+      "loss": 0.1738,
+      "step": 15355
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.678848379324752e-10,
+      "loss": 0.185,
+      "step": 15356
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.477632866539683e-10,
+      "loss": 0.1632,
+      "step": 15357
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.28083954218117e-10,
+      "loss": 0.1592,
+      "step": 15358
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.0884684149311573e-10,
+      "loss": 0.1681,
+      "step": 15359
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.9005194933161575e-10,
+      "loss": 0.168,
+      "step": 15360
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.7169927856406385e-10,
+      "loss": 0.1645,
+      "step": 15361
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.537888300009229e-10,
+      "loss": 0.1566,
+      "step": 15362
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.363206044371126e-10,
+      "loss": 0.13,
+      "step": 15363
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.1929460264201737e-10,
+      "loss": 0.1549,
+      "step": 15364
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.0271082537058905e-10,
+      "loss": 0.1608,
+      "step": 15365
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.865692733555747e-10,
+      "loss": 0.1558,
+      "step": 15366
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.708699473108478e-10,
+      "loss": 0.176,
+      "step": 15367
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.5561284793140796e-10,
+      "loss": 0.1419,
+      "step": 15368
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.407979758900503e-10,
+      "loss": 0.1545,
+      "step": 15369
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.2642533184402683e-10,
+      "loss": 0.177,
+      "step": 15370
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.1249491642838516e-10,
+      "loss": 0.1691,
+      "step": 15371
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.9900673025818885e-10,
+      "loss": 0.1742,
+      "step": 15372
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8596077393073786e-10,
+      "loss": 0.148,
+      "step": 15373
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.733570480233482e-10,
+      "loss": 0.1649,
+      "step": 15374
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.6119555309335179e-10,
+      "loss": 0.1677,
+      "step": 15375
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.4947628967698636e-10,
+      "loss": 0.1804,
+      "step": 15376
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.3819925829494652e-10,
+      "loss": 0.1841,
+      "step": 15377
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.2736445944461217e-10,
+      "loss": 0.1407,
+      "step": 15378
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.1697189360448946e-10,
+      "loss": 0.1512,
+      "step": 15379
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.0702156123532093e-10,
+      "loss": 0.1793,
+      "step": 15380
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 9.751346277675488e-11,
+      "loss": 0.1579,
+      "step": 15381
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 8.844759864956587e-11,
+      "loss": 0.1651,
+      "step": 15382
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.982396925343417e-11,
+      "loss": 0.1724,
+      "step": 15383
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.164257497138672e-11,
+      "loss": 0.1589,
+      "step": 15384
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 6.390341616535622e-11,
+      "loss": 0.1529,
+      "step": 15385
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 5.6606493176181164e-11,
+      "loss": 0.1517,
+      "step": 15386
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.9751806326936434e-11,
+      "loss": 0.1694,
+      "step": 15387
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.333935592071292e-11,
+      "loss": 0.1714,
+      "step": 15388
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.736914224172772e-11,
+      "loss": 0.1578,
+      "step": 15389
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.18411655542139e-11,
+      "loss": 0.1683,
+      "step": 15390
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.675542610131032e-11,
+      "loss": 0.1666,
+      "step": 15391
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.211192410950247e-11,
+      "loss": 0.1632,
+      "step": 15392
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.7910659783071383e-11,
+      "loss": 0.1644,
+      "step": 15393
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.4151633307424307e-11,
+      "loss": 0.1424,
+      "step": 15394
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.0834844850204918e-11,
+      "loss": 0.148,
+      "step": 15395
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.960294556852433e-12,
+      "loss": 0.1321,
+      "step": 15396
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 5.527982555042499e-12,
+      "loss": 0.1718,
+      "step": 15397
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.5379089524667508e-12,
+      "loss": 0.2021,
+      "step": 15398
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.9900738368328066e-12,
+      "loss": 0.1615,
+      "step": 15399
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 8.844772769744936e-13,
+      "loss": 0.175,
+      "step": 15400
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.211193217416252e-13,
+      "loss": 0.1494,
+      "step": 15401
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0,
+      "loss": 0.1207,
+      "step": 15402
+    },
+    {
+      "epoch": 3.0,
+      "step": 15402,
+      "total_flos": 1.000413420838296e+19,
+      "train_loss": 0.2018385535608428,
+      "train_runtime": 75011.5183,
+      "train_samples_per_second": 26.278,
+      "train_steps_per_second": 0.205
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 15402,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 500,
+  "total_flos": 1.000413420838296e+19,
+  "train_batch_size": 16,
+  "trial_name": null,
+  "trial_params": null
+}