diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,75664 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.0267009811806336,
+  "global_step": 12600,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.7869907076483205e-08,
+      "loss": 1.3311,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.573981415296641e-08,
+      "loss": 1.2747,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.360972122944961e-08,
+      "loss": 1.3177,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.147962830593282e-08,
+      "loss": 1.3363,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.934953538241601e-08,
+      "loss": 1.3094,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0721944245889921e-07,
+      "loss": 1.3446,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2508934953538243e-07,
+      "loss": 1.3245,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4295925661186564e-07,
+      "loss": 1.314,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6082916368834881e-07,
+      "loss": 1.3228,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.7869907076483202e-07,
+      "loss": 1.356,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9656897784131522e-07,
+      "loss": 1.3309,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1443888491779843e-07,
+      "loss": 1.3708,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3230879199428163e-07,
+      "loss": 1.3475,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.5017869907076486e-07,
+      "loss": 1.3228,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6804860614724804e-07,
+      "loss": 1.318,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.859185132237313e-07,
+      "loss": 1.3065,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.0378842030021445e-07,
+      "loss": 1.3126,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.2165832737669763e-07,
+      "loss": 1.3531,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.3952823445318086e-07,
+      "loss": 1.3228,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.5739814152966404e-07,
+      "loss": 1.3044,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.7526804860614727e-07,
+      "loss": 1.2969,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.9313795568263045e-07,
+      "loss": 1.3571,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.110078627591137e-07,
+      "loss": 1.3193,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.2887776983559686e-07,
+      "loss": 1.3143,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.467476769120801e-07,
+      "loss": 1.3104,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.6461758398856327e-07,
+      "loss": 1.2896,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.824874910650465e-07,
+      "loss": 1.3158,
+      "step": 27
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.003573981415297e-07,
+      "loss": 1.3124,
+      "step": 28
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.182273052180129e-07,
+      "loss": 1.325,
+      "step": 29
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.360972122944961e-07,
+      "loss": 1.2971,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.539671193709793e-07,
+      "loss": 1.3398,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.718370264474625e-07,
+      "loss": 1.3366,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.897069335239457e-07,
+      "loss": 1.2915,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.075768406004289e-07,
+      "loss": 1.2777,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.25446747676912e-07,
+      "loss": 1.3366,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.433166547533953e-07,
+      "loss": 1.3109,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.611865618298786e-07,
+      "loss": 1.3322,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.790564689063617e-07,
+      "loss": 1.3109,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.96926375982845e-07,
+      "loss": 1.2986,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.147962830593281e-07,
+      "loss": 1.259,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.326661901358113e-07,
+      "loss": 1.2982,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.505360972122945e-07,
+      "loss": 1.3381,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.684060042887778e-07,
+      "loss": 1.3006,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.862759113652609e-07,
+      "loss": 1.3631,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.041458184417441e-07,
+      "loss": 1.3011,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.220157255182274e-07,
+      "loss": 1.334,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.398856325947106e-07,
+      "loss": 1.3089,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.577555396711937e-07,
+      "loss": 1.3338,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.756254467476769e-07,
+      "loss": 1.3062,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.934953538241602e-07,
+      "loss": 1.2835,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.113652609006434e-07,
+      "loss": 1.3544,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.292351679771265e-07,
+      "loss": 1.2955,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.471050750536098e-07,
+      "loss": 1.3373,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.64974982130093e-07,
+      "loss": 1.3418,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.828448892065761e-07,
+      "loss": 1.2992,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0007147962830595e-06,
+      "loss": 1.33,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0185847033595426e-06,
+      "loss": 1.3172,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0364546104360257e-06,
+      "loss": 1.3297,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.054324517512509e-06,
+      "loss": 1.3191,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0721944245889922e-06,
+      "loss": 1.3161,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0900643316654753e-06,
+      "loss": 1.352,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1079342387419586e-06,
+      "loss": 1.3045,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1258041458184418e-06,
+      "loss": 1.3481,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.143674052894925e-06,
+      "loss": 1.3327,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1615439599714082e-06,
+      "loss": 1.3112,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1794138670478913e-06,
+      "loss": 1.3358,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1972837741243747e-06,
+      "loss": 1.293,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2151536812008578e-06,
+      "loss": 1.3143,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2330235882773411e-06,
+      "loss": 1.3179,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.250893495353824e-06,
+      "loss": 1.329,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2687634024303074e-06,
+      "loss": 1.2964,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2866333095067905e-06,
+      "loss": 1.3006,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3045032165832739e-06,
+      "loss": 1.2901,
+      "step": 73
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3223731236597572e-06,
+      "loss": 1.3213,
+      "step": 74
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3402430307362401e-06,
+      "loss": 1.3303,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3581129378127234e-06,
+      "loss": 1.3209,
+      "step": 76
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3759828448892066e-06,
+      "loss": 1.3187,
+      "step": 77
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.39385275196569e-06,
+      "loss": 1.3308,
+      "step": 78
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.411722659042173e-06,
+      "loss": 1.2648,
+      "step": 79
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4295925661186562e-06,
+      "loss": 1.3361,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4474624731951395e-06,
+      "loss": 1.3306,
+      "step": 81
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4653323802716226e-06,
+      "loss": 1.3448,
+      "step": 82
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.483202287348106e-06,
+      "loss": 1.2941,
+      "step": 83
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.501072194424589e-06,
+      "loss": 1.336,
+      "step": 84
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5189421015010722e-06,
+      "loss": 1.3207,
+      "step": 85
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5368120085775555e-06,
+      "loss": 1.3056,
+      "step": 86
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5546819156540387e-06,
+      "loss": 1.3332,
+      "step": 87
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5725518227305218e-06,
+      "loss": 1.3102,
+      "step": 88
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5904217298070051e-06,
+      "loss": 1.3316,
+      "step": 89
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6082916368834883e-06,
+      "loss": 1.2885,
+      "step": 90
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6261615439599716e-06,
+      "loss": 1.3236,
+      "step": 91
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6440314510364547e-06,
+      "loss": 1.3414,
+      "step": 92
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.661901358112938e-06,
+      "loss": 1.3101,
+      "step": 93
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6797712651894212e-06,
+      "loss": 1.2998,
+      "step": 94
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.697641172265904e-06,
+      "loss": 1.3342,
+      "step": 95
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7155110793423874e-06,
+      "loss": 1.2903,
+      "step": 96
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7333809864188706e-06,
+      "loss": 1.3448,
+      "step": 97
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7512508934953539e-06,
+      "loss": 1.2998,
+      "step": 98
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.769120800571837e-06,
+      "loss": 1.3339,
+      "step": 99
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.7869907076483204e-06,
+      "loss": 1.3096,
+      "step": 100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8048606147248035e-06,
+      "loss": 1.3058,
+      "step": 101
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8227305218012868e-06,
+      "loss": 1.3275,
+      "step": 102
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8406004288777702e-06,
+      "loss": 1.3292,
+      "step": 103
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.858470335954253e-06,
+      "loss": 1.3164,
+      "step": 104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8763402430307362e-06,
+      "loss": 1.3167,
+      "step": 105
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.8942101501072195e-06,
+      "loss": 1.3385,
+      "step": 106
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9120800571837027e-06,
+      "loss": 1.3147,
+      "step": 107
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.929949964260186e-06,
+      "loss": 1.3152,
+      "step": 108
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9478198713366693e-06,
+      "loss": 1.3587,
+      "step": 109
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9656897784131522e-06,
+      "loss": 1.3155,
+      "step": 110
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9835596854896356e-06,
+      "loss": 1.304,
+      "step": 111
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.001429592566119e-06,
+      "loss": 1.3308,
+      "step": 112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.019299499642602e-06,
+      "loss": 1.3294,
+      "step": 113
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.037169406719085e-06,
+      "loss": 1.3299,
+      "step": 114
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.055039313795568e-06,
+      "loss": 1.3133,
+      "step": 115
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0729092208720514e-06,
+      "loss": 1.3308,
+      "step": 116
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.0907791279485348e-06,
+      "loss": 1.3112,
+      "step": 117
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.108649035025018e-06,
+      "loss": 1.2878,
+      "step": 118
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1265189421015014e-06,
+      "loss": 1.3416,
+      "step": 119
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1443888491779843e-06,
+      "loss": 1.304,
+      "step": 120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1622587562544677e-06,
+      "loss": 1.2928,
+      "step": 121
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1801286633309506e-06,
+      "loss": 1.3293,
+      "step": 122
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.197998570407434e-06,
+      "loss": 1.2826,
+      "step": 123
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2158684774839173e-06,
+      "loss": 1.301,
+      "step": 124
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2337383845604e-06,
+      "loss": 1.312,
+      "step": 125
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.2516082916368835e-06,
+      "loss": 1.3009,
+      "step": 126
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.269478198713367e-06,
+      "loss": 1.3131,
+      "step": 127
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.28734810578985e-06,
+      "loss": 1.3006,
+      "step": 128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.305218012866333e-06,
+      "loss": 1.3112,
+      "step": 129
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3230879199428164e-06,
+      "loss": 1.2985,
+      "step": 130
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3409578270192994e-06,
+      "loss": 1.3199,
+      "step": 131
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3588277340957827e-06,
+      "loss": 1.3119,
+      "step": 132
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.376697641172266e-06,
+      "loss": 1.2986,
+      "step": 133
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.3945675482487494e-06,
+      "loss": 1.3632,
+      "step": 134
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.4124374553252323e-06,
+      "loss": 1.333,
+      "step": 135
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.4303073624017156e-06,
+      "loss": 1.3368,
+      "step": 136
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.448177269478199e-06,
+      "loss": 1.3321,
+      "step": 137
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.4660471765546823e-06,
+      "loss": 1.282,
+      "step": 138
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.483917083631165e-06,
+      "loss": 1.3092,
+      "step": 139
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.501786990707648e-06,
+      "loss": 1.3267,
+      "step": 140
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5196568977841315e-06,
+      "loss": 1.3019,
+      "step": 141
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.537526804860615e-06,
+      "loss": 1.3066,
+      "step": 142
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.555396711937098e-06,
+      "loss": 1.3127,
+      "step": 143
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.573266619013581e-06,
+      "loss": 1.3433,
+      "step": 144
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.5911365260900644e-06,
+      "loss": 1.3027,
+      "step": 145
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6090064331665477e-06,
+      "loss": 1.3089,
+      "step": 146
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.626876340243031e-06,
+      "loss": 1.3305,
+      "step": 147
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6447462473195144e-06,
+      "loss": 1.3156,
+      "step": 148
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6626161543959973e-06,
+      "loss": 1.3156,
+      "step": 149
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6804860614724802e-06,
+      "loss": 1.2411,
+      "step": 150
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.6983559685489636e-06,
+      "loss": 1.2647,
+      "step": 151
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.716225875625447e-06,
+      "loss": 1.2908,
+      "step": 152
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7340957827019302e-06,
+      "loss": 1.3034,
+      "step": 153
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.751965689778413e-06,
+      "loss": 1.3014,
+      "step": 154
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7698355968548965e-06,
+      "loss": 1.3159,
+      "step": 155
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.78770550393138e-06,
+      "loss": 1.3279,
+      "step": 156
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.805575411007863e-06,
+      "loss": 1.2894,
+      "step": 157
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.823445318084346e-06,
+      "loss": 1.3192,
+      "step": 158
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.841315225160829e-06,
+      "loss": 1.3096,
+      "step": 159
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.8591851322373123e-06,
+      "loss": 1.3367,
+      "step": 160
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.8770550393137956e-06,
+      "loss": 1.3285,
+      "step": 161
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.894924946390279e-06,
+      "loss": 1.2608,
+      "step": 162
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9127948534667623e-06,
+      "loss": 1.2976,
+      "step": 163
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9306647605432452e-06,
+      "loss": 1.3256,
+      "step": 164
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9485346676197286e-06,
+      "loss": 1.3102,
+      "step": 165
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.966404574696212e-06,
+      "loss": 1.3164,
+      "step": 166
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.984274481772695e-06,
+      "loss": 1.277,
+      "step": 167
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.002144388849178e-06,
+      "loss": 1.2895,
+      "step": 168
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.020014295925661e-06,
+      "loss": 1.2886,
+      "step": 169
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.0378842030021444e-06,
+      "loss": 1.3098,
+      "step": 170
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.0557541100786277e-06,
+      "loss": 1.3072,
+      "step": 171
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.073624017155111e-06,
+      "loss": 1.3106,
+      "step": 172
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.091493924231594e-06,
+      "loss": 1.2352,
+      "step": 173
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.1093638313080773e-06,
+      "loss": 1.2943,
+      "step": 174
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.1272337383845607e-06,
+      "loss": 1.2695,
+      "step": 175
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.1451036454610436e-06,
+      "loss": 1.2875,
+      "step": 176
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.1629735525375273e-06,
+      "loss": 1.3564,
+      "step": 177
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.1808434596140103e-06,
+      "loss": 1.2981,
+      "step": 178
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.1987133666904936e-06,
+      "loss": 1.2791,
+      "step": 179
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.2165832737669765e-06,
+      "loss": 1.3136,
+      "step": 180
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.2344531808434594e-06,
+      "loss": 1.3273,
+      "step": 181
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.252323087919943e-06,
+      "loss": 1.3263,
+      "step": 182
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.270192994996426e-06,
+      "loss": 1.2881,
+      "step": 183
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.2880629020729094e-06,
+      "loss": 1.2912,
+      "step": 184
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.3059328091493923e-06,
+      "loss": 1.3155,
+      "step": 185
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.323802716225876e-06,
+      "loss": 1.2984,
+      "step": 186
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.341672623302359e-06,
+      "loss": 1.2682,
+      "step": 187
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.3595425303788424e-06,
+      "loss": 1.291,
+      "step": 188
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.3774124374553253e-06,
+      "loss": 1.3157,
+      "step": 189
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.395282344531808e-06,
+      "loss": 1.2627,
+      "step": 190
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.413152251608292e-06,
+      "loss": 1.3069,
+      "step": 191
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.431022158684775e-06,
+      "loss": 1.3252,
+      "step": 192
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.448892065761258e-06,
+      "loss": 1.3398,
+      "step": 193
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.466761972837741e-06,
+      "loss": 1.2941,
+      "step": 194
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.484631879914225e-06,
+      "loss": 1.3165,
+      "step": 195
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.5025017869907078e-06,
+      "loss": 1.3138,
+      "step": 196
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.520371694067191e-06,
+      "loss": 1.2743,
+      "step": 197
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.538241601143674e-06,
+      "loss": 1.2929,
+      "step": 198
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.556111508220157e-06,
+      "loss": 1.3057,
+      "step": 199
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.5739814152966407e-06,
+      "loss": 1.3031,
+      "step": 200
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.5918513223731236e-06,
+      "loss": 1.2887,
+      "step": 201
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.609721229449607e-06,
+      "loss": 1.3213,
+      "step": 202
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.62759113652609e-06,
+      "loss": 1.2569,
+      "step": 203
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.6454610436025736e-06,
+      "loss": 1.3051,
+      "step": 204
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.6633309506790565e-06,
+      "loss": 1.2867,
+      "step": 205
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.6812008577555403e-06,
+      "loss": 1.2942,
+      "step": 206
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.6990707648320232e-06,
+      "loss": 1.2692,
+      "step": 207
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.716940671908506e-06,
+      "loss": 1.336,
+      "step": 208
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.7348105789849895e-06,
+      "loss": 1.323,
+      "step": 209
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.7526804860614724e-06,
+      "loss": 1.3458,
+      "step": 210
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.770550393137956e-06,
+      "loss": 1.3197,
+      "step": 211
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.788420300214439e-06,
+      "loss": 1.3014,
+      "step": 212
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.8062902072909224e-06,
+      "loss": 1.3073,
+      "step": 213
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.824160114367405e-06,
+      "loss": 1.2671,
+      "step": 214
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.842030021443889e-06,
+      "loss": 1.3185,
+      "step": 215
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.859899928520372e-06,
+      "loss": 1.3332,
+      "step": 216
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.8777698355968545e-06,
+      "loss": 1.3304,
+      "step": 217
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.895639742673339e-06,
+      "loss": 1.3228,
+      "step": 218
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.913509649749821e-06,
+      "loss": 1.3311,
+      "step": 219
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9313795568263045e-06,
+      "loss": 1.2859,
+      "step": 220
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.949249463902788e-06,
+      "loss": 1.2873,
+      "step": 221
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.967119370979271e-06,
+      "loss": 1.2869,
+      "step": 222
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9849892780557545e-06,
+      "loss": 1.2945,
+      "step": 223
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.002859185132238e-06,
+      "loss": 1.282,
+      "step": 224
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.02072909220872e-06,
+      "loss": 1.2778,
+      "step": 225
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.038598999285204e-06,
+      "loss": 1.2973,
+      "step": 226
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.056468906361687e-06,
+      "loss": 1.2918,
+      "step": 227
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.07433881343817e-06,
+      "loss": 1.2634,
+      "step": 228
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.092208720514654e-06,
+      "loss": 1.3298,
+      "step": 229
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.110078627591136e-06,
+      "loss": 1.3184,
+      "step": 230
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.12794853466762e-06,
+      "loss": 1.3103,
+      "step": 231
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.145818441744103e-06,
+      "loss": 1.2807,
+      "step": 232
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.163688348820587e-06,
+      "loss": 1.251,
+      "step": 233
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.1815582558970695e-06,
+      "loss": 1.2933,
+      "step": 234
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.199428162973552e-06,
+      "loss": 1.3123,
+      "step": 235
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.217298070050036e-06,
+      "loss": 1.2784,
+      "step": 236
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.235167977126519e-06,
+      "loss": 1.2947,
+      "step": 237
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.253037884203003e-06,
+      "loss": 1.2899,
+      "step": 238
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.270907791279485e-06,
+      "loss": 1.2834,
+      "step": 239
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.288777698355969e-06,
+      "loss": 1.2918,
+      "step": 240
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.306647605432452e-06,
+      "loss": 1.2996,
+      "step": 241
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.324517512508935e-06,
+      "loss": 1.2803,
+      "step": 242
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.342387419585419e-06,
+      "loss": 1.3017,
+      "step": 243
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.360257326661901e-06,
+      "loss": 1.2768,
+      "step": 244
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.3781272337383845e-06,
+      "loss": 1.2861,
+      "step": 245
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.395997140814868e-06,
+      "loss": 1.2511,
+      "step": 246
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.413867047891351e-06,
+      "loss": 1.2705,
+      "step": 247
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.4317369549678345e-06,
+      "loss": 1.2991,
+      "step": 248
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.449606862044318e-06,
+      "loss": 1.3133,
+      "step": 249
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.4674767691208e-06,
+      "loss": 1.2876,
+      "step": 250
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.4853466761972845e-06,
+      "loss": 1.3146,
+      "step": 251
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.503216583273767e-06,
+      "loss": 1.3313,
+      "step": 252
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.52108649035025e-06,
+      "loss": 1.3216,
+      "step": 253
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.538956397426734e-06,
+      "loss": 1.2879,
+      "step": 254
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.556826304503216e-06,
+      "loss": 1.2619,
+      "step": 255
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.5746962115797e-06,
+      "loss": 1.2839,
+      "step": 256
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.592566118656183e-06,
+      "loss": 1.3157,
+      "step": 257
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.610436025732666e-06,
+      "loss": 1.2801,
+      "step": 258
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.6283059328091495e-06,
+      "loss": 1.2967,
+      "step": 259
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.646175839885633e-06,
+      "loss": 1.2878,
+      "step": 260
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.664045746962116e-06,
+      "loss": 1.2626,
+      "step": 261
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.681915654038599e-06,
+      "loss": 1.3109,
+      "step": 262
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.699785561115083e-06,
+      "loss": 1.2718,
+      "step": 263
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.717655468191565e-06,
+      "loss": 1.2959,
+      "step": 264
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.735525375268049e-06,
+      "loss": 1.2596,
+      "step": 265
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.753395282344532e-06,
+      "loss": 1.3023,
+      "step": 266
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.771265189421015e-06,
+      "loss": 1.3038,
+      "step": 267
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.789135096497499e-06,
+      "loss": 1.288,
+      "step": 268
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.807005003573982e-06,
+      "loss": 1.2996,
+      "step": 269
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.8248749106504646e-06,
+      "loss": 1.2679,
+      "step": 270
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.842744817726948e-06,
+      "loss": 1.2623,
+      "step": 271
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.860614724803431e-06,
+      "loss": 1.2795,
+      "step": 272
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.8784846318799146e-06,
+      "loss": 1.279,
+      "step": 273
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.896354538956398e-06,
+      "loss": 1.3225,
+      "step": 274
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.91422444603288e-06,
+      "loss": 1.3103,
+      "step": 275
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9320943531093646e-06,
+      "loss": 1.2772,
+      "step": 276
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.949964260185847e-06,
+      "loss": 1.2746,
+      "step": 277
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.96783416726233e-06,
+      "loss": 1.2997,
+      "step": 278
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.985704074338814e-06,
+      "loss": 1.3199,
+      "step": 279
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.003573981415296e-06,
+      "loss": 1.3241,
+      "step": 280
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.02144388849178e-06,
+      "loss": 1.2745,
+      "step": 281
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.039313795568263e-06,
+      "loss": 1.3157,
+      "step": 282
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.057183702644746e-06,
+      "loss": 1.3073,
+      "step": 283
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.07505360972123e-06,
+      "loss": 1.313,
+      "step": 284
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.092923516797713e-06,
+      "loss": 1.3024,
+      "step": 285
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.110793423874196e-06,
+      "loss": 1.3027,
+      "step": 286
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.12866333095068e-06,
+      "loss": 1.3159,
+      "step": 287
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.146533238027162e-06,
+      "loss": 1.2804,
+      "step": 288
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.164403145103645e-06,
+      "loss": 1.3282,
+      "step": 289
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.182273052180129e-06,
+      "loss": 1.2678,
+      "step": 290
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.200142959256612e-06,
+      "loss": 1.308,
+      "step": 291
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.2180128663330954e-06,
+      "loss": 1.324,
+      "step": 292
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.235882773409578e-06,
+      "loss": 1.2483,
+      "step": 293
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.253752680486062e-06,
+      "loss": 1.2946,
+      "step": 294
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.271622587562545e-06,
+      "loss": 1.3016,
+      "step": 295
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.289492494639029e-06,
+      "loss": 1.2712,
+      "step": 296
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.307362401715511e-06,
+      "loss": 1.2998,
+      "step": 297
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.325232308791995e-06,
+      "loss": 1.319,
+      "step": 298
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.343102215868478e-06,
+      "loss": 1.3037,
+      "step": 299
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.3609721229449604e-06,
+      "loss": 1.3104,
+      "step": 300
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.378842030021445e-06,
+      "loss": 1.3472,
+      "step": 301
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.396711937097927e-06,
+      "loss": 1.3841,
+      "step": 302
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.4145818441744104e-06,
+      "loss": 1.3363,
+      "step": 303
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.432451751250894e-06,
+      "loss": 1.3194,
+      "step": 304
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.450321658327377e-06,
+      "loss": 1.2785,
+      "step": 305
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.4681915654038604e-06,
+      "loss": 1.2852,
+      "step": 306
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.486061472480343e-06,
+      "loss": 1.3232,
+      "step": 307
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.503931379556826e-06,
+      "loss": 1.2575,
+      "step": 308
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.52180128663331e-06,
+      "loss": 1.2704,
+      "step": 309
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.539671193709793e-06,
+      "loss": 1.2869,
+      "step": 310
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.557541100786276e-06,
+      "loss": 1.3061,
+      "step": 311
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.57541100786276e-06,
+      "loss": 1.3053,
+      "step": 312
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.593280914939242e-06,
+      "loss": 1.3043,
+      "step": 313
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.611150822015726e-06,
+      "loss": 1.3381,
+      "step": 314
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.629020729092209e-06,
+      "loss": 1.2816,
+      "step": 315
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.646890636168692e-06,
+      "loss": 1.2238,
+      "step": 316
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.6647605432451755e-06,
+      "loss": 1.3512,
+      "step": 317
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.682630450321658e-06,
+      "loss": 1.2995,
+      "step": 318
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.700500357398142e-06,
+      "loss": 1.284,
+      "step": 319
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.718370264474625e-06,
+      "loss": 1.2719,
+      "step": 320
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.736240171551109e-06,
+      "loss": 1.2989,
+      "step": 321
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.754110078627591e-06,
+      "loss": 1.291,
+      "step": 322
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.771979985704075e-06,
+      "loss": 1.3331,
+      "step": 323
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.789849892780558e-06,
+      "loss": 1.2573,
+      "step": 324
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.8077197998570405e-06,
+      "loss": 1.2356,
+      "step": 325
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.825589706933525e-06,
+      "loss": 1.272,
+      "step": 326
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.843459614010007e-06,
+      "loss": 1.3054,
+      "step": 327
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.8613295210864905e-06,
+      "loss": 1.2857,
+      "step": 328
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.879199428162974e-06,
+      "loss": 1.2563,
+      "step": 329
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.897069335239457e-06,
+      "loss": 1.2792,
+      "step": 330
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.9149392423159405e-06,
+      "loss": 1.33,
+      "step": 331
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.932809149392424e-06,
+      "loss": 1.3026,
+      "step": 332
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.950679056468906e-06,
+      "loss": 1.2833,
+      "step": 333
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.96854896354539e-06,
+      "loss": 1.312,
+      "step": 334
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 5.986418870621873e-06,
+      "loss": 1.264,
+      "step": 335
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.004288777698356e-06,
+      "loss": 1.2641,
+      "step": 336
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.02215868477484e-06,
+      "loss": 1.3083,
+      "step": 337
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.040028591851322e-06,
+      "loss": 1.2959,
+      "step": 338
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.057898498927806e-06,
+      "loss": 1.2977,
+      "step": 339
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.075768406004289e-06,
+      "loss": 1.3287,
+      "step": 340
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.093638313080772e-06,
+      "loss": 1.2572,
+      "step": 341
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.1115082201572555e-06,
+      "loss": 1.2815,
+      "step": 342
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.129378127233738e-06,
+      "loss": 1.3003,
+      "step": 343
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.147248034310222e-06,
+      "loss": 1.2659,
+      "step": 344
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.165117941386705e-06,
+      "loss": 1.2852,
+      "step": 345
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.182987848463188e-06,
+      "loss": 1.2703,
+      "step": 346
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.200857755539671e-06,
+      "loss": 1.3353,
+      "step": 347
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.218727662616155e-06,
+      "loss": 1.2647,
+      "step": 348
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.236597569692638e-06,
+      "loss": 1.2977,
+      "step": 349
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.254467476769121e-06,
+      "loss": 1.2731,
+      "step": 350
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.272337383845605e-06,
+      "loss": 1.3399,
+      "step": 351
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.290207290922087e-06,
+      "loss": 1.2886,
+      "step": 352
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.30807719799857e-06,
+      "loss": 1.2801,
+      "step": 353
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.325947105075055e-06,
+      "loss": 1.3154,
+      "step": 354
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.343817012151537e-06,
+      "loss": 1.3123,
+      "step": 355
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.3616869192280205e-06,
+      "loss": 1.2587,
+      "step": 356
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.379556826304503e-06,
+      "loss": 1.3325,
+      "step": 357
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.397426733380987e-06,
+      "loss": 1.3091,
+      "step": 358
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.4152966404574705e-06,
+      "loss": 1.2544,
+      "step": 359
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.433166547533953e-06,
+      "loss": 1.2818,
+      "step": 360
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.451036454610436e-06,
+      "loss": 1.3043,
+      "step": 361
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.468906361686919e-06,
+      "loss": 1.3186,
+      "step": 362
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.486776268763403e-06,
+      "loss": 1.2685,
+      "step": 363
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.504646175839886e-06,
+      "loss": 1.2613,
+      "step": 364
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.522516082916369e-06,
+      "loss": 1.3362,
+      "step": 365
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.540385989992852e-06,
+      "loss": 1.2985,
+      "step": 366
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.558255897069336e-06,
+      "loss": 1.2951,
+      "step": 367
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.576125804145819e-06,
+      "loss": 1.2516,
+      "step": 368
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.593995711222302e-06,
+      "loss": 1.2369,
+      "step": 369
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.611865618298785e-06,
+      "loss": 1.2485,
+      "step": 370
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.629735525375268e-06,
+      "loss": 1.2665,
+      "step": 371
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.647605432451752e-06,
+      "loss": 1.3374,
+      "step": 372
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.665475339528235e-06,
+      "loss": 1.2774,
+      "step": 373
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.683345246604718e-06,
+      "loss": 1.294,
+      "step": 374
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.7012151536812005e-06,
+      "loss": 1.296,
+      "step": 375
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.719085060757685e-06,
+      "loss": 1.3009,
+      "step": 376
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.736954967834168e-06,
+      "loss": 1.3116,
+      "step": 377
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.7548248749106505e-06,
+      "loss": 1.2949,
+      "step": 378
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.772694781987134e-06,
+      "loss": 1.2791,
+      "step": 379
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.790564689063616e-06,
+      "loss": 1.292,
+      "step": 380
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.8084345961401006e-06,
+      "loss": 1.2748,
+      "step": 381
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.826304503216584e-06,
+      "loss": 1.2927,
+      "step": 382
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.844174410293066e-06,
+      "loss": 1.3051,
+      "step": 383
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.86204431736955e-06,
+      "loss": 1.3452,
+      "step": 384
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.879914224446034e-06,
+      "loss": 1.3211,
+      "step": 385
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.897784131522516e-06,
+      "loss": 1.3121,
+      "step": 386
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.915654038599e-06,
+      "loss": 1.2964,
+      "step": 387
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.933523945675482e-06,
+      "loss": 1.2752,
+      "step": 388
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.9513938527519656e-06,
+      "loss": 1.2863,
+      "step": 389
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.96926375982845e-06,
+      "loss": 1.2533,
+      "step": 390
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 6.987133666904932e-06,
+      "loss": 1.3023,
+      "step": 391
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.0050035739814156e-06,
+      "loss": 1.3006,
+      "step": 392
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.022873481057898e-06,
+      "loss": 1.2711,
+      "step": 393
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.040743388134382e-06,
+      "loss": 1.2881,
+      "step": 394
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.058613295210866e-06,
+      "loss": 1.2906,
+      "step": 395
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.076483202287348e-06,
+      "loss": 1.3122,
+      "step": 396
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.094353109363831e-06,
+      "loss": 1.2587,
+      "step": 397
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.112223016440314e-06,
+      "loss": 1.2675,
+      "step": 398
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.130092923516798e-06,
+      "loss": 1.2614,
+      "step": 399
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.147962830593281e-06,
+      "loss": 1.2808,
+      "step": 400
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.165832737669764e-06,
+      "loss": 1.2878,
+      "step": 401
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.183702644746247e-06,
+      "loss": 1.3085,
+      "step": 402
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.2015725518227314e-06,
+      "loss": 1.2989,
+      "step": 403
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.219442458899214e-06,
+      "loss": 1.3449,
+      "step": 404
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.237312365975697e-06,
+      "loss": 1.2923,
+      "step": 405
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.25518227305218e-06,
+      "loss": 1.2619,
+      "step": 406
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.273052180128663e-06,
+      "loss": 1.3009,
+      "step": 407
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.290922087205147e-06,
+      "loss": 1.298,
+      "step": 408
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.308791994281631e-06,
+      "loss": 1.3174,
+      "step": 409
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.326661901358113e-06,
+      "loss": 1.2822,
+      "step": 410
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.344531808434596e-06,
+      "loss": 1.2673,
+      "step": 411
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.362401715511081e-06,
+      "loss": 1.2849,
+      "step": 412
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.380271622587563e-06,
+      "loss": 1.2843,
+      "step": 413
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.3981415296640464e-06,
+      "loss": 1.3044,
+      "step": 414
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.416011436740529e-06,
+      "loss": 1.2938,
+      "step": 415
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.433881343817012e-06,
+      "loss": 1.2709,
+      "step": 416
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.4517512508934965e-06,
+      "loss": 1.2732,
+      "step": 417
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.469621157969979e-06,
+      "loss": 1.2629,
+      "step": 418
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.487491065046462e-06,
+      "loss": 1.2831,
+      "step": 419
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.505360972122945e-06,
+      "loss": 1.3241,
+      "step": 420
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.523230879199429e-06,
+      "loss": 1.2871,
+      "step": 421
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.541100786275912e-06,
+      "loss": 1.2841,
+      "step": 422
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.558970693352395e-06,
+      "loss": 1.2908,
+      "step": 423
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.576840600428878e-06,
+      "loss": 1.2817,
+      "step": 424
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.594710507505361e-06,
+      "loss": 1.2896,
+      "step": 425
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.612580414581845e-06,
+      "loss": 1.2943,
+      "step": 426
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.630450321658328e-06,
+      "loss": 1.3184,
+      "step": 427
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.64832022873481e-06,
+      "loss": 1.2523,
+      "step": 428
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.666190135811293e-06,
+      "loss": 1.3176,
+      "step": 429
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.684060042887777e-06,
+      "loss": 1.2859,
+      "step": 430
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.701929949964261e-06,
+      "loss": 1.273,
+      "step": 431
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.719799857040744e-06,
+      "loss": 1.3158,
+      "step": 432
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.737669764117226e-06,
+      "loss": 1.2726,
+      "step": 433
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.755539671193709e-06,
+      "loss": 1.3118,
+      "step": 434
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.773409578270193e-06,
+      "loss": 1.2774,
+      "step": 435
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.791279485346677e-06,
+      "loss": 1.2682,
+      "step": 436
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.80914939242316e-06,
+      "loss": 1.2974,
+      "step": 437
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.827019299499642e-06,
+      "loss": 1.2562,
+      "step": 438
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.844889206576126e-06,
+      "loss": 1.2418,
+      "step": 439
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.862759113652609e-06,
+      "loss": 1.235,
+      "step": 440
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.880629020729093e-06,
+      "loss": 1.3164,
+      "step": 441
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.898498927805576e-06,
+      "loss": 1.2702,
+      "step": 442
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.916368834882058e-06,
+      "loss": 1.2969,
+      "step": 443
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.934238741958542e-06,
+      "loss": 1.2558,
+      "step": 444
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.952108649035025e-06,
+      "loss": 1.2535,
+      "step": 445
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.969978556111509e-06,
+      "loss": 1.2331,
+      "step": 446
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 7.987848463187991e-06,
+      "loss": 1.2659,
+      "step": 447
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.005718370264476e-06,
+      "loss": 1.2845,
+      "step": 448
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.023588277340958e-06,
+      "loss": 1.2955,
+      "step": 449
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.04145818441744e-06,
+      "loss": 1.3014,
+      "step": 450
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.059328091493925e-06,
+      "loss": 1.3262,
+      "step": 451
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.077197998570407e-06,
+      "loss": 1.291,
+      "step": 452
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.095067905646891e-06,
+      "loss": 1.2878,
+      "step": 453
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.112937812723374e-06,
+      "loss": 1.2647,
+      "step": 454
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.130807719799856e-06,
+      "loss": 1.2322,
+      "step": 455
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.14867762687634e-06,
+      "loss": 1.264,
+      "step": 456
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.166547533952825e-06,
+      "loss": 1.2707,
+      "step": 457
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.184417441029307e-06,
+      "loss": 1.2418,
+      "step": 458
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.20228734810579e-06,
+      "loss": 1.2571,
+      "step": 459
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.220157255182272e-06,
+      "loss": 1.2579,
+      "step": 460
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.238027162258757e-06,
+      "loss": 1.2653,
+      "step": 461
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.25589706933524e-06,
+      "loss": 1.3259,
+      "step": 462
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.273766976411723e-06,
+      "loss": 1.2844,
+      "step": 463
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.291636883488206e-06,
+      "loss": 1.3323,
+      "step": 464
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.309506790564688e-06,
+      "loss": 1.2524,
+      "step": 465
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.327376697641174e-06,
+      "loss": 1.2795,
+      "step": 466
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.345246604717657e-06,
+      "loss": 1.2789,
+      "step": 467
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.363116511794139e-06,
+      "loss": 1.2956,
+      "step": 468
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.380986418870622e-06,
+      "loss": 1.2677,
+      "step": 469
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.398856325947104e-06,
+      "loss": 1.2616,
+      "step": 470
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.41672623302359e-06,
+      "loss": 1.2734,
+      "step": 471
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.434596140100072e-06,
+      "loss": 1.2652,
+      "step": 472
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.452466047176555e-06,
+      "loss": 1.3297,
+      "step": 473
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.470335954253037e-06,
+      "loss": 1.3042,
+      "step": 474
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.488205861329522e-06,
+      "loss": 1.3091,
+      "step": 475
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.506075768406006e-06,
+      "loss": 1.2987,
+      "step": 476
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.523945675482488e-06,
+      "loss": 1.2571,
+      "step": 477
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.54181558255897e-06,
+      "loss": 1.2886,
+      "step": 478
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.559685489635453e-06,
+      "loss": 1.2912,
+      "step": 479
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.577555396711937e-06,
+      "loss": 1.2553,
+      "step": 480
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.595425303788422e-06,
+      "loss": 1.3095,
+      "step": 481
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.613295210864904e-06,
+      "loss": 1.3299,
+      "step": 482
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.631165117941387e-06,
+      "loss": 1.3017,
+      "step": 483
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.64903502501787e-06,
+      "loss": 1.3067,
+      "step": 484
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.666904932094353e-06,
+      "loss": 1.299,
+      "step": 485
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.684774839170837e-06,
+      "loss": 1.2796,
+      "step": 486
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.70264474624732e-06,
+      "loss": 1.3038,
+      "step": 487
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.720514653323802e-06,
+      "loss": 1.3122,
+      "step": 488
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.738384560400287e-06,
+      "loss": 1.2539,
+      "step": 489
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.756254467476769e-06,
+      "loss": 1.2825,
+      "step": 490
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.774124374553253e-06,
+      "loss": 1.2952,
+      "step": 491
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.791994281629736e-06,
+      "loss": 1.2811,
+      "step": 492
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.80986418870622e-06,
+      "loss": 1.334,
+      "step": 493
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.827734095782702e-06,
+      "loss": 1.2992,
+      "step": 494
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.845604002859185e-06,
+      "loss": 1.3421,
+      "step": 495
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.863473909935669e-06,
+      "loss": 1.2985,
+      "step": 496
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.881343817012152e-06,
+      "loss": 1.2816,
+      "step": 497
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.899213724088636e-06,
+      "loss": 1.2574,
+      "step": 498
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.917083631165118e-06,
+      "loss": 1.2923,
+      "step": 499
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.9349535382416e-06,
+      "loss": 1.2607,
+      "step": 500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.952823445318085e-06,
+      "loss": 1.2751,
+      "step": 501
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.970693352394569e-06,
+      "loss": 1.2515,
+      "step": 502
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 8.988563259471052e-06,
+      "loss": 1.2684,
+      "step": 503
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.006433166547534e-06,
+      "loss": 1.3062,
+      "step": 504
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.024303073624017e-06,
+      "loss": 1.2744,
+      "step": 505
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.0421729807005e-06,
+      "loss": 1.289,
+      "step": 506
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.060042887776985e-06,
+      "loss": 1.289,
+      "step": 507
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.077912794853467e-06,
+      "loss": 1.3049,
+      "step": 508
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.09578270192995e-06,
+      "loss": 1.2613,
+      "step": 509
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.113652609006432e-06,
+      "loss": 1.2412,
+      "step": 510
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.131522516082917e-06,
+      "loss": 1.2594,
+      "step": 511
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.1493924231594e-06,
+      "loss": 1.2713,
+      "step": 512
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.167262330235883e-06,
+      "loss": 1.2977,
+      "step": 513
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.185132237312366e-06,
+      "loss": 1.2764,
+      "step": 514
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.203002144388848e-06,
+      "loss": 1.2366,
+      "step": 515
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.220872051465332e-06,
+      "loss": 1.2721,
+      "step": 516
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.238741958541817e-06,
+      "loss": 1.273,
+      "step": 517
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.256611865618299e-06,
+      "loss": 1.2802,
+      "step": 518
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.274481772694782e-06,
+      "loss": 1.2777,
+      "step": 519
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.292351679771266e-06,
+      "loss": 1.2856,
+      "step": 520
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.310221586847748e-06,
+      "loss": 1.3008,
+      "step": 521
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.328091493924232e-06,
+      "loss": 1.3023,
+      "step": 522
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.345961401000715e-06,
+      "loss": 1.2543,
+      "step": 523
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.363831308077197e-06,
+      "loss": 1.2751,
+      "step": 524
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.381701215153682e-06,
+      "loss": 1.2846,
+      "step": 525
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.399571122230166e-06,
+      "loss": 1.2559,
+      "step": 526
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.417441029306648e-06,
+      "loss": 1.3193,
+      "step": 527
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.43531093638313e-06,
+      "loss": 1.3255,
+      "step": 528
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.453180843459615e-06,
+      "loss": 1.3163,
+      "step": 529
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.471050750536097e-06,
+      "loss": 1.2849,
+      "step": 530
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.488920657612582e-06,
+      "loss": 1.2906,
+      "step": 531
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.506790564689064e-06,
+      "loss": 1.2559,
+      "step": 532
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.524660471765547e-06,
+      "loss": 1.2906,
+      "step": 533
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.54253037884203e-06,
+      "loss": 1.2784,
+      "step": 534
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.560400285918513e-06,
+      "loss": 1.2762,
+      "step": 535
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.578270192994997e-06,
+      "loss": 1.2735,
+      "step": 536
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.59614010007148e-06,
+      "loss": 1.2583,
+      "step": 537
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.614010007147964e-06,
+      "loss": 1.2705,
+      "step": 538
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.631879914224447e-06,
+      "loss": 1.2699,
+      "step": 539
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.649749821300929e-06,
+      "loss": 1.2287,
+      "step": 540
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.667619728377413e-06,
+      "loss": 1.2865,
+      "step": 541
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.685489635453896e-06,
+      "loss": 1.2862,
+      "step": 542
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.70335954253038e-06,
+      "loss": 1.2525,
+      "step": 543
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.721229449606862e-06,
+      "loss": 1.2683,
+      "step": 544
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.739099356683345e-06,
+      "loss": 1.248,
+      "step": 545
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.756969263759829e-06,
+      "loss": 1.2788,
+      "step": 546
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.774839170836313e-06,
+      "loss": 1.2878,
+      "step": 547
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.792709077912796e-06,
+      "loss": 1.2648,
+      "step": 548
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.810578984989278e-06,
+      "loss": 1.2881,
+      "step": 549
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.82844889206576e-06,
+      "loss": 1.2386,
+      "step": 550
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.846318799142245e-06,
+      "loss": 1.2503,
+      "step": 551
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.864188706218729e-06,
+      "loss": 1.2824,
+      "step": 552
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.882058613295212e-06,
+      "loss": 1.3043,
+      "step": 553
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.899928520371694e-06,
+      "loss": 1.2716,
+      "step": 554
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.917798427448177e-06,
+      "loss": 1.3102,
+      "step": 555
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.93566833452466e-06,
+      "loss": 1.2857,
+      "step": 556
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.953538241601145e-06,
+      "loss": 1.2455,
+      "step": 557
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.971408148677627e-06,
+      "loss": 1.3131,
+      "step": 558
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.98927805575411e-06,
+      "loss": 1.2714,
+      "step": 559
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.0007147962830592e-05,
+      "loss": 1.3239,
+      "step": 560
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.0025017869907077e-05,
+      "loss": 1.2926,
+      "step": 561
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.004288777698356e-05,
+      "loss": 1.2739,
+      "step": 562
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.0060757684060043e-05,
+      "loss": 1.2677,
+      "step": 563
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.0078627591136526e-05,
+      "loss": 1.2904,
+      "step": 564
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.009649749821301e-05,
+      "loss": 1.3223,
+      "step": 565
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.0114367405289492e-05,
+      "loss": 1.3149,
+      "step": 566
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.0132237312365977e-05,
+      "loss": 1.2175,
+      "step": 567
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.015010721944246e-05,
+      "loss": 1.2788,
+      "step": 568
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.0167977126518942e-05,
+      "loss": 1.2463,
+      "step": 569
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.0185847033595426e-05,
+      "loss": 1.265,
+      "step": 570
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.0203716940671908e-05,
+      "loss": 1.2591,
+      "step": 571
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.0221586847748393e-05,
+      "loss": 1.2659,
+      "step": 572
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.0239456754824875e-05,
+      "loss": 1.2638,
+      "step": 573
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.025732666190136e-05,
+      "loss": 1.2503,
+      "step": 574
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.0275196568977842e-05,
+      "loss": 1.2786,
+      "step": 575
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.0293066476054324e-05,
+      "loss": 1.2436,
+      "step": 576
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.0310936383130808e-05,
+      "loss": 1.2379,
+      "step": 577
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.032880629020729e-05,
+      "loss": 1.2275,
+      "step": 578
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.0346676197283775e-05,
+      "loss": 1.2727,
+      "step": 579
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.0364546104360258e-05,
+      "loss": 1.2798,
+      "step": 580
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.038241601143674e-05,
+      "loss": 1.2899,
+      "step": 581
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.0400285918513224e-05,
+      "loss": 1.2998,
+      "step": 582
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.0418155825589708e-05,
+      "loss": 1.2594,
+      "step": 583
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.0436025732666191e-05,
+      "loss": 1.2523,
+      "step": 584
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.0453895639742673e-05,
+      "loss": 1.2979,
+      "step": 585
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.0471765546819156e-05,
+      "loss": 1.3323,
+      "step": 586
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.048963545389564e-05,
+      "loss": 1.2601,
+      "step": 587
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.0507505360972124e-05,
+      "loss": 1.2296,
+      "step": 588
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.0525375268048607e-05,
+      "loss": 1.2673,
+      "step": 589
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.054324517512509e-05,
+      "loss": 1.2483,
+      "step": 590
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.0561115082201573e-05,
+      "loss": 1.3114,
+      "step": 591
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.0578984989278058e-05,
+      "loss": 1.2917,
+      "step": 592
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.059685489635454e-05,
+      "loss": 1.2966,
+      "step": 593
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.0614724803431023e-05,
+      "loss": 1.2832,
+      "step": 594
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.0632594710507505e-05,
+      "loss": 1.3025,
+      "step": 595
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.065046461758399e-05,
+      "loss": 1.2403,
+      "step": 596
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.0668334524660473e-05,
+      "loss": 1.2877,
+      "step": 597
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.0686204431736956e-05,
+      "loss": 1.2648,
+      "step": 598
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.0704074338813438e-05,
+      "loss": 1.2838,
+      "step": 599
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.0721944245889921e-05,
+      "loss": 1.2465,
+      "step": 600
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.0739814152966405e-05,
+      "loss": 1.2739,
+      "step": 601
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.075768406004289e-05,
+      "loss": 1.2731,
+      "step": 602
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.0775553967119372e-05,
+      "loss": 1.2732,
+      "step": 603
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.0793423874195854e-05,
+      "loss": 1.2891,
+      "step": 604
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.0811293781272337e-05,
+      "loss": 1.2922,
+      "step": 605
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.0829163688348821e-05,
+      "loss": 1.2563,
+      "step": 606
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.0847033595425305e-05,
+      "loss": 1.2486,
+      "step": 607
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.0864903502501788e-05,
+      "loss": 1.2712,
+      "step": 608
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.088277340957827e-05,
+      "loss": 1.25,
+      "step": 609
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.0900643316654754e-05,
+      "loss": 1.261,
+      "step": 610
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.0918513223731237e-05,
+      "loss": 1.2952,
+      "step": 611
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.0936383130807721e-05,
+      "loss": 1.2915,
+      "step": 612
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.0954253037884203e-05,
+      "loss": 1.2975,
+      "step": 613
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.0972122944960686e-05,
+      "loss": 1.2874,
+      "step": 614
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.098999285203717e-05,
+      "loss": 1.2983,
+      "step": 615
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1007862759113653e-05,
+      "loss": 1.3123,
+      "step": 616
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1025732666190137e-05,
+      "loss": 1.3213,
+      "step": 617
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.104360257326662e-05,
+      "loss": 1.3241,
+      "step": 618
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1061472480343103e-05,
+      "loss": 1.24,
+      "step": 619
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1079342387419586e-05,
+      "loss": 1.2575,
+      "step": 620
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1097212294496068e-05,
+      "loss": 1.2754,
+      "step": 621
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1115082201572553e-05,
+      "loss": 1.268,
+      "step": 622
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1132952108649035e-05,
+      "loss": 1.2918,
+      "step": 623
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.115082201572552e-05,
+      "loss": 1.283,
+      "step": 624
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1168691922802002e-05,
+      "loss": 1.2815,
+      "step": 625
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1186561829878484e-05,
+      "loss": 1.281,
+      "step": 626
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1204431736954968e-05,
+      "loss": 1.289,
+      "step": 627
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1222301644031453e-05,
+      "loss": 1.2478,
+      "step": 628
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1240171551107935e-05,
+      "loss": 1.2526,
+      "step": 629
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1258041458184418e-05,
+      "loss": 1.3221,
+      "step": 630
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.12759113652609e-05,
+      "loss": 1.2831,
+      "step": 631
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1293781272337384e-05,
+      "loss": 1.2468,
+      "step": 632
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1311651179413868e-05,
+      "loss": 1.2834,
+      "step": 633
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1329521086490351e-05,
+      "loss": 1.2805,
+      "step": 634
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1347390993566833e-05,
+      "loss": 1.2952,
+      "step": 635
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1365260900643316e-05,
+      "loss": 1.3007,
+      "step": 636
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1383130807719802e-05,
+      "loss": 1.2648,
+      "step": 637
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1401000714796284e-05,
+      "loss": 1.2828,
+      "step": 638
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1418870621872767e-05,
+      "loss": 1.2719,
+      "step": 639
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.143674052894925e-05,
+      "loss": 1.262,
+      "step": 640
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1454610436025732e-05,
+      "loss": 1.2497,
+      "step": 641
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1472480343102218e-05,
+      "loss": 1.2557,
+      "step": 642
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.14903502501787e-05,
+      "loss": 1.2632,
+      "step": 643
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1508220157255183e-05,
+      "loss": 1.272,
+      "step": 644
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1526090064331665e-05,
+      "loss": 1.2664,
+      "step": 645
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.154395997140815e-05,
+      "loss": 1.2577,
+      "step": 646
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1561829878484633e-05,
+      "loss": 1.2679,
+      "step": 647
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1579699785561116e-05,
+      "loss": 1.2675,
+      "step": 648
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1597569692637598e-05,
+      "loss": 1.2867,
+      "step": 649
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1615439599714081e-05,
+      "loss": 1.2651,
+      "step": 650
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.1633309506790565e-05,
+      "loss": 1.2564,
+      "step": 651
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.165117941386705e-05,
+      "loss": 1.2657,
+      "step": 652
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1669049320943532e-05,
+      "loss": 1.2788,
+      "step": 653
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1686919228020014e-05,
+      "loss": 1.25,
+      "step": 654
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1704789135096498e-05,
+      "loss": 1.304,
+      "step": 655
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1722659042172981e-05,
+      "loss": 1.3171,
+      "step": 656
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1740528949249465e-05,
+      "loss": 1.2778,
+      "step": 657
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1758398856325948e-05,
+      "loss": 1.2857,
+      "step": 658
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.177626876340243e-05,
+      "loss": 1.2773,
+      "step": 659
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1794138670478914e-05,
+      "loss": 1.2775,
+      "step": 660
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1812008577555397e-05,
+      "loss": 1.2891,
+      "step": 661
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1829878484631881e-05,
+      "loss": 1.265,
+      "step": 662
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1847748391708363e-05,
+      "loss": 1.264,
+      "step": 663
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1865618298784848e-05,
+      "loss": 1.2774,
+      "step": 664
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.188348820586133e-05,
+      "loss": 1.2825,
+      "step": 665
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1901358112937813e-05,
+      "loss": 1.3032,
+      "step": 666
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1919228020014297e-05,
+      "loss": 1.2651,
+      "step": 667
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.193709792709078e-05,
+      "loss": 1.2805,
+      "step": 668
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1954967834167263e-05,
+      "loss": 1.232,
+      "step": 669
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1972837741243746e-05,
+      "loss": 1.2805,
+      "step": 670
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.1990707648320228e-05,
+      "loss": 1.2776,
+      "step": 671
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.2008577555396713e-05,
+      "loss": 1.2818,
+      "step": 672
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.2026447462473197e-05,
+      "loss": 1.2745,
+      "step": 673
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.204431736954968e-05,
+      "loss": 1.2914,
+      "step": 674
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.2062187276626162e-05,
+      "loss": 1.2545,
+      "step": 675
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.2080057183702644e-05,
+      "loss": 1.314,
+      "step": 676
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.2097927090779128e-05,
+      "loss": 1.283,
+      "step": 677
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.2115796997855613e-05,
+      "loss": 1.2412,
+      "step": 678
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.2133666904932095e-05,
+      "loss": 1.2567,
+      "step": 679
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.2151536812008578e-05,
+      "loss": 1.2804,
+      "step": 680
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.216940671908506e-05,
+      "loss": 1.2641,
+      "step": 681
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.2187276626161544e-05,
+      "loss": 1.3005,
+      "step": 682
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.2205146533238029e-05,
+      "loss": 1.2596,
+      "step": 683
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.2223016440314511e-05,
+      "loss": 1.2937,
+      "step": 684
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.2240886347390993e-05,
+      "loss": 1.2413,
+      "step": 685
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.2258756254467476e-05,
+      "loss": 1.269,
+      "step": 686
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.227662616154396e-05,
+      "loss": 1.2979,
+      "step": 687
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.2294496068620444e-05,
+      "loss": 1.2737,
+      "step": 688
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.2312365975696927e-05,
+      "loss": 1.2917,
+      "step": 689
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.233023588277341e-05,
+      "loss": 1.2944,
+      "step": 690
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.2348105789849894e-05,
+      "loss": 1.2882,
+      "step": 691
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.2365975696926376e-05,
+      "loss": 1.2733,
+      "step": 692
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.238384560400286e-05,
+      "loss": 1.3092,
+      "step": 693
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.2401715511079343e-05,
+      "loss": 1.2701,
+      "step": 694
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.2419585418155825e-05,
+      "loss": 1.301,
+      "step": 695
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.243745532523231e-05,
+      "loss": 1.2548,
+      "step": 696
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.2455325232308792e-05,
+      "loss": 1.2777,
+      "step": 697
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.2473195139385276e-05,
+      "loss": 1.2436,
+      "step": 698
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.2491065046461759e-05,
+      "loss": 1.2692,
+      "step": 699
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.2508934953538243e-05,
+      "loss": 1.2853,
+      "step": 700
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.2526804860614723e-05,
+      "loss": 1.2548,
+      "step": 701
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.254467476769121e-05,
+      "loss": 1.2849,
+      "step": 702
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.2562544674767694e-05,
+      "loss": 1.2972,
+      "step": 703
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.2580414581844174e-05,
+      "loss": 1.3267,
+      "step": 704
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.2598284488920659e-05,
+      "loss": 1.2805,
+      "step": 705
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.261615439599714e-05,
+      "loss": 1.2546,
+      "step": 706
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.2634024303073625e-05,
+      "loss": 1.3002,
+      "step": 707
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.265189421015011e-05,
+      "loss": 1.2782,
+      "step": 708
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.266976411722659e-05,
+      "loss": 1.2978,
+      "step": 709
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.2687634024303074e-05,
+      "loss": 1.2776,
+      "step": 710
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.2705503931379559e-05,
+      "loss": 1.2986,
+      "step": 711
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.2723373838456041e-05,
+      "loss": 1.2612,
+      "step": 712
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.2741243745532525e-05,
+      "loss": 1.2937,
+      "step": 713
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.2759113652609006e-05,
+      "loss": 1.2726,
+      "step": 714
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.277698355968549e-05,
+      "loss": 1.3159,
+      "step": 715
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.2794853466761974e-05,
+      "loss": 1.2951,
+      "step": 716
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.2812723373838457e-05,
+      "loss": 1.2802,
+      "step": 717
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.2830593280914941e-05,
+      "loss": 1.2778,
+      "step": 718
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.2848463187991422e-05,
+      "loss": 1.2667,
+      "step": 719
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.2866333095067906e-05,
+      "loss": 1.2439,
+      "step": 720
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.288420300214439e-05,
+      "loss": 1.2838,
+      "step": 721
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.2902072909220873e-05,
+      "loss": 1.3305,
+      "step": 722
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.2919942816297357e-05,
+      "loss": 1.2609,
+      "step": 723
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.2937812723373838e-05,
+      "loss": 1.2846,
+      "step": 724
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.2955682630450322e-05,
+      "loss": 1.2951,
+      "step": 725
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.2973552537526806e-05,
+      "loss": 1.3005,
+      "step": 726
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.2991422444603289e-05,
+      "loss": 1.2965,
+      "step": 727
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.3009292351679773e-05,
+      "loss": 1.3011,
+      "step": 728
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.3027162258756257e-05,
+      "loss": 1.2524,
+      "step": 729
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.3045032165832738e-05,
+      "loss": 1.2784,
+      "step": 730
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.3062902072909222e-05,
+      "loss": 1.2816,
+      "step": 731
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.3080771979985704e-05,
+      "loss": 1.2804,
+      "step": 732
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.3098641887062189e-05,
+      "loss": 1.2532,
+      "step": 733
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.3116511794138673e-05,
+      "loss": 1.2667,
+      "step": 734
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.3134381701215154e-05,
+      "loss": 1.2942,
+      "step": 735
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.3152251608291638e-05,
+      "loss": 1.2843,
+      "step": 736
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.317012151536812e-05,
+      "loss": 1.2645,
+      "step": 737
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.3187991422444604e-05,
+      "loss": 1.2929,
+      "step": 738
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.3205861329521089e-05,
+      "loss": 1.2979,
+      "step": 739
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.322373123659757e-05,
+      "loss": 1.2541,
+      "step": 740
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.3241601143674054e-05,
+      "loss": 1.2775,
+      "step": 741
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.3259471050750536e-05,
+      "loss": 1.2561,
+      "step": 742
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.327734095782702e-05,
+      "loss": 1.2549,
+      "step": 743
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.3295210864903504e-05,
+      "loss": 1.287,
+      "step": 744
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.3313080771979985e-05,
+      "loss": 1.268,
+      "step": 745
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.333095067905647e-05,
+      "loss": 1.2454,
+      "step": 746
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.3348820586132954e-05,
+      "loss": 1.2728,
+      "step": 747
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.3366690493209436e-05,
+      "loss": 1.2637,
+      "step": 748
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.338456040028592e-05,
+      "loss": 1.2819,
+      "step": 749
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.3402430307362401e-05,
+      "loss": 1.3117,
+      "step": 750
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.3420300214438885e-05,
+      "loss": 1.2883,
+      "step": 751
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.343817012151537e-05,
+      "loss": 1.269,
+      "step": 752
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.3456040028591852e-05,
+      "loss": 1.2942,
+      "step": 753
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.3473909935668336e-05,
+      "loss": 1.2785,
+      "step": 754
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.3491779842744817e-05,
+      "loss": 1.2982,
+      "step": 755
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.3509649749821301e-05,
+      "loss": 1.3044,
+      "step": 756
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.3527519656897785e-05,
+      "loss": 1.2666,
+      "step": 757
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.3545389563974268e-05,
+      "loss": 1.2636,
+      "step": 758
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.3563259471050752e-05,
+      "loss": 1.2623,
+      "step": 759
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.3581129378127233e-05,
+      "loss": 1.2836,
+      "step": 760
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.3598999285203717e-05,
+      "loss": 1.2966,
+      "step": 761
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.3616869192280201e-05,
+      "loss": 1.2716,
+      "step": 762
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.3634739099356684e-05,
+      "loss": 1.3085,
+      "step": 763
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.3652609006433168e-05,
+      "loss": 1.2564,
+      "step": 764
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.3670478913509652e-05,
+      "loss": 1.2514,
+      "step": 765
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.3688348820586133e-05,
+      "loss": 1.3113,
+      "step": 766
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.3706218727662617e-05,
+      "loss": 1.2721,
+      "step": 767
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.37240886347391e-05,
+      "loss": 1.2818,
+      "step": 768
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.3741958541815584e-05,
+      "loss": 1.2831,
+      "step": 769
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.3759828448892068e-05,
+      "loss": 1.2461,
+      "step": 770
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.3777698355968549e-05,
+      "loss": 1.2475,
+      "step": 771
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.3795568263045033e-05,
+      "loss": 1.2678,
+      "step": 772
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.3813438170121515e-05,
+      "loss": 1.3052,
+      "step": 773
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.3831308077198e-05,
+      "loss": 1.2807,
+      "step": 774
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.3849177984274484e-05,
+      "loss": 1.2956,
+      "step": 775
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.3867047891350964e-05,
+      "loss": 1.2815,
+      "step": 776
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.3884917798427449e-05,
+      "loss": 1.2195,
+      "step": 777
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.3902787705503931e-05,
+      "loss": 1.2824,
+      "step": 778
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.3920657612580415e-05,
+      "loss": 1.2556,
+      "step": 779
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.39385275196569e-05,
+      "loss": 1.2882,
+      "step": 780
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.395639742673338e-05,
+      "loss": 1.2827,
+      "step": 781
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.3974267333809864e-05,
+      "loss": 1.2715,
+      "step": 782
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.3992137240886349e-05,
+      "loss": 1.2737,
+      "step": 783
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.4010007147962831e-05,
+      "loss": 1.2177,
+      "step": 784
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.4027877055039315e-05,
+      "loss": 1.2365,
+      "step": 785
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.4045746962115796e-05,
+      "loss": 1.2748,
+      "step": 786
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.406361686919228e-05,
+      "loss": 1.2871,
+      "step": 787
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.4081486776268764e-05,
+      "loss": 1.2931,
+      "step": 788
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.4099356683345247e-05,
+      "loss": 1.2701,
+      "step": 789
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.4117226590421731e-05,
+      "loss": 1.3061,
+      "step": 790
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.4135096497498212e-05,
+      "loss": 1.2985,
+      "step": 791
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.4152966404574696e-05,
+      "loss": 1.291,
+      "step": 792
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.417083631165118e-05,
+      "loss": 1.2854,
+      "step": 793
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.4188706218727663e-05,
+      "loss": 1.2422,
+      "step": 794
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.4206576125804147e-05,
+      "loss": 1.3061,
+      "step": 795
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.4224446032880628e-05,
+      "loss": 1.2632,
+      "step": 796
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.4242315939957112e-05,
+      "loss": 1.2564,
+      "step": 797
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.4260185847033596e-05,
+      "loss": 1.2986,
+      "step": 798
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.4278055754110079e-05,
+      "loss": 1.2636,
+      "step": 799
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.4295925661186563e-05,
+      "loss": 1.2556,
+      "step": 800
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.4313795568263047e-05,
+      "loss": 1.277,
+      "step": 801
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.4331665475339528e-05,
+      "loss": 1.2635,
+      "step": 802
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.4349535382416012e-05,
+      "loss": 1.2695,
+      "step": 803
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.4367405289492494e-05,
+      "loss": 1.2671,
+      "step": 804
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.4385275196568979e-05,
+      "loss": 1.2689,
+      "step": 805
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.4403145103645463e-05,
+      "loss": 1.2654,
+      "step": 806
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.4421015010721944e-05,
+      "loss": 1.2521,
+      "step": 807
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.4438884917798428e-05,
+      "loss": 1.2981,
+      "step": 808
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.445675482487491e-05,
+      "loss": 1.2446,
+      "step": 809
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.4474624731951395e-05,
+      "loss": 1.2553,
+      "step": 810
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.4492494639027879e-05,
+      "loss": 1.3102,
+      "step": 811
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.451036454610436e-05,
+      "loss": 1.2855,
+      "step": 812
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.4528234453180845e-05,
+      "loss": 1.2541,
+      "step": 813
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.4546104360257326e-05,
+      "loss": 1.2894,
+      "step": 814
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.456397426733381e-05,
+      "loss": 1.2612,
+      "step": 815
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.4581844174410295e-05,
+      "loss": 1.2623,
+      "step": 816
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.4599714081486775e-05,
+      "loss": 1.2732,
+      "step": 817
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.4617583988563261e-05,
+      "loss": 1.2561,
+      "step": 818
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.4635453895639745e-05,
+      "loss": 1.2622,
+      "step": 819
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.4653323802716226e-05,
+      "loss": 1.2561,
+      "step": 820
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.467119370979271e-05,
+      "loss": 1.2819,
+      "step": 821
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.4689063616869191e-05,
+      "loss": 1.2412,
+      "step": 822
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.4706933523945677e-05,
+      "loss": 1.2952,
+      "step": 823
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.4724803431022161e-05,
+      "loss": 1.2505,
+      "step": 824
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.4742673338098642e-05,
+      "loss": 1.2527,
+      "step": 825
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.4760543245175126e-05,
+      "loss": 1.2968,
+      "step": 826
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.4778413152251607e-05,
+      "loss": 1.319,
+      "step": 827
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.4796283059328093e-05,
+      "loss": 1.2727,
+      "step": 828
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.4814152966404577e-05,
+      "loss": 1.2851,
+      "step": 829
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.4832022873481058e-05,
+      "loss": 1.3092,
+      "step": 830
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.4849892780557542e-05,
+      "loss": 1.2626,
+      "step": 831
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.4867762687634025e-05,
+      "loss": 1.2641,
+      "step": 832
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.4885632594710509e-05,
+      "loss": 1.2847,
+      "step": 833
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.4903502501786993e-05,
+      "loss": 1.2804,
+      "step": 834
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.4921372408863474e-05,
+      "loss": 1.2269,
+      "step": 835
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.4939242315939958e-05,
+      "loss": 1.2421,
+      "step": 836
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.4957112223016442e-05,
+      "loss": 1.2365,
+      "step": 837
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.4974982130092925e-05,
+      "loss": 1.2726,
+      "step": 838
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.4992852037169409e-05,
+      "loss": 1.2609,
+      "step": 839
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.501072194424589e-05,
+      "loss": 1.2836,
+      "step": 840
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.5028591851322374e-05,
+      "loss": 1.2851,
+      "step": 841
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.5046461758398858e-05,
+      "loss": 1.2913,
+      "step": 842
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.506433166547534e-05,
+      "loss": 1.2733,
+      "step": 843
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.5082201572551825e-05,
+      "loss": 1.2321,
+      "step": 844
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.5100071479628305e-05,
+      "loss": 1.2664,
+      "step": 845
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.511794138670479e-05,
+      "loss": 1.2607,
+      "step": 846
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.5135811293781274e-05,
+      "loss": 1.2576,
+      "step": 847
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.5153681200857756e-05,
+      "loss": 1.2839,
+      "step": 848
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.517155110793424e-05,
+      "loss": 1.2748,
+      "step": 849
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.5189421015010721e-05,
+      "loss": 1.2458,
+      "step": 850
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.5207290922087205e-05,
+      "loss": 1.2773,
+      "step": 851
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.522516082916369e-05,
+      "loss": 1.279,
+      "step": 852
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.5243030736240172e-05,
+      "loss": 1.2362,
+      "step": 853
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.5260900643316656e-05,
+      "loss": 1.2795,
+      "step": 854
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.527877055039314e-05,
+      "loss": 1.2411,
+      "step": 855
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.529664045746962e-05,
+      "loss": 1.2717,
+      "step": 856
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.5314510364546105e-05,
+      "loss": 1.2548,
+      "step": 857
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.5332380271622586e-05,
+      "loss": 1.3166,
+      "step": 858
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.535025017869907e-05,
+      "loss": 1.2479,
+      "step": 859
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.5368120085775555e-05,
+      "loss": 1.2602,
+      "step": 860
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.538598999285204e-05,
+      "loss": 1.2904,
+      "step": 861
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.5403859899928523e-05,
+      "loss": 1.2754,
+      "step": 862
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.5421729807005004e-05,
+      "loss": 1.281,
+      "step": 863
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.5439599714081488e-05,
+      "loss": 1.2873,
+      "step": 864
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.5457469621157972e-05,
+      "loss": 1.2158,
+      "step": 865
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.5475339528234453e-05,
+      "loss": 1.2644,
+      "step": 866
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.5493209435310937e-05,
+      "loss": 1.2562,
+      "step": 867
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.5511079342387418e-05,
+      "loss": 1.2622,
+      "step": 868
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.5528949249463902e-05,
+      "loss": 1.2714,
+      "step": 869
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.5546819156540386e-05,
+      "loss": 1.2289,
+      "step": 870
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.556468906361687e-05,
+      "loss": 1.2577,
+      "step": 871
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.5582558970693355e-05,
+      "loss": 1.2838,
+      "step": 872
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.560042887776984e-05,
+      "loss": 1.2703,
+      "step": 873
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.561829878484632e-05,
+      "loss": 1.2612,
+      "step": 874
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.5636168691922804e-05,
+      "loss": 1.2589,
+      "step": 875
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.5654038598999285e-05,
+      "loss": 1.2817,
+      "step": 876
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.567190850607577e-05,
+      "loss": 1.2206,
+      "step": 877
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.5689778413152253e-05,
+      "loss": 1.2392,
+      "step": 878
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.5707648320228734e-05,
+      "loss": 1.2809,
+      "step": 879
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.5725518227305218e-05,
+      "loss": 1.2673,
+      "step": 880
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.5743388134381702e-05,
+      "loss": 1.2429,
+      "step": 881
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.5761258041458186e-05,
+      "loss": 1.2498,
+      "step": 882
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.577912794853467e-05,
+      "loss": 1.2566,
+      "step": 883
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.579699785561115e-05,
+      "loss": 1.3132,
+      "step": 884
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.5814867762687635e-05,
+      "loss": 1.2605,
+      "step": 885
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.5832737669764116e-05,
+      "loss": 1.2865,
+      "step": 886
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.58506075768406e-05,
+      "loss": 1.2661,
+      "step": 887
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.5868477483917085e-05,
+      "loss": 1.2403,
+      "step": 888
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.5886347390993565e-05,
+      "loss": 1.2484,
+      "step": 889
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.590421729807005e-05,
+      "loss": 1.2736,
+      "step": 890
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.5922087205146534e-05,
+      "loss": 1.2549,
+      "step": 891
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.5939957112223018e-05,
+      "loss": 1.2824,
+      "step": 892
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.5957827019299502e-05,
+      "loss": 1.2508,
+      "step": 893
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.5975696926375983e-05,
+      "loss": 1.2715,
+      "step": 894
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.5993566833452467e-05,
+      "loss": 1.2609,
+      "step": 895
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.601143674052895e-05,
+      "loss": 1.2775,
+      "step": 896
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.6029306647605432e-05,
+      "loss": 1.2829,
+      "step": 897
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.6047176554681916e-05,
+      "loss": 1.3052,
+      "step": 898
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.6065046461758397e-05,
+      "loss": 1.2969,
+      "step": 899
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.608291636883488e-05,
+      "loss": 1.2497,
+      "step": 900
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.6100786275911365e-05,
+      "loss": 1.2682,
+      "step": 901
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.611865618298785e-05,
+      "loss": 1.2727,
+      "step": 902
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.6136526090064334e-05,
+      "loss": 1.2552,
+      "step": 903
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.6154395997140815e-05,
+      "loss": 1.2682,
+      "step": 904
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.61722659042173e-05,
+      "loss": 1.2673,
+      "step": 905
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.6190135811293783e-05,
+      "loss": 1.2767,
+      "step": 906
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.6208005718370264e-05,
+      "loss": 1.3004,
+      "step": 907
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.6225875625446748e-05,
+      "loss": 1.2734,
+      "step": 908
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.6243745532523232e-05,
+      "loss": 1.2616,
+      "step": 909
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.6261615439599713e-05,
+      "loss": 1.2918,
+      "step": 910
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.6279485346676197e-05,
+      "loss": 1.2553,
+      "step": 911
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.629735525375268e-05,
+      "loss": 1.2728,
+      "step": 912
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.6315225160829166e-05,
+      "loss": 1.2991,
+      "step": 913
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.633309506790565e-05,
+      "loss": 1.2502,
+      "step": 914
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.635096497498213e-05,
+      "loss": 1.2597,
+      "step": 915
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.6368834882058615e-05,
+      "loss": 1.2761,
+      "step": 916
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.6386704789135095e-05,
+      "loss": 1.2687,
+      "step": 917
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.640457469621158e-05,
+      "loss": 1.2889,
+      "step": 918
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.6422444603288064e-05,
+      "loss": 1.2634,
+      "step": 919
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.6440314510364545e-05,
+      "loss": 1.2827,
+      "step": 920
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.645818441744103e-05,
+      "loss": 1.266,
+      "step": 921
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.6476054324517513e-05,
+      "loss": 1.2754,
+      "step": 922
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.6493924231593997e-05,
+      "loss": 1.2874,
+      "step": 923
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.651179413867048e-05,
+      "loss": 1.252,
+      "step": 924
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.6529664045746962e-05,
+      "loss": 1.3072,
+      "step": 925
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.6547533952823446e-05,
+      "loss": 1.2786,
+      "step": 926
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.656540385989993e-05,
+      "loss": 1.2682,
+      "step": 927
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.658327376697641e-05,
+      "loss": 1.2798,
+      "step": 928
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.6601143674052896e-05,
+      "loss": 1.2822,
+      "step": 929
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.6619013581129376e-05,
+      "loss": 1.2809,
+      "step": 930
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.663688348820586e-05,
+      "loss": 1.2485,
+      "step": 931
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.6654753395282348e-05,
+      "loss": 1.2807,
+      "step": 932
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.667262330235883e-05,
+      "loss": 1.2548,
+      "step": 933
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.6690493209435313e-05,
+      "loss": 1.2422,
+      "step": 934
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.6708363116511794e-05,
+      "loss": 1.2818,
+      "step": 935
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.6726233023588278e-05,
+      "loss": 1.2656,
+      "step": 936
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.6744102930664762e-05,
+      "loss": 1.2418,
+      "step": 937
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.6761972837741243e-05,
+      "loss": 1.2277,
+      "step": 938
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.6779842744817727e-05,
+      "loss": 1.2299,
+      "step": 939
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.6797712651894208e-05,
+      "loss": 1.2814,
+      "step": 940
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.6815582558970696e-05,
+      "loss": 1.285,
+      "step": 941
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.683345246604718e-05,
+      "loss": 1.2208,
+      "step": 942
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.685132237312366e-05,
+      "loss": 1.279,
+      "step": 943
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.6869192280200145e-05,
+      "loss": 1.274,
+      "step": 944
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.688706218727663e-05,
+      "loss": 1.2727,
+      "step": 945
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.690493209435311e-05,
+      "loss": 1.2588,
+      "step": 946
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.6922802001429594e-05,
+      "loss": 1.2658,
+      "step": 947
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.6940671908506075e-05,
+      "loss": 1.2423,
+      "step": 948
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.695854181558256e-05,
+      "loss": 1.2505,
+      "step": 949
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.6976411722659043e-05,
+      "loss": 1.2332,
+      "step": 950
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.6994281629735527e-05,
+      "loss": 1.2863,
+      "step": 951
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.701215153681201e-05,
+      "loss": 1.3024,
+      "step": 952
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.7030021443888492e-05,
+      "loss": 1.2885,
+      "step": 953
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.7047891350964976e-05,
+      "loss": 1.2761,
+      "step": 954
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.706576125804146e-05,
+      "loss": 1.249,
+      "step": 955
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.708363116511794e-05,
+      "loss": 1.2499,
+      "step": 956
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.7101501072194426e-05,
+      "loss": 1.2817,
+      "step": 957
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.7119370979270906e-05,
+      "loss": 1.3008,
+      "step": 958
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.713724088634739e-05,
+      "loss": 1.2464,
+      "step": 959
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.7155110793423875e-05,
+      "loss": 1.289,
+      "step": 960
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.717298070050036e-05,
+      "loss": 1.305,
+      "step": 961
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.7190850607576843e-05,
+      "loss": 1.2417,
+      "step": 962
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.7208720514653327e-05,
+      "loss": 1.2559,
+      "step": 963
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7226590421729808e-05,
+      "loss": 1.249,
+      "step": 964
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7244460328806292e-05,
+      "loss": 1.2547,
+      "step": 965
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7262330235882773e-05,
+      "loss": 1.2432,
+      "step": 966
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7280200142959257e-05,
+      "loss": 1.2706,
+      "step": 967
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.729807005003574e-05,
+      "loss": 1.2976,
+      "step": 968
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7315939957112222e-05,
+      "loss": 1.2717,
+      "step": 969
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7333809864188706e-05,
+      "loss": 1.2624,
+      "step": 970
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.735167977126519e-05,
+      "loss": 1.2732,
+      "step": 971
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7369549678341675e-05,
+      "loss": 1.2667,
+      "step": 972
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.738741958541816e-05,
+      "loss": 1.235,
+      "step": 973
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.740528949249464e-05,
+      "loss": 1.2718,
+      "step": 974
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7423159399571124e-05,
+      "loss": 1.2728,
+      "step": 975
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7441029306647605e-05,
+      "loss": 1.2714,
+      "step": 976
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.745889921372409e-05,
+      "loss": 1.3086,
+      "step": 977
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7476769120800573e-05,
+      "loss": 1.2294,
+      "step": 978
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7494639027877054e-05,
+      "loss": 1.3049,
+      "step": 979
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7512508934953538e-05,
+      "loss": 1.2852,
+      "step": 980
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7530378842030022e-05,
+      "loss": 1.2737,
+      "step": 981
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7548248749106506e-05,
+      "loss": 1.2777,
+      "step": 982
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.756611865618299e-05,
+      "loss": 1.2924,
+      "step": 983
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.758398856325947e-05,
+      "loss": 1.2939,
+      "step": 984
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7601858470335956e-05,
+      "loss": 1.2878,
+      "step": 985
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.761972837741244e-05,
+      "loss": 1.3003,
+      "step": 986
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.763759828448892e-05,
+      "loss": 1.2613,
+      "step": 987
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7655468191565405e-05,
+      "loss": 1.2957,
+      "step": 988
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7673338098641886e-05,
+      "loss": 1.2825,
+      "step": 989
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.769120800571837e-05,
+      "loss": 1.2796,
+      "step": 990
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7709077912794854e-05,
+      "loss": 1.2795,
+      "step": 991
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7726947819871338e-05,
+      "loss": 1.3086,
+      "step": 992
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7744817726947822e-05,
+      "loss": 1.2873,
+      "step": 993
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7762687634024303e-05,
+      "loss": 1.2357,
+      "step": 994
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7780557541100787e-05,
+      "loss": 1.2631,
+      "step": 995
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.779842744817727e-05,
+      "loss": 1.2471,
+      "step": 996
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7816297355253752e-05,
+      "loss": 1.2421,
+      "step": 997
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7834167262330236e-05,
+      "loss": 1.2675,
+      "step": 998
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.785203716940672e-05,
+      "loss": 1.2816,
+      "step": 999
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.78699070764832e-05,
+      "loss": 1.2397,
+      "step": 1000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7887776983559686e-05,
+      "loss": 1.2195,
+      "step": 1001
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.790564689063617e-05,
+      "loss": 1.2552,
+      "step": 1002
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7923516797712654e-05,
+      "loss": 1.2711,
+      "step": 1003
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7941386704789138e-05,
+      "loss": 1.2621,
+      "step": 1004
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.795925661186562e-05,
+      "loss": 1.2878,
+      "step": 1005
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7977126518942103e-05,
+      "loss": 1.3012,
+      "step": 1006
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.7994996426018584e-05,
+      "loss": 1.2563,
+      "step": 1007
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8012866333095068e-05,
+      "loss": 1.2405,
+      "step": 1008
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8030736240171552e-05,
+      "loss": 1.2675,
+      "step": 1009
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8048606147248033e-05,
+      "loss": 1.2611,
+      "step": 1010
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8066476054324517e-05,
+      "loss": 1.2579,
+      "step": 1011
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8084345961401e-05,
+      "loss": 1.2838,
+      "step": 1012
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8102215868477486e-05,
+      "loss": 1.2768,
+      "step": 1013
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.812008577555397e-05,
+      "loss": 1.2634,
+      "step": 1014
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.813795568263045e-05,
+      "loss": 1.2547,
+      "step": 1015
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8155825589706935e-05,
+      "loss": 1.258,
+      "step": 1016
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.817369549678342e-05,
+      "loss": 1.2509,
+      "step": 1017
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.81915654038599e-05,
+      "loss": 1.2908,
+      "step": 1018
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8209435310936384e-05,
+      "loss": 1.2703,
+      "step": 1019
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8227305218012865e-05,
+      "loss": 1.2909,
+      "step": 1020
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.824517512508935e-05,
+      "loss": 1.2822,
+      "step": 1021
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8263045032165833e-05,
+      "loss": 1.2313,
+      "step": 1022
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8280914939242317e-05,
+      "loss": 1.2953,
+      "step": 1023
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.82987848463188e-05,
+      "loss": 1.2633,
+      "step": 1024
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.8316654753395282e-05,
+      "loss": 1.2562,
+      "step": 1025
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8334524660471766e-05,
+      "loss": 1.2632,
+      "step": 1026
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.835239456754825e-05,
+      "loss": 1.2545,
+      "step": 1027
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.837026447462473e-05,
+      "loss": 1.2705,
+      "step": 1028
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8388134381701216e-05,
+      "loss": 1.267,
+      "step": 1029
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8406004288777696e-05,
+      "loss": 1.2599,
+      "step": 1030
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.842387419585418e-05,
+      "loss": 1.2535,
+      "step": 1031
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8441744102930665e-05,
+      "loss": 1.265,
+      "step": 1032
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.845961401000715e-05,
+      "loss": 1.2566,
+      "step": 1033
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8477483917083633e-05,
+      "loss": 1.2256,
+      "step": 1034
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8495353824160117e-05,
+      "loss": 1.2126,
+      "step": 1035
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8513223731236598e-05,
+      "loss": 1.2659,
+      "step": 1036
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8531093638313082e-05,
+      "loss": 1.2579,
+      "step": 1037
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8548963545389563e-05,
+      "loss": 1.2695,
+      "step": 1038
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8566833452466047e-05,
+      "loss": 1.293,
+      "step": 1039
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.858470335954253e-05,
+      "loss": 1.2459,
+      "step": 1040
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8602573266619012e-05,
+      "loss": 1.2377,
+      "step": 1041
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8620443173695496e-05,
+      "loss": 1.3188,
+      "step": 1042
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.863831308077198e-05,
+      "loss": 1.2694,
+      "step": 1043
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8656182987848465e-05,
+      "loss": 1.2533,
+      "step": 1044
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.867405289492495e-05,
+      "loss": 1.2646,
+      "step": 1045
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.869192280200143e-05,
+      "loss": 1.2536,
+      "step": 1046
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8709792709077914e-05,
+      "loss": 1.2698,
+      "step": 1047
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8727662616154395e-05,
+      "loss": 1.2792,
+      "step": 1048
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.874553252323088e-05,
+      "loss": 1.2695,
+      "step": 1049
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8763402430307363e-05,
+      "loss": 1.264,
+      "step": 1050
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8781272337383844e-05,
+      "loss": 1.2795,
+      "step": 1051
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.879914224446033e-05,
+      "loss": 1.2936,
+      "step": 1052
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8817012151536816e-05,
+      "loss": 1.2648,
+      "step": 1053
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8834882058613297e-05,
+      "loss": 1.2949,
+      "step": 1054
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.885275196568978e-05,
+      "loss": 1.2719,
+      "step": 1055
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.887062187276626e-05,
+      "loss": 1.3175,
+      "step": 1056
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8888491779842746e-05,
+      "loss": 1.2488,
+      "step": 1057
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.890636168691923e-05,
+      "loss": 1.2614,
+      "step": 1058
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.892423159399571e-05,
+      "loss": 1.2244,
+      "step": 1059
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8942101501072195e-05,
+      "loss": 1.2199,
+      "step": 1060
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8959971408148676e-05,
+      "loss": 1.2843,
+      "step": 1061
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8977841315225163e-05,
+      "loss": 1.2579,
+      "step": 1062
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8995711222301647e-05,
+      "loss": 1.2353,
+      "step": 1063
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9013581129378128e-05,
+      "loss": 1.2641,
+      "step": 1064
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9031451036454612e-05,
+      "loss": 1.2755,
+      "step": 1065
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9049320943531093e-05,
+      "loss": 1.2517,
+      "step": 1066
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9067190850607577e-05,
+      "loss": 1.2807,
+      "step": 1067
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.908506075768406e-05,
+      "loss": 1.275,
+      "step": 1068
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9102930664760542e-05,
+      "loss": 1.2289,
+      "step": 1069
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9120800571837027e-05,
+      "loss": 1.2609,
+      "step": 1070
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.913867047891351e-05,
+      "loss": 1.2549,
+      "step": 1071
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9156540385989995e-05,
+      "loss": 1.2532,
+      "step": 1072
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.917441029306648e-05,
+      "loss": 1.2492,
+      "step": 1073
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.919228020014296e-05,
+      "loss": 1.2948,
+      "step": 1074
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9210150107219444e-05,
+      "loss": 1.2636,
+      "step": 1075
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9228020014295928e-05,
+      "loss": 1.2311,
+      "step": 1076
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.924588992137241e-05,
+      "loss": 1.2471,
+      "step": 1077
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9263759828448893e-05,
+      "loss": 1.2839,
+      "step": 1078
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9281629735525374e-05,
+      "loss": 1.3036,
+      "step": 1079
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9299499642601858e-05,
+      "loss": 1.3034,
+      "step": 1080
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9317369549678342e-05,
+      "loss": 1.3004,
+      "step": 1081
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9335239456754827e-05,
+      "loss": 1.2597,
+      "step": 1082
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.935310936383131e-05,
+      "loss": 1.2747,
+      "step": 1083
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.937097927090779e-05,
+      "loss": 1.2388,
+      "step": 1084
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9388849177984276e-05,
+      "loss": 1.2485,
+      "step": 1085
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.940671908506076e-05,
+      "loss": 1.2825,
+      "step": 1086
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.942458899213724e-05,
+      "loss": 1.2352,
+      "step": 1087
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9442458899213725e-05,
+      "loss": 1.2616,
+      "step": 1088
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.946032880629021e-05,
+      "loss": 1.2742,
+      "step": 1089
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.947819871336669e-05,
+      "loss": 1.2894,
+      "step": 1090
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9496068620443174e-05,
+      "loss": 1.2582,
+      "step": 1091
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9513938527519658e-05,
+      "loss": 1.2321,
+      "step": 1092
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9531808434596142e-05,
+      "loss": 1.2857,
+      "step": 1093
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9549678341672627e-05,
+      "loss": 1.254,
+      "step": 1094
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9567548248749107e-05,
+      "loss": 1.2516,
+      "step": 1095
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.958541815582559e-05,
+      "loss": 1.2579,
+      "step": 1096
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9603288062902072e-05,
+      "loss": 1.2503,
+      "step": 1097
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9621157969978557e-05,
+      "loss": 1.2233,
+      "step": 1098
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.963902787705504e-05,
+      "loss": 1.2491,
+      "step": 1099
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.965689778413152e-05,
+      "loss": 1.281,
+      "step": 1100
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9674767691208006e-05,
+      "loss": 1.2833,
+      "step": 1101
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.969263759828449e-05,
+      "loss": 1.2935,
+      "step": 1102
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9710507505360974e-05,
+      "loss": 1.2405,
+      "step": 1103
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9728377412437458e-05,
+      "loss": 1.2544,
+      "step": 1104
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.974624731951394e-05,
+      "loss": 1.2572,
+      "step": 1105
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9764117226590423e-05,
+      "loss": 1.33,
+      "step": 1106
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9781987133666907e-05,
+      "loss": 1.2499,
+      "step": 1107
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9799857040743388e-05,
+      "loss": 1.2878,
+      "step": 1108
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9817726947819872e-05,
+      "loss": 1.2704,
+      "step": 1109
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9835596854896353e-05,
+      "loss": 1.2464,
+      "step": 1110
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9853466761972837e-05,
+      "loss": 1.2708,
+      "step": 1111
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.987133666904932e-05,
+      "loss": 1.22,
+      "step": 1112
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9889206576125806e-05,
+      "loss": 1.2749,
+      "step": 1113
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.990707648320229e-05,
+      "loss": 1.2586,
+      "step": 1114
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.992494639027877e-05,
+      "loss": 1.2548,
+      "step": 1115
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9942816297355255e-05,
+      "loss": 1.2274,
+      "step": 1116
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.996068620443174e-05,
+      "loss": 1.2696,
+      "step": 1117
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.997855611150822e-05,
+      "loss": 1.2601,
+      "step": 1118
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9996426018584704e-05,
+      "loss": 1.2934,
+      "step": 1119
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0014295925661185e-05,
+      "loss": 1.2294,
+      "step": 1120
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.003216583273767e-05,
+      "loss": 1.2204,
+      "step": 1121
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0050035739814153e-05,
+      "loss": 1.2618,
+      "step": 1122
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0067905646890637e-05,
+      "loss": 1.2667,
+      "step": 1123
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.008577555396712e-05,
+      "loss": 1.2659,
+      "step": 1124
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0103645461043606e-05,
+      "loss": 1.2289,
+      "step": 1125
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0121515368120087e-05,
+      "loss": 1.2927,
+      "step": 1126
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.013938527519657e-05,
+      "loss": 1.2642,
+      "step": 1127
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.015725518227305e-05,
+      "loss": 1.237,
+      "step": 1128
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0175125089349536e-05,
+      "loss": 1.2885,
+      "step": 1129
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.019299499642602e-05,
+      "loss": 1.2697,
+      "step": 1130
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.02108649035025e-05,
+      "loss": 1.2549,
+      "step": 1131
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0228734810578985e-05,
+      "loss": 1.2418,
+      "step": 1132
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.024660471765547e-05,
+      "loss": 1.2655,
+      "step": 1133
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0264474624731953e-05,
+      "loss": 1.2385,
+      "step": 1134
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0282344531808438e-05,
+      "loss": 1.231,
+      "step": 1135
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.030021443888492e-05,
+      "loss": 1.2851,
+      "step": 1136
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0318084345961402e-05,
+      "loss": 1.2545,
+      "step": 1137
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0335954253037883e-05,
+      "loss": 1.2663,
+      "step": 1138
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0353824160114367e-05,
+      "loss": 1.249,
+      "step": 1139
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.037169406719085e-05,
+      "loss": 1.2186,
+      "step": 1140
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0389563974267332e-05,
+      "loss": 1.2969,
+      "step": 1141
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0407433881343817e-05,
+      "loss": 1.2766,
+      "step": 1142
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.04253037884203e-05,
+      "loss": 1.2478,
+      "step": 1143
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0443173695496785e-05,
+      "loss": 1.2777,
+      "step": 1144
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.046104360257327e-05,
+      "loss": 1.2636,
+      "step": 1145
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.047891350964975e-05,
+      "loss": 1.2586,
+      "step": 1146
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0496783416726234e-05,
+      "loss": 1.2406,
+      "step": 1147
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.051465332380272e-05,
+      "loss": 1.2638,
+      "step": 1148
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.05325232308792e-05,
+      "loss": 1.2532,
+      "step": 1149
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.0550393137955683e-05,
+      "loss": 1.2901,
+      "step": 1150
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0568263045032164e-05,
+      "loss": 1.2788,
+      "step": 1151
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.058613295210865e-05,
+      "loss": 1.2594,
+      "step": 1152
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0604002859185132e-05,
+      "loss": 1.2638,
+      "step": 1153
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0621872766261617e-05,
+      "loss": 1.2023,
+      "step": 1154
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.06397426733381e-05,
+      "loss": 1.2805,
+      "step": 1155
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.065761258041458e-05,
+      "loss": 1.2757,
+      "step": 1156
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0675482487491066e-05,
+      "loss": 1.2756,
+      "step": 1157
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.069335239456755e-05,
+      "loss": 1.2422,
+      "step": 1158
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.071122230164403e-05,
+      "loss": 1.2608,
+      "step": 1159
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0729092208720515e-05,
+      "loss": 1.2838,
+      "step": 1160
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0746962115797e-05,
+      "loss": 1.2676,
+      "step": 1161
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.076483202287348e-05,
+      "loss": 1.2739,
+      "step": 1162
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0782701929949968e-05,
+      "loss": 1.2823,
+      "step": 1163
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.080057183702645e-05,
+      "loss": 1.2822,
+      "step": 1164
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0818441744102933e-05,
+      "loss": 1.2287,
+      "step": 1165
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0836311651179417e-05,
+      "loss": 1.2662,
+      "step": 1166
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0854181558255898e-05,
+      "loss": 1.2562,
+      "step": 1167
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0872051465332382e-05,
+      "loss": 1.2993,
+      "step": 1168
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0889921372408862e-05,
+      "loss": 1.2834,
+      "step": 1169
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0907791279485347e-05,
+      "loss": 1.2701,
+      "step": 1170
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.092566118656183e-05,
+      "loss": 1.2438,
+      "step": 1171
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.094353109363831e-05,
+      "loss": 1.2447,
+      "step": 1172
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.09614010007148e-05,
+      "loss": 1.2661,
+      "step": 1173
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.097927090779128e-05,
+      "loss": 1.2461,
+      "step": 1174
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.0997140814867764e-05,
+      "loss": 1.2625,
+      "step": 1175
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.101501072194425e-05,
+      "loss": 1.2531,
+      "step": 1176
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.103288062902073e-05,
+      "loss": 1.2773,
+      "step": 1177
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.1050750536097213e-05,
+      "loss": 1.2876,
+      "step": 1178
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.1068620443173698e-05,
+      "loss": 1.2849,
+      "step": 1179
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.108649035025018e-05,
+      "loss": 1.2606,
+      "step": 1180
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.1104360257326663e-05,
+      "loss": 1.2923,
+      "step": 1181
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.1122230164403147e-05,
+      "loss": 1.2933,
+      "step": 1182
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.114010007147963e-05,
+      "loss": 1.2546,
+      "step": 1183
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.1157969978556115e-05,
+      "loss": 1.2665,
+      "step": 1184
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.1175839885632596e-05,
+      "loss": 1.2695,
+      "step": 1185
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.119370979270908e-05,
+      "loss": 1.2719,
+      "step": 1186
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.121157969978556e-05,
+      "loss": 1.2702,
+      "step": 1187
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.1229449606862045e-05,
+      "loss": 1.2313,
+      "step": 1188
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.124731951393853e-05,
+      "loss": 1.2307,
+      "step": 1189
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.126518942101501e-05,
+      "loss": 1.25,
+      "step": 1190
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.1283059328091494e-05,
+      "loss": 1.2539,
+      "step": 1191
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.130092923516798e-05,
+      "loss": 1.2707,
+      "step": 1192
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.1318799142244463e-05,
+      "loss": 1.2909,
+      "step": 1193
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.1336669049320947e-05,
+      "loss": 1.2844,
+      "step": 1194
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.1354538956397428e-05,
+      "loss": 1.28,
+      "step": 1195
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.1372408863473912e-05,
+      "loss": 1.2754,
+      "step": 1196
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.1390278770550396e-05,
+      "loss": 1.2721,
+      "step": 1197
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.1408148677626877e-05,
+      "loss": 1.2763,
+      "step": 1198
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.142601858470336e-05,
+      "loss": 1.257,
+      "step": 1199
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.1443888491779842e-05,
+      "loss": 1.2819,
+      "step": 1200
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.1461758398856326e-05,
+      "loss": 1.2943,
+      "step": 1201
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.147962830593281e-05,
+      "loss": 1.2635,
+      "step": 1202
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.1497498213009294e-05,
+      "loss": 1.2817,
+      "step": 1203
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.151536812008578e-05,
+      "loss": 1.2654,
+      "step": 1204
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.153323802716226e-05,
+      "loss": 1.2715,
+      "step": 1205
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.1551107934238743e-05,
+      "loss": 1.2785,
+      "step": 1206
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.1568977841315228e-05,
+      "loss": 1.2354,
+      "step": 1207
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.158684774839171e-05,
+      "loss": 1.2457,
+      "step": 1208
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.1604717655468193e-05,
+      "loss": 1.2607,
+      "step": 1209
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.1622587562544673e-05,
+      "loss": 1.2644,
+      "step": 1210
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.1640457469621158e-05,
+      "loss": 1.2843,
+      "step": 1211
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.1658327376697642e-05,
+      "loss": 1.2987,
+      "step": 1212
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.1676197283774126e-05,
+      "loss": 1.2337,
+      "step": 1213
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.169406719085061e-05,
+      "loss": 1.2544,
+      "step": 1214
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.1711937097927094e-05,
+      "loss": 1.2967,
+      "step": 1215
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.1729807005003575e-05,
+      "loss": 1.2665,
+      "step": 1216
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.174767691208006e-05,
+      "loss": 1.2442,
+      "step": 1217
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.176554681915654e-05,
+      "loss": 1.2454,
+      "step": 1218
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.1783416726233024e-05,
+      "loss": 1.2772,
+      "step": 1219
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.180128663330951e-05,
+      "loss": 1.2379,
+      "step": 1220
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.181915654038599e-05,
+      "loss": 1.2405,
+      "step": 1221
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.1837026447462473e-05,
+      "loss": 1.2716,
+      "step": 1222
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.1854896354538958e-05,
+      "loss": 1.2474,
+      "step": 1223
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.1872766261615442e-05,
+      "loss": 1.247,
+      "step": 1224
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.1890636168691926e-05,
+      "loss": 1.2824,
+      "step": 1225
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.1908506075768407e-05,
+      "loss": 1.2915,
+      "step": 1226
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.192637598284489e-05,
+      "loss": 1.2879,
+      "step": 1227
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.1944245889921372e-05,
+      "loss": 1.2322,
+      "step": 1228
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.1962115796997856e-05,
+      "loss": 1.2697,
+      "step": 1229
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.197998570407434e-05,
+      "loss": 1.2599,
+      "step": 1230
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.199785561115082e-05,
+      "loss": 1.2842,
+      "step": 1231
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.2015725518227305e-05,
+      "loss": 1.2375,
+      "step": 1232
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.203359542530379e-05,
+      "loss": 1.2683,
+      "step": 1233
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.2051465332380273e-05,
+      "loss": 1.2123,
+      "step": 1234
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.2069335239456758e-05,
+      "loss": 1.2494,
+      "step": 1235
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.208720514653324e-05,
+      "loss": 1.2117,
+      "step": 1236
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.2105075053609723e-05,
+      "loss": 1.2404,
+      "step": 1237
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.2122944960686207e-05,
+      "loss": 1.221,
+      "step": 1238
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.2140814867762688e-05,
+      "loss": 1.2842,
+      "step": 1239
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.2158684774839172e-05,
+      "loss": 1.2622,
+      "step": 1240
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.2176554681915653e-05,
+      "loss": 1.2626,
+      "step": 1241
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.2194424588992137e-05,
+      "loss": 1.2418,
+      "step": 1242
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.221229449606862e-05,
+      "loss": 1.2834,
+      "step": 1243
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.2230164403145105e-05,
+      "loss": 1.2702,
+      "step": 1244
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.224803431022159e-05,
+      "loss": 1.2398,
+      "step": 1245
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.226590421729807e-05,
+      "loss": 1.2575,
+      "step": 1246
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.2283774124374554e-05,
+      "loss": 1.2617,
+      "step": 1247
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.230164403145104e-05,
+      "loss": 1.2315,
+      "step": 1248
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.231951393852752e-05,
+      "loss": 1.2878,
+      "step": 1249
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.2337383845604003e-05,
+      "loss": 1.2442,
+      "step": 1250
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.2355253752680488e-05,
+      "loss": 1.2743,
+      "step": 1251
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.237312365975697e-05,
+      "loss": 1.2399,
+      "step": 1252
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.2390993566833453e-05,
+      "loss": 1.2322,
+      "step": 1253
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.2408863473909937e-05,
+      "loss": 1.263,
+      "step": 1254
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.242673338098642e-05,
+      "loss": 1.2703,
+      "step": 1255
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.2444603288062905e-05,
+      "loss": 1.2773,
+      "step": 1256
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.2462473195139386e-05,
+      "loss": 1.2244,
+      "step": 1257
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.248034310221587e-05,
+      "loss": 1.2677,
+      "step": 1258
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.249821300929235e-05,
+      "loss": 1.2842,
+      "step": 1259
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.2516082916368835e-05,
+      "loss": 1.2812,
+      "step": 1260
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.253395282344532e-05,
+      "loss": 1.2656,
+      "step": 1261
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.25518227305218e-05,
+      "loss": 1.2507,
+      "step": 1262
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.2569692637598284e-05,
+      "loss": 1.304,
+      "step": 1263
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.258756254467477e-05,
+      "loss": 1.2551,
+      "step": 1264
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.2605432451751253e-05,
+      "loss": 1.2924,
+      "step": 1265
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.2623302358827737e-05,
+      "loss": 1.2498,
+      "step": 1266
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.2641172265904218e-05,
+      "loss": 1.2817,
+      "step": 1267
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.2659042172980702e-05,
+      "loss": 1.2555,
+      "step": 1268
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.2676912080057186e-05,
+      "loss": 1.2622,
+      "step": 1269
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.2694781987133667e-05,
+      "loss": 1.2372,
+      "step": 1270
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.271265189421015e-05,
+      "loss": 1.2794,
+      "step": 1271
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.2730521801286632e-05,
+      "loss": 1.2353,
+      "step": 1272
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.2748391708363116e-05,
+      "loss": 1.2231,
+      "step": 1273
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.2766261615439604e-05,
+      "loss": 1.2811,
+      "step": 1274
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.2784131522516084e-05,
+      "loss": 1.2436,
+      "step": 1275
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.280200142959257e-05,
+      "loss": 1.278,
+      "step": 1276
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.281987133666905e-05,
+      "loss": 1.2656,
+      "step": 1277
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.2837741243745534e-05,
+      "loss": 1.28,
+      "step": 1278
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.2855611150822018e-05,
+      "loss": 1.2265,
+      "step": 1279
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.28734810578985e-05,
+      "loss": 1.2737,
+      "step": 1280
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.2891350964974983e-05,
+      "loss": 1.2663,
+      "step": 1281
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.2909220872051463e-05,
+      "loss": 1.2427,
+      "step": 1282
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.2927090779127948e-05,
+      "loss": 1.2425,
+      "step": 1283
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.2944960686204435e-05,
+      "loss": 1.2686,
+      "step": 1284
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.2962830593280916e-05,
+      "loss": 1.2669,
+      "step": 1285
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.29807005003574e-05,
+      "loss": 1.2443,
+      "step": 1286
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.2998570407433884e-05,
+      "loss": 1.2516,
+      "step": 1287
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.3016440314510365e-05,
+      "loss": 1.2684,
+      "step": 1288
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.303431022158685e-05,
+      "loss": 1.2436,
+      "step": 1289
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.305218012866333e-05,
+      "loss": 1.2649,
+      "step": 1290
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.3070050035739814e-05,
+      "loss": 1.2585,
+      "step": 1291
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.30879199428163e-05,
+      "loss": 1.2822,
+      "step": 1292
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.3105789849892783e-05,
+      "loss": 1.2945,
+      "step": 1293
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.3123659756969267e-05,
+      "loss": 1.2684,
+      "step": 1294
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.3141529664045748e-05,
+      "loss": 1.2589,
+      "step": 1295
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.3159399571122232e-05,
+      "loss": 1.2574,
+      "step": 1296
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.3177269478198716e-05,
+      "loss": 1.2836,
+      "step": 1297
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.3195139385275197e-05,
+      "loss": 1.2513,
+      "step": 1298
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.321300929235168e-05,
+      "loss": 1.2599,
+      "step": 1299
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.3230879199428162e-05,
+      "loss": 1.2589,
+      "step": 1300
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.3248749106504646e-05,
+      "loss": 1.2335,
+      "step": 1301
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.326661901358113e-05,
+      "loss": 1.2386,
+      "step": 1302
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.3284488920657614e-05,
+      "loss": 1.2731,
+      "step": 1303
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.33023588277341e-05,
+      "loss": 1.2364,
+      "step": 1304
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.3320228734810583e-05,
+      "loss": 1.2738,
+      "step": 1305
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.3338098641887064e-05,
+      "loss": 1.2639,
+      "step": 1306
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.3355968548963548e-05,
+      "loss": 1.2529,
+      "step": 1307
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.337383845604003e-05,
+      "loss": 1.2371,
+      "step": 1308
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.3391708363116513e-05,
+      "loss": 1.2802,
+      "step": 1309
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.3409578270192997e-05,
+      "loss": 1.2923,
+      "step": 1310
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.3427448177269478e-05,
+      "loss": 1.2866,
+      "step": 1311
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.3445318084345962e-05,
+      "loss": 1.2199,
+      "step": 1312
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.3463187991422446e-05,
+      "loss": 1.2557,
+      "step": 1313
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.348105789849893e-05,
+      "loss": 1.2272,
+      "step": 1314
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.3498927805575414e-05,
+      "loss": 1.2409,
+      "step": 1315
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.3516797712651895e-05,
+      "loss": 1.2503,
+      "step": 1316
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.353466761972838e-05,
+      "loss": 1.264,
+      "step": 1317
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.355253752680486e-05,
+      "loss": 1.2306,
+      "step": 1318
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.3570407433881344e-05,
+      "loss": 1.2619,
+      "step": 1319
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.358827734095783e-05,
+      "loss": 1.2859,
+      "step": 1320
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.360614724803431e-05,
+      "loss": 1.3136,
+      "step": 1321
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.3624017155110794e-05,
+      "loss": 1.2808,
+      "step": 1322
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.3641887062187278e-05,
+      "loss": 1.2617,
+      "step": 1323
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.3659756969263762e-05,
+      "loss": 1.2648,
+      "step": 1324
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.3677626876340246e-05,
+      "loss": 1.2297,
+      "step": 1325
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.3695496783416727e-05,
+      "loss": 1.2269,
+      "step": 1326
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.371336669049321e-05,
+      "loss": 1.2493,
+      "step": 1327
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.3731236597569695e-05,
+      "loss": 1.2557,
+      "step": 1328
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.3749106504646176e-05,
+      "loss": 1.2323,
+      "step": 1329
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.376697641172266e-05,
+      "loss": 1.2997,
+      "step": 1330
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.378484631879914e-05,
+      "loss": 1.2773,
+      "step": 1331
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.3802716225875625e-05,
+      "loss": 1.2385,
+      "step": 1332
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.382058613295211e-05,
+      "loss": 1.2697,
+      "step": 1333
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.3838456040028594e-05,
+      "loss": 1.2652,
+      "step": 1334
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.3856325947105078e-05,
+      "loss": 1.3037,
+      "step": 1335
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.387419585418156e-05,
+      "loss": 1.2442,
+      "step": 1336
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.3892065761258043e-05,
+      "loss": 1.2647,
+      "step": 1337
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.3909935668334527e-05,
+      "loss": 1.2592,
+      "step": 1338
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.3927805575411008e-05,
+      "loss": 1.2468,
+      "step": 1339
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.3945675482487492e-05,
+      "loss": 1.2387,
+      "step": 1340
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.3963545389563976e-05,
+      "loss": 1.23,
+      "step": 1341
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.3981415296640457e-05,
+      "loss": 1.2536,
+      "step": 1342
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.399928520371694e-05,
+      "loss": 1.2537,
+      "step": 1343
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.4017155110793425e-05,
+      "loss": 1.2337,
+      "step": 1344
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.403502501786991e-05,
+      "loss": 1.2347,
+      "step": 1345
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.4052894924946394e-05,
+      "loss": 1.2507,
+      "step": 1346
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.4070764832022874e-05,
+      "loss": 1.2849,
+      "step": 1347
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.408863473909936e-05,
+      "loss": 1.2452,
+      "step": 1348
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.410650464617584e-05,
+      "loss": 1.2675,
+      "step": 1349
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.4124374553252324e-05,
+      "loss": 1.2766,
+      "step": 1350
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.4142244460328808e-05,
+      "loss": 1.2689,
+      "step": 1351
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.416011436740529e-05,
+      "loss": 1.2857,
+      "step": 1352
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.4177984274481773e-05,
+      "loss": 1.2561,
+      "step": 1353
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.4195854181558257e-05,
+      "loss": 1.2641,
+      "step": 1354
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.421372408863474e-05,
+      "loss": 1.2548,
+      "step": 1355
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.4231593995711225e-05,
+      "loss": 1.2588,
+      "step": 1356
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.4249463902787706e-05,
+      "loss": 1.2693,
+      "step": 1357
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.426733380986419e-05,
+      "loss": 1.2757,
+      "step": 1358
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.4285203716940674e-05,
+      "loss": 1.2483,
+      "step": 1359
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.4303073624017155e-05,
+      "loss": 1.2662,
+      "step": 1360
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.432094353109364e-05,
+      "loss": 1.262,
+      "step": 1361
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.433881343817012e-05,
+      "loss": 1.282,
+      "step": 1362
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.4356683345246604e-05,
+      "loss": 1.3008,
+      "step": 1363
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.437455325232309e-05,
+      "loss": 1.2506,
+      "step": 1364
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.4392423159399573e-05,
+      "loss": 1.2432,
+      "step": 1365
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.4410293066476057e-05,
+      "loss": 1.2583,
+      "step": 1366
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.4428162973552538e-05,
+      "loss": 1.2781,
+      "step": 1367
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.4446032880629022e-05,
+      "loss": 1.2744,
+      "step": 1368
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.4463902787705506e-05,
+      "loss": 1.2538,
+      "step": 1369
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.4481772694781987e-05,
+      "loss": 1.2553,
+      "step": 1370
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.449964260185847e-05,
+      "loss": 1.2953,
+      "step": 1371
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.4517512508934952e-05,
+      "loss": 1.2494,
+      "step": 1372
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.4535382416011436e-05,
+      "loss": 1.2749,
+      "step": 1373
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.455325232308792e-05,
+      "loss": 1.2501,
+      "step": 1374
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.4571122230164404e-05,
+      "loss": 1.2674,
+      "step": 1375
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.458899213724089e-05,
+      "loss": 1.2545,
+      "step": 1376
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.4606862044317373e-05,
+      "loss": 1.2521,
+      "step": 1377
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.4624731951393854e-05,
+      "loss": 1.2738,
+      "step": 1378
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.4642601858470338e-05,
+      "loss": 1.2512,
+      "step": 1379
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.466047176554682e-05,
+      "loss": 1.2468,
+      "step": 1380
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.4678341672623303e-05,
+      "loss": 1.232,
+      "step": 1381
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.4696211579699787e-05,
+      "loss": 1.2461,
+      "step": 1382
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.4714081486776268e-05,
+      "loss": 1.2712,
+      "step": 1383
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.4731951393852752e-05,
+      "loss": 1.2173,
+      "step": 1384
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.4749821300929236e-05,
+      "loss": 1.2622,
+      "step": 1385
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.476769120800572e-05,
+      "loss": 1.2308,
+      "step": 1386
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.4785561115082205e-05,
+      "loss": 1.2508,
+      "step": 1387
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.4803431022158685e-05,
+      "loss": 1.2602,
+      "step": 1388
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.482130092923517e-05,
+      "loss": 1.2517,
+      "step": 1389
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.483917083631165e-05,
+      "loss": 1.2708,
+      "step": 1390
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.4857040743388135e-05,
+      "loss": 1.2454,
+      "step": 1391
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.487491065046462e-05,
+      "loss": 1.2441,
+      "step": 1392
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.48927805575411e-05,
+      "loss": 1.2578,
+      "step": 1393
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.4910650464617584e-05,
+      "loss": 1.2482,
+      "step": 1394
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.492852037169407e-05,
+      "loss": 1.2471,
+      "step": 1395
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.4946390278770552e-05,
+      "loss": 1.2582,
+      "step": 1396
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.4964260185847036e-05,
+      "loss": 1.2619,
+      "step": 1397
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.4982130092923517e-05,
+      "loss": 1.2425,
+      "step": 1398
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.5e-05,
+      "loss": 1.2641,
+      "step": 1399
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.5017869907076485e-05,
+      "loss": 1.281,
+      "step": 1400
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.503573981415297e-05,
+      "loss": 1.28,
+      "step": 1401
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.5053609721229447e-05,
+      "loss": 1.2312,
+      "step": 1402
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.507147962830593e-05,
+      "loss": 1.2722,
+      "step": 1403
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.508934953538242e-05,
+      "loss": 1.2847,
+      "step": 1404
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.5107219442458903e-05,
+      "loss": 1.2854,
+      "step": 1405
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.5125089349535387e-05,
+      "loss": 1.2529,
+      "step": 1406
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.5142959256611865e-05,
+      "loss": 1.2496,
+      "step": 1407
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.516082916368835e-05,
+      "loss": 1.2607,
+      "step": 1408
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.5178699070764833e-05,
+      "loss": 1.2722,
+      "step": 1409
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.5196568977841317e-05,
+      "loss": 1.2621,
+      "step": 1410
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.52144388849178e-05,
+      "loss": 1.2391,
+      "step": 1411
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.523230879199428e-05,
+      "loss": 1.2479,
+      "step": 1412
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.5250178699070763e-05,
+      "loss": 1.2276,
+      "step": 1413
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.526804860614725e-05,
+      "loss": 1.2615,
+      "step": 1414
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.5285918513223735e-05,
+      "loss": 1.2901,
+      "step": 1415
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.530378842030022e-05,
+      "loss": 1.2741,
+      "step": 1416
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.5321658327376703e-05,
+      "loss": 1.2659,
+      "step": 1417
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.533952823445318e-05,
+      "loss": 1.2551,
+      "step": 1418
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.5357398141529665e-05,
+      "loss": 1.2615,
+      "step": 1419
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.537526804860615e-05,
+      "loss": 1.2477,
+      "step": 1420
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.5393137955682633e-05,
+      "loss": 1.2547,
+      "step": 1421
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.5411007862759117e-05,
+      "loss": 1.2437,
+      "step": 1422
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.5428877769835598e-05,
+      "loss": 1.2584,
+      "step": 1423
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.5446747676912082e-05,
+      "loss": 1.277,
+      "step": 1424
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.5464617583988566e-05,
+      "loss": 1.2472,
+      "step": 1425
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.548248749106505e-05,
+      "loss": 1.2623,
+      "step": 1426
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.5500357398141535e-05,
+      "loss": 1.2718,
+      "step": 1427
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.5518227305218012e-05,
+      "loss": 1.2703,
+      "step": 1428
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.5536097212294496e-05,
+      "loss": 1.2524,
+      "step": 1429
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.555396711937098e-05,
+      "loss": 1.2382,
+      "step": 1430
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.5571837026447465e-05,
+      "loss": 1.28,
+      "step": 1431
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.558970693352395e-05,
+      "loss": 1.2713,
+      "step": 1432
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.560757684060043e-05,
+      "loss": 1.2717,
+      "step": 1433
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.5625446747676914e-05,
+      "loss": 1.2977,
+      "step": 1434
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.5643316654753398e-05,
+      "loss": 1.2392,
+      "step": 1435
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.5661186561829882e-05,
+      "loss": 1.2588,
+      "step": 1436
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.5679056468906366e-05,
+      "loss": 1.2694,
+      "step": 1437
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.5696926375982844e-05,
+      "loss": 1.2814,
+      "step": 1438
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.5714796283059328e-05,
+      "loss": 1.1955,
+      "step": 1439
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.5732666190135812e-05,
+      "loss": 1.288,
+      "step": 1440
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.5750536097212296e-05,
+      "loss": 1.2568,
+      "step": 1441
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.576840600428878e-05,
+      "loss": 1.2605,
+      "step": 1442
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.578627591136526e-05,
+      "loss": 1.2234,
+      "step": 1443
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.5804145818441745e-05,
+      "loss": 1.2629,
+      "step": 1444
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.582201572551823e-05,
+      "loss": 1.2538,
+      "step": 1445
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.5839885632594714e-05,
+      "loss": 1.2843,
+      "step": 1446
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.5857755539671198e-05,
+      "loss": 1.2364,
+      "step": 1447
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.5875625446747675e-05,
+      "loss": 1.2472,
+      "step": 1448
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.589349535382416e-05,
+      "loss": 1.2612,
+      "step": 1449
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.5911365260900644e-05,
+      "loss": 1.2867,
+      "step": 1450
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.5929235167977128e-05,
+      "loss": 1.2468,
+      "step": 1451
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.5947105075053612e-05,
+      "loss": 1.2606,
+      "step": 1452
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.5964974982130096e-05,
+      "loss": 1.2679,
+      "step": 1453
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.5982844889206577e-05,
+      "loss": 1.2282,
+      "step": 1454
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.600071479628306e-05,
+      "loss": 1.2644,
+      "step": 1455
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.6018584703359545e-05,
+      "loss": 1.3288,
+      "step": 1456
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.603645461043603e-05,
+      "loss": 1.2724,
+      "step": 1457
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.6054324517512514e-05,
+      "loss": 1.2498,
+      "step": 1458
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.607219442458899e-05,
+      "loss": 1.2739,
+      "step": 1459
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.6090064331665475e-05,
+      "loss": 1.2956,
+      "step": 1460
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.610793423874196e-05,
+      "loss": 1.2358,
+      "step": 1461
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.6125804145818444e-05,
+      "loss": 1.2882,
+      "step": 1462
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.6143674052894928e-05,
+      "loss": 1.2697,
+      "step": 1463
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.616154395997141e-05,
+      "loss": 1.2796,
+      "step": 1464
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.6179413867047893e-05,
+      "loss": 1.2496,
+      "step": 1465
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.6197283774124377e-05,
+      "loss": 1.2525,
+      "step": 1466
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.621515368120086e-05,
+      "loss": 1.2469,
+      "step": 1467
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.6233023588277346e-05,
+      "loss": 1.2677,
+      "step": 1468
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.6250893495353823e-05,
+      "loss": 1.2324,
+      "step": 1469
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.6268763402430307e-05,
+      "loss": 1.2218,
+      "step": 1470
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.628663330950679e-05,
+      "loss": 1.2372,
+      "step": 1471
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.6304503216583275e-05,
+      "loss": 1.2484,
+      "step": 1472
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.632237312365976e-05,
+      "loss": 1.2611,
+      "step": 1473
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.634024303073624e-05,
+      "loss": 1.2452,
+      "step": 1474
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.6358112937812725e-05,
+      "loss": 1.2539,
+      "step": 1475
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.637598284488921e-05,
+      "loss": 1.2566,
+      "step": 1476
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.6393852751965693e-05,
+      "loss": 1.2783,
+      "step": 1477
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.6411722659042177e-05,
+      "loss": 1.2462,
+      "step": 1478
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.6429592566118655e-05,
+      "loss": 1.2844,
+      "step": 1479
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.644746247319514e-05,
+      "loss": 1.2533,
+      "step": 1480
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.6465332380271623e-05,
+      "loss": 1.2441,
+      "step": 1481
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.6483202287348107e-05,
+      "loss": 1.2838,
+      "step": 1482
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.650107219442459e-05,
+      "loss": 1.2627,
+      "step": 1483
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.6518942101501072e-05,
+      "loss": 1.2635,
+      "step": 1484
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.6536812008577556e-05,
+      "loss": 1.2505,
+      "step": 1485
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.655468191565404e-05,
+      "loss": 1.286,
+      "step": 1486
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.6572551822730525e-05,
+      "loss": 1.2683,
+      "step": 1487
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.659042172980701e-05,
+      "loss": 1.2552,
+      "step": 1488
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.6608291636883493e-05,
+      "loss": 1.2666,
+      "step": 1489
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.662616154395997e-05,
+      "loss": 1.2613,
+      "step": 1490
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.6644031451036455e-05,
+      "loss": 1.2778,
+      "step": 1491
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.666190135811294e-05,
+      "loss": 1.2571,
+      "step": 1492
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.6679771265189423e-05,
+      "loss": 1.2629,
+      "step": 1493
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.6697641172265907e-05,
+      "loss": 1.266,
+      "step": 1494
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.6715511079342388e-05,
+      "loss": 1.2278,
+      "step": 1495
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.6733380986418872e-05,
+      "loss": 1.2542,
+      "step": 1496
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.6751250893495356e-05,
+      "loss": 1.2439,
+      "step": 1497
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.676912080057184e-05,
+      "loss": 1.2262,
+      "step": 1498
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.6786990707648325e-05,
+      "loss": 1.2294,
+      "step": 1499
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.6804860614724802e-05,
+      "loss": 1.2729,
+      "step": 1500
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.6822730521801286e-05,
+      "loss": 1.2656,
+      "step": 1501
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.684060042887777e-05,
+      "loss": 1.2695,
+      "step": 1502
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.6858470335954255e-05,
+      "loss": 1.2383,
+      "step": 1503
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.687634024303074e-05,
+      "loss": 1.2368,
+      "step": 1504
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.689421015010722e-05,
+      "loss": 1.2575,
+      "step": 1505
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.6912080057183704e-05,
+      "loss": 1.2495,
+      "step": 1506
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.6929949964260188e-05,
+      "loss": 1.2473,
+      "step": 1507
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.6947819871336672e-05,
+      "loss": 1.2823,
+      "step": 1508
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.6965689778413156e-05,
+      "loss": 1.2784,
+      "step": 1509
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.6983559685489634e-05,
+      "loss": 1.2844,
+      "step": 1510
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7001429592566118e-05,
+      "loss": 1.2619,
+      "step": 1511
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7019299499642602e-05,
+      "loss": 1.2775,
+      "step": 1512
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7037169406719086e-05,
+      "loss": 1.2553,
+      "step": 1513
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.705503931379557e-05,
+      "loss": 1.2551,
+      "step": 1514
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.707290922087205e-05,
+      "loss": 1.2241,
+      "step": 1515
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7090779127948536e-05,
+      "loss": 1.2719,
+      "step": 1516
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.710864903502502e-05,
+      "loss": 1.2402,
+      "step": 1517
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7126518942101504e-05,
+      "loss": 1.2536,
+      "step": 1518
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7144388849177988e-05,
+      "loss": 1.2397,
+      "step": 1519
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7162258756254465e-05,
+      "loss": 1.2823,
+      "step": 1520
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.718012866333095e-05,
+      "loss": 1.2336,
+      "step": 1521
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7197998570407434e-05,
+      "loss": 1.2933,
+      "step": 1522
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.7215868477483918e-05,
+      "loss": 1.2509,
+      "step": 1523
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7233738384560402e-05,
+      "loss": 1.2463,
+      "step": 1524
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7251608291636886e-05,
+      "loss": 1.3022,
+      "step": 1525
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7269478198713367e-05,
+      "loss": 1.2741,
+      "step": 1526
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.728734810578985e-05,
+      "loss": 1.2663,
+      "step": 1527
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7305218012866336e-05,
+      "loss": 1.2249,
+      "step": 1528
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.732308791994282e-05,
+      "loss": 1.2504,
+      "step": 1529
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7340957827019304e-05,
+      "loss": 1.2666,
+      "step": 1530
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.735882773409578e-05,
+      "loss": 1.2765,
+      "step": 1531
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7376697641172266e-05,
+      "loss": 1.2755,
+      "step": 1532
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.739456754824875e-05,
+      "loss": 1.2331,
+      "step": 1533
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7412437455325234e-05,
+      "loss": 1.2489,
+      "step": 1534
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7430307362401718e-05,
+      "loss": 1.2234,
+      "step": 1535
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.74481772694782e-05,
+      "loss": 1.2383,
+      "step": 1536
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7466047176554683e-05,
+      "loss": 1.2713,
+      "step": 1537
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7483917083631167e-05,
+      "loss": 1.251,
+      "step": 1538
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.750178699070765e-05,
+      "loss": 1.2535,
+      "step": 1539
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7519656897784136e-05,
+      "loss": 1.25,
+      "step": 1540
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7537526804860613e-05,
+      "loss": 1.2291,
+      "step": 1541
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7555396711937097e-05,
+      "loss": 1.2238,
+      "step": 1542
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.757326661901358e-05,
+      "loss": 1.2484,
+      "step": 1543
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7591136526090066e-05,
+      "loss": 1.272,
+      "step": 1544
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.760900643316655e-05,
+      "loss": 1.2844,
+      "step": 1545
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.762687634024303e-05,
+      "loss": 1.243,
+      "step": 1546
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7644746247319515e-05,
+      "loss": 1.2628,
+      "step": 1547
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7662616154396e-05,
+      "loss": 1.2717,
+      "step": 1548
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7680486061472483e-05,
+      "loss": 1.2468,
+      "step": 1549
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7698355968548967e-05,
+      "loss": 1.2825,
+      "step": 1550
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7716225875625445e-05,
+      "loss": 1.2617,
+      "step": 1551
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.773409578270193e-05,
+      "loss": 1.265,
+      "step": 1552
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7751965689778413e-05,
+      "loss": 1.2489,
+      "step": 1553
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7769835596854897e-05,
+      "loss": 1.3052,
+      "step": 1554
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.778770550393138e-05,
+      "loss": 1.2846,
+      "step": 1555
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7805575411007862e-05,
+      "loss": 1.2272,
+      "step": 1556
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7823445318084346e-05,
+      "loss": 1.2576,
+      "step": 1557
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.784131522516083e-05,
+      "loss": 1.2237,
+      "step": 1558
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7859185132237315e-05,
+      "loss": 1.237,
+      "step": 1559
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.78770550393138e-05,
+      "loss": 1.2465,
+      "step": 1560
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7894924946390283e-05,
+      "loss": 1.2381,
+      "step": 1561
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.791279485346676e-05,
+      "loss": 1.2351,
+      "step": 1562
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7930664760543245e-05,
+      "loss": 1.2419,
+      "step": 1563
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.794853466761973e-05,
+      "loss": 1.259,
+      "step": 1564
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7966404574696213e-05,
+      "loss": 1.2487,
+      "step": 1565
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.7984274481772697e-05,
+      "loss": 1.244,
+      "step": 1566
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.8002144388849178e-05,
+      "loss": 1.2712,
+      "step": 1567
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.8020014295925662e-05,
+      "loss": 1.236,
+      "step": 1568
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.8037884203002146e-05,
+      "loss": 1.229,
+      "step": 1569
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.805575411007863e-05,
+      "loss": 1.2691,
+      "step": 1570
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.8073624017155115e-05,
+      "loss": 1.1974,
+      "step": 1571
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.8091493924231592e-05,
+      "loss": 1.2831,
+      "step": 1572
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.8109363831308076e-05,
+      "loss": 1.2716,
+      "step": 1573
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.812723373838456e-05,
+      "loss": 1.2549,
+      "step": 1574
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.8145103645461045e-05,
+      "loss": 1.2424,
+      "step": 1575
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.816297355253753e-05,
+      "loss": 1.2342,
+      "step": 1576
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.818084345961401e-05,
+      "loss": 1.2525,
+      "step": 1577
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.8198713366690494e-05,
+      "loss": 1.2603,
+      "step": 1578
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.8216583273766978e-05,
+      "loss": 1.2406,
+      "step": 1579
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.8234453180843462e-05,
+      "loss": 1.2512,
+      "step": 1580
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.8252323087919946e-05,
+      "loss": 1.2754,
+      "step": 1581
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.8270192994996424e-05,
+      "loss": 1.2247,
+      "step": 1582
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.8288062902072908e-05,
+      "loss": 1.2603,
+      "step": 1583
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.8305932809149392e-05,
+      "loss": 1.2384,
+      "step": 1584
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.8323802716225876e-05,
+      "loss": 1.2601,
+      "step": 1585
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.834167262330236e-05,
+      "loss": 1.2608,
+      "step": 1586
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.835954253037884e-05,
+      "loss": 1.2566,
+      "step": 1587
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.8377412437455326e-05,
+      "loss": 1.2681,
+      "step": 1588
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.839528234453181e-05,
+      "loss": 1.2671,
+      "step": 1589
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.8413152251608294e-05,
+      "loss": 1.28,
+      "step": 1590
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.8431022158684778e-05,
+      "loss": 1.2797,
+      "step": 1591
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.8448892065761256e-05,
+      "loss": 1.2791,
+      "step": 1592
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.846676197283774e-05,
+      "loss": 1.289,
+      "step": 1593
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.8484631879914224e-05,
+      "loss": 1.2608,
+      "step": 1594
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.8502501786990708e-05,
+      "loss": 1.2555,
+      "step": 1595
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.8520371694067192e-05,
+      "loss": 1.2641,
+      "step": 1596
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.8538241601143677e-05,
+      "loss": 1.2601,
+      "step": 1597
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.8556111508220157e-05,
+      "loss": 1.2542,
+      "step": 1598
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.857398141529664e-05,
+      "loss": 1.2557,
+      "step": 1599
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.8591851322373126e-05,
+      "loss": 1.2541,
+      "step": 1600
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.860972122944961e-05,
+      "loss": 1.2928,
+      "step": 1601
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.8627591136526094e-05,
+      "loss": 1.2463,
+      "step": 1602
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.864546104360257e-05,
+      "loss": 1.2259,
+      "step": 1603
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.8663330950679056e-05,
+      "loss": 1.228,
+      "step": 1604
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.868120085775554e-05,
+      "loss": 1.2529,
+      "step": 1605
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.8699070764832024e-05,
+      "loss": 1.2861,
+      "step": 1606
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.871694067190851e-05,
+      "loss": 1.2657,
+      "step": 1607
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.873481057898499e-05,
+      "loss": 1.2205,
+      "step": 1608
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.8752680486061473e-05,
+      "loss": 1.242,
+      "step": 1609
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.8770550393137957e-05,
+      "loss": 1.2533,
+      "step": 1610
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.878842030021444e-05,
+      "loss": 1.2455,
+      "step": 1611
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.8806290207290926e-05,
+      "loss": 1.2218,
+      "step": 1612
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.8824160114367403e-05,
+      "loss": 1.2699,
+      "step": 1613
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.8842030021443887e-05,
+      "loss": 1.2543,
+      "step": 1614
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.885989992852037e-05,
+      "loss": 1.2555,
+      "step": 1615
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.8877769835596856e-05,
+      "loss": 1.2687,
+      "step": 1616
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.8895639742673343e-05,
+      "loss": 1.228,
+      "step": 1617
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.891350964974982e-05,
+      "loss": 1.261,
+      "step": 1618
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.8931379556826305e-05,
+      "loss": 1.2482,
+      "step": 1619
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.894924946390279e-05,
+      "loss": 1.2422,
+      "step": 1620
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.8967119370979273e-05,
+      "loss": 1.2652,
+      "step": 1621
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.8984989278055757e-05,
+      "loss": 1.2581,
+      "step": 1622
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9002859185132235e-05,
+      "loss": 1.2403,
+      "step": 1623
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.902072909220872e-05,
+      "loss": 1.256,
+      "step": 1624
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9038598999285203e-05,
+      "loss": 1.264,
+      "step": 1625
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.905646890636169e-05,
+      "loss": 1.2401,
+      "step": 1626
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9074338813438175e-05,
+      "loss": 1.2558,
+      "step": 1627
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9092208720514652e-05,
+      "loss": 1.2545,
+      "step": 1628
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9110078627591137e-05,
+      "loss": 1.2672,
+      "step": 1629
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.912794853466762e-05,
+      "loss": 1.2443,
+      "step": 1630
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9145818441744105e-05,
+      "loss": 1.2393,
+      "step": 1631
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.916368834882059e-05,
+      "loss": 1.2086,
+      "step": 1632
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9181558255897073e-05,
+      "loss": 1.2734,
+      "step": 1633
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.919942816297355e-05,
+      "loss": 1.2197,
+      "step": 1634
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9217298070050035e-05,
+      "loss": 1.2181,
+      "step": 1635
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9235167977126522e-05,
+      "loss": 1.2577,
+      "step": 1636
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9253037884203007e-05,
+      "loss": 1.2716,
+      "step": 1637
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.927090779127949e-05,
+      "loss": 1.2575,
+      "step": 1638
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9288777698355968e-05,
+      "loss": 1.2528,
+      "step": 1639
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9306647605432452e-05,
+      "loss": 1.258,
+      "step": 1640
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9324517512508937e-05,
+      "loss": 1.2579,
+      "step": 1641
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.934238741958542e-05,
+      "loss": 1.2466,
+      "step": 1642
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9360257326661905e-05,
+      "loss": 1.2575,
+      "step": 1643
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9378127233738382e-05,
+      "loss": 1.2873,
+      "step": 1644
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.939599714081487e-05,
+      "loss": 1.2986,
+      "step": 1645
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9413867047891354e-05,
+      "loss": 1.2453,
+      "step": 1646
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.9431736954967838e-05,
+      "loss": 1.293,
+      "step": 1647
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9449606862044322e-05,
+      "loss": 1.281,
+      "step": 1648
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.94674767691208e-05,
+      "loss": 1.2675,
+      "step": 1649
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9485346676197284e-05,
+      "loss": 1.2367,
+      "step": 1650
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9503216583273768e-05,
+      "loss": 1.2422,
+      "step": 1651
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9521086490350252e-05,
+      "loss": 1.2553,
+      "step": 1652
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9538956397426737e-05,
+      "loss": 1.254,
+      "step": 1653
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9556826304503214e-05,
+      "loss": 1.2307,
+      "step": 1654
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.95746962115797e-05,
+      "loss": 1.2389,
+      "step": 1655
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9592566118656186e-05,
+      "loss": 1.2108,
+      "step": 1656
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.961043602573267e-05,
+      "loss": 1.2496,
+      "step": 1657
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9628305932809154e-05,
+      "loss": 1.2308,
+      "step": 1658
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.964617583988563e-05,
+      "loss": 1.2646,
+      "step": 1659
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9664045746962116e-05,
+      "loss": 1.2322,
+      "step": 1660
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.96819156540386e-05,
+      "loss": 1.2289,
+      "step": 1661
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9699785561115084e-05,
+      "loss": 1.2552,
+      "step": 1662
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9717655468191568e-05,
+      "loss": 1.2553,
+      "step": 1663
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.973552537526805e-05,
+      "loss": 1.2367,
+      "step": 1664
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9753395282344533e-05,
+      "loss": 1.2828,
+      "step": 1665
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9771265189421017e-05,
+      "loss": 1.2918,
+      "step": 1666
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.97891350964975e-05,
+      "loss": 1.2553,
+      "step": 1667
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9807005003573986e-05,
+      "loss": 1.2589,
+      "step": 1668
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.982487491065047e-05,
+      "loss": 1.2583,
+      "step": 1669
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9842744817726947e-05,
+      "loss": 1.2511,
+      "step": 1670
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.986061472480343e-05,
+      "loss": 1.2532,
+      "step": 1671
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9878484631879916e-05,
+      "loss": 1.2616,
+      "step": 1672
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.98963545389564e-05,
+      "loss": 1.2645,
+      "step": 1673
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9914224446032884e-05,
+      "loss": 1.2754,
+      "step": 1674
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9932094353109365e-05,
+      "loss": 1.2482,
+      "step": 1675
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.994996426018585e-05,
+      "loss": 1.2772,
+      "step": 1676
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9967834167262333e-05,
+      "loss": 1.2384,
+      "step": 1677
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.9985704074338817e-05,
+      "loss": 1.2089,
+      "step": 1678
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.00035739814153e-05,
+      "loss": 1.2348,
+      "step": 1679
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.002144388849178e-05,
+      "loss": 1.2655,
+      "step": 1680
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.0039313795568263e-05,
+      "loss": 1.2842,
+      "step": 1681
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.0057183702644747e-05,
+      "loss": 1.2503,
+      "step": 1682
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.007505360972123e-05,
+      "loss": 1.2519,
+      "step": 1683
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.0092923516797716e-05,
+      "loss": 1.2376,
+      "step": 1684
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.0110793423874197e-05,
+      "loss": 1.25,
+      "step": 1685
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.012866333095068e-05,
+      "loss": 1.245,
+      "step": 1686
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.0146533238027165e-05,
+      "loss": 1.2809,
+      "step": 1687
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.016440314510365e-05,
+      "loss": 1.2555,
+      "step": 1688
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.0182273052180133e-05,
+      "loss": 1.2785,
+      "step": 1689
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.020014295925661e-05,
+      "loss": 1.2277,
+      "step": 1690
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.0218012866333095e-05,
+      "loss": 1.3165,
+      "step": 1691
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.023588277340958e-05,
+      "loss": 1.2565,
+      "step": 1692
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.0253752680486063e-05,
+      "loss": 1.2425,
+      "step": 1693
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.0271622587562547e-05,
+      "loss": 1.2547,
+      "step": 1694
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.0289492494639028e-05,
+      "loss": 1.2559,
+      "step": 1695
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.0307362401715512e-05,
+      "loss": 1.2393,
+      "step": 1696
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.0325232308791997e-05,
+      "loss": 1.2678,
+      "step": 1697
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.034310221586848e-05,
+      "loss": 1.2256,
+      "step": 1698
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.0360972122944965e-05,
+      "loss": 1.2489,
+      "step": 1699
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.0378842030021442e-05,
+      "loss": 1.2129,
+      "step": 1700
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.0396711937097927e-05,
+      "loss": 1.2432,
+      "step": 1701
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.041458184417441e-05,
+      "loss": 1.2624,
+      "step": 1702
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.0432451751250895e-05,
+      "loss": 1.2306,
+      "step": 1703
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.045032165832738e-05,
+      "loss": 1.2458,
+      "step": 1704
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.0468191565403863e-05,
+      "loss": 1.2812,
+      "step": 1705
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.0486061472480344e-05,
+      "loss": 1.2514,
+      "step": 1706
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.050393137955683e-05,
+      "loss": 1.2127,
+      "step": 1707
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.052180128663331e-05,
+      "loss": 1.2539,
+      "step": 1708
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.05396711937098e-05,
+      "loss": 1.2753,
+      "step": 1709
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.055754110078628e-05,
+      "loss": 1.2866,
+      "step": 1710
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.057541100786276e-05,
+      "loss": 1.2719,
+      "step": 1711
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.059328091493924e-05,
+      "loss": 1.2696,
+      "step": 1712
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.061115082201573e-05,
+      "loss": 1.27,
+      "step": 1713
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.062902072909221e-05,
+      "loss": 1.2694,
+      "step": 1714
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.0646890636168695e-05,
+      "loss": 1.2638,
+      "step": 1715
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.066476054324517e-05,
+      "loss": 1.2469,
+      "step": 1716
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.0682630450321657e-05,
+      "loss": 1.2327,
+      "step": 1717
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.070050035739814e-05,
+      "loss": 1.2561,
+      "step": 1718
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.0718370264474625e-05,
+      "loss": 1.2486,
+      "step": 1719
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.073624017155111e-05,
+      "loss": 1.2358,
+      "step": 1720
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.075411007862759e-05,
+      "loss": 1.2856,
+      "step": 1721
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.077197998570408e-05,
+      "loss": 1.2764,
+      "step": 1722
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.078984989278056e-05,
+      "loss": 1.2859,
+      "step": 1723
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.0807719799857046e-05,
+      "loss": 1.2908,
+      "step": 1724
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.082558970693353e-05,
+      "loss": 1.2756,
+      "step": 1725
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.084345961401001e-05,
+      "loss": 1.2669,
+      "step": 1726
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.086132952108649e-05,
+      "loss": 1.2535,
+      "step": 1727
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.0879199428162976e-05,
+      "loss": 1.2474,
+      "step": 1728
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.089706933523946e-05,
+      "loss": 1.2769,
+      "step": 1729
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.0914939242315944e-05,
+      "loss": 1.2652,
+      "step": 1730
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.093280914939242e-05,
+      "loss": 1.2709,
+      "step": 1731
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.0950679056468906e-05,
+      "loss": 1.2752,
+      "step": 1732
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.096854896354539e-05,
+      "loss": 1.248,
+      "step": 1733
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.0986418870621874e-05,
+      "loss": 1.2342,
+      "step": 1734
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.100428877769836e-05,
+      "loss": 1.2642,
+      "step": 1735
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.1022158684774836e-05,
+      "loss": 1.2737,
+      "step": 1736
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.104002859185132e-05,
+      "loss": 1.2774,
+      "step": 1737
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.1057898498927804e-05,
+      "loss": 1.2318,
+      "step": 1738
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.107576840600429e-05,
+      "loss": 1.2335,
+      "step": 1739
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.109363831308077e-05,
+      "loss": 1.2467,
+      "step": 1740
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.111150822015726e-05,
+      "loss": 1.2668,
+      "step": 1741
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.112937812723374e-05,
+      "loss": 1.2657,
+      "step": 1742
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.1147248034310225e-05,
+      "loss": 1.3057,
+      "step": 1743
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.116511794138671e-05,
+      "loss": 1.2654,
+      "step": 1744
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.1182987848463193e-05,
+      "loss": 1.2562,
+      "step": 1745
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.120085775553968e-05,
+      "loss": 1.2455,
+      "step": 1746
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.1218727662616155e-05,
+      "loss": 1.2432,
+      "step": 1747
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.123659756969264e-05,
+      "loss": 1.2855,
+      "step": 1748
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.1254467476769123e-05,
+      "loss": 1.2757,
+      "step": 1749
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.127233738384561e-05,
+      "loss": 1.2757,
+      "step": 1750
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.129020729092209e-05,
+      "loss": 1.2584,
+      "step": 1751
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.130807719799857e-05,
+      "loss": 1.2075,
+      "step": 1752
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.132594710507505e-05,
+      "loss": 1.2526,
+      "step": 1753
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.134381701215154e-05,
+      "loss": 1.2749,
+      "step": 1754
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.136168691922802e-05,
+      "loss": 1.2411,
+      "step": 1755
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.1379556826304506e-05,
+      "loss": 1.3008,
+      "step": 1756
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.139742673338098e-05,
+      "loss": 1.2555,
+      "step": 1757
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.141529664045747e-05,
+      "loss": 1.2667,
+      "step": 1758
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.143316654753395e-05,
+      "loss": 1.2435,
+      "step": 1759
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.1451036454610436e-05,
+      "loss": 1.2621,
+      "step": 1760
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.146890636168692e-05,
+      "loss": 1.2525,
+      "step": 1761
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.1486776268763404e-05,
+      "loss": 1.2312,
+      "step": 1762
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.150464617583989e-05,
+      "loss": 1.2757,
+      "step": 1763
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.152251608291637e-05,
+      "loss": 1.2574,
+      "step": 1764
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.154038598999286e-05,
+      "loss": 1.2509,
+      "step": 1765
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.155825589706934e-05,
+      "loss": 1.2808,
+      "step": 1766
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.157612580414582e-05,
+      "loss": 1.2918,
+      "step": 1767
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.15939957112223e-05,
+      "loss": 1.2814,
+      "step": 1768
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.161186561829879e-05,
+      "loss": 1.2757,
+      "step": 1769
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.162973552537527e-05,
+      "loss": 1.244,
+      "step": 1770
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.1647605432451755e-05,
+      "loss": 1.2969,
+      "step": 1771
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.166547533952823e-05,
+      "loss": 1.2522,
+      "step": 1772
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.168334524660472e-05,
+      "loss": 1.2424,
+      "step": 1773
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.17012151536812e-05,
+      "loss": 1.2688,
+      "step": 1774
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.1719085060757685e-05,
+      "loss": 1.2698,
+      "step": 1775
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.173695496783417e-05,
+      "loss": 1.2757,
+      "step": 1776
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.1754824874910653e-05,
+      "loss": 1.2474,
+      "step": 1777
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.177269478198713e-05,
+      "loss": 1.2664,
+      "step": 1778
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.1790564689063615e-05,
+      "loss": 1.228,
+      "step": 1779
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.18084345961401e-05,
+      "loss": 1.2868,
+      "step": 1780
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.1826304503216583e-05,
+      "loss": 1.2267,
+      "step": 1781
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.184417441029307e-05,
+      "loss": 1.2723,
+      "step": 1782
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.186204431736955e-05,
+      "loss": 1.2725,
+      "step": 1783
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.1879914224446036e-05,
+      "loss": 1.2966,
+      "step": 1784
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.189778413152252e-05,
+      "loss": 1.2483,
+      "step": 1785
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.1915654038599004e-05,
+      "loss": 1.2606,
+      "step": 1786
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.193352394567549e-05,
+      "loss": 1.2467,
+      "step": 1787
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.1951393852751966e-05,
+      "loss": 1.2737,
+      "step": 1788
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.196926375982845e-05,
+      "loss": 1.2378,
+      "step": 1789
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.1987133666904934e-05,
+      "loss": 1.2246,
+      "step": 1790
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.200500357398142e-05,
+      "loss": 1.2646,
+      "step": 1791
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.20228734810579e-05,
+      "loss": 1.2229,
+      "step": 1792
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.204074338813438e-05,
+      "loss": 1.2891,
+      "step": 1793
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.2058613295210864e-05,
+      "loss": 1.2399,
+      "step": 1794
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.207648320228735e-05,
+      "loss": 1.24,
+      "step": 1795
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.209435310936383e-05,
+      "loss": 1.2617,
+      "step": 1796
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.211222301644032e-05,
+      "loss": 1.2817,
+      "step": 1797
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.2130092923516794e-05,
+      "loss": 1.2515,
+      "step": 1798
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.214796283059328e-05,
+      "loss": 1.2342,
+      "step": 1799
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.216583273766976e-05,
+      "loss": 1.2529,
+      "step": 1800
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.218370264474625e-05,
+      "loss": 1.2258,
+      "step": 1801
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.220157255182273e-05,
+      "loss": 1.2499,
+      "step": 1802
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.2219442458899215e-05,
+      "loss": 1.27,
+      "step": 1803
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.22373123659757e-05,
+      "loss": 1.2642,
+      "step": 1804
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.2255182273052183e-05,
+      "loss": 1.2336,
+      "step": 1805
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.227305218012867e-05,
+      "loss": 1.2627,
+      "step": 1806
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.229092208720515e-05,
+      "loss": 1.2741,
+      "step": 1807
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.230879199428163e-05,
+      "loss": 1.234,
+      "step": 1808
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.2326661901358113e-05,
+      "loss": 1.2387,
+      "step": 1809
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.23445318084346e-05,
+      "loss": 1.2237,
+      "step": 1810
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.236240171551108e-05,
+      "loss": 1.2582,
+      "step": 1811
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.2380271622587566e-05,
+      "loss": 1.275,
+      "step": 1812
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.239814152966405e-05,
+      "loss": 1.2854,
+      "step": 1813
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.241601143674053e-05,
+      "loss": 1.2702,
+      "step": 1814
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.243388134381701e-05,
+      "loss": 1.2918,
+      "step": 1815
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.2451751250893496e-05,
+      "loss": 1.2538,
+      "step": 1816
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.246962115796998e-05,
+      "loss": 1.2575,
+      "step": 1817
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.2487491065046464e-05,
+      "loss": 1.2496,
+      "step": 1818
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.250536097212294e-05,
+      "loss": 1.2832,
+      "step": 1819
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.2523230879199426e-05,
+      "loss": 1.2335,
+      "step": 1820
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.254110078627591e-05,
+      "loss": 1.2776,
+      "step": 1821
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.2558970693352394e-05,
+      "loss": 1.2733,
+      "step": 1822
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.257684060042888e-05,
+      "loss": 1.2634,
+      "step": 1823
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.259471050750536e-05,
+      "loss": 1.2543,
+      "step": 1824
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.261258041458185e-05,
+      "loss": 1.3087,
+      "step": 1825
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.263045032165833e-05,
+      "loss": 1.2423,
+      "step": 1826
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.2648320228734815e-05,
+      "loss": 1.2592,
+      "step": 1827
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.26661901358113e-05,
+      "loss": 1.2374,
+      "step": 1828
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.268406004288778e-05,
+      "loss": 1.2454,
+      "step": 1829
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.270192994996426e-05,
+      "loss": 1.262,
+      "step": 1830
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.2719799857040745e-05,
+      "loss": 1.2531,
+      "step": 1831
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.273766976411723e-05,
+      "loss": 1.2734,
+      "step": 1832
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.2755539671193714e-05,
+      "loss": 1.2267,
+      "step": 1833
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.277340957827019e-05,
+      "loss": 1.2555,
+      "step": 1834
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2791279485346675e-05,
+      "loss": 1.2391,
+      "step": 1835
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.280914939242316e-05,
+      "loss": 1.2332,
+      "step": 1836
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2827019299499643e-05,
+      "loss": 1.2809,
+      "step": 1837
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.284488920657613e-05,
+      "loss": 1.2685,
+      "step": 1838
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2862759113652605e-05,
+      "loss": 1.2382,
+      "step": 1839
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.288062902072909e-05,
+      "loss": 1.2346,
+      "step": 1840
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2898498927805573e-05,
+      "loss": 1.2586,
+      "step": 1841
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.291636883488206e-05,
+      "loss": 1.2557,
+      "step": 1842
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.293423874195854e-05,
+      "loss": 1.259,
+      "step": 1843
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2952108649035026e-05,
+      "loss": 1.2473,
+      "step": 1844
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.296997855611151e-05,
+      "loss": 1.2732,
+      "step": 1845
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.2987848463187994e-05,
+      "loss": 1.2605,
+      "step": 1846
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.300571837026448e-05,
+      "loss": 1.2481,
+      "step": 1847
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.302358827734096e-05,
+      "loss": 1.2538,
+      "step": 1848
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.304145818441745e-05,
+      "loss": 1.2595,
+      "step": 1849
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.3059328091493924e-05,
+      "loss": 1.2376,
+      "step": 1850
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.307719799857041e-05,
+      "loss": 1.2374,
+      "step": 1851
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.309506790564689e-05,
+      "loss": 1.2258,
+      "step": 1852
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.311293781272338e-05,
+      "loss": 1.2439,
+      "step": 1853
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.313080771979986e-05,
+      "loss": 1.213,
+      "step": 1854
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.314867762687634e-05,
+      "loss": 1.2638,
+      "step": 1855
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.316654753395282e-05,
+      "loss": 1.2661,
+      "step": 1856
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.318441744102931e-05,
+      "loss": 1.2612,
+      "step": 1857
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.320228734810579e-05,
+      "loss": 1.2678,
+      "step": 1858
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.3220157255182275e-05,
+      "loss": 1.2522,
+      "step": 1859
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.323802716225875e-05,
+      "loss": 1.2503,
+      "step": 1860
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.325589706933524e-05,
+      "loss": 1.3016,
+      "step": 1861
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.327376697641172e-05,
+      "loss": 1.2647,
+      "step": 1862
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.329163688348821e-05,
+      "loss": 1.2543,
+      "step": 1863
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.3309506790564696e-05,
+      "loss": 1.2675,
+      "step": 1864
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.3327376697641174e-05,
+      "loss": 1.2627,
+      "step": 1865
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.334524660471766e-05,
+      "loss": 1.2548,
+      "step": 1866
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.336311651179414e-05,
+      "loss": 1.2639,
+      "step": 1867
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.3380986418870626e-05,
+      "loss": 1.2266,
+      "step": 1868
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.339885632594711e-05,
+      "loss": 1.2426,
+      "step": 1869
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.341672623302359e-05,
+      "loss": 1.2379,
+      "step": 1870
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.343459614010007e-05,
+      "loss": 1.2637,
+      "step": 1871
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.3452466047176556e-05,
+      "loss": 1.2705,
+      "step": 1872
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.347033595425304e-05,
+      "loss": 1.2991,
+      "step": 1873
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.3488205861329524e-05,
+      "loss": 1.2515,
+      "step": 1874
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.3506075768406e-05,
+      "loss": 1.2582,
+      "step": 1875
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.3523945675482486e-05,
+      "loss": 1.2421,
+      "step": 1876
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.354181558255897e-05,
+      "loss": 1.2959,
+      "step": 1877
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.3559685489635454e-05,
+      "loss": 1.263,
+      "step": 1878
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.357755539671194e-05,
+      "loss": 1.2229,
+      "step": 1879
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.3595425303788416e-05,
+      "loss": 1.2276,
+      "step": 1880
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.36132952108649e-05,
+      "loss": 1.2478,
+      "step": 1881
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.363116511794139e-05,
+      "loss": 1.2664,
+      "step": 1882
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.3649035025017875e-05,
+      "loss": 1.26,
+      "step": 1883
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.366690493209436e-05,
+      "loss": 1.2401,
+      "step": 1884
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.3684774839170844e-05,
+      "loss": 1.2754,
+      "step": 1885
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.370264474624732e-05,
+      "loss": 1.2602,
+      "step": 1886
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.3720514653323805e-05,
+      "loss": 1.2742,
+      "step": 1887
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.373838456040029e-05,
+      "loss": 1.2323,
+      "step": 1888
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.3756254467476774e-05,
+      "loss": 1.2347,
+      "step": 1889
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.377412437455326e-05,
+      "loss": 1.2502,
+      "step": 1890
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.3791994281629735e-05,
+      "loss": 1.2549,
+      "step": 1891
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.380986418870622e-05,
+      "loss": 1.2626,
+      "step": 1892
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.3827734095782704e-05,
+      "loss": 1.2176,
+      "step": 1893
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.384560400285919e-05,
+      "loss": 1.2457,
+      "step": 1894
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.386347390993567e-05,
+      "loss": 1.2598,
+      "step": 1895
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.388134381701215e-05,
+      "loss": 1.2837,
+      "step": 1896
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.3899213724088634e-05,
+      "loss": 1.2858,
+      "step": 1897
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.391708363116512e-05,
+      "loss": 1.2392,
+      "step": 1898
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.39349535382416e-05,
+      "loss": 1.2692,
+      "step": 1899
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.3952823445318086e-05,
+      "loss": 1.2698,
+      "step": 1900
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.397069335239457e-05,
+      "loss": 1.2308,
+      "step": 1901
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.3988563259471054e-05,
+      "loss": 1.2507,
+      "step": 1902
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.400643316654754e-05,
+      "loss": 1.2594,
+      "step": 1903
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.402430307362402e-05,
+      "loss": 1.2327,
+      "step": 1904
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.404217298070051e-05,
+      "loss": 1.2612,
+      "step": 1905
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4060042887776984e-05,
+      "loss": 1.2431,
+      "step": 1906
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.407791279485347e-05,
+      "loss": 1.2433,
+      "step": 1907
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.409578270192995e-05,
+      "loss": 1.2354,
+      "step": 1908
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.411365260900644e-05,
+      "loss": 1.2396,
+      "step": 1909
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.413152251608292e-05,
+      "loss": 1.268,
+      "step": 1910
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.41493924231594e-05,
+      "loss": 1.2943,
+      "step": 1911
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.416726233023588e-05,
+      "loss": 1.2571,
+      "step": 1912
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.418513223731237e-05,
+      "loss": 1.3042,
+      "step": 1913
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.420300214438885e-05,
+      "loss": 1.2539,
+      "step": 1914
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4220872051465335e-05,
+      "loss": 1.2531,
+      "step": 1915
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.423874195854181e-05,
+      "loss": 1.2266,
+      "step": 1916
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.42566118656183e-05,
+      "loss": 1.2553,
+      "step": 1917
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.427448177269478e-05,
+      "loss": 1.2845,
+      "step": 1918
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4292351679771265e-05,
+      "loss": 1.273,
+      "step": 1919
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.431022158684775e-05,
+      "loss": 1.2629,
+      "step": 1920
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4328091493924234e-05,
+      "loss": 1.2495,
+      "step": 1921
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.434596140100072e-05,
+      "loss": 1.2102,
+      "step": 1922
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.43638313080772e-05,
+      "loss": 1.2708,
+      "step": 1923
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4381701215153686e-05,
+      "loss": 1.2679,
+      "step": 1924
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.439957112223017e-05,
+      "loss": 1.2631,
+      "step": 1925
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4417441029306655e-05,
+      "loss": 1.2701,
+      "step": 1926
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.443531093638313e-05,
+      "loss": 1.2251,
+      "step": 1927
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4453180843459616e-05,
+      "loss": 1.2596,
+      "step": 1928
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.44710507505361e-05,
+      "loss": 1.2693,
+      "step": 1929
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4488920657612585e-05,
+      "loss": 1.2746,
+      "step": 1930
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.450679056468907e-05,
+      "loss": 1.251,
+      "step": 1931
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4524660471765546e-05,
+      "loss": 1.2611,
+      "step": 1932
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.454253037884203e-05,
+      "loss": 1.2664,
+      "step": 1933
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4560400285918514e-05,
+      "loss": 1.3142,
+      "step": 1934
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4578270192995e-05,
+      "loss": 1.2612,
+      "step": 1935
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.459614010007148e-05,
+      "loss": 1.2737,
+      "step": 1936
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.461401000714796e-05,
+      "loss": 1.27,
+      "step": 1937
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4631879914224444e-05,
+      "loss": 1.2793,
+      "step": 1938
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.464974982130093e-05,
+      "loss": 1.2672,
+      "step": 1939
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.466761972837741e-05,
+      "loss": 1.3015,
+      "step": 1940
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.46854896354539e-05,
+      "loss": 1.2482,
+      "step": 1941
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.470335954253038e-05,
+      "loss": 1.2427,
+      "step": 1942
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4721229449606865e-05,
+      "loss": 1.2327,
+      "step": 1943
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.473909935668335e-05,
+      "loss": 1.2594,
+      "step": 1944
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4756969263759834e-05,
+      "loss": 1.2586,
+      "step": 1945
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.477483917083632e-05,
+      "loss": 1.277,
+      "step": 1946
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4792709077912795e-05,
+      "loss": 1.2702,
+      "step": 1947
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.481057898498928e-05,
+      "loss": 1.2766,
+      "step": 1948
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4828448892065764e-05,
+      "loss": 1.2273,
+      "step": 1949
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.484631879914225e-05,
+      "loss": 1.2618,
+      "step": 1950
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.486418870621873e-05,
+      "loss": 1.2317,
+      "step": 1951
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.488205861329521e-05,
+      "loss": 1.2462,
+      "step": 1952
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4899928520371694e-05,
+      "loss": 1.2645,
+      "step": 1953
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.491779842744818e-05,
+      "loss": 1.2914,
+      "step": 1954
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.493566833452466e-05,
+      "loss": 1.2593,
+      "step": 1955
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.4953538241601146e-05,
+      "loss": 1.2498,
+      "step": 1956
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.497140814867763e-05,
+      "loss": 1.2618,
+      "step": 1957
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.498927805575411e-05,
+      "loss": 1.2452,
+      "step": 1958
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.500714796283059e-05,
+      "loss": 1.2569,
+      "step": 1959
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.5025017869907076e-05,
+      "loss": 1.2369,
+      "step": 1960
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.504288777698356e-05,
+      "loss": 1.2682,
+      "step": 1961
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.5060757684060045e-05,
+      "loss": 1.2706,
+      "step": 1962
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.507862759113653e-05,
+      "loss": 1.2783,
+      "step": 1963
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.509649749821301e-05,
+      "loss": 1.2399,
+      "step": 1964
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.51143674052895e-05,
+      "loss": 1.2528,
+      "step": 1965
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.513223731236598e-05,
+      "loss": 1.2486,
+      "step": 1966
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.5150107219442465e-05,
+      "loss": 1.2561,
+      "step": 1967
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.516797712651894e-05,
+      "loss": 1.2504,
+      "step": 1968
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.518584703359543e-05,
+      "loss": 1.2835,
+      "step": 1969
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.520371694067191e-05,
+      "loss": 1.2108,
+      "step": 1970
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.5221586847748395e-05,
+      "loss": 1.2454,
+      "step": 1971
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.523945675482488e-05,
+      "loss": 1.2525,
+      "step": 1972
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.525732666190136e-05,
+      "loss": 1.2903,
+      "step": 1973
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.527519656897784e-05,
+      "loss": 1.2556,
+      "step": 1974
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.5293066476054325e-05,
+      "loss": 1.2707,
+      "step": 1975
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.531093638313081e-05,
+      "loss": 1.2848,
+      "step": 1976
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.5328806290207294e-05,
+      "loss": 1.2594,
+      "step": 1977
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.534667619728377e-05,
+      "loss": 1.2518,
+      "step": 1978
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.5364546104360255e-05,
+      "loss": 1.2622,
+      "step": 1979
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.538241601143674e-05,
+      "loss": 1.287,
+      "step": 1980
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.5400285918513224e-05,
+      "loss": 1.2672,
+      "step": 1981
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.541815582558971e-05,
+      "loss": 1.2889,
+      "step": 1982
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.543602573266619e-05,
+      "loss": 1.2577,
+      "step": 1983
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.5453895639742676e-05,
+      "loss": 1.2612,
+      "step": 1984
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.547176554681916e-05,
+      "loss": 1.2621,
+      "step": 1985
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.5489635453895645e-05,
+      "loss": 1.2362,
+      "step": 1986
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.550750536097213e-05,
+      "loss": 1.2692,
+      "step": 1987
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.5525375268048606e-05,
+      "loss": 1.2746,
+      "step": 1988
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.554324517512509e-05,
+      "loss": 1.2808,
+      "step": 1989
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.5561115082201575e-05,
+      "loss": 1.2674,
+      "step": 1990
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.557898498927806e-05,
+      "loss": 1.2694,
+      "step": 1991
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.559685489635454e-05,
+      "loss": 1.2799,
+      "step": 1992
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.561472480343103e-05,
+      "loss": 1.2744,
+      "step": 1993
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.5632594710507505e-05,
+      "loss": 1.2444,
+      "step": 1994
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.565046461758399e-05,
+      "loss": 1.2557,
+      "step": 1995
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.566833452466047e-05,
+      "loss": 1.2434,
+      "step": 1996
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.568620443173696e-05,
+      "loss": 1.2497,
+      "step": 1997
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.570407433881344e-05,
+      "loss": 1.2429,
+      "step": 1998
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.572194424588992e-05,
+      "loss": 1.2469,
+      "step": 1999
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.57398141529664e-05,
+      "loss": 1.2613,
+      "step": 2000
+    },
+    {
+      "epoch": 0.32,
+      "eval_loss": 1.2876265048980713,
+      "eval_runtime": 61.5413,
+      "eval_samples_per_second": 162.493,
+      "eval_steps_per_second": 0.227,
+      "step": 2000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.575768406004289e-05,
+      "loss": 1.2436,
+      "step": 2001
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.577555396711937e-05,
+      "loss": 1.2577,
+      "step": 2002
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.5793423874195855e-05,
+      "loss": 1.2403,
+      "step": 2003
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.581129378127234e-05,
+      "loss": 1.2289,
+      "step": 2004
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.5829163688348824e-05,
+      "loss": 1.2408,
+      "step": 2005
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.584703359542531e-05,
+      "loss": 1.2401,
+      "step": 2006
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.586490350250179e-05,
+      "loss": 1.2454,
+      "step": 2007
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.5882773409578276e-05,
+      "loss": 1.2649,
+      "step": 2008
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.5900643316654754e-05,
+      "loss": 1.2402,
+      "step": 2009
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.591851322373124e-05,
+      "loss": 1.2792,
+      "step": 2010
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.593638313080772e-05,
+      "loss": 1.2907,
+      "step": 2011
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.5954253037884206e-05,
+      "loss": 1.2343,
+      "step": 2012
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.597212294496069e-05,
+      "loss": 1.2753,
+      "step": 2013
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.598999285203717e-05,
+      "loss": 1.2439,
+      "step": 2014
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.600786275911365e-05,
+      "loss": 1.2726,
+      "step": 2015
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.6025732666190136e-05,
+      "loss": 1.2381,
+      "step": 2016
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.604360257326662e-05,
+      "loss": 1.2452,
+      "step": 2017
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.6061472480343105e-05,
+      "loss": 1.2367,
+      "step": 2018
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.607934238741958e-05,
+      "loss": 1.2524,
+      "step": 2019
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.6097212294496066e-05,
+      "loss": 1.2768,
+      "step": 2020
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.611508220157255e-05,
+      "loss": 1.2669,
+      "step": 2021
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.6132952108649035e-05,
+      "loss": 1.2079,
+      "step": 2022
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.615082201572552e-05,
+      "loss": 1.233,
+      "step": 2023
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.6168691922802e-05,
+      "loss": 1.2744,
+      "step": 2024
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.618656182987849e-05,
+      "loss": 1.2657,
+      "step": 2025
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.620443173695497e-05,
+      "loss": 1.2512,
+      "step": 2026
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.6222301644031455e-05,
+      "loss": 1.2805,
+      "step": 2027
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.624017155110794e-05,
+      "loss": 1.2386,
+      "step": 2028
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.6258041458184424e-05,
+      "loss": 1.2513,
+      "step": 2029
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.62759113652609e-05,
+      "loss": 1.278,
+      "step": 2030
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.6293781272337385e-05,
+      "loss": 1.2743,
+      "step": 2031
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.631165117941387e-05,
+      "loss": 1.2464,
+      "step": 2032
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.6329521086490354e-05,
+      "loss": 1.2615,
+      "step": 2033
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.634739099356684e-05,
+      "loss": 1.2473,
+      "step": 2034
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.6365260900643315e-05,
+      "loss": 1.2339,
+      "step": 2035
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.63831308077198e-05,
+      "loss": 1.2624,
+      "step": 2036
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.6401000714796284e-05,
+      "loss": 1.2797,
+      "step": 2037
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.641887062187277e-05,
+      "loss": 1.2444,
+      "step": 2038
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.643674052894925e-05,
+      "loss": 1.2487,
+      "step": 2039
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.645461043602573e-05,
+      "loss": 1.2764,
+      "step": 2040
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.6472480343102214e-05,
+      "loss": 1.2282,
+      "step": 2041
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.64903502501787e-05,
+      "loss": 1.2544,
+      "step": 2042
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.650822015725518e-05,
+      "loss": 1.2344,
+      "step": 2043
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.6526090064331666e-05,
+      "loss": 1.2507,
+      "step": 2044
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.654395997140815e-05,
+      "loss": 1.279,
+      "step": 2045
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.6561829878484635e-05,
+      "loss": 1.2527,
+      "step": 2046
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.657969978556112e-05,
+      "loss": 1.2676,
+      "step": 2047
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.65975696926376e-05,
+      "loss": 1.2838,
+      "step": 2048
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.661543959971409e-05,
+      "loss": 1.2762,
+      "step": 2049
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.6633309506790565e-05,
+      "loss": 1.2397,
+      "step": 2050
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.665117941386705e-05,
+      "loss": 1.2433,
+      "step": 2051
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.666904932094353e-05,
+      "loss": 1.2842,
+      "step": 2052
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.668691922802002e-05,
+      "loss": 1.2309,
+      "step": 2053
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.67047891350965e-05,
+      "loss": 1.2883,
+      "step": 2054
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.672265904217298e-05,
+      "loss": 1.2892,
+      "step": 2055
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.674052894924946e-05,
+      "loss": 1.2336,
+      "step": 2056
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.675839885632595e-05,
+      "loss": 1.252,
+      "step": 2057
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.677626876340243e-05,
+      "loss": 1.2516,
+      "step": 2058
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.6794138670478915e-05,
+      "loss": 1.2746,
+      "step": 2059
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.681200857755539e-05,
+      "loss": 1.2514,
+      "step": 2060
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.682987848463188e-05,
+      "loss": 1.2682,
+      "step": 2061
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.684774839170836e-05,
+      "loss": 1.2612,
+      "step": 2062
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.6865618298784845e-05,
+      "loss": 1.255,
+      "step": 2063
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.688348820586133e-05,
+      "loss": 1.2799,
+      "step": 2064
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.6901358112937814e-05,
+      "loss": 1.2411,
+      "step": 2065
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.69192280200143e-05,
+      "loss": 1.2591,
+      "step": 2066
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.693709792709078e-05,
+      "loss": 1.2828,
+      "step": 2067
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.6954967834167266e-05,
+      "loss": 1.2296,
+      "step": 2068
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.697283774124375e-05,
+      "loss": 1.245,
+      "step": 2069
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.6990707648320235e-05,
+      "loss": 1.2641,
+      "step": 2070
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.700857755539671e-05,
+      "loss": 1.2459,
+      "step": 2071
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.7026447462473196e-05,
+      "loss": 1.254,
+      "step": 2072
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.704431736954968e-05,
+      "loss": 1.2563,
+      "step": 2073
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.7062187276626165e-05,
+      "loss": 1.2493,
+      "step": 2074
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.708005718370265e-05,
+      "loss": 1.2623,
+      "step": 2075
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.7097927090779126e-05,
+      "loss": 1.2535,
+      "step": 2076
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.711579699785561e-05,
+      "loss": 1.2247,
+      "step": 2077
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.7133666904932095e-05,
+      "loss": 1.2784,
+      "step": 2078
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.715153681200858e-05,
+      "loss": 1.2336,
+      "step": 2079
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.716940671908506e-05,
+      "loss": 1.234,
+      "step": 2080
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.718727662616154e-05,
+      "loss": 1.2034,
+      "step": 2081
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.7205146533238025e-05,
+      "loss": 1.2607,
+      "step": 2082
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.722301644031451e-05,
+      "loss": 1.21,
+      "step": 2083
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.724088634739099e-05,
+      "loss": 1.2438,
+      "step": 2084
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.7258756254467484e-05,
+      "loss": 1.2653,
+      "step": 2085
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.727662616154396e-05,
+      "loss": 1.2228,
+      "step": 2086
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.7294496068620446e-05,
+      "loss": 1.2959,
+      "step": 2087
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.731236597569693e-05,
+      "loss": 1.2293,
+      "step": 2088
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.7330235882773414e-05,
+      "loss": 1.2205,
+      "step": 2089
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.73481057898499e-05,
+      "loss": 1.2871,
+      "step": 2090
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.7365975696926375e-05,
+      "loss": 1.2205,
+      "step": 2091
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.738384560400286e-05,
+      "loss": 1.2745,
+      "step": 2092
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.7401715511079344e-05,
+      "loss": 1.2345,
+      "step": 2093
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.741958541815583e-05,
+      "loss": 1.2807,
+      "step": 2094
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.743745532523231e-05,
+      "loss": 1.258,
+      "step": 2095
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.745532523230879e-05,
+      "loss": 1.2542,
+      "step": 2096
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.7473195139385274e-05,
+      "loss": 1.2297,
+      "step": 2097
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.749106504646176e-05,
+      "loss": 1.2299,
+      "step": 2098
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.750893495353824e-05,
+      "loss": 1.2763,
+      "step": 2099
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.7526804860614726e-05,
+      "loss": 1.2713,
+      "step": 2100
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.754467476769121e-05,
+      "loss": 1.271,
+      "step": 2101
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.756254467476769e-05,
+      "loss": 1.2642,
+      "step": 2102
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.758041458184417e-05,
+      "loss": 1.2419,
+      "step": 2103
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.759828448892066e-05,
+      "loss": 1.2992,
+      "step": 2104
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.761615439599715e-05,
+      "loss": 1.2622,
+      "step": 2105
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.763402430307363e-05,
+      "loss": 1.218,
+      "step": 2106
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.765189421015011e-05,
+      "loss": 1.2363,
+      "step": 2107
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.766976411722659e-05,
+      "loss": 1.248,
+      "step": 2108
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.768763402430308e-05,
+      "loss": 1.2734,
+      "step": 2109
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.770550393137956e-05,
+      "loss": 1.237,
+      "step": 2110
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.7723373838456046e-05,
+      "loss": 1.2505,
+      "step": 2111
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.774124374553252e-05,
+      "loss": 1.2242,
+      "step": 2112
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.775911365260901e-05,
+      "loss": 1.2665,
+      "step": 2113
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.777698355968549e-05,
+      "loss": 1.229,
+      "step": 2114
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.7794853466761976e-05,
+      "loss": 1.2817,
+      "step": 2115
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.781272337383846e-05,
+      "loss": 1.2524,
+      "step": 2116
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.783059328091494e-05,
+      "loss": 1.2664,
+      "step": 2117
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.784846318799142e-05,
+      "loss": 1.2453,
+      "step": 2118
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.7866333095067906e-05,
+      "loss": 1.2748,
+      "step": 2119
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.788420300214439e-05,
+      "loss": 1.2629,
+      "step": 2120
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.7902072909220874e-05,
+      "loss": 1.2507,
+      "step": 2121
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.791994281629735e-05,
+      "loss": 1.2726,
+      "step": 2122
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.793781272337384e-05,
+      "loss": 1.2671,
+      "step": 2123
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.7955682630450326e-05,
+      "loss": 1.2148,
+      "step": 2124
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.797355253752681e-05,
+      "loss": 1.2602,
+      "step": 2125
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.7991422444603295e-05,
+      "loss": 1.2513,
+      "step": 2126
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.800929235167977e-05,
+      "loss": 1.2453,
+      "step": 2127
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.8027162258756256e-05,
+      "loss": 1.2731,
+      "step": 2128
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.804503216583274e-05,
+      "loss": 1.2468,
+      "step": 2129
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.8062902072909225e-05,
+      "loss": 1.2728,
+      "step": 2130
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.808077197998571e-05,
+      "loss": 1.231,
+      "step": 2131
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.8098641887062186e-05,
+      "loss": 1.2795,
+      "step": 2132
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.811651179413867e-05,
+      "loss": 1.2482,
+      "step": 2133
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.8134381701215155e-05,
+      "loss": 1.2615,
+      "step": 2134
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.815225160829164e-05,
+      "loss": 1.2476,
+      "step": 2135
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.817012151536812e-05,
+      "loss": 1.2349,
+      "step": 2136
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.818799142244461e-05,
+      "loss": 1.2845,
+      "step": 2137
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.8205861329521085e-05,
+      "loss": 1.2634,
+      "step": 2138
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.822373123659757e-05,
+      "loss": 1.2676,
+      "step": 2139
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.824160114367405e-05,
+      "loss": 1.2766,
+      "step": 2140
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.825947105075054e-05,
+      "loss": 1.2432,
+      "step": 2141
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.827734095782702e-05,
+      "loss": 1.2651,
+      "step": 2142
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.8295210864903506e-05,
+      "loss": 1.2407,
+      "step": 2143
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.831308077197999e-05,
+      "loss": 1.2704,
+      "step": 2144
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.8330950679056474e-05,
+      "loss": 1.2416,
+      "step": 2145
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.834882058613296e-05,
+      "loss": 1.2411,
+      "step": 2146
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.836669049320944e-05,
+      "loss": 1.2305,
+      "step": 2147
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.838456040028592e-05,
+      "loss": 1.2667,
+      "step": 2148
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.8402430307362404e-05,
+      "loss": 1.2297,
+      "step": 2149
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.842030021443889e-05,
+      "loss": 1.2751,
+      "step": 2150
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.843817012151537e-05,
+      "loss": 1.2175,
+      "step": 2151
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.8456040028591857e-05,
+      "loss": 1.2612,
+      "step": 2152
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.8473909935668334e-05,
+      "loss": 1.2508,
+      "step": 2153
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.849177984274482e-05,
+      "loss": 1.3006,
+      "step": 2154
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.85096497498213e-05,
+      "loss": 1.2673,
+      "step": 2155
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.8527519656897786e-05,
+      "loss": 1.2668,
+      "step": 2156
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.854538956397427e-05,
+      "loss": 1.2774,
+      "step": 2157
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.856325947105075e-05,
+      "loss": 1.2493,
+      "step": 2158
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.858112937812723e-05,
+      "loss": 1.2723,
+      "step": 2159
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.8598999285203716e-05,
+      "loss": 1.2684,
+      "step": 2160
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.86168691922802e-05,
+      "loss": 1.2604,
+      "step": 2161
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.8634739099356685e-05,
+      "loss": 1.251,
+      "step": 2162
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.865260900643317e-05,
+      "loss": 1.2712,
+      "step": 2163
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.867047891350965e-05,
+      "loss": 1.2602,
+      "step": 2164
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.868834882058614e-05,
+      "loss": 1.2518,
+      "step": 2165
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.870621872766262e-05,
+      "loss": 1.2735,
+      "step": 2166
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.8724088634739106e-05,
+      "loss": 1.2698,
+      "step": 2167
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.874195854181558e-05,
+      "loss": 1.2637,
+      "step": 2168
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.875982844889207e-05,
+      "loss": 1.2307,
+      "step": 2169
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.877769835596855e-05,
+      "loss": 1.2706,
+      "step": 2170
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.8795568263045036e-05,
+      "loss": 1.2193,
+      "step": 2171
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.881343817012152e-05,
+      "loss": 1.2453,
+      "step": 2172
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.8831308077198004e-05,
+      "loss": 1.2739,
+      "step": 2173
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.884917798427448e-05,
+      "loss": 1.2351,
+      "step": 2174
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.8867047891350966e-05,
+      "loss": 1.2455,
+      "step": 2175
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.888491779842745e-05,
+      "loss": 1.224,
+      "step": 2176
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.8902787705503934e-05,
+      "loss": 1.2506,
+      "step": 2177
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.892065761258042e-05,
+      "loss": 1.2795,
+      "step": 2178
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.8938527519656896e-05,
+      "loss": 1.2357,
+      "step": 2179
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.895639742673338e-05,
+      "loss": 1.309,
+      "step": 2180
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.8974267333809864e-05,
+      "loss": 1.2819,
+      "step": 2181
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.899213724088635e-05,
+      "loss": 1.2742,
+      "step": 2182
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.901000714796283e-05,
+      "loss": 1.2801,
+      "step": 2183
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.9027877055039317e-05,
+      "loss": 1.2657,
+      "step": 2184
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.90457469621158e-05,
+      "loss": 1.2694,
+      "step": 2185
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.9063616869192285e-05,
+      "loss": 1.2189,
+      "step": 2186
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.908148677626877e-05,
+      "loss": 1.2488,
+      "step": 2187
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.909935668334525e-05,
+      "loss": 1.2477,
+      "step": 2188
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.911722659042173e-05,
+      "loss": 1.2621,
+      "step": 2189
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.9135096497498215e-05,
+      "loss": 1.2673,
+      "step": 2190
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.91529664045747e-05,
+      "loss": 1.2368,
+      "step": 2191
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.917083631165118e-05,
+      "loss": 1.2646,
+      "step": 2192
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.918870621872767e-05,
+      "loss": 1.2745,
+      "step": 2193
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.9206576125804145e-05,
+      "loss": 1.2613,
+      "step": 2194
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.922444603288063e-05,
+      "loss": 1.2262,
+      "step": 2195
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.924231593995711e-05,
+      "loss": 1.2207,
+      "step": 2196
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.92601858470336e-05,
+      "loss": 1.2675,
+      "step": 2197
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.927805575411008e-05,
+      "loss": 1.254,
+      "step": 2198
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.929592566118656e-05,
+      "loss": 1.2642,
+      "step": 2199
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.931379556826304e-05,
+      "loss": 1.2782,
+      "step": 2200
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.933166547533953e-05,
+      "loss": 1.245,
+      "step": 2201
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.934953538241601e-05,
+      "loss": 1.2354,
+      "step": 2202
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.9367405289492496e-05,
+      "loss": 1.2736,
+      "step": 2203
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.938527519656898e-05,
+      "loss": 1.2375,
+      "step": 2204
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.9403145103645464e-05,
+      "loss": 1.2182,
+      "step": 2205
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.942101501072195e-05,
+      "loss": 1.2639,
+      "step": 2206
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.943888491779843e-05,
+      "loss": 1.2556,
+      "step": 2207
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.9456754824874917e-05,
+      "loss": 1.2394,
+      "step": 2208
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.94746247319514e-05,
+      "loss": 1.2758,
+      "step": 2209
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.949249463902788e-05,
+      "loss": 1.2635,
+      "step": 2210
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.951036454610436e-05,
+      "loss": 1.2611,
+      "step": 2211
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.9528234453180847e-05,
+      "loss": 1.2932,
+      "step": 2212
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.954610436025733e-05,
+      "loss": 1.2517,
+      "step": 2213
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.9563974267333815e-05,
+      "loss": 1.2543,
+      "step": 2214
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.958184417441029e-05,
+      "loss": 1.2616,
+      "step": 2215
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.9599714081486777e-05,
+      "loss": 1.2308,
+      "step": 2216
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.961758398856326e-05,
+      "loss": 1.2128,
+      "step": 2217
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.9635453895639745e-05,
+      "loss": 1.2637,
+      "step": 2218
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.965332380271623e-05,
+      "loss": 1.2503,
+      "step": 2219
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.9671193709792706e-05,
+      "loss": 1.2453,
+      "step": 2220
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.968906361686919e-05,
+      "loss": 1.2351,
+      "step": 2221
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.9706933523945675e-05,
+      "loss": 1.2486,
+      "step": 2222
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.972480343102216e-05,
+      "loss": 1.2336,
+      "step": 2223
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.974267333809864e-05,
+      "loss": 1.2194,
+      "step": 2224
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.976054324517513e-05,
+      "loss": 1.2481,
+      "step": 2225
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.977841315225161e-05,
+      "loss": 1.2547,
+      "step": 2226
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.9796283059328096e-05,
+      "loss": 1.2748,
+      "step": 2227
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.981415296640458e-05,
+      "loss": 1.2384,
+      "step": 2228
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.9832022873481064e-05,
+      "loss": 1.2473,
+      "step": 2229
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.984989278055754e-05,
+      "loss": 1.2287,
+      "step": 2230
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.9867762687634026e-05,
+      "loss": 1.2464,
+      "step": 2231
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.988563259471051e-05,
+      "loss": 1.2147,
+      "step": 2232
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.9903502501786994e-05,
+      "loss": 1.215,
+      "step": 2233
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.992137240886348e-05,
+      "loss": 1.2008,
+      "step": 2234
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.9939242315939956e-05,
+      "loss": 1.2659,
+      "step": 2235
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.995711222301644e-05,
+      "loss": 1.2624,
+      "step": 2236
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.9974982130092924e-05,
+      "loss": 1.2426,
+      "step": 2237
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.999285203716941e-05,
+      "loss": 1.2428,
+      "step": 2238
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.001072194424589e-05,
+      "loss": 1.2509,
+      "step": 2239
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.002859185132237e-05,
+      "loss": 1.2612,
+      "step": 2240
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.0046461758398854e-05,
+      "loss": 1.28,
+      "step": 2241
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.006433166547534e-05,
+      "loss": 1.2608,
+      "step": 2242
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.008220157255182e-05,
+      "loss": 1.251,
+      "step": 2243
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.0100071479628307e-05,
+      "loss": 1.2561,
+      "step": 2244
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.011794138670479e-05,
+      "loss": 1.2693,
+      "step": 2245
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.0135811293781275e-05,
+      "loss": 1.2504,
+      "step": 2246
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.015368120085776e-05,
+      "loss": 1.2827,
+      "step": 2247
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.017155110793424e-05,
+      "loss": 1.2622,
+      "step": 2248
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.018942101501073e-05,
+      "loss": 1.2689,
+      "step": 2249
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.020729092208721e-05,
+      "loss": 1.2532,
+      "step": 2250
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.022516082916369e-05,
+      "loss": 1.2535,
+      "step": 2251
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.024303073624017e-05,
+      "loss": 1.27,
+      "step": 2252
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.026090064331666e-05,
+      "loss": 1.2334,
+      "step": 2253
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.027877055039314e-05,
+      "loss": 1.2627,
+      "step": 2254
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.0296640457469626e-05,
+      "loss": 1.2677,
+      "step": 2255
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.03145103645461e-05,
+      "loss": 1.2606,
+      "step": 2256
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.033238027162259e-05,
+      "loss": 1.2236,
+      "step": 2257
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.035025017869907e-05,
+      "loss": 1.2215,
+      "step": 2258
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.0368120085775556e-05,
+      "loss": 1.2678,
+      "step": 2259
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.038598999285204e-05,
+      "loss": 1.2419,
+      "step": 2260
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.040385989992852e-05,
+      "loss": 1.2697,
+      "step": 2261
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.0421729807005e-05,
+      "loss": 1.2736,
+      "step": 2262
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.0439599714081486e-05,
+      "loss": 1.268,
+      "step": 2263
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.045746962115797e-05,
+      "loss": 1.2611,
+      "step": 2264
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.0475339528234454e-05,
+      "loss": 1.2306,
+      "step": 2265
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.049320943531094e-05,
+      "loss": 1.2558,
+      "step": 2266
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.051107934238742e-05,
+      "loss": 1.2575,
+      "step": 2267
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.052894924946391e-05,
+      "loss": 1.2621,
+      "step": 2268
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.054681915654039e-05,
+      "loss": 1.2102,
+      "step": 2269
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.0564689063616875e-05,
+      "loss": 1.2523,
+      "step": 2270
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.058255897069335e-05,
+      "loss": 1.2588,
+      "step": 2271
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.060042887776984e-05,
+      "loss": 1.2776,
+      "step": 2272
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.061829878484632e-05,
+      "loss": 1.2423,
+      "step": 2273
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.0636168691922805e-05,
+      "loss": 1.2328,
+      "step": 2274
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.065403859899929e-05,
+      "loss": 1.2563,
+      "step": 2275
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.0671908506075767e-05,
+      "loss": 1.2617,
+      "step": 2276
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.068977841315225e-05,
+      "loss": 1.2438,
+      "step": 2277
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.0707648320228735e-05,
+      "loss": 1.253,
+      "step": 2278
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.072551822730522e-05,
+      "loss": 1.2576,
+      "step": 2279
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.07433881343817e-05,
+      "loss": 1.2449,
+      "step": 2280
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.076125804145819e-05,
+      "loss": 1.2507,
+      "step": 2281
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.0779127948534665e-05,
+      "loss": 1.2195,
+      "step": 2282
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.079699785561115e-05,
+      "loss": 1.2561,
+      "step": 2283
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.081486776268763e-05,
+      "loss": 1.2209,
+      "step": 2284
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.083273766976412e-05,
+      "loss": 1.203,
+      "step": 2285
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.08506075768406e-05,
+      "loss": 1.23,
+      "step": 2286
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.0868477483917086e-05,
+      "loss": 1.2328,
+      "step": 2287
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.088634739099357e-05,
+      "loss": 1.289,
+      "step": 2288
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.0904217298070054e-05,
+      "loss": 1.2643,
+      "step": 2289
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.092208720514654e-05,
+      "loss": 1.2383,
+      "step": 2290
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.093995711222302e-05,
+      "loss": 1.2761,
+      "step": 2291
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.09578270192995e-05,
+      "loss": 1.248,
+      "step": 2292
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.0975696926375984e-05,
+      "loss": 1.2269,
+      "step": 2293
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.099356683345247e-05,
+      "loss": 1.2588,
+      "step": 2294
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.101143674052895e-05,
+      "loss": 1.242,
+      "step": 2295
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.102930664760544e-05,
+      "loss": 1.2585,
+      "step": 2296
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.1047176554681914e-05,
+      "loss": 1.2952,
+      "step": 2297
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.10650464617584e-05,
+      "loss": 1.2145,
+      "step": 2298
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.108291636883488e-05,
+      "loss": 1.2637,
+      "step": 2299
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.110078627591137e-05,
+      "loss": 1.2615,
+      "step": 2300
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.111865618298785e-05,
+      "loss": 1.2267,
+      "step": 2301
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.113652609006433e-05,
+      "loss": 1.2563,
+      "step": 2302
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.115439599714081e-05,
+      "loss": 1.2176,
+      "step": 2303
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.11722659042173e-05,
+      "loss": 1.2639,
+      "step": 2304
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.119013581129378e-05,
+      "loss": 1.3028,
+      "step": 2305
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.1208005718370265e-05,
+      "loss": 1.2623,
+      "step": 2306
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.122587562544675e-05,
+      "loss": 1.24,
+      "step": 2307
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.124374553252323e-05,
+      "loss": 1.2472,
+      "step": 2308
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.126161543959972e-05,
+      "loss": 1.235,
+      "step": 2309
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.12794853466762e-05,
+      "loss": 1.2325,
+      "step": 2310
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.1297355253752686e-05,
+      "loss": 1.276,
+      "step": 2311
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.131522516082916e-05,
+      "loss": 1.3138,
+      "step": 2312
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.133309506790565e-05,
+      "loss": 1.221,
+      "step": 2313
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.135096497498213e-05,
+      "loss": 1.2931,
+      "step": 2314
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.1368834882058616e-05,
+      "loss": 1.2368,
+      "step": 2315
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.13867047891351e-05,
+      "loss": 1.2458,
+      "step": 2316
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.1404574696211584e-05,
+      "loss": 1.2379,
+      "step": 2317
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.142244460328806e-05,
+      "loss": 1.2465,
+      "step": 2318
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.1440314510364546e-05,
+      "loss": 1.2277,
+      "step": 2319
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.145818441744103e-05,
+      "loss": 1.2564,
+      "step": 2320
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.1476054324517514e-05,
+      "loss": 1.2766,
+      "step": 2321
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.1493924231594e-05,
+      "loss": 1.2236,
+      "step": 2322
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.1511794138670476e-05,
+      "loss": 1.2347,
+      "step": 2323
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.152966404574696e-05,
+      "loss": 1.2349,
+      "step": 2324
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.1547533952823444e-05,
+      "loss": 1.2741,
+      "step": 2325
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.1565403859899935e-05,
+      "loss": 1.222,
+      "step": 2326
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.158327376697642e-05,
+      "loss": 1.2466,
+      "step": 2327
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.16011436740529e-05,
+      "loss": 1.2843,
+      "step": 2328
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.161901358112938e-05,
+      "loss": 1.2306,
+      "step": 2329
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.1636883488205865e-05,
+      "loss": 1.2378,
+      "step": 2330
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.165475339528235e-05,
+      "loss": 1.3089,
+      "step": 2331
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.1672623302358833e-05,
+      "loss": 1.3081,
+      "step": 2332
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.169049320943531e-05,
+      "loss": 1.2437,
+      "step": 2333
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.1708363116511795e-05,
+      "loss": 1.2334,
+      "step": 2334
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.172623302358828e-05,
+      "loss": 1.2823,
+      "step": 2335
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.1744102930664763e-05,
+      "loss": 1.2325,
+      "step": 2336
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.176197283774125e-05,
+      "loss": 1.2582,
+      "step": 2337
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.1779842744817725e-05,
+      "loss": 1.289,
+      "step": 2338
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.179771265189421e-05,
+      "loss": 1.2379,
+      "step": 2339
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.181558255897069e-05,
+      "loss": 1.2727,
+      "step": 2340
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.183345246604718e-05,
+      "loss": 1.2425,
+      "step": 2341
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.185132237312366e-05,
+      "loss": 1.267,
+      "step": 2342
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.186919228020014e-05,
+      "loss": 1.2195,
+      "step": 2343
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.188706218727662e-05,
+      "loss": 1.2558,
+      "step": 2344
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.1904932094353114e-05,
+      "loss": 1.2509,
+      "step": 2345
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.19228020014296e-05,
+      "loss": 1.2876,
+      "step": 2346
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.194067190850608e-05,
+      "loss": 1.2445,
+      "step": 2347
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.195854181558256e-05,
+      "loss": 1.2905,
+      "step": 2348
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.1976411722659044e-05,
+      "loss": 1.2563,
+      "step": 2349
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.199428162973553e-05,
+      "loss": 1.2609,
+      "step": 2350
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.201215153681201e-05,
+      "loss": 1.2286,
+      "step": 2351
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.20300214438885e-05,
+      "loss": 1.2714,
+      "step": 2352
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.204789135096498e-05,
+      "loss": 1.2005,
+      "step": 2353
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.206576125804146e-05,
+      "loss": 1.306,
+      "step": 2354
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.208363116511794e-05,
+      "loss": 1.2799,
+      "step": 2355
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.210150107219443e-05,
+      "loss": 1.2369,
+      "step": 2356
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.211937097927091e-05,
+      "loss": 1.254,
+      "step": 2357
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.2137240886347395e-05,
+      "loss": 1.262,
+      "step": 2358
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.215511079342387e-05,
+      "loss": 1.2328,
+      "step": 2359
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.217298070050036e-05,
+      "loss": 1.2532,
+      "step": 2360
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.219085060757684e-05,
+      "loss": 1.2613,
+      "step": 2361
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.2208720514653325e-05,
+      "loss": 1.2747,
+      "step": 2362
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.222659042172981e-05,
+      "loss": 1.2622,
+      "step": 2363
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.2244460328806293e-05,
+      "loss": 1.2437,
+      "step": 2364
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.226233023588278e-05,
+      "loss": 1.2794,
+      "step": 2365
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.228020014295926e-05,
+      "loss": 1.2758,
+      "step": 2366
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.2298070050035746e-05,
+      "loss": 1.2649,
+      "step": 2367
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.231593995711223e-05,
+      "loss": 1.2536,
+      "step": 2368
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.233380986418871e-05,
+      "loss": 1.293,
+      "step": 2369
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.235167977126519e-05,
+      "loss": 1.2301,
+      "step": 2370
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.2369549678341676e-05,
+      "loss": 1.2633,
+      "step": 2371
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.238741958541816e-05,
+      "loss": 1.2721,
+      "step": 2372
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.2405289492494644e-05,
+      "loss": 1.2797,
+      "step": 2373
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.242315939957112e-05,
+      "loss": 1.2707,
+      "step": 2374
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.2441029306647606e-05,
+      "loss": 1.2616,
+      "step": 2375
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.245889921372409e-05,
+      "loss": 1.2696,
+      "step": 2376
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.2476769120800574e-05,
+      "loss": 1.2765,
+      "step": 2377
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.249463902787706e-05,
+      "loss": 1.2675,
+      "step": 2378
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.2512508934953536e-05,
+      "loss": 1.2546,
+      "step": 2379
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.253037884203002e-05,
+      "loss": 1.2579,
+      "step": 2380
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.2548248749106504e-05,
+      "loss": 1.2519,
+      "step": 2381
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.256611865618299e-05,
+      "loss": 1.2391,
+      "step": 2382
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.258398856325947e-05,
+      "loss": 1.2524,
+      "step": 2383
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.260185847033596e-05,
+      "loss": 1.2512,
+      "step": 2384
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.261972837741244e-05,
+      "loss": 1.2416,
+      "step": 2385
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.2637598284488925e-05,
+      "loss": 1.2382,
+      "step": 2386
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.265546819156541e-05,
+      "loss": 1.219,
+      "step": 2387
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.2673338098641894e-05,
+      "loss": 1.2351,
+      "step": 2388
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.269120800571838e-05,
+      "loss": 1.2517,
+      "step": 2389
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.2709077912794855e-05,
+      "loss": 1.2472,
+      "step": 2390
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.272694781987134e-05,
+      "loss": 1.234,
+      "step": 2391
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.2744817726947823e-05,
+      "loss": 1.2379,
+      "step": 2392
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.276268763402431e-05,
+      "loss": 1.2307,
+      "step": 2393
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.278055754110079e-05,
+      "loss": 1.2332,
+      "step": 2394
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.279842744817727e-05,
+      "loss": 1.2715,
+      "step": 2395
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.2816297355253753e-05,
+      "loss": 1.2285,
+      "step": 2396
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.283416726233024e-05,
+      "loss": 1.2363,
+      "step": 2397
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.285203716940672e-05,
+      "loss": 1.2485,
+      "step": 2398
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.2869907076483206e-05,
+      "loss": 1.2207,
+      "step": 2399
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.2887776983559683e-05,
+      "loss": 1.2428,
+      "step": 2400
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.290564689063617e-05,
+      "loss": 1.2001,
+      "step": 2401
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.292351679771265e-05,
+      "loss": 1.2771,
+      "step": 2402
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.2941386704789136e-05,
+      "loss": 1.2683,
+      "step": 2403
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.295925661186562e-05,
+      "loss": 1.2261,
+      "step": 2404
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.2977126518942104e-05,
+      "loss": 1.2744,
+      "step": 2405
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.299499642601859e-05,
+      "loss": 1.2341,
+      "step": 2406
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.301286633309507e-05,
+      "loss": 1.2261,
+      "step": 2407
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.303073624017156e-05,
+      "loss": 1.2717,
+      "step": 2408
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.304860614724804e-05,
+      "loss": 1.2419,
+      "step": 2409
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.306647605432452e-05,
+      "loss": 1.2614,
+      "step": 2410
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3084345961401e-05,
+      "loss": 1.2343,
+      "step": 2411
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.310221586847749e-05,
+      "loss": 1.2425,
+      "step": 2412
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.312008577555397e-05,
+      "loss": 1.2385,
+      "step": 2413
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3137955682630455e-05,
+      "loss": 1.2711,
+      "step": 2414
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.315582558970693e-05,
+      "loss": 1.2582,
+      "step": 2415
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.317369549678342e-05,
+      "loss": 1.2698,
+      "step": 2416
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.31915654038599e-05,
+      "loss": 1.263,
+      "step": 2417
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3209435310936385e-05,
+      "loss": 1.2425,
+      "step": 2418
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.322730521801287e-05,
+      "loss": 1.2504,
+      "step": 2419
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.324517512508935e-05,
+      "loss": 1.2233,
+      "step": 2420
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.326304503216583e-05,
+      "loss": 1.2383,
+      "step": 2421
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3280914939242315e-05,
+      "loss": 1.2431,
+      "step": 2422
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.32987848463188e-05,
+      "loss": 1.2287,
+      "step": 2423
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3316654753395284e-05,
+      "loss": 1.2271,
+      "step": 2424
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.333452466047177e-05,
+      "loss": 1.2386,
+      "step": 2425
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.335239456754825e-05,
+      "loss": 1.2075,
+      "step": 2426
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3370264474624736e-05,
+      "loss": 1.2985,
+      "step": 2427
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.338813438170122e-05,
+      "loss": 1.2489,
+      "step": 2428
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3406004288777704e-05,
+      "loss": 1.2644,
+      "step": 2429
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.342387419585419e-05,
+      "loss": 1.2458,
+      "step": 2430
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3441744102930666e-05,
+      "loss": 1.2613,
+      "step": 2431
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.345961401000715e-05,
+      "loss": 1.2165,
+      "step": 2432
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3477483917083634e-05,
+      "loss": 1.2493,
+      "step": 2433
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.349535382416012e-05,
+      "loss": 1.2625,
+      "step": 2434
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.35132237312366e-05,
+      "loss": 1.2472,
+      "step": 2435
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.353109363831308e-05,
+      "loss": 1.2385,
+      "step": 2436
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3548963545389564e-05,
+      "loss": 1.2211,
+      "step": 2437
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.356683345246605e-05,
+      "loss": 1.2001,
+      "step": 2438
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.358470335954253e-05,
+      "loss": 1.2647,
+      "step": 2439
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.360257326661902e-05,
+      "loss": 1.2139,
+      "step": 2440
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3620443173695494e-05,
+      "loss": 1.276,
+      "step": 2441
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.363831308077198e-05,
+      "loss": 1.285,
+      "step": 2442
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.365618298784846e-05,
+      "loss": 1.2623,
+      "step": 2443
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.367405289492495e-05,
+      "loss": 1.2726,
+      "step": 2444
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.369192280200143e-05,
+      "loss": 1.224,
+      "step": 2445
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3709792709077915e-05,
+      "loss": 1.2677,
+      "step": 2446
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.37276626161544e-05,
+      "loss": 1.2325,
+      "step": 2447
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3745532523230884e-05,
+      "loss": 1.2831,
+      "step": 2448
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.376340243030737e-05,
+      "loss": 1.2581,
+      "step": 2449
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.378127233738385e-05,
+      "loss": 1.231,
+      "step": 2450
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.379914224446033e-05,
+      "loss": 1.2696,
+      "step": 2451
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3817012151536814e-05,
+      "loss": 1.2841,
+      "step": 2452
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.38348820586133e-05,
+      "loss": 1.2893,
+      "step": 2453
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.385275196568978e-05,
+      "loss": 1.2511,
+      "step": 2454
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3870621872766266e-05,
+      "loss": 1.2649,
+      "step": 2455
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3888491779842744e-05,
+      "loss": 1.2753,
+      "step": 2456
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.390636168691923e-05,
+      "loss": 1.2248,
+      "step": 2457
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.392423159399571e-05,
+      "loss": 1.237,
+      "step": 2458
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3942101501072196e-05,
+      "loss": 1.2303,
+      "step": 2459
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.395997140814868e-05,
+      "loss": 1.2684,
+      "step": 2460
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3977841315225164e-05,
+      "loss": 1.2618,
+      "step": 2461
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.399571122230164e-05,
+      "loss": 1.2555,
+      "step": 2462
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.4013581129378126e-05,
+      "loss": 1.2493,
+      "step": 2463
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.403145103645461e-05,
+      "loss": 1.2886,
+      "step": 2464
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.4049320943531094e-05,
+      "loss": 1.2704,
+      "step": 2465
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.406719085060758e-05,
+      "loss": 1.3006,
+      "step": 2466
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.408506075768406e-05,
+      "loss": 1.2757,
+      "step": 2467
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.410293066476055e-05,
+      "loss": 1.2469,
+      "step": 2468
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.412080057183703e-05,
+      "loss": 1.3096,
+      "step": 2469
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.4138670478913515e-05,
+      "loss": 1.2791,
+      "step": 2470
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.415654038599e-05,
+      "loss": 1.2831,
+      "step": 2471
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.417441029306648e-05,
+      "loss": 1.296,
+      "step": 2472
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.419228020014296e-05,
+      "loss": 1.2274,
+      "step": 2473
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.4210150107219445e-05,
+      "loss": 1.2384,
+      "step": 2474
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.422802001429593e-05,
+      "loss": 1.2552,
+      "step": 2475
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.4245889921372414e-05,
+      "loss": 1.2528,
+      "step": 2476
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.426375982844889e-05,
+      "loss": 1.266,
+      "step": 2477
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.4281629735525375e-05,
+      "loss": 1.2691,
+      "step": 2478
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.429949964260186e-05,
+      "loss": 1.2323,
+      "step": 2479
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.4317369549678344e-05,
+      "loss": 1.2717,
+      "step": 2480
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.433523945675483e-05,
+      "loss": 1.244,
+      "step": 2481
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.4353109363831305e-05,
+      "loss": 1.2115,
+      "step": 2482
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.437097927090779e-05,
+      "loss": 1.257,
+      "step": 2483
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.4388849177984274e-05,
+      "loss": 1.2392,
+      "step": 2484
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.440671908506076e-05,
+      "loss": 1.2342,
+      "step": 2485
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.442458899213724e-05,
+      "loss": 1.2607,
+      "step": 2486
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.4442458899213726e-05,
+      "loss": 1.2522,
+      "step": 2487
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.446032880629021e-05,
+      "loss": 1.2144,
+      "step": 2488
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.4478198713366694e-05,
+      "loss": 1.233,
+      "step": 2489
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.449606862044318e-05,
+      "loss": 1.2751,
+      "step": 2490
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.451393852751966e-05,
+      "loss": 1.2405,
+      "step": 2491
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.453180843459614e-05,
+      "loss": 1.2704,
+      "step": 2492
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.4549678341672624e-05,
+      "loss": 1.2402,
+      "step": 2493
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.456754824874911e-05,
+      "loss": 1.2762,
+      "step": 2494
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.458541815582559e-05,
+      "loss": 1.2398,
+      "step": 2495
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.460328806290208e-05,
+      "loss": 1.2619,
+      "step": 2496
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.462115796997856e-05,
+      "loss": 1.2468,
+      "step": 2497
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.463902787705504e-05,
+      "loss": 1.2691,
+      "step": 2498
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.465689778413152e-05,
+      "loss": 1.2666,
+      "step": 2499
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.467476769120801e-05,
+      "loss": 1.2428,
+      "step": 2500
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.469263759828449e-05,
+      "loss": 1.2458,
+      "step": 2501
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.4710507505360975e-05,
+      "loss": 1.2807,
+      "step": 2502
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.472837741243745e-05,
+      "loss": 1.2549,
+      "step": 2503
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.474624731951394e-05,
+      "loss": 1.2535,
+      "step": 2504
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.476411722659042e-05,
+      "loss": 1.2919,
+      "step": 2505
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.4781987133666905e-05,
+      "loss": 1.2791,
+      "step": 2506
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.479985704074339e-05,
+      "loss": 1.259,
+      "step": 2507
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.4817726947819874e-05,
+      "loss": 1.2496,
+      "step": 2508
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.483559685489636e-05,
+      "loss": 1.2542,
+      "step": 2509
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.485346676197284e-05,
+      "loss": 1.2642,
+      "step": 2510
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.4871336669049326e-05,
+      "loss": 1.2549,
+      "step": 2511
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.488920657612581e-05,
+      "loss": 1.2668,
+      "step": 2512
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.490707648320229e-05,
+      "loss": 1.2492,
+      "step": 2513
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.492494639027877e-05,
+      "loss": 1.2585,
+      "step": 2514
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.4942816297355256e-05,
+      "loss": 1.2171,
+      "step": 2515
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.496068620443174e-05,
+      "loss": 1.271,
+      "step": 2516
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.4978556111508225e-05,
+      "loss": 1.2343,
+      "step": 2517
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.49964260185847e-05,
+      "loss": 1.2702,
+      "step": 2518
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.5014295925661186e-05,
+      "loss": 1.2857,
+      "step": 2519
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.503216583273767e-05,
+      "loss": 1.2761,
+      "step": 2520
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.5050035739814154e-05,
+      "loss": 1.2675,
+      "step": 2521
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.506790564689064e-05,
+      "loss": 1.2375,
+      "step": 2522
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.5085775553967116e-05,
+      "loss": 1.2321,
+      "step": 2523
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.51036454610436e-05,
+      "loss": 1.2385,
+      "step": 2524
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.5121515368120084e-05,
+      "loss": 1.2631,
+      "step": 2525
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.513938527519657e-05,
+      "loss": 1.2746,
+      "step": 2526
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.515725518227305e-05,
+      "loss": 1.285,
+      "step": 2527
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.517512508934954e-05,
+      "loss": 1.237,
+      "step": 2528
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.519299499642602e-05,
+      "loss": 1.221,
+      "step": 2529
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.5210864903502505e-05,
+      "loss": 1.2641,
+      "step": 2530
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.522873481057899e-05,
+      "loss": 1.2292,
+      "step": 2531
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.5246604717655474e-05,
+      "loss": 1.2728,
+      "step": 2532
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.526447462473196e-05,
+      "loss": 1.2795,
+      "step": 2533
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.5282344531808435e-05,
+      "loss": 1.2721,
+      "step": 2534
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.530021443888492e-05,
+      "loss": 1.2308,
+      "step": 2535
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.5318084345961404e-05,
+      "loss": 1.2482,
+      "step": 2536
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.533595425303789e-05,
+      "loss": 1.2649,
+      "step": 2537
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.535382416011437e-05,
+      "loss": 1.2737,
+      "step": 2538
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.537169406719085e-05,
+      "loss": 1.2372,
+      "step": 2539
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.5389563974267334e-05,
+      "loss": 1.2194,
+      "step": 2540
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.540743388134382e-05,
+      "loss": 1.2785,
+      "step": 2541
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.54253037884203e-05,
+      "loss": 1.286,
+      "step": 2542
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.5443173695496786e-05,
+      "loss": 1.2448,
+      "step": 2543
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.5461043602573264e-05,
+      "loss": 1.2199,
+      "step": 2544
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.547891350964975e-05,
+      "loss": 1.2566,
+      "step": 2545
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.549678341672623e-05,
+      "loss": 1.2449,
+      "step": 2546
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.5514653323802716e-05,
+      "loss": 1.2274,
+      "step": 2547
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.553252323087921e-05,
+      "loss": 1.2789,
+      "step": 2548
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.5550393137955685e-05,
+      "loss": 1.2389,
+      "step": 2549
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.556826304503217e-05,
+      "loss": 1.2758,
+      "step": 2550
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.558613295210865e-05,
+      "loss": 1.2489,
+      "step": 2551
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.560400285918514e-05,
+      "loss": 1.2396,
+      "step": 2552
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.562187276626162e-05,
+      "loss": 1.2188,
+      "step": 2553
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.56397426733381e-05,
+      "loss": 1.2667,
+      "step": 2554
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.565761258041458e-05,
+      "loss": 1.2665,
+      "step": 2555
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.567548248749107e-05,
+      "loss": 1.2852,
+      "step": 2556
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.569335239456755e-05,
+      "loss": 1.278,
+      "step": 2557
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.5711222301644035e-05,
+      "loss": 1.291,
+      "step": 2558
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.572909220872051e-05,
+      "loss": 1.2752,
+      "step": 2559
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.5746962115797e-05,
+      "loss": 1.2408,
+      "step": 2560
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.576483202287348e-05,
+      "loss": 1.253,
+      "step": 2561
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.5782701929949965e-05,
+      "loss": 1.2614,
+      "step": 2562
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.580057183702645e-05,
+      "loss": 1.2609,
+      "step": 2563
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.581844174410293e-05,
+      "loss": 1.2535,
+      "step": 2564
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.583631165117941e-05,
+      "loss": 1.2759,
+      "step": 2565
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.5854181558255895e-05,
+      "loss": 1.2468,
+      "step": 2566
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.5872051465332386e-05,
+      "loss": 1.2233,
+      "step": 2567
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.588992137240887e-05,
+      "loss": 1.2706,
+      "step": 2568
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.5907791279485355e-05,
+      "loss": 1.2425,
+      "step": 2569
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.592566118656183e-05,
+      "loss": 1.236,
+      "step": 2570
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.5943531093638316e-05,
+      "loss": 1.3052,
+      "step": 2571
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.59614010007148e-05,
+      "loss": 1.2515,
+      "step": 2572
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.5979270907791285e-05,
+      "loss": 1.2786,
+      "step": 2573
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.599714081486777e-05,
+      "loss": 1.2459,
+      "step": 2574
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.6015010721944246e-05,
+      "loss": 1.2575,
+      "step": 2575
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.603288062902073e-05,
+      "loss": 1.2728,
+      "step": 2576
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.6050750536097215e-05,
+      "loss": 1.2875,
+      "step": 2577
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.60686204431737e-05,
+      "loss": 1.2602,
+      "step": 2578
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.608649035025018e-05,
+      "loss": 1.2235,
+      "step": 2579
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.610436025732666e-05,
+      "loss": 1.2783,
+      "step": 2580
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.6122230164403145e-05,
+      "loss": 1.257,
+      "step": 2581
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.614010007147963e-05,
+      "loss": 1.2224,
+      "step": 2582
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.615796997855611e-05,
+      "loss": 1.2721,
+      "step": 2583
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.61758398856326e-05,
+      "loss": 1.2181,
+      "step": 2584
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.6193709792709074e-05,
+      "loss": 1.2695,
+      "step": 2585
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.6211579699785565e-05,
+      "loss": 1.2638,
+      "step": 2586
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.622944960686205e-05,
+      "loss": 1.2191,
+      "step": 2587
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.6247319513938534e-05,
+      "loss": 1.2504,
+      "step": 2588
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.626518942101502e-05,
+      "loss": 1.2068,
+      "step": 2589
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.6283059328091495e-05,
+      "loss": 1.255,
+      "step": 2590
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.630092923516798e-05,
+      "loss": 1.222,
+      "step": 2591
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.6318799142244464e-05,
+      "loss": 1.2727,
+      "step": 2592
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.633666904932095e-05,
+      "loss": 1.2485,
+      "step": 2593
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.635453895639743e-05,
+      "loss": 1.2839,
+      "step": 2594
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.637240886347391e-05,
+      "loss": 1.2313,
+      "step": 2595
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.6390278770550394e-05,
+      "loss": 1.2308,
+      "step": 2596
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.640814867762688e-05,
+      "loss": 1.2491,
+      "step": 2597
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.642601858470336e-05,
+      "loss": 1.249,
+      "step": 2598
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.6443888491779846e-05,
+      "loss": 1.2384,
+      "step": 2599
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.6461758398856324e-05,
+      "loss": 1.2493,
+      "step": 2600
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.647962830593281e-05,
+      "loss": 1.3057,
+      "step": 2601
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.649749821300929e-05,
+      "loss": 1.28,
+      "step": 2602
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.6515368120085776e-05,
+      "loss": 1.2286,
+      "step": 2603
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.653323802716226e-05,
+      "loss": 1.2576,
+      "step": 2604
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.6551107934238745e-05,
+      "loss": 1.2787,
+      "step": 2605
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.656897784131523e-05,
+      "loss": 1.2465,
+      "step": 2606
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.658684774839171e-05,
+      "loss": 1.2903,
+      "step": 2607
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.66047176554682e-05,
+      "loss": 1.2609,
+      "step": 2608
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.662258756254468e-05,
+      "loss": 1.223,
+      "step": 2609
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.6640457469621166e-05,
+      "loss": 1.2616,
+      "step": 2610
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.665832737669764e-05,
+      "loss": 1.2297,
+      "step": 2611
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.667619728377413e-05,
+      "loss": 1.2813,
+      "step": 2612
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.669406719085061e-05,
+      "loss": 1.2533,
+      "step": 2613
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.6711937097927096e-05,
+      "loss": 1.2637,
+      "step": 2614
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.672980700500358e-05,
+      "loss": 1.3108,
+      "step": 2615
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.674767691208006e-05,
+      "loss": 1.2614,
+      "step": 2616
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.676554681915654e-05,
+      "loss": 1.2386,
+      "step": 2617
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.6783416726233025e-05,
+      "loss": 1.2412,
+      "step": 2618
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.680128663330951e-05,
+      "loss": 1.2395,
+      "step": 2619
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.6819156540385994e-05,
+      "loss": 1.2805,
+      "step": 2620
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.683702644746247e-05,
+      "loss": 1.242,
+      "step": 2621
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.6854896354538955e-05,
+      "loss": 1.2955,
+      "step": 2622
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.687276626161544e-05,
+      "loss": 1.2679,
+      "step": 2623
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.6890636168691924e-05,
+      "loss": 1.2446,
+      "step": 2624
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.690850607576841e-05,
+      "loss": 1.2506,
+      "step": 2625
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.692637598284489e-05,
+      "loss": 1.2432,
+      "step": 2626
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.6944245889921376e-05,
+      "loss": 1.2583,
+      "step": 2627
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.696211579699786e-05,
+      "loss": 1.2403,
+      "step": 2628
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.6979985704074345e-05,
+      "loss": 1.2376,
+      "step": 2629
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.699785561115083e-05,
+      "loss": 1.258,
+      "step": 2630
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.7015725518227306e-05,
+      "loss": 1.2835,
+      "step": 2631
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.703359542530379e-05,
+      "loss": 1.2638,
+      "step": 2632
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.7051465332380275e-05,
+      "loss": 1.2353,
+      "step": 2633
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.706933523945676e-05,
+      "loss": 1.2698,
+      "step": 2634
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.708720514653324e-05,
+      "loss": 1.2466,
+      "step": 2635
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.710507505360972e-05,
+      "loss": 1.2495,
+      "step": 2636
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.7122944960686205e-05,
+      "loss": 1.2791,
+      "step": 2637
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.714081486776269e-05,
+      "loss": 1.252,
+      "step": 2638
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.715868477483917e-05,
+      "loss": 1.2868,
+      "step": 2639
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.717655468191566e-05,
+      "loss": 1.2701,
+      "step": 2640
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.719442458899214e-05,
+      "loss": 1.2127,
+      "step": 2641
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.721229449606862e-05,
+      "loss": 1.2366,
+      "step": 2642
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.72301644031451e-05,
+      "loss": 1.2754,
+      "step": 2643
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.724803431022159e-05,
+      "loss": 1.2697,
+      "step": 2644
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.726590421729807e-05,
+      "loss": 1.2864,
+      "step": 2645
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.7283774124374556e-05,
+      "loss": 1.2447,
+      "step": 2646
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.730164403145104e-05,
+      "loss": 1.2112,
+      "step": 2647
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.7319513938527524e-05,
+      "loss": 1.2629,
+      "step": 2648
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.733738384560401e-05,
+      "loss": 1.2812,
+      "step": 2649
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.735525375268049e-05,
+      "loss": 1.2726,
+      "step": 2650
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.7373123659756976e-05,
+      "loss": 1.2657,
+      "step": 2651
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.7390993566833454e-05,
+      "loss": 1.2711,
+      "step": 2652
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.740886347390994e-05,
+      "loss": 1.2272,
+      "step": 2653
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.742673338098642e-05,
+      "loss": 1.269,
+      "step": 2654
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.7444603288062906e-05,
+      "loss": 1.2685,
+      "step": 2655
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.746247319513939e-05,
+      "loss": 1.2779,
+      "step": 2656
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.748034310221587e-05,
+      "loss": 1.2619,
+      "step": 2657
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.749821300929235e-05,
+      "loss": 1.2402,
+      "step": 2658
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.7516082916368836e-05,
+      "loss": 1.2999,
+      "step": 2659
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.753395282344532e-05,
+      "loss": 1.2654,
+      "step": 2660
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.7551822730521805e-05,
+      "loss": 1.2331,
+      "step": 2661
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.756969263759828e-05,
+      "loss": 1.2927,
+      "step": 2662
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.7587562544674766e-05,
+      "loss": 1.2725,
+      "step": 2663
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.760543245175125e-05,
+      "loss": 1.2766,
+      "step": 2664
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.7623302358827735e-05,
+      "loss": 1.2483,
+      "step": 2665
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.764117226590422e-05,
+      "loss": 1.2959,
+      "step": 2666
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.76590421729807e-05,
+      "loss": 1.2418,
+      "step": 2667
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.767691208005719e-05,
+      "loss": 1.2305,
+      "step": 2668
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.769478198713367e-05,
+      "loss": 1.2548,
+      "step": 2669
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.7712651894210156e-05,
+      "loss": 1.2578,
+      "step": 2670
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.773052180128664e-05,
+      "loss": 1.2774,
+      "step": 2671
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.774839170836312e-05,
+      "loss": 1.2467,
+      "step": 2672
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.77662616154396e-05,
+      "loss": 1.2722,
+      "step": 2673
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.7784131522516086e-05,
+      "loss": 1.2605,
+      "step": 2674
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.780200142959257e-05,
+      "loss": 1.2562,
+      "step": 2675
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.7819871336669054e-05,
+      "loss": 1.2302,
+      "step": 2676
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.783774124374554e-05,
+      "loss": 1.2791,
+      "step": 2677
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.7855611150822016e-05,
+      "loss": 1.2649,
+      "step": 2678
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.78734810578985e-05,
+      "loss": 1.2599,
+      "step": 2679
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.7891350964974984e-05,
+      "loss": 1.2388,
+      "step": 2680
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.790922087205147e-05,
+      "loss": 1.2534,
+      "step": 2681
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.792709077912795e-05,
+      "loss": 1.2653,
+      "step": 2682
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.794496068620443e-05,
+      "loss": 1.2681,
+      "step": 2683
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.7962830593280914e-05,
+      "loss": 1.2736,
+      "step": 2684
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.79807005003574e-05,
+      "loss": 1.2784,
+      "step": 2685
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.799857040743388e-05,
+      "loss": 1.3109,
+      "step": 2686
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.8016440314510366e-05,
+      "loss": 1.2697,
+      "step": 2687
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.803431022158685e-05,
+      "loss": 1.2859,
+      "step": 2688
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.8052180128663335e-05,
+      "loss": 1.275,
+      "step": 2689
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.807005003573982e-05,
+      "loss": 1.2526,
+      "step": 2690
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.80879199428163e-05,
+      "loss": 1.2863,
+      "step": 2691
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.810578984989279e-05,
+      "loss": 1.2368,
+      "step": 2692
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.8123659756969265e-05,
+      "loss": 1.2679,
+      "step": 2693
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.814152966404575e-05,
+      "loss": 1.2573,
+      "step": 2694
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.815939957112223e-05,
+      "loss": 1.252,
+      "step": 2695
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.817726947819872e-05,
+      "loss": 1.2613,
+      "step": 2696
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.81951393852752e-05,
+      "loss": 1.2578,
+      "step": 2697
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.821300929235168e-05,
+      "loss": 1.2678,
+      "step": 2698
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.823087919942816e-05,
+      "loss": 1.253,
+      "step": 2699
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.824874910650465e-05,
+      "loss": 1.2351,
+      "step": 2700
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.826661901358113e-05,
+      "loss": 1.2496,
+      "step": 2701
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.8284488920657616e-05,
+      "loss": 1.2346,
+      "step": 2702
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.830235882773409e-05,
+      "loss": 1.2513,
+      "step": 2703
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.832022873481058e-05,
+      "loss": 1.2514,
+      "step": 2704
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.833809864188706e-05,
+      "loss": 1.2798,
+      "step": 2705
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.8355968548963546e-05,
+      "loss": 1.2581,
+      "step": 2706
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.837383845604003e-05,
+      "loss": 1.2418,
+      "step": 2707
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.8391708363116514e-05,
+      "loss": 1.2578,
+      "step": 2708
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.8409578270193e-05,
+      "loss": 1.2773,
+      "step": 2709
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.842744817726948e-05,
+      "loss": 1.2608,
+      "step": 2710
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.8445318084345966e-05,
+      "loss": 1.2577,
+      "step": 2711
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.846318799142245e-05,
+      "loss": 1.2518,
+      "step": 2712
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.8481057898498935e-05,
+      "loss": 1.2681,
+      "step": 2713
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.849892780557541e-05,
+      "loss": 1.2709,
+      "step": 2714
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.8516797712651896e-05,
+      "loss": 1.2636,
+      "step": 2715
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.853466761972838e-05,
+      "loss": 1.2527,
+      "step": 2716
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.8552537526804865e-05,
+      "loss": 1.2676,
+      "step": 2717
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.857040743388135e-05,
+      "loss": 1.2775,
+      "step": 2718
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.8588277340957826e-05,
+      "loss": 1.24,
+      "step": 2719
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.860614724803431e-05,
+      "loss": 1.2576,
+      "step": 2720
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.8624017155110795e-05,
+      "loss": 1.2467,
+      "step": 2721
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.864188706218728e-05,
+      "loss": 1.2272,
+      "step": 2722
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.865975696926376e-05,
+      "loss": 1.2769,
+      "step": 2723
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.867762687634024e-05,
+      "loss": 1.3195,
+      "step": 2724
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.8695496783416725e-05,
+      "loss": 1.2612,
+      "step": 2725
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.871336669049321e-05,
+      "loss": 1.2981,
+      "step": 2726
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.873123659756969e-05,
+      "loss": 1.2723,
+      "step": 2727
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.874910650464618e-05,
+      "loss": 1.2711,
+      "step": 2728
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.876697641172266e-05,
+      "loss": 1.2858,
+      "step": 2729
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.8784846318799146e-05,
+      "loss": 1.3021,
+      "step": 2730
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.880271622587563e-05,
+      "loss": 1.2642,
+      "step": 2731
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.8820586132952114e-05,
+      "loss": 1.2355,
+      "step": 2732
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.88384560400286e-05,
+      "loss": 1.234,
+      "step": 2733
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.8856325947105076e-05,
+      "loss": 1.2862,
+      "step": 2734
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.887419585418156e-05,
+      "loss": 1.2772,
+      "step": 2735
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.8892065761258044e-05,
+      "loss": 1.246,
+      "step": 2736
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.890993566833453e-05,
+      "loss": 1.2352,
+      "step": 2737
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.892780557541101e-05,
+      "loss": 1.2614,
+      "step": 2738
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.894567548248749e-05,
+      "loss": 1.2775,
+      "step": 2739
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.8963545389563974e-05,
+      "loss": 1.2813,
+      "step": 2740
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.898141529664046e-05,
+      "loss": 1.2502,
+      "step": 2741
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.899928520371694e-05,
+      "loss": 1.2312,
+      "step": 2742
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.9017155110793426e-05,
+      "loss": 1.2547,
+      "step": 2743
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.9035025017869904e-05,
+      "loss": 1.2381,
+      "step": 2744
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.905289492494639e-05,
+      "loss": 1.2542,
+      "step": 2745
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.907076483202287e-05,
+      "loss": 1.2757,
+      "step": 2746
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.9088634739099356e-05,
+      "loss": 1.2223,
+      "step": 2747
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.910650464617584e-05,
+      "loss": 1.2891,
+      "step": 2748
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.9124374553252325e-05,
+      "loss": 1.2614,
+      "step": 2749
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.914224446032881e-05,
+      "loss": 1.2394,
+      "step": 2750
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.916011436740529e-05,
+      "loss": 1.2498,
+      "step": 2751
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.917798427448178e-05,
+      "loss": 1.2543,
+      "step": 2752
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.919585418155826e-05,
+      "loss": 1.2474,
+      "step": 2753
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.9213724088634746e-05,
+      "loss": 1.2779,
+      "step": 2754
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.923159399571122e-05,
+      "loss": 1.2252,
+      "step": 2755
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.924946390278771e-05,
+      "loss": 1.2671,
+      "step": 2756
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.926733380986419e-05,
+      "loss": 1.2505,
+      "step": 2757
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.9285203716940676e-05,
+      "loss": 1.2472,
+      "step": 2758
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.930307362401716e-05,
+      "loss": 1.2487,
+      "step": 2759
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.932094353109364e-05,
+      "loss": 1.2462,
+      "step": 2760
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.933881343817012e-05,
+      "loss": 1.2801,
+      "step": 2761
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.9356683345246606e-05,
+      "loss": 1.2337,
+      "step": 2762
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.937455325232309e-05,
+      "loss": 1.2767,
+      "step": 2763
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.9392423159399574e-05,
+      "loss": 1.2346,
+      "step": 2764
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.941029306647605e-05,
+      "loss": 1.2379,
+      "step": 2765
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.9428162973552536e-05,
+      "loss": 1.259,
+      "step": 2766
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.944603288062902e-05,
+      "loss": 1.2473,
+      "step": 2767
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.9463902787705504e-05,
+      "loss": 1.2594,
+      "step": 2768
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.948177269478199e-05,
+      "loss": 1.2835,
+      "step": 2769
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.949964260185847e-05,
+      "loss": 1.2934,
+      "step": 2770
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.9517512508934957e-05,
+      "loss": 1.2626,
+      "step": 2771
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.953538241601144e-05,
+      "loss": 1.2595,
+      "step": 2772
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.9553252323087925e-05,
+      "loss": 1.2504,
+      "step": 2773
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.957112223016441e-05,
+      "loss": 1.2764,
+      "step": 2774
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.9588992137240886e-05,
+      "loss": 1.239,
+      "step": 2775
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.960686204431737e-05,
+      "loss": 1.2748,
+      "step": 2776
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.9624731951393855e-05,
+      "loss": 1.2782,
+      "step": 2777
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.964260185847034e-05,
+      "loss": 1.2906,
+      "step": 2778
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.966047176554682e-05,
+      "loss": 1.2671,
+      "step": 2779
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.96783416726233e-05,
+      "loss": 1.2745,
+      "step": 2780
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.9696211579699785e-05,
+      "loss": 1.2356,
+      "step": 2781
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.971408148677627e-05,
+      "loss": 1.2965,
+      "step": 2782
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.973195139385275e-05,
+      "loss": 1.2769,
+      "step": 2783
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.974982130092924e-05,
+      "loss": 1.2361,
+      "step": 2784
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.976769120800572e-05,
+      "loss": 1.2661,
+      "step": 2785
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.97855611150822e-05,
+      "loss": 1.278,
+      "step": 2786
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.980343102215868e-05,
+      "loss": 1.2689,
+      "step": 2787
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.982130092923517e-05,
+      "loss": 1.2828,
+      "step": 2788
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.983917083631166e-05,
+      "loss": 1.2955,
+      "step": 2789
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.985704074338814e-05,
+      "loss": 1.2246,
+      "step": 2790
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.987491065046462e-05,
+      "loss": 1.2837,
+      "step": 2791
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.9892780557541104e-05,
+      "loss": 1.2476,
+      "step": 2792
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.991065046461759e-05,
+      "loss": 1.2722,
+      "step": 2793
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.992852037169407e-05,
+      "loss": 1.3033,
+      "step": 2794
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.994639027877056e-05,
+      "loss": 1.2862,
+      "step": 2795
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.9964260185847034e-05,
+      "loss": 1.2991,
+      "step": 2796
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.998213009292352e-05,
+      "loss": 1.2514,
+      "step": 2797
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5e-05,
+      "loss": 1.278,
+      "step": 2798
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.0017869907076487e-05,
+      "loss": 1.2738,
+      "step": 2799
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.003573981415297e-05,
+      "loss": 1.2711,
+      "step": 2800
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.0053609721229455e-05,
+      "loss": 1.2457,
+      "step": 2801
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.007147962830594e-05,
+      "loss": 1.2585,
+      "step": 2802
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.008934953538242e-05,
+      "loss": 1.2593,
+      "step": 2803
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.0107219442458894e-05,
+      "loss": 1.2753,
+      "step": 2804
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.012508934953538e-05,
+      "loss": 1.2513,
+      "step": 2805
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.014295925661186e-05,
+      "loss": 1.2484,
+      "step": 2806
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.0160829163688346e-05,
+      "loss": 1.2118,
+      "step": 2807
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.017869907076484e-05,
+      "loss": 1.2203,
+      "step": 2808
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.019656897784132e-05,
+      "loss": 1.2533,
+      "step": 2809
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.0214438884917806e-05,
+      "loss": 1.2111,
+      "step": 2810
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.023230879199429e-05,
+      "loss": 1.2575,
+      "step": 2811
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.0250178699070774e-05,
+      "loss": 1.194,
+      "step": 2812
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.026804860614726e-05,
+      "loss": 1.2965,
+      "step": 2813
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.028591851322373e-05,
+      "loss": 1.2627,
+      "step": 2814
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.030378842030021e-05,
+      "loss": 1.2645,
+      "step": 2815
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.03216583273767e-05,
+      "loss": 1.2361,
+      "step": 2816
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.033952823445318e-05,
+      "loss": 1.2732,
+      "step": 2817
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.0357398141529666e-05,
+      "loss": 1.2271,
+      "step": 2818
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.037526804860615e-05,
+      "loss": 1.268,
+      "step": 2819
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.0393137955682634e-05,
+      "loss": 1.2407,
+      "step": 2820
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.041100786275912e-05,
+      "loss": 1.2383,
+      "step": 2821
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.04288777698356e-05,
+      "loss": 1.2722,
+      "step": 2822
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.044674767691209e-05,
+      "loss": 1.2585,
+      "step": 2823
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.046461758398856e-05,
+      "loss": 1.2613,
+      "step": 2824
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.048248749106504e-05,
+      "loss": 1.2481,
+      "step": 2825
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.0500357398141526e-05,
+      "loss": 1.2885,
+      "step": 2826
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.051822730521802e-05,
+      "loss": 1.257,
+      "step": 2827
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.05360972122945e-05,
+      "loss": 1.245,
+      "step": 2828
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.0553967119370985e-05,
+      "loss": 1.2705,
+      "step": 2829
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.057183702644747e-05,
+      "loss": 1.2277,
+      "step": 2830
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.058970693352395e-05,
+      "loss": 1.2352,
+      "step": 2831
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.060757684060044e-05,
+      "loss": 1.2958,
+      "step": 2832
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.062544674767692e-05,
+      "loss": 1.2524,
+      "step": 2833
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.0643316654753406e-05,
+      "loss": 1.2583,
+      "step": 2834
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.0661186561829877e-05,
+      "loss": 1.2377,
+      "step": 2835
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.067905646890636e-05,
+      "loss": 1.2233,
+      "step": 2836
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.0696926375982845e-05,
+      "loss": 1.245,
+      "step": 2837
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.071479628305933e-05,
+      "loss": 1.2695,
+      "step": 2838
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.073266619013581e-05,
+      "loss": 1.2294,
+      "step": 2839
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.07505360972123e-05,
+      "loss": 1.2234,
+      "step": 2840
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.076840600428878e-05,
+      "loss": 1.2389,
+      "step": 2841
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.0786275911365266e-05,
+      "loss": 1.2701,
+      "step": 2842
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.080414581844175e-05,
+      "loss": 1.263,
+      "step": 2843
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.0822015725518234e-05,
+      "loss": 1.2986,
+      "step": 2844
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.0839885632594705e-05,
+      "loss": 1.2524,
+      "step": 2845
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.0857755539671196e-05,
+      "loss": 1.2542,
+      "step": 2846
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.087562544674768e-05,
+      "loss": 1.2612,
+      "step": 2847
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.0893495353824164e-05,
+      "loss": 1.2871,
+      "step": 2848
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.091136526090065e-05,
+      "loss": 1.2749,
+      "step": 2849
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.092923516797713e-05,
+      "loss": 1.2334,
+      "step": 2850
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.094710507505362e-05,
+      "loss": 1.2226,
+      "step": 2851
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.09649749821301e-05,
+      "loss": 1.2614,
+      "step": 2852
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.0982844889206585e-05,
+      "loss": 1.2519,
+      "step": 2853
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.100071479628307e-05,
+      "loss": 1.2802,
+      "step": 2854
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.101858470335954e-05,
+      "loss": 1.24,
+      "step": 2855
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.1036454610436024e-05,
+      "loss": 1.2738,
+      "step": 2856
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.105432451751251e-05,
+      "loss": 1.2564,
+      "step": 2857
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.107219442458899e-05,
+      "loss": 1.2525,
+      "step": 2858
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.109006433166548e-05,
+      "loss": 1.2433,
+      "step": 2859
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.110793423874196e-05,
+      "loss": 1.2258,
+      "step": 2860
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.1125804145818445e-05,
+      "loss": 1.2776,
+      "step": 2861
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.114367405289493e-05,
+      "loss": 1.2772,
+      "step": 2862
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.116154395997141e-05,
+      "loss": 1.2907,
+      "step": 2863
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.11794138670479e-05,
+      "loss": 1.2741,
+      "step": 2864
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.1197283774124375e-05,
+      "loss": 1.2688,
+      "step": 2865
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.121515368120086e-05,
+      "loss": 1.2547,
+      "step": 2866
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.123302358827734e-05,
+      "loss": 1.2691,
+      "step": 2867
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.125089349535383e-05,
+      "loss": 1.2216,
+      "step": 2868
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.126876340243031e-05,
+      "loss": 1.2767,
+      "step": 2869
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.1286633309506796e-05,
+      "loss": 1.2613,
+      "step": 2870
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.130450321658328e-05,
+      "loss": 1.2749,
+      "step": 2871
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.1322373123659764e-05,
+      "loss": 1.2669,
+      "step": 2872
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.134024303073625e-05,
+      "loss": 1.2745,
+      "step": 2873
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.135811293781273e-05,
+      "loss": 1.2636,
+      "step": 2874
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.137598284488922e-05,
+      "loss": 1.2528,
+      "step": 2875
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.139385275196569e-05,
+      "loss": 1.2836,
+      "step": 2876
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.141172265904217e-05,
+      "loss": 1.2389,
+      "step": 2877
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.1429592566118656e-05,
+      "loss": 1.2294,
+      "step": 2878
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.144746247319514e-05,
+      "loss": 1.2218,
+      "step": 2879
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.1465332380271624e-05,
+      "loss": 1.2543,
+      "step": 2880
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.148320228734811e-05,
+      "loss": 1.2608,
+      "step": 2881
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.150107219442459e-05,
+      "loss": 1.2677,
+      "step": 2882
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.151894210150108e-05,
+      "loss": 1.2678,
+      "step": 2883
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.153681200857756e-05,
+      "loss": 1.265,
+      "step": 2884
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.1554681915654045e-05,
+      "loss": 1.2329,
+      "step": 2885
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.157255182273052e-05,
+      "loss": 1.2571,
+      "step": 2886
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.159042172980701e-05,
+      "loss": 1.2194,
+      "step": 2887
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.160829163688349e-05,
+      "loss": 1.2669,
+      "step": 2888
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.1626161543959975e-05,
+      "loss": 1.2522,
+      "step": 2889
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.164403145103646e-05,
+      "loss": 1.254,
+      "step": 2890
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.1661901358112943e-05,
+      "loss": 1.2268,
+      "step": 2891
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.167977126518943e-05,
+      "loss": 1.2669,
+      "step": 2892
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.169764117226591e-05,
+      "loss": 1.2431,
+      "step": 2893
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.1715511079342396e-05,
+      "loss": 1.2462,
+      "step": 2894
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.173338098641888e-05,
+      "loss": 1.2479,
+      "step": 2895
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.175125089349535e-05,
+      "loss": 1.2427,
+      "step": 2896
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.1769120800571835e-05,
+      "loss": 1.2382,
+      "step": 2897
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.178699070764832e-05,
+      "loss": 1.2247,
+      "step": 2898
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.18048606147248e-05,
+      "loss": 1.2183,
+      "step": 2899
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.182273052180129e-05,
+      "loss": 1.2534,
+      "step": 2900
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.184060042887777e-05,
+      "loss": 1.2569,
+      "step": 2901
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.1858470335954256e-05,
+      "loss": 1.263,
+      "step": 2902
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.187634024303074e-05,
+      "loss": 1.2318,
+      "step": 2903
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.1894210150107224e-05,
+      "loss": 1.2289,
+      "step": 2904
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.191208005718371e-05,
+      "loss": 1.2235,
+      "step": 2905
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.192994996426019e-05,
+      "loss": 1.2561,
+      "step": 2906
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.194781987133667e-05,
+      "loss": 1.2908,
+      "step": 2907
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.1965689778413154e-05,
+      "loss": 1.2721,
+      "step": 2908
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.198355968548964e-05,
+      "loss": 1.2534,
+      "step": 2909
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.200142959256612e-05,
+      "loss": 1.2711,
+      "step": 2910
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.201929949964261e-05,
+      "loss": 1.2021,
+      "step": 2911
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.203716940671909e-05,
+      "loss": 1.2392,
+      "step": 2912
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.2055039313795575e-05,
+      "loss": 1.2697,
+      "step": 2913
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.207290922087206e-05,
+      "loss": 1.2477,
+      "step": 2914
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.2090779127948544e-05,
+      "loss": 1.2774,
+      "step": 2915
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.210864903502503e-05,
+      "loss": 1.2726,
+      "step": 2916
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.21265189421015e-05,
+      "loss": 1.2572,
+      "step": 2917
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.214438884917798e-05,
+      "loss": 1.2409,
+      "step": 2918
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.216225875625447e-05,
+      "loss": 1.258,
+      "step": 2919
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.218012866333095e-05,
+      "loss": 1.2496,
+      "step": 2920
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.2197998570407435e-05,
+      "loss": 1.2493,
+      "step": 2921
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.221586847748392e-05,
+      "loss": 1.2965,
+      "step": 2922
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.2233738384560403e-05,
+      "loss": 1.2941,
+      "step": 2923
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.225160829163689e-05,
+      "loss": 1.2563,
+      "step": 2924
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.226947819871337e-05,
+      "loss": 1.2671,
+      "step": 2925
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.2287348105789856e-05,
+      "loss": 1.2489,
+      "step": 2926
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.230521801286633e-05,
+      "loss": 1.2562,
+      "step": 2927
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.232308791994282e-05,
+      "loss": 1.2366,
+      "step": 2928
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.23409578270193e-05,
+      "loss": 1.2541,
+      "step": 2929
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.2358827734095786e-05,
+      "loss": 1.2691,
+      "step": 2930
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.237669764117227e-05,
+      "loss": 1.2757,
+      "step": 2931
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.2394567548248754e-05,
+      "loss": 1.2715,
+      "step": 2932
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.241243745532524e-05,
+      "loss": 1.2615,
+      "step": 2933
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.243030736240172e-05,
+      "loss": 1.2676,
+      "step": 2934
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.244817726947821e-05,
+      "loss": 1.2333,
+      "step": 2935
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.246604717655469e-05,
+      "loss": 1.2644,
+      "step": 2936
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.248391708363116e-05,
+      "loss": 1.24,
+      "step": 2937
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.2501786990707646e-05,
+      "loss": 1.2947,
+      "step": 2938
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.251965689778413e-05,
+      "loss": 1.2601,
+      "step": 2939
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.2537526804860614e-05,
+      "loss": 1.2594,
+      "step": 2940
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.25553967119371e-05,
+      "loss": 1.2403,
+      "step": 2941
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.257326661901358e-05,
+      "loss": 1.2505,
+      "step": 2942
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.259113652609007e-05,
+      "loss": 1.247,
+      "step": 2943
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.260900643316655e-05,
+      "loss": 1.2935,
+      "step": 2944
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.2626876340243035e-05,
+      "loss": 1.2428,
+      "step": 2945
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.264474624731952e-05,
+      "loss": 1.2601,
+      "step": 2946
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.2662616154396004e-05,
+      "loss": 1.2641,
+      "step": 2947
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.268048606147248e-05,
+      "loss": 1.2745,
+      "step": 2948
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.2698355968548965e-05,
+      "loss": 1.2446,
+      "step": 2949
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.271622587562545e-05,
+      "loss": 1.2894,
+      "step": 2950
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.2734095782701933e-05,
+      "loss": 1.2526,
+      "step": 2951
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.275196568977842e-05,
+      "loss": 1.2797,
+      "step": 2952
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.27698355968549e-05,
+      "loss": 1.2779,
+      "step": 2953
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.2787705503931386e-05,
+      "loss": 1.2591,
+      "step": 2954
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.280557541100787e-05,
+      "loss": 1.2673,
+      "step": 2955
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.2823445318084354e-05,
+      "loss": 1.2469,
+      "step": 2956
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.284131522516084e-05,
+      "loss": 1.306,
+      "step": 2957
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.285918513223731e-05,
+      "loss": 1.3099,
+      "step": 2958
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.287705503931379e-05,
+      "loss": 1.2766,
+      "step": 2959
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.289492494639028e-05,
+      "loss": 1.291,
+      "step": 2960
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.291279485346676e-05,
+      "loss": 1.2562,
+      "step": 2961
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.2930664760543246e-05,
+      "loss": 1.2318,
+      "step": 2962
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.294853466761973e-05,
+      "loss": 1.2628,
+      "step": 2963
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.2966404574696214e-05,
+      "loss": 1.2993,
+      "step": 2964
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.29842744817727e-05,
+      "loss": 1.282,
+      "step": 2965
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.300214438884918e-05,
+      "loss": 1.2228,
+      "step": 2966
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.302001429592567e-05,
+      "loss": 1.3021,
+      "step": 2967
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.3037884203002144e-05,
+      "loss": 1.2515,
+      "step": 2968
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.305575411007863e-05,
+      "loss": 1.2512,
+      "step": 2969
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.307362401715511e-05,
+      "loss": 1.2474,
+      "step": 2970
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.30914939242316e-05,
+      "loss": 1.2387,
+      "step": 2971
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.310936383130808e-05,
+      "loss": 1.2872,
+      "step": 2972
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.3127233738384565e-05,
+      "loss": 1.2246,
+      "step": 2973
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.314510364546105e-05,
+      "loss": 1.2571,
+      "step": 2974
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.3162973552537534e-05,
+      "loss": 1.2524,
+      "step": 2975
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.318084345961402e-05,
+      "loss": 1.2443,
+      "step": 2976
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.31987133666905e-05,
+      "loss": 1.259,
+      "step": 2977
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.3216583273766986e-05,
+      "loss": 1.2642,
+      "step": 2978
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.323445318084346e-05,
+      "loss": 1.2456,
+      "step": 2979
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.325232308791994e-05,
+      "loss": 1.2741,
+      "step": 2980
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.3270192994996425e-05,
+      "loss": 1.2734,
+      "step": 2981
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.328806290207291e-05,
+      "loss": 1.2653,
+      "step": 2982
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.3305932809149393e-05,
+      "loss": 1.2125,
+      "step": 2983
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.332380271622588e-05,
+      "loss": 1.2576,
+      "step": 2984
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.334167262330236e-05,
+      "loss": 1.2909,
+      "step": 2985
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.3359542530378846e-05,
+      "loss": 1.2252,
+      "step": 2986
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.337741243745533e-05,
+      "loss": 1.2608,
+      "step": 2987
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.3395282344531814e-05,
+      "loss": 1.2702,
+      "step": 2988
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.341315225160829e-05,
+      "loss": 1.2788,
+      "step": 2989
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.3431022158684776e-05,
+      "loss": 1.2404,
+      "step": 2990
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.344889206576126e-05,
+      "loss": 1.2318,
+      "step": 2991
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.3466761972837744e-05,
+      "loss": 1.2345,
+      "step": 2992
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.348463187991423e-05,
+      "loss": 1.2559,
+      "step": 2993
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.350250178699071e-05,
+      "loss": 1.2278,
+      "step": 2994
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.35203716940672e-05,
+      "loss": 1.2286,
+      "step": 2995
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.353824160114368e-05,
+      "loss": 1.3027,
+      "step": 2996
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.3556111508220165e-05,
+      "loss": 1.2894,
+      "step": 2997
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.357398141529665e-05,
+      "loss": 1.2522,
+      "step": 2998
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.359185132237312e-05,
+      "loss": 1.2618,
+      "step": 2999
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.3609721229449604e-05,
+      "loss": 1.252,
+      "step": 3000
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.362759113652609e-05,
+      "loss": 1.2268,
+      "step": 3001
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.364546104360257e-05,
+      "loss": 1.2536,
+      "step": 3002
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.366333095067906e-05,
+      "loss": 1.2759,
+      "step": 3003
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.368120085775554e-05,
+      "loss": 1.2997,
+      "step": 3004
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.3699070764832025e-05,
+      "loss": 1.2577,
+      "step": 3005
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.371694067190851e-05,
+      "loss": 1.2427,
+      "step": 3006
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.3734810578984994e-05,
+      "loss": 1.2539,
+      "step": 3007
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.375268048606148e-05,
+      "loss": 1.2711,
+      "step": 3008
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.3770550393137955e-05,
+      "loss": 1.2909,
+      "step": 3009
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.378842030021444e-05,
+      "loss": 1.235,
+      "step": 3010
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.3806290207290924e-05,
+      "loss": 1.2781,
+      "step": 3011
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.382416011436741e-05,
+      "loss": 1.2353,
+      "step": 3012
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.384203002144389e-05,
+      "loss": 1.2486,
+      "step": 3013
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.3859899928520376e-05,
+      "loss": 1.2476,
+      "step": 3014
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.387776983559686e-05,
+      "loss": 1.2569,
+      "step": 3015
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3895639742673344e-05,
+      "loss": 1.253,
+      "step": 3016
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.391350964974983e-05,
+      "loss": 1.2168,
+      "step": 3017
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.393137955682631e-05,
+      "loss": 1.2614,
+      "step": 3018
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.39492494639028e-05,
+      "loss": 1.2522,
+      "step": 3019
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.396711937097927e-05,
+      "loss": 1.2447,
+      "step": 3020
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.398498927805575e-05,
+      "loss": 1.2493,
+      "step": 3021
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4002859185132236e-05,
+      "loss": 1.252,
+      "step": 3022
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.402072909220872e-05,
+      "loss": 1.2594,
+      "step": 3023
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4038598999285204e-05,
+      "loss": 1.2664,
+      "step": 3024
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.405646890636169e-05,
+      "loss": 1.3048,
+      "step": 3025
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.407433881343817e-05,
+      "loss": 1.2747,
+      "step": 3026
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.409220872051466e-05,
+      "loss": 1.2683,
+      "step": 3027
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.411007862759114e-05,
+      "loss": 1.2574,
+      "step": 3028
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4127948534667625e-05,
+      "loss": 1.242,
+      "step": 3029
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.41458184417441e-05,
+      "loss": 1.2913,
+      "step": 3030
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.416368834882059e-05,
+      "loss": 1.2781,
+      "step": 3031
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.418155825589707e-05,
+      "loss": 1.2709,
+      "step": 3032
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4199428162973555e-05,
+      "loss": 1.2275,
+      "step": 3033
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.421729807005004e-05,
+      "loss": 1.2763,
+      "step": 3034
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4235167977126524e-05,
+      "loss": 1.2472,
+      "step": 3035
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.425303788420301e-05,
+      "loss": 1.2304,
+      "step": 3036
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.427090779127949e-05,
+      "loss": 1.2854,
+      "step": 3037
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4288777698355976e-05,
+      "loss": 1.2361,
+      "step": 3038
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.430664760543246e-05,
+      "loss": 1.2113,
+      "step": 3039
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.432451751250893e-05,
+      "loss": 1.2705,
+      "step": 3040
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4342387419585415e-05,
+      "loss": 1.2646,
+      "step": 3041
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.43602573266619e-05,
+      "loss": 1.3048,
+      "step": 3042
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4378127233738384e-05,
+      "loss": 1.2477,
+      "step": 3043
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.439599714081487e-05,
+      "loss": 1.3054,
+      "step": 3044
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.441386704789135e-05,
+      "loss": 1.2749,
+      "step": 3045
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4431736954967836e-05,
+      "loss": 1.2689,
+      "step": 3046
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.444960686204432e-05,
+      "loss": 1.2423,
+      "step": 3047
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4467476769120804e-05,
+      "loss": 1.2853,
+      "step": 3048
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.448534667619729e-05,
+      "loss": 1.2979,
+      "step": 3049
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.450321658327377e-05,
+      "loss": 1.2901,
+      "step": 3050
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.452108649035025e-05,
+      "loss": 1.2673,
+      "step": 3051
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4538956397426734e-05,
+      "loss": 1.2635,
+      "step": 3052
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.455682630450322e-05,
+      "loss": 1.259,
+      "step": 3053
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.45746962115797e-05,
+      "loss": 1.2695,
+      "step": 3054
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.459256611865619e-05,
+      "loss": 1.275,
+      "step": 3055
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.461043602573267e-05,
+      "loss": 1.2329,
+      "step": 3056
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4628305932809155e-05,
+      "loss": 1.2448,
+      "step": 3057
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.464617583988564e-05,
+      "loss": 1.2409,
+      "step": 3058
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4664045746962124e-05,
+      "loss": 1.2095,
+      "step": 3059
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.468191565403861e-05,
+      "loss": 1.2517,
+      "step": 3060
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.469978556111508e-05,
+      "loss": 1.2546,
+      "step": 3061
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.471765546819156e-05,
+      "loss": 1.2513,
+      "step": 3062
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.473552537526805e-05,
+      "loss": 1.2613,
+      "step": 3063
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.475339528234453e-05,
+      "loss": 1.2182,
+      "step": 3064
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4771265189421015e-05,
+      "loss": 1.2723,
+      "step": 3065
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.47891350964975e-05,
+      "loss": 1.2607,
+      "step": 3066
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4807005003573984e-05,
+      "loss": 1.2536,
+      "step": 3067
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.482487491065047e-05,
+      "loss": 1.2384,
+      "step": 3068
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.484274481772695e-05,
+      "loss": 1.2471,
+      "step": 3069
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4860614724803436e-05,
+      "loss": 1.2715,
+      "step": 3070
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4878484631879914e-05,
+      "loss": 1.2695,
+      "step": 3071
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.48963545389564e-05,
+      "loss": 1.2622,
+      "step": 3072
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.491422444603288e-05,
+      "loss": 1.2462,
+      "step": 3073
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4932094353109366e-05,
+      "loss": 1.259,
+      "step": 3074
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.494996426018585e-05,
+      "loss": 1.2739,
+      "step": 3075
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4967834167262334e-05,
+      "loss": 1.2932,
+      "step": 3076
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.498570407433882e-05,
+      "loss": 1.2898,
+      "step": 3077
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.50035739814153e-05,
+      "loss": 1.2649,
+      "step": 3078
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.502144388849179e-05,
+      "loss": 1.2593,
+      "step": 3079
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.503931379556827e-05,
+      "loss": 1.2523,
+      "step": 3080
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.505718370264474e-05,
+      "loss": 1.2846,
+      "step": 3081
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.5075053609721226e-05,
+      "loss": 1.2647,
+      "step": 3082
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.509292351679771e-05,
+      "loss": 1.2486,
+      "step": 3083
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.5110793423874194e-05,
+      "loss": 1.2629,
+      "step": 3084
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.512866333095068e-05,
+      "loss": 1.2581,
+      "step": 3085
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.514653323802716e-05,
+      "loss": 1.3124,
+      "step": 3086
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.516440314510365e-05,
+      "loss": 1.2487,
+      "step": 3087
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.518227305218013e-05,
+      "loss": 1.2567,
+      "step": 3088
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.5200142959256615e-05,
+      "loss": 1.2693,
+      "step": 3089
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.52180128663331e-05,
+      "loss": 1.3001,
+      "step": 3090
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.5235882773409584e-05,
+      "loss": 1.2802,
+      "step": 3091
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.525375268048606e-05,
+      "loss": 1.3085,
+      "step": 3092
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.5271622587562545e-05,
+      "loss": 1.2749,
+      "step": 3093
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.528949249463903e-05,
+      "loss": 1.2879,
+      "step": 3094
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.5307362401715514e-05,
+      "loss": 1.2506,
+      "step": 3095
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.5325232308792e-05,
+      "loss": 1.2649,
+      "step": 3096
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.534310221586848e-05,
+      "loss": 1.2627,
+      "step": 3097
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.5360972122944966e-05,
+      "loss": 1.22,
+      "step": 3098
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.537884203002145e-05,
+      "loss": 1.2172,
+      "step": 3099
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.5396711937097935e-05,
+      "loss": 1.2544,
+      "step": 3100
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.541458184417442e-05,
+      "loss": 1.2478,
+      "step": 3101
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.543245175125089e-05,
+      "loss": 1.2592,
+      "step": 3102
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.5450321658327374e-05,
+      "loss": 1.2374,
+      "step": 3103
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.546819156540386e-05,
+      "loss": 1.2521,
+      "step": 3104
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.548606147248034e-05,
+      "loss": 1.2605,
+      "step": 3105
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.5503931379556826e-05,
+      "loss": 1.2172,
+      "step": 3106
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.552180128663331e-05,
+      "loss": 1.249,
+      "step": 3107
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.5539671193709794e-05,
+      "loss": 1.2893,
+      "step": 3108
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.555754110078628e-05,
+      "loss": 1.2588,
+      "step": 3109
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.557541100786276e-05,
+      "loss": 1.2281,
+      "step": 3110
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.559328091493925e-05,
+      "loss": 1.233,
+      "step": 3111
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.5611150822015724e-05,
+      "loss": 1.3051,
+      "step": 3112
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.562902072909221e-05,
+      "loss": 1.3048,
+      "step": 3113
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.564689063616869e-05,
+      "loss": 1.2432,
+      "step": 3114
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.566476054324518e-05,
+      "loss": 1.276,
+      "step": 3115
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.568263045032166e-05,
+      "loss": 1.2831,
+      "step": 3116
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.5700500357398145e-05,
+      "loss": 1.262,
+      "step": 3117
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.571837026447463e-05,
+      "loss": 1.245,
+      "step": 3118
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.5736240171551114e-05,
+      "loss": 1.2534,
+      "step": 3119
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.57541100786276e-05,
+      "loss": 1.2958,
+      "step": 3120
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.577197998570408e-05,
+      "loss": 1.2716,
+      "step": 3121
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.5789849892780566e-05,
+      "loss": 1.2428,
+      "step": 3122
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.580771979985704e-05,
+      "loss": 1.2872,
+      "step": 3123
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.582558970693352e-05,
+      "loss": 1.2967,
+      "step": 3124
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.5843459614010005e-05,
+      "loss": 1.2828,
+      "step": 3125
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.586132952108649e-05,
+      "loss": 1.2651,
+      "step": 3126
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.5879199428162974e-05,
+      "loss": 1.2628,
+      "step": 3127
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.589706933523946e-05,
+      "loss": 1.2773,
+      "step": 3128
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.591493924231594e-05,
+      "loss": 1.3105,
+      "step": 3129
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.5932809149392426e-05,
+      "loss": 1.2565,
+      "step": 3130
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.595067905646891e-05,
+      "loss": 1.2423,
+      "step": 3131
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.5968548963545395e-05,
+      "loss": 1.2479,
+      "step": 3132
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.598641887062187e-05,
+      "loss": 1.269,
+      "step": 3133
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.6004288777698356e-05,
+      "loss": 1.2682,
+      "step": 3134
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.602215868477484e-05,
+      "loss": 1.2699,
+      "step": 3135
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.6040028591851325e-05,
+      "loss": 1.2764,
+      "step": 3136
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.605789849892781e-05,
+      "loss": 1.2562,
+      "step": 3137
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.607576840600429e-05,
+      "loss": 1.2949,
+      "step": 3138
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.609363831308078e-05,
+      "loss": 1.2762,
+      "step": 3139
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.611150822015726e-05,
+      "loss": 1.2433,
+      "step": 3140
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.6129378127233745e-05,
+      "loss": 1.2765,
+      "step": 3141
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.614724803431023e-05,
+      "loss": 1.2153,
+      "step": 3142
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.61651179413867e-05,
+      "loss": 1.2637,
+      "step": 3143
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.6182987848463184e-05,
+      "loss": 1.2349,
+      "step": 3144
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.620085775553967e-05,
+      "loss": 1.2373,
+      "step": 3145
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.621872766261615e-05,
+      "loss": 1.2392,
+      "step": 3146
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.623659756969264e-05,
+      "loss": 1.2921,
+      "step": 3147
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.625446747676912e-05,
+      "loss": 1.2514,
+      "step": 3148
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.6272337383845605e-05,
+      "loss": 1.2543,
+      "step": 3149
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.629020729092209e-05,
+      "loss": 1.2767,
+      "step": 3150
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.6308077197998574e-05,
+      "loss": 1.2327,
+      "step": 3151
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.632594710507506e-05,
+      "loss": 1.2448,
+      "step": 3152
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.6343817012151535e-05,
+      "loss": 1.2673,
+      "step": 3153
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.636168691922802e-05,
+      "loss": 1.301,
+      "step": 3154
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.6379556826304504e-05,
+      "loss": 1.3119,
+      "step": 3155
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.639742673338099e-05,
+      "loss": 1.2784,
+      "step": 3156
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.641529664045747e-05,
+      "loss": 1.288,
+      "step": 3157
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.6433166547533956e-05,
+      "loss": 1.2193,
+      "step": 3158
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.645103645461044e-05,
+      "loss": 1.243,
+      "step": 3159
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.6468906361686925e-05,
+      "loss": 1.2868,
+      "step": 3160
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.648677626876341e-05,
+      "loss": 1.2602,
+      "step": 3161
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.650464617583989e-05,
+      "loss": 1.246,
+      "step": 3162
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.652251608291638e-05,
+      "loss": 1.2786,
+      "step": 3163
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.654038598999285e-05,
+      "loss": 1.2558,
+      "step": 3164
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.655825589706933e-05,
+      "loss": 1.2514,
+      "step": 3165
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.6576125804145816e-05,
+      "loss": 1.2447,
+      "step": 3166
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.65939957112223e-05,
+      "loss": 1.2594,
+      "step": 3167
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.6611865618298785e-05,
+      "loss": 1.2524,
+      "step": 3168
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.662973552537527e-05,
+      "loss": 1.2638,
+      "step": 3169
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.664760543245175e-05,
+      "loss": 1.2144,
+      "step": 3170
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.666547533952824e-05,
+      "loss": 1.2653,
+      "step": 3171
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.668334524660472e-05,
+      "loss": 1.2608,
+      "step": 3172
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.6701215153681205e-05,
+      "loss": 1.2897,
+      "step": 3173
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.671908506075768e-05,
+      "loss": 1.3099,
+      "step": 3174
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.673695496783417e-05,
+      "loss": 1.2933,
+      "step": 3175
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.675482487491065e-05,
+      "loss": 1.273,
+      "step": 3176
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.6772694781987135e-05,
+      "loss": 1.2438,
+      "step": 3177
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.679056468906362e-05,
+      "loss": 1.2624,
+      "step": 3178
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.6808434596140104e-05,
+      "loss": 1.2843,
+      "step": 3179
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.682630450321659e-05,
+      "loss": 1.2772,
+      "step": 3180
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.684417441029307e-05,
+      "loss": 1.2788,
+      "step": 3181
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.6862044317369556e-05,
+      "loss": 1.2908,
+      "step": 3182
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.687991422444604e-05,
+      "loss": 1.2816,
+      "step": 3183
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.689778413152251e-05,
+      "loss": 1.2763,
+      "step": 3184
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.6915654038598995e-05,
+      "loss": 1.2834,
+      "step": 3185
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.693352394567548e-05,
+      "loss": 1.2767,
+      "step": 3186
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.6951393852751964e-05,
+      "loss": 1.2704,
+      "step": 3187
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.696926375982845e-05,
+      "loss": 1.2152,
+      "step": 3188
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.698713366690493e-05,
+      "loss": 1.2702,
+      "step": 3189
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.7005003573981416e-05,
+      "loss": 1.2622,
+      "step": 3190
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.70228734810579e-05,
+      "loss": 1.2564,
+      "step": 3191
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.7040743388134385e-05,
+      "loss": 1.2492,
+      "step": 3192
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.705861329521087e-05,
+      "loss": 1.2282,
+      "step": 3193
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.707648320228735e-05,
+      "loss": 1.2405,
+      "step": 3194
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.709435310936383e-05,
+      "loss": 1.3001,
+      "step": 3195
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.7112223016440315e-05,
+      "loss": 1.2306,
+      "step": 3196
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.71300929235168e-05,
+      "loss": 1.231,
+      "step": 3197
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.714796283059328e-05,
+      "loss": 1.2648,
+      "step": 3198
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.716583273766977e-05,
+      "loss": 1.2816,
+      "step": 3199
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.718370264474625e-05,
+      "loss": 1.2726,
+      "step": 3200
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.7201572551822736e-05,
+      "loss": 1.259,
+      "step": 3201
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.721944245889922e-05,
+      "loss": 1.2859,
+      "step": 3202
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.7237312365975704e-05,
+      "loss": 1.2677,
+      "step": 3203
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.725518227305219e-05,
+      "loss": 1.2592,
+      "step": 3204
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.727305218012866e-05,
+      "loss": 1.2584,
+      "step": 3205
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.729092208720514e-05,
+      "loss": 1.251,
+      "step": 3206
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.730879199428163e-05,
+      "loss": 1.2197,
+      "step": 3207
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.732666190135811e-05,
+      "loss": 1.2366,
+      "step": 3208
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.7344531808434595e-05,
+      "loss": 1.3106,
+      "step": 3209
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.736240171551108e-05,
+      "loss": 1.266,
+      "step": 3210
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.7380271622587564e-05,
+      "loss": 1.2309,
+      "step": 3211
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.739814152966405e-05,
+      "loss": 1.2425,
+      "step": 3212
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.741601143674053e-05,
+      "loss": 1.2321,
+      "step": 3213
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.743388134381702e-05,
+      "loss": 1.2257,
+      "step": 3214
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.7451751250893494e-05,
+      "loss": 1.2217,
+      "step": 3215
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.746962115796998e-05,
+      "loss": 1.2414,
+      "step": 3216
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.748749106504646e-05,
+      "loss": 1.2305,
+      "step": 3217
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.7505360972122946e-05,
+      "loss": 1.2691,
+      "step": 3218
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.752323087919943e-05,
+      "loss": 1.2469,
+      "step": 3219
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.7541100786275915e-05,
+      "loss": 1.2454,
+      "step": 3220
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.75589706933524e-05,
+      "loss": 1.2677,
+      "step": 3221
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.757684060042888e-05,
+      "loss": 1.2289,
+      "step": 3222
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.759471050750537e-05,
+      "loss": 1.2286,
+      "step": 3223
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.761258041458185e-05,
+      "loss": 1.2879,
+      "step": 3224
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.763045032165832e-05,
+      "loss": 1.2426,
+      "step": 3225
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.7648320228734806e-05,
+      "loss": 1.252,
+      "step": 3226
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.766619013581129e-05,
+      "loss": 1.282,
+      "step": 3227
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.7684060042887775e-05,
+      "loss": 1.2416,
+      "step": 3228
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.770192994996426e-05,
+      "loss": 1.2555,
+      "step": 3229
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.771979985704074e-05,
+      "loss": 1.2733,
+      "step": 3230
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.773766976411723e-05,
+      "loss": 1.2923,
+      "step": 3231
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.775553967119371e-05,
+      "loss": 1.2377,
+      "step": 3232
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.77734095782702e-05,
+      "loss": 1.2292,
+      "step": 3233
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.7791279485346686e-05,
+      "loss": 1.2705,
+      "step": 3234
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.780914939242317e-05,
+      "loss": 1.2648,
+      "step": 3235
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.782701929949964e-05,
+      "loss": 1.2946,
+      "step": 3236
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.7844889206576125e-05,
+      "loss": 1.2831,
+      "step": 3237
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.786275911365261e-05,
+      "loss": 1.2785,
+      "step": 3238
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.7880629020729094e-05,
+      "loss": 1.2498,
+      "step": 3239
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.789849892780558e-05,
+      "loss": 1.2763,
+      "step": 3240
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.791636883488206e-05,
+      "loss": 1.2288,
+      "step": 3241
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.7934238741958546e-05,
+      "loss": 1.2543,
+      "step": 3242
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.795210864903503e-05,
+      "loss": 1.2603,
+      "step": 3243
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.7969978556111515e-05,
+      "loss": 1.2467,
+      "step": 3244
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.7987848463188e-05,
+      "loss": 1.2678,
+      "step": 3245
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.800571837026447e-05,
+      "loss": 1.2768,
+      "step": 3246
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.8023588277340954e-05,
+      "loss": 1.3083,
+      "step": 3247
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.804145818441744e-05,
+      "loss": 1.2751,
+      "step": 3248
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.805932809149392e-05,
+      "loss": 1.2721,
+      "step": 3249
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.8077197998570406e-05,
+      "loss": 1.26,
+      "step": 3250
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.809506790564689e-05,
+      "loss": 1.2612,
+      "step": 3251
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.811293781272338e-05,
+      "loss": 1.2669,
+      "step": 3252
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.8130807719799866e-05,
+      "loss": 1.2675,
+      "step": 3253
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.814867762687635e-05,
+      "loss": 1.2703,
+      "step": 3254
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.8166547533952834e-05,
+      "loss": 1.2676,
+      "step": 3255
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.8184417441029305e-05,
+      "loss": 1.2762,
+      "step": 3256
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.820228734810579e-05,
+      "loss": 1.2755,
+      "step": 3257
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.822015725518227e-05,
+      "loss": 1.2254,
+      "step": 3258
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.823802716225876e-05,
+      "loss": 1.2457,
+      "step": 3259
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.825589706933524e-05,
+      "loss": 1.275,
+      "step": 3260
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.8273766976411726e-05,
+      "loss": 1.2975,
+      "step": 3261
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.829163688348821e-05,
+      "loss": 1.2535,
+      "step": 3262
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5.8309506790564694e-05,
+      "loss": 1.2792,
+      "step": 3263
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.832737669764118e-05,
+      "loss": 1.2761,
+      "step": 3264
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.834524660471766e-05,
+      "loss": 1.2509,
+      "step": 3265
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.8363116511794146e-05,
+      "loss": 1.2424,
+      "step": 3266
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.838098641887062e-05,
+      "loss": 1.2499,
+      "step": 3267
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.83988563259471e-05,
+      "loss": 1.287,
+      "step": 3268
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.8416726233023585e-05,
+      "loss": 1.2204,
+      "step": 3269
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.843459614010007e-05,
+      "loss": 1.2251,
+      "step": 3270
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.845246604717656e-05,
+      "loss": 1.2346,
+      "step": 3271
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.8470335954253045e-05,
+      "loss": 1.2379,
+      "step": 3272
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.848820586132953e-05,
+      "loss": 1.2531,
+      "step": 3273
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.850607576840601e-05,
+      "loss": 1.267,
+      "step": 3274
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.85239456754825e-05,
+      "loss": 1.2498,
+      "step": 3275
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.854181558255898e-05,
+      "loss": 1.2718,
+      "step": 3276
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.855968548963545e-05,
+      "loss": 1.2905,
+      "step": 3277
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.8577555396711936e-05,
+      "loss": 1.3016,
+      "step": 3278
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.859542530378842e-05,
+      "loss": 1.2595,
+      "step": 3279
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.8613295210864905e-05,
+      "loss": 1.2867,
+      "step": 3280
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.863116511794139e-05,
+      "loss": 1.3015,
+      "step": 3281
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.864903502501787e-05,
+      "loss": 1.2952,
+      "step": 3282
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.866690493209436e-05,
+      "loss": 1.244,
+      "step": 3283
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.868477483917084e-05,
+      "loss": 1.2716,
+      "step": 3284
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.8702644746247326e-05,
+      "loss": 1.2661,
+      "step": 3285
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.872051465332381e-05,
+      "loss": 1.2587,
+      "step": 3286
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.873838456040028e-05,
+      "loss": 1.2921,
+      "step": 3287
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.8756254467476765e-05,
+      "loss": 1.2531,
+      "step": 3288
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.877412437455325e-05,
+      "loss": 1.2776,
+      "step": 3289
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.879199428162974e-05,
+      "loss": 1.2665,
+      "step": 3290
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.8809864188706224e-05,
+      "loss": 1.2576,
+      "step": 3291
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.882773409578271e-05,
+      "loss": 1.2655,
+      "step": 3292
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.884560400285919e-05,
+      "loss": 1.2626,
+      "step": 3293
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.8863473909935677e-05,
+      "loss": 1.2673,
+      "step": 3294
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.888134381701216e-05,
+      "loss": 1.2268,
+      "step": 3295
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.8899213724088645e-05,
+      "loss": 1.254,
+      "step": 3296
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.8917083631165116e-05,
+      "loss": 1.2411,
+      "step": 3297
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.89349535382416e-05,
+      "loss": 1.3062,
+      "step": 3298
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.8952823445318084e-05,
+      "loss": 1.2798,
+      "step": 3299
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.897069335239457e-05,
+      "loss": 1.2453,
+      "step": 3300
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.898856325947105e-05,
+      "loss": 1.2521,
+      "step": 3301
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.9006433166547536e-05,
+      "loss": 1.2502,
+      "step": 3302
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.902430307362402e-05,
+      "loss": 1.2752,
+      "step": 3303
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.9042172980700505e-05,
+      "loss": 1.265,
+      "step": 3304
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.906004288777699e-05,
+      "loss": 1.2792,
+      "step": 3305
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.907791279485347e-05,
+      "loss": 1.2791,
+      "step": 3306
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.909578270192996e-05,
+      "loss": 1.2545,
+      "step": 3307
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.911365260900643e-05,
+      "loss": 1.2944,
+      "step": 3308
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.913152251608292e-05,
+      "loss": 1.2537,
+      "step": 3309
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.91493924231594e-05,
+      "loss": 1.2695,
+      "step": 3310
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.916726233023589e-05,
+      "loss": 1.2658,
+      "step": 3311
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.918513223731237e-05,
+      "loss": 1.2851,
+      "step": 3312
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.9203002144388856e-05,
+      "loss": 1.2524,
+      "step": 3313
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.922087205146534e-05,
+      "loss": 1.299,
+      "step": 3314
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.9238741958541824e-05,
+      "loss": 1.2688,
+      "step": 3315
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.925661186561831e-05,
+      "loss": 1.2961,
+      "step": 3316
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.927448177269479e-05,
+      "loss": 1.2585,
+      "step": 3317
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.929235167977126e-05,
+      "loss": 1.2731,
+      "step": 3318
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.931022158684775e-05,
+      "loss": 1.2643,
+      "step": 3319
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.932809149392423e-05,
+      "loss": 1.2694,
+      "step": 3320
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.9345961401000716e-05,
+      "loss": 1.2667,
+      "step": 3321
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.93638313080772e-05,
+      "loss": 1.2532,
+      "step": 3322
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.9381701215153684e-05,
+      "loss": 1.2601,
+      "step": 3323
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.939957112223017e-05,
+      "loss": 1.2267,
+      "step": 3324
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.941744102930665e-05,
+      "loss": 1.2511,
+      "step": 3325
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 5.9435310936383137e-05,
+      "loss": 1.2199,
+      "step": 3326
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.945318084345962e-05,
+      "loss": 1.2425,
+      "step": 3327
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.94710507505361e-05,
+      "loss": 1.2435,
+      "step": 3328
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.948892065761258e-05,
+      "loss": 1.2763,
+      "step": 3329
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.9506790564689067e-05,
+      "loss": 1.2713,
+      "step": 3330
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.952466047176555e-05,
+      "loss": 1.2529,
+      "step": 3331
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.9542530378842035e-05,
+      "loss": 1.272,
+      "step": 3332
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.956040028591852e-05,
+      "loss": 1.2772,
+      "step": 3333
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.9578270192995e-05,
+      "loss": 1.2591,
+      "step": 3334
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.959614010007149e-05,
+      "loss": 1.2843,
+      "step": 3335
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.961401000714797e-05,
+      "loss": 1.2712,
+      "step": 3336
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.9631879914224456e-05,
+      "loss": 1.262,
+      "step": 3337
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.964974982130094e-05,
+      "loss": 1.2928,
+      "step": 3338
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.966761972837741e-05,
+      "loss": 1.2715,
+      "step": 3339
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.9685489635453895e-05,
+      "loss": 1.2507,
+      "step": 3340
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.970335954253038e-05,
+      "loss": 1.2168,
+      "step": 3341
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.972122944960686e-05,
+      "loss": 1.2811,
+      "step": 3342
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.973909935668335e-05,
+      "loss": 1.2389,
+      "step": 3343
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.975696926375983e-05,
+      "loss": 1.2467,
+      "step": 3344
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.9774839170836316e-05,
+      "loss": 1.2573,
+      "step": 3345
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.97927090779128e-05,
+      "loss": 1.2478,
+      "step": 3346
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.9810578984989284e-05,
+      "loss": 1.233,
+      "step": 3347
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.982844889206577e-05,
+      "loss": 1.2787,
+      "step": 3348
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.9846318799142246e-05,
+      "loss": 1.2698,
+      "step": 3349
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.986418870621873e-05,
+      "loss": 1.2894,
+      "step": 3350
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.9882058613295214e-05,
+      "loss": 1.2843,
+      "step": 3351
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.98999285203717e-05,
+      "loss": 1.2746,
+      "step": 3352
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.991779842744818e-05,
+      "loss": 1.276,
+      "step": 3353
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.9935668334524667e-05,
+      "loss": 1.2451,
+      "step": 3354
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.995353824160115e-05,
+      "loss": 1.2807,
+      "step": 3355
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.9971408148677635e-05,
+      "loss": 1.268,
+      "step": 3356
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 5.998927805575412e-05,
+      "loss": 1.249,
+      "step": 3357
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.00071479628306e-05,
+      "loss": 1.265,
+      "step": 3358
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.0025017869907074e-05,
+      "loss": 1.2734,
+      "step": 3359
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.004288777698356e-05,
+      "loss": 1.2634,
+      "step": 3360
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.006075768406004e-05,
+      "loss": 1.284,
+      "step": 3361
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.0078627591136527e-05,
+      "loss": 1.2544,
+      "step": 3362
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.009649749821301e-05,
+      "loss": 1.2634,
+      "step": 3363
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.0114367405289495e-05,
+      "loss": 1.2495,
+      "step": 3364
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.013223731236598e-05,
+      "loss": 1.263,
+      "step": 3365
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.015010721944246e-05,
+      "loss": 1.2828,
+      "step": 3366
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.016797712651895e-05,
+      "loss": 1.243,
+      "step": 3367
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.018584703359543e-05,
+      "loss": 1.2967,
+      "step": 3368
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.020371694067191e-05,
+      "loss": 1.2547,
+      "step": 3369
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.022158684774839e-05,
+      "loss": 1.2599,
+      "step": 3370
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.023945675482488e-05,
+      "loss": 1.2621,
+      "step": 3371
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.025732666190136e-05,
+      "loss": 1.2664,
+      "step": 3372
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.0275196568977846e-05,
+      "loss": 1.2774,
+      "step": 3373
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.029306647605433e-05,
+      "loss": 1.2921,
+      "step": 3374
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.0310936383130814e-05,
+      "loss": 1.2626,
+      "step": 3375
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.03288062902073e-05,
+      "loss": 1.2511,
+      "step": 3376
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.034667619728378e-05,
+      "loss": 1.2576,
+      "step": 3377
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.036454610436027e-05,
+      "loss": 1.266,
+      "step": 3378
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.038241601143675e-05,
+      "loss": 1.2504,
+      "step": 3379
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.040028591851322e-05,
+      "loss": 1.2584,
+      "step": 3380
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.0418155825589706e-05,
+      "loss": 1.2586,
+      "step": 3381
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.043602573266619e-05,
+      "loss": 1.2945,
+      "step": 3382
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.0453895639742674e-05,
+      "loss": 1.3,
+      "step": 3383
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.047176554681916e-05,
+      "loss": 1.2834,
+      "step": 3384
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.048963545389564e-05,
+      "loss": 1.2597,
+      "step": 3385
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.0507505360972127e-05,
+      "loss": 1.2851,
+      "step": 3386
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.052537526804861e-05,
+      "loss": 1.2365,
+      "step": 3387
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 6.0543245175125095e-05,
+      "loss": 1.2654,
+      "step": 3388
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.056111508220158e-05,
+      "loss": 1.2475,
+      "step": 3389
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.0578984989278057e-05,
+      "loss": 1.2498,
+      "step": 3390
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.059685489635454e-05,
+      "loss": 1.2805,
+      "step": 3391
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.0614724803431025e-05,
+      "loss": 1.2914,
+      "step": 3392
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.063259471050751e-05,
+      "loss": 1.275,
+      "step": 3393
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.065046461758399e-05,
+      "loss": 1.262,
+      "step": 3394
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.066833452466048e-05,
+      "loss": 1.2545,
+      "step": 3395
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.068620443173696e-05,
+      "loss": 1.2633,
+      "step": 3396
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.0704074338813446e-05,
+      "loss": 1.2867,
+      "step": 3397
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.072194424588993e-05,
+      "loss": 1.2395,
+      "step": 3398
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.0739814152966414e-05,
+      "loss": 1.2887,
+      "step": 3399
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.0757684060042885e-05,
+      "loss": 1.231,
+      "step": 3400
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.077555396711937e-05,
+      "loss": 1.2675,
+      "step": 3401
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.079342387419585e-05,
+      "loss": 1.2645,
+      "step": 3402
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.081129378127234e-05,
+      "loss": 1.2339,
+      "step": 3403
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.082916368834882e-05,
+      "loss": 1.2685,
+      "step": 3404
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.0847033595425306e-05,
+      "loss": 1.2954,
+      "step": 3405
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.086490350250179e-05,
+      "loss": 1.2558,
+      "step": 3406
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.0882773409578274e-05,
+      "loss": 1.2715,
+      "step": 3407
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.090064331665476e-05,
+      "loss": 1.2812,
+      "step": 3408
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.091851322373124e-05,
+      "loss": 1.3016,
+      "step": 3409
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.093638313080773e-05,
+      "loss": 1.262,
+      "step": 3410
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.0954253037884204e-05,
+      "loss": 1.2765,
+      "step": 3411
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.097212294496069e-05,
+      "loss": 1.2863,
+      "step": 3412
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.098999285203717e-05,
+      "loss": 1.2691,
+      "step": 3413
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.100786275911366e-05,
+      "loss": 1.2406,
+      "step": 3414
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.102573266619014e-05,
+      "loss": 1.2657,
+      "step": 3415
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.104360257326663e-05,
+      "loss": 1.2438,
+      "step": 3416
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.10614724803431e-05,
+      "loss": 1.2759,
+      "step": 3417
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.10793423874196e-05,
+      "loss": 1.2866,
+      "step": 3418
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.109721229449607e-05,
+      "loss": 1.2666,
+      "step": 3419
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.111508220157256e-05,
+      "loss": 1.2235,
+      "step": 3420
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.113295210864904e-05,
+      "loss": 1.2548,
+      "step": 3421
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.115082201572552e-05,
+      "loss": 1.285,
+      "step": 3422
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.116869192280201e-05,
+      "loss": 1.2604,
+      "step": 3423
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.118656182987848e-05,
+      "loss": 1.2895,
+      "step": 3424
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.120443173695498e-05,
+      "loss": 1.2471,
+      "step": 3425
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.122230164403145e-05,
+      "loss": 1.2638,
+      "step": 3426
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.124017155110794e-05,
+      "loss": 1.2686,
+      "step": 3427
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.125804145818442e-05,
+      "loss": 1.2316,
+      "step": 3428
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.127591136526091e-05,
+      "loss": 1.2768,
+      "step": 3429
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.129378127233739e-05,
+      "loss": 1.2622,
+      "step": 3430
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.131165117941387e-05,
+      "loss": 1.286,
+      "step": 3431
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.132952108649034e-05,
+      "loss": 1.2924,
+      "step": 3432
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.134739099356684e-05,
+      "loss": 1.2643,
+      "step": 3433
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.136526090064331e-05,
+      "loss": 1.2687,
+      "step": 3434
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.13831308077198e-05,
+      "loss": 1.2579,
+      "step": 3435
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.140100071479628e-05,
+      "loss": 1.3197,
+      "step": 3436
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.141887062187277e-05,
+      "loss": 1.2732,
+      "step": 3437
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.143674052894925e-05,
+      "loss": 1.289,
+      "step": 3438
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.145461043602574e-05,
+      "loss": 1.2503,
+      "step": 3439
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.147248034310222e-05,
+      "loss": 1.2566,
+      "step": 3440
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.14903502501787e-05,
+      "loss": 1.2596,
+      "step": 3441
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.150822015725519e-05,
+      "loss": 1.2996,
+      "step": 3442
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.152609006433166e-05,
+      "loss": 1.2547,
+      "step": 3443
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.154395997140816e-05,
+      "loss": 1.2664,
+      "step": 3444
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.156182987848463e-05,
+      "loss": 1.2854,
+      "step": 3445
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.157969978556112e-05,
+      "loss": 1.2833,
+      "step": 3446
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.15975696926376e-05,
+      "loss": 1.2836,
+      "step": 3447
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.161543959971409e-05,
+      "loss": 1.2515,
+      "step": 3448
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.163330950679057e-05,
+      "loss": 1.2522,
+      "step": 3449
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 6.165117941386706e-05,
+      "loss": 1.2391,
+      "step": 3450
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.166904932094354e-05,
+      "loss": 1.2928,
+      "step": 3451
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.168691922802001e-05,
+      "loss": 1.2673,
+      "step": 3452
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.170478913509649e-05,
+      "loss": 1.3072,
+      "step": 3453
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.172265904217298e-05,
+      "loss": 1.2671,
+      "step": 3454
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.174052894924946e-05,
+      "loss": 1.291,
+      "step": 3455
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.175839885632595e-05,
+      "loss": 1.2547,
+      "step": 3456
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.177626876340243e-05,
+      "loss": 1.2686,
+      "step": 3457
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.179413867047892e-05,
+      "loss": 1.2553,
+      "step": 3458
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.18120085775554e-05,
+      "loss": 1.2724,
+      "step": 3459
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.182987848463189e-05,
+      "loss": 1.2759,
+      "step": 3460
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.184774839170837e-05,
+      "loss": 1.2811,
+      "step": 3461
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.186561829878484e-05,
+      "loss": 1.2756,
+      "step": 3462
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.188348820586133e-05,
+      "loss": 1.2874,
+      "step": 3463
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.190135811293781e-05,
+      "loss": 1.2608,
+      "step": 3464
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.19192280200143e-05,
+      "loss": 1.2343,
+      "step": 3465
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.193709792709078e-05,
+      "loss": 1.2694,
+      "step": 3466
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.195496783416727e-05,
+      "loss": 1.2251,
+      "step": 3467
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.197283774124375e-05,
+      "loss": 1.2685,
+      "step": 3468
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.199070764832024e-05,
+      "loss": 1.2737,
+      "step": 3469
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.200857755539672e-05,
+      "loss": 1.2699,
+      "step": 3470
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.202644746247321e-05,
+      "loss": 1.296,
+      "step": 3471
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.204431736954967e-05,
+      "loss": 1.2467,
+      "step": 3472
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.206218727662616e-05,
+      "loss": 1.2941,
+      "step": 3473
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.208005718370264e-05,
+      "loss": 1.2851,
+      "step": 3474
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.209792709077913e-05,
+      "loss": 1.245,
+      "step": 3475
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.211579699785561e-05,
+      "loss": 1.2605,
+      "step": 3476
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.21336669049321e-05,
+      "loss": 1.2574,
+      "step": 3477
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.215153681200858e-05,
+      "loss": 1.3026,
+      "step": 3478
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.216940671908507e-05,
+      "loss": 1.2882,
+      "step": 3479
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.218727662616155e-05,
+      "loss": 1.2796,
+      "step": 3480
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.220514653323804e-05,
+      "loss": 1.2666,
+      "step": 3481
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.222301644031451e-05,
+      "loss": 1.2173,
+      "step": 3482
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.224088634739099e-05,
+      "loss": 1.2759,
+      "step": 3483
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.225875625446748e-05,
+      "loss": 1.2171,
+      "step": 3484
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.227662616154396e-05,
+      "loss": 1.2847,
+      "step": 3485
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.229449606862045e-05,
+      "loss": 1.2778,
+      "step": 3486
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.231236597569693e-05,
+      "loss": 1.2812,
+      "step": 3487
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.233023588277342e-05,
+      "loss": 1.2351,
+      "step": 3488
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.23481057898499e-05,
+      "loss": 1.2766,
+      "step": 3489
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.236597569692639e-05,
+      "loss": 1.2904,
+      "step": 3490
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.238384560400286e-05,
+      "loss": 1.259,
+      "step": 3491
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.240171551107936e-05,
+      "loss": 1.2481,
+      "step": 3492
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.241958541815582e-05,
+      "loss": 1.3003,
+      "step": 3493
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.243745532523231e-05,
+      "loss": 1.2589,
+      "step": 3494
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.245532523230879e-05,
+      "loss": 1.2825,
+      "step": 3495
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.247319513938528e-05,
+      "loss": 1.296,
+      "step": 3496
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.249106504646176e-05,
+      "loss": 1.2866,
+      "step": 3497
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.250893495353825e-05,
+      "loss": 1.2564,
+      "step": 3498
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.252680486061472e-05,
+      "loss": 1.2995,
+      "step": 3499
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.254467476769122e-05,
+      "loss": 1.2619,
+      "step": 3500
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.256254467476769e-05,
+      "loss": 1.2661,
+      "step": 3501
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.258041458184418e-05,
+      "loss": 1.2505,
+      "step": 3502
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.259828448892066e-05,
+      "loss": 1.261,
+      "step": 3503
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.261615439599714e-05,
+      "loss": 1.2814,
+      "step": 3504
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.263402430307363e-05,
+      "loss": 1.2467,
+      "step": 3505
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.26518942101501e-05,
+      "loss": 1.2631,
+      "step": 3506
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.26697641172266e-05,
+      "loss": 1.2786,
+      "step": 3507
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.268763402430308e-05,
+      "loss": 1.2903,
+      "step": 3508
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.270550393137957e-05,
+      "loss": 1.2708,
+      "step": 3509
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.272337383845604e-05,
+      "loss": 1.2732,
+      "step": 3510
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.274124374553253e-05,
+      "loss": 1.2432,
+      "step": 3511
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 6.275911365260901e-05,
+      "loss": 1.2524,
+      "step": 3512
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.277698355968549e-05,
+      "loss": 1.2756,
+      "step": 3513
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.279485346676197e-05,
+      "loss": 1.2718,
+      "step": 3514
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.281272337383846e-05,
+      "loss": 1.2705,
+      "step": 3515
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.283059328091493e-05,
+      "loss": 1.2696,
+      "step": 3516
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.284846318799143e-05,
+      "loss": 1.2565,
+      "step": 3517
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.28663330950679e-05,
+      "loss": 1.2427,
+      "step": 3518
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.28842030021444e-05,
+      "loss": 1.3082,
+      "step": 3519
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.290207290922087e-05,
+      "loss": 1.2309,
+      "step": 3520
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.291994281629736e-05,
+      "loss": 1.3092,
+      "step": 3521
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.293781272337384e-05,
+      "loss": 1.2833,
+      "step": 3522
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.295568263045033e-05,
+      "loss": 1.2607,
+      "step": 3523
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.297355253752681e-05,
+      "loss": 1.2451,
+      "step": 3524
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.299142244460329e-05,
+      "loss": 1.2808,
+      "step": 3525
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.300929235167978e-05,
+      "loss": 1.2689,
+      "step": 3526
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.302716225875625e-05,
+      "loss": 1.2524,
+      "step": 3527
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.304503216583275e-05,
+      "loss": 1.2596,
+      "step": 3528
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.306290207290922e-05,
+      "loss": 1.294,
+      "step": 3529
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.308077197998571e-05,
+      "loss": 1.2787,
+      "step": 3530
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.309864188706219e-05,
+      "loss": 1.239,
+      "step": 3531
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.311651179413868e-05,
+      "loss": 1.2951,
+      "step": 3532
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.313438170121516e-05,
+      "loss": 1.2534,
+      "step": 3533
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.315225160829164e-05,
+      "loss": 1.2824,
+      "step": 3534
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.317012151536811e-05,
+      "loss": 1.2735,
+      "step": 3535
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.31879914224446e-05,
+      "loss": 1.2539,
+      "step": 3536
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.320586132952108e-05,
+      "loss": 1.2695,
+      "step": 3537
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.322373123659757e-05,
+      "loss": 1.2544,
+      "step": 3538
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.324160114367405e-05,
+      "loss": 1.2883,
+      "step": 3539
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.325947105075054e-05,
+      "loss": 1.2573,
+      "step": 3540
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.327734095782702e-05,
+      "loss": 1.2365,
+      "step": 3541
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.329521086490351e-05,
+      "loss": 1.237,
+      "step": 3542
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.331308077197999e-05,
+      "loss": 1.2534,
+      "step": 3543
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.333095067905647e-05,
+      "loss": 1.2326,
+      "step": 3544
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.334882058613296e-05,
+      "loss": 1.2745,
+      "step": 3545
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.336669049320943e-05,
+      "loss": 1.268,
+      "step": 3546
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.338456040028592e-05,
+      "loss": 1.2566,
+      "step": 3547
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.34024303073624e-05,
+      "loss": 1.3075,
+      "step": 3548
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.342030021443889e-05,
+      "loss": 1.2835,
+      "step": 3549
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.343817012151537e-05,
+      "loss": 1.2859,
+      "step": 3550
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.345604002859186e-05,
+      "loss": 1.2806,
+      "step": 3551
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.347390993566834e-05,
+      "loss": 1.2729,
+      "step": 3552
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.349177984274483e-05,
+      "loss": 1.2542,
+      "step": 3553
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.350964974982131e-05,
+      "loss": 1.2996,
+      "step": 3554
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.352751965689778e-05,
+      "loss": 1.2892,
+      "step": 3555
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.354538956397426e-05,
+      "loss": 1.292,
+      "step": 3556
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.356325947105075e-05,
+      "loss": 1.2959,
+      "step": 3557
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.358112937812723e-05,
+      "loss": 1.2526,
+      "step": 3558
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.359899928520372e-05,
+      "loss": 1.2848,
+      "step": 3559
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.36168691922802e-05,
+      "loss": 1.282,
+      "step": 3560
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.363473909935669e-05,
+      "loss": 1.2484,
+      "step": 3561
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.365260900643317e-05,
+      "loss": 1.25,
+      "step": 3562
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.367047891350966e-05,
+      "loss": 1.2765,
+      "step": 3563
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.368834882058614e-05,
+      "loss": 1.2779,
+      "step": 3564
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.370621872766261e-05,
+      "loss": 1.2555,
+      "step": 3565
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.37240886347391e-05,
+      "loss": 1.289,
+      "step": 3566
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.374195854181558e-05,
+      "loss": 1.243,
+      "step": 3567
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.375982844889207e-05,
+      "loss": 1.2387,
+      "step": 3568
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.377769835596855e-05,
+      "loss": 1.28,
+      "step": 3569
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.379556826304504e-05,
+      "loss": 1.2598,
+      "step": 3570
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.381343817012152e-05,
+      "loss": 1.2735,
+      "step": 3571
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.383130807719801e-05,
+      "loss": 1.2967,
+      "step": 3572
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.384917798427449e-05,
+      "loss": 1.2718,
+      "step": 3573
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 6.386704789135098e-05,
+      "loss": 1.2842,
+      "step": 3574
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.388491779842744e-05,
+      "loss": 1.2827,
+      "step": 3575
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.390278770550393e-05,
+      "loss": 1.2664,
+      "step": 3576
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.392065761258041e-05,
+      "loss": 1.2676,
+      "step": 3577
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.39385275196569e-05,
+      "loss": 1.263,
+      "step": 3578
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.395639742673338e-05,
+      "loss": 1.2655,
+      "step": 3579
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.397426733380987e-05,
+      "loss": 1.2477,
+      "step": 3580
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.399213724088635e-05,
+      "loss": 1.272,
+      "step": 3581
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.401000714796284e-05,
+      "loss": 1.2667,
+      "step": 3582
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.402787705503931e-05,
+      "loss": 1.2346,
+      "step": 3583
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.40457469621158e-05,
+      "loss": 1.263,
+      "step": 3584
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.406361686919228e-05,
+      "loss": 1.2728,
+      "step": 3585
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.408148677626876e-05,
+      "loss": 1.2946,
+      "step": 3586
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.409935668334525e-05,
+      "loss": 1.2412,
+      "step": 3587
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.411722659042173e-05,
+      "loss": 1.2506,
+      "step": 3588
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.413509649749822e-05,
+      "loss": 1.3008,
+      "step": 3589
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.41529664045747e-05,
+      "loss": 1.2726,
+      "step": 3590
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.417083631165119e-05,
+      "loss": 1.2506,
+      "step": 3591
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.418870621872767e-05,
+      "loss": 1.2682,
+      "step": 3592
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.420657612580416e-05,
+      "loss": 1.2792,
+      "step": 3593
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.422444603288063e-05,
+      "loss": 1.2541,
+      "step": 3594
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.424231593995712e-05,
+      "loss": 1.2469,
+      "step": 3595
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.426018584703359e-05,
+      "loss": 1.2442,
+      "step": 3596
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.427805575411008e-05,
+      "loss": 1.2712,
+      "step": 3597
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.429592566118656e-05,
+      "loss": 1.2922,
+      "step": 3598
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.431379556826305e-05,
+      "loss": 1.2953,
+      "step": 3599
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.433166547533953e-05,
+      "loss": 1.2889,
+      "step": 3600
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.434953538241602e-05,
+      "loss": 1.3193,
+      "step": 3601
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.43674052894925e-05,
+      "loss": 1.2571,
+      "step": 3602
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.438527519656898e-05,
+      "loss": 1.2633,
+      "step": 3603
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.440314510364546e-05,
+      "loss": 1.2669,
+      "step": 3604
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.442101501072195e-05,
+      "loss": 1.2956,
+      "step": 3605
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.443888491779843e-05,
+      "loss": 1.269,
+      "step": 3606
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.445675482487491e-05,
+      "loss": 1.2881,
+      "step": 3607
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.44746247319514e-05,
+      "loss": 1.2675,
+      "step": 3608
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.449249463902788e-05,
+      "loss": 1.2742,
+      "step": 3609
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.451036454610437e-05,
+      "loss": 1.2563,
+      "step": 3610
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.452823445318084e-05,
+      "loss": 1.2686,
+      "step": 3611
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.454610436025734e-05,
+      "loss": 1.2838,
+      "step": 3612
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.456397426733381e-05,
+      "loss": 1.2668,
+      "step": 3613
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.45818441744103e-05,
+      "loss": 1.2882,
+      "step": 3614
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.459971408148678e-05,
+      "loss": 1.2521,
+      "step": 3615
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.461758398856326e-05,
+      "loss": 1.2627,
+      "step": 3616
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.463545389563974e-05,
+      "loss": 1.244,
+      "step": 3617
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.465332380271623e-05,
+      "loss": 1.2406,
+      "step": 3618
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.46711937097927e-05,
+      "loss": 1.2781,
+      "step": 3619
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.46890636168692e-05,
+      "loss": 1.29,
+      "step": 3620
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.470693352394567e-05,
+      "loss": 1.261,
+      "step": 3621
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.472480343102216e-05,
+      "loss": 1.2413,
+      "step": 3622
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.474267333809864e-05,
+      "loss": 1.2738,
+      "step": 3623
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.476054324517513e-05,
+      "loss": 1.2763,
+      "step": 3624
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.477841315225161e-05,
+      "loss": 1.2896,
+      "step": 3625
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.47962830593281e-05,
+      "loss": 1.2471,
+      "step": 3626
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.481415296640458e-05,
+      "loss": 1.2584,
+      "step": 3627
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.483202287348106e-05,
+      "loss": 1.241,
+      "step": 3628
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.484989278055755e-05,
+      "loss": 1.2619,
+      "step": 3629
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.486776268763402e-05,
+      "loss": 1.2214,
+      "step": 3630
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.488563259471051e-05,
+      "loss": 1.253,
+      "step": 3631
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.490350250178699e-05,
+      "loss": 1.291,
+      "step": 3632
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.492137240886348e-05,
+      "loss": 1.2713,
+      "step": 3633
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.493924231593996e-05,
+      "loss": 1.2776,
+      "step": 3634
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.495711222301645e-05,
+      "loss": 1.2446,
+      "step": 3635
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 6.497498213009293e-05,
+      "loss": 1.304,
+      "step": 3636
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.49928520371694e-05,
+      "loss": 1.2286,
+      "step": 3637
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.501072194424588e-05,
+      "loss": 1.2894,
+      "step": 3638
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.502859185132237e-05,
+      "loss": 1.2712,
+      "step": 3639
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.504646175839885e-05,
+      "loss": 1.2732,
+      "step": 3640
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.506433166547534e-05,
+      "loss": 1.2666,
+      "step": 3641
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.508220157255182e-05,
+      "loss": 1.2408,
+      "step": 3642
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.510007147962831e-05,
+      "loss": 1.3098,
+      "step": 3643
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.511794138670479e-05,
+      "loss": 1.2282,
+      "step": 3644
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.513581129378128e-05,
+      "loss": 1.2675,
+      "step": 3645
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.515368120085776e-05,
+      "loss": 1.2756,
+      "step": 3646
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.517155110793423e-05,
+      "loss": 1.2473,
+      "step": 3647
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.518942101501073e-05,
+      "loss": 1.2487,
+      "step": 3648
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.52072909220872e-05,
+      "loss": 1.2927,
+      "step": 3649
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.52251608291637e-05,
+      "loss": 1.3046,
+      "step": 3650
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.524303073624017e-05,
+      "loss": 1.27,
+      "step": 3651
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.526090064331666e-05,
+      "loss": 1.2782,
+      "step": 3652
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.527877055039314e-05,
+      "loss": 1.2583,
+      "step": 3653
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.529664045746963e-05,
+      "loss": 1.259,
+      "step": 3654
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.531451036454611e-05,
+      "loss": 1.2821,
+      "step": 3655
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.53323802716226e-05,
+      "loss": 1.2841,
+      "step": 3656
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.535025017869906e-05,
+      "loss": 1.2637,
+      "step": 3657
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.536812008577555e-05,
+      "loss": 1.2657,
+      "step": 3658
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.538598999285203e-05,
+      "loss": 1.2335,
+      "step": 3659
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.540385989992852e-05,
+      "loss": 1.2919,
+      "step": 3660
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.5421729807005e-05,
+      "loss": 1.2855,
+      "step": 3661
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.543959971408149e-05,
+      "loss": 1.2941,
+      "step": 3662
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.545746962115797e-05,
+      "loss": 1.266,
+      "step": 3663
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.547533952823446e-05,
+      "loss": 1.2487,
+      "step": 3664
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.549320943531094e-05,
+      "loss": 1.2255,
+      "step": 3665
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.551107934238743e-05,
+      "loss": 1.2422,
+      "step": 3666
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.55289492494639e-05,
+      "loss": 1.2789,
+      "step": 3667
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.554681915654038e-05,
+      "loss": 1.2761,
+      "step": 3668
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.556468906361687e-05,
+      "loss": 1.2858,
+      "step": 3669
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.558255897069335e-05,
+      "loss": 1.2631,
+      "step": 3670
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.560042887776984e-05,
+      "loss": 1.2743,
+      "step": 3671
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.561829878484632e-05,
+      "loss": 1.2952,
+      "step": 3672
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.563616869192281e-05,
+      "loss": 1.2571,
+      "step": 3673
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.565403859899929e-05,
+      "loss": 1.2691,
+      "step": 3674
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.567190850607578e-05,
+      "loss": 1.2726,
+      "step": 3675
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.568977841315226e-05,
+      "loss": 1.283,
+      "step": 3676
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.570764832022875e-05,
+      "loss": 1.2498,
+      "step": 3677
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.572551822730521e-05,
+      "loss": 1.2784,
+      "step": 3678
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.57433881343817e-05,
+      "loss": 1.212,
+      "step": 3679
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.576125804145818e-05,
+      "loss": 1.256,
+      "step": 3680
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.577912794853467e-05,
+      "loss": 1.2351,
+      "step": 3681
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.579699785561115e-05,
+      "loss": 1.2488,
+      "step": 3682
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.581486776268764e-05,
+      "loss": 1.2586,
+      "step": 3683
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.583273766976412e-05,
+      "loss": 1.267,
+      "step": 3684
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.58506075768406e-05,
+      "loss": 1.286,
+      "step": 3685
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.586847748391708e-05,
+      "loss": 1.2538,
+      "step": 3686
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.588634739099357e-05,
+      "loss": 1.2273,
+      "step": 3687
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.590421729807005e-05,
+      "loss": 1.2799,
+      "step": 3688
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.592208720514653e-05,
+      "loss": 1.2811,
+      "step": 3689
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.593995711222302e-05,
+      "loss": 1.2319,
+      "step": 3690
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.59578270192995e-05,
+      "loss": 1.2654,
+      "step": 3691
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.597569692637599e-05,
+      "loss": 1.2512,
+      "step": 3692
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.599356683345247e-05,
+      "loss": 1.2509,
+      "step": 3693
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.601143674052896e-05,
+      "loss": 1.2678,
+      "step": 3694
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.602930664760543e-05,
+      "loss": 1.2809,
+      "step": 3695
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.604717655468193e-05,
+      "loss": 1.251,
+      "step": 3696
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.60650464617584e-05,
+      "loss": 1.296,
+      "step": 3697
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.60829163688349e-05,
+      "loss": 1.281,
+      "step": 3698
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 6.610078627591136e-05,
+      "loss": 1.2461,
+      "step": 3699
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.611865618298785e-05,
+      "loss": 1.2945,
+      "step": 3700
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.613652609006433e-05,
+      "loss": 1.3108,
+      "step": 3701
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.615439599714082e-05,
+      "loss": 1.2701,
+      "step": 3702
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.61722659042173e-05,
+      "loss": 1.279,
+      "step": 3703
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.619013581129379e-05,
+      "loss": 1.3085,
+      "step": 3704
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.620800571837026e-05,
+      "loss": 1.2435,
+      "step": 3705
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.622587562544675e-05,
+      "loss": 1.2765,
+      "step": 3706
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.624374553252324e-05,
+      "loss": 1.2516,
+      "step": 3707
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.626161543959972e-05,
+      "loss": 1.2493,
+      "step": 3708
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.62794853466762e-05,
+      "loss": 1.2635,
+      "step": 3709
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.629735525375268e-05,
+      "loss": 1.239,
+      "step": 3710
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.631522516082917e-05,
+      "loss": 1.2688,
+      "step": 3711
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.633309506790565e-05,
+      "loss": 1.2389,
+      "step": 3712
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.635096497498214e-05,
+      "loss": 1.2047,
+      "step": 3713
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.636883488205861e-05,
+      "loss": 1.2511,
+      "step": 3714
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.63867047891351e-05,
+      "loss": 1.2579,
+      "step": 3715
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.640457469621158e-05,
+      "loss": 1.2466,
+      "step": 3716
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.642244460328807e-05,
+      "loss": 1.2394,
+      "step": 3717
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.644031451036455e-05,
+      "loss": 1.255,
+      "step": 3718
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.645818441744103e-05,
+      "loss": 1.2564,
+      "step": 3719
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.64760543245175e-05,
+      "loss": 1.2917,
+      "step": 3720
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.6493924231594e-05,
+      "loss": 1.2643,
+      "step": 3721
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.651179413867047e-05,
+      "loss": 1.252,
+      "step": 3722
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.652966404574696e-05,
+      "loss": 1.2908,
+      "step": 3723
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.654753395282344e-05,
+      "loss": 1.258,
+      "step": 3724
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.656540385989993e-05,
+      "loss": 1.3048,
+      "step": 3725
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.658327376697642e-05,
+      "loss": 1.264,
+      "step": 3726
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.66011436740529e-05,
+      "loss": 1.2949,
+      "step": 3727
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.661901358112939e-05,
+      "loss": 1.2447,
+      "step": 3728
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.663688348820586e-05,
+      "loss": 1.3037,
+      "step": 3729
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.665475339528235e-05,
+      "loss": 1.2816,
+      "step": 3730
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.667262330235882e-05,
+      "loss": 1.2819,
+      "step": 3731
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.669049320943532e-05,
+      "loss": 1.2375,
+      "step": 3732
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.670836311651179e-05,
+      "loss": 1.2843,
+      "step": 3733
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.672623302358828e-05,
+      "loss": 1.2658,
+      "step": 3734
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.674410293066476e-05,
+      "loss": 1.2633,
+      "step": 3735
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.676197283774125e-05,
+      "loss": 1.2988,
+      "step": 3736
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.677984274481773e-05,
+      "loss": 1.2743,
+      "step": 3737
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.679771265189422e-05,
+      "loss": 1.2941,
+      "step": 3738
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.68155825589707e-05,
+      "loss": 1.2691,
+      "step": 3739
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.683345246604718e-05,
+      "loss": 1.2603,
+      "step": 3740
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.685132237312365e-05,
+      "loss": 1.2374,
+      "step": 3741
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.686919228020014e-05,
+      "loss": 1.2633,
+      "step": 3742
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.688706218727662e-05,
+      "loss": 1.2513,
+      "step": 3743
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.690493209435311e-05,
+      "loss": 1.2453,
+      "step": 3744
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.69228020014296e-05,
+      "loss": 1.2905,
+      "step": 3745
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.694067190850608e-05,
+      "loss": 1.2658,
+      "step": 3746
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.695854181558257e-05,
+      "loss": 1.3022,
+      "step": 3747
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.697641172265905e-05,
+      "loss": 1.3142,
+      "step": 3748
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.699428162973554e-05,
+      "loss": 1.2538,
+      "step": 3749
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.7012151536812e-05,
+      "loss": 1.2848,
+      "step": 3750
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.70300214438885e-05,
+      "loss": 1.288,
+      "step": 3751
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.704789135096497e-05,
+      "loss": 1.268,
+      "step": 3752
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.706576125804146e-05,
+      "loss": 1.2592,
+      "step": 3753
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.708363116511794e-05,
+      "loss": 1.2762,
+      "step": 3754
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.710150107219443e-05,
+      "loss": 1.2677,
+      "step": 3755
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.711937097927091e-05,
+      "loss": 1.2508,
+      "step": 3756
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.71372408863474e-05,
+      "loss": 1.2779,
+      "step": 3757
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.715511079342388e-05,
+      "loss": 1.3053,
+      "step": 3758
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.717298070050037e-05,
+      "loss": 1.2679,
+      "step": 3759
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.719085060757683e-05,
+      "loss": 1.2788,
+      "step": 3760
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 6.720872051465332e-05,
+      "loss": 1.2812,
+      "step": 3761
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.72265904217298e-05,
+      "loss": 1.2754,
+      "step": 3762
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.724446032880629e-05,
+      "loss": 1.2825,
+      "step": 3763
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.726233023588278e-05,
+      "loss": 1.2231,
+      "step": 3764
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.728020014295926e-05,
+      "loss": 1.2791,
+      "step": 3765
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.729807005003575e-05,
+      "loss": 1.2576,
+      "step": 3766
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.731593995711223e-05,
+      "loss": 1.3142,
+      "step": 3767
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.733380986418872e-05,
+      "loss": 1.2419,
+      "step": 3768
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.73516797712652e-05,
+      "loss": 1.2524,
+      "step": 3769
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.736954967834169e-05,
+      "loss": 1.2486,
+      "step": 3770
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.738741958541815e-05,
+      "loss": 1.2565,
+      "step": 3771
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.740528949249464e-05,
+      "loss": 1.295,
+      "step": 3772
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.742315939957112e-05,
+      "loss": 1.2422,
+      "step": 3773
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.744102930664761e-05,
+      "loss": 1.2812,
+      "step": 3774
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.745889921372409e-05,
+      "loss": 1.2788,
+      "step": 3775
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.747676912080058e-05,
+      "loss": 1.2759,
+      "step": 3776
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.749463902787706e-05,
+      "loss": 1.2664,
+      "step": 3777
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.751250893495355e-05,
+      "loss": 1.2784,
+      "step": 3778
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.753037884203002e-05,
+      "loss": 1.2927,
+      "step": 3779
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.754824874910652e-05,
+      "loss": 1.2533,
+      "step": 3780
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.756611865618298e-05,
+      "loss": 1.2911,
+      "step": 3781
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.758398856325947e-05,
+      "loss": 1.2759,
+      "step": 3782
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.760185847033596e-05,
+      "loss": 1.2784,
+      "step": 3783
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.761972837741244e-05,
+      "loss": 1.2925,
+      "step": 3784
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.763759828448893e-05,
+      "loss": 1.2503,
+      "step": 3785
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.765546819156541e-05,
+      "loss": 1.2581,
+      "step": 3786
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.76733380986419e-05,
+      "loss": 1.2861,
+      "step": 3787
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.769120800571838e-05,
+      "loss": 1.2449,
+      "step": 3788
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.770907791279487e-05,
+      "loss": 1.3214,
+      "step": 3789
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.772694781987134e-05,
+      "loss": 1.2624,
+      "step": 3790
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.774481772694782e-05,
+      "loss": 1.2253,
+      "step": 3791
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.77626876340243e-05,
+      "loss": 1.2467,
+      "step": 3792
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.778055754110079e-05,
+      "loss": 1.268,
+      "step": 3793
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.779842744817727e-05,
+      "loss": 1.2748,
+      "step": 3794
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.781629735525376e-05,
+      "loss": 1.2586,
+      "step": 3795
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.783416726233024e-05,
+      "loss": 1.2862,
+      "step": 3796
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.785203716940673e-05,
+      "loss": 1.3149,
+      "step": 3797
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.78699070764832e-05,
+      "loss": 1.266,
+      "step": 3798
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.78877769835597e-05,
+      "loss": 1.2549,
+      "step": 3799
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.790564689063617e-05,
+      "loss": 1.2254,
+      "step": 3800
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.792351679771265e-05,
+      "loss": 1.243,
+      "step": 3801
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.794138670478914e-05,
+      "loss": 1.2519,
+      "step": 3802
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.795925661186562e-05,
+      "loss": 1.2982,
+      "step": 3803
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.797712651894211e-05,
+      "loss": 1.2995,
+      "step": 3804
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.799499642601859e-05,
+      "loss": 1.2759,
+      "step": 3805
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.801286633309508e-05,
+      "loss": 1.2879,
+      "step": 3806
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.803073624017155e-05,
+      "loss": 1.289,
+      "step": 3807
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.804860614724805e-05,
+      "loss": 1.2612,
+      "step": 3808
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.806647605432452e-05,
+      "loss": 1.2442,
+      "step": 3809
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.808434596140101e-05,
+      "loss": 1.2452,
+      "step": 3810
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.810221586847749e-05,
+      "loss": 1.2786,
+      "step": 3811
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.812008577555397e-05,
+      "loss": 1.2638,
+      "step": 3812
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.813795568263045e-05,
+      "loss": 1.2552,
+      "step": 3813
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.815582558970694e-05,
+      "loss": 1.2759,
+      "step": 3814
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.817369549678341e-05,
+      "loss": 1.2567,
+      "step": 3815
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.81915654038599e-05,
+      "loss": 1.2707,
+      "step": 3816
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.820943531093638e-05,
+      "loss": 1.2537,
+      "step": 3817
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.822730521801287e-05,
+      "loss": 1.2887,
+      "step": 3818
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.824517512508935e-05,
+      "loss": 1.2515,
+      "step": 3819
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.826304503216584e-05,
+      "loss": 1.2263,
+      "step": 3820
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.828091493924232e-05,
+      "loss": 1.2696,
+      "step": 3821
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.82987848463188e-05,
+      "loss": 1.3016,
+      "step": 3822
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.831665475339529e-05,
+      "loss": 1.224,
+      "step": 3823
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.833452466047177e-05,
+      "loss": 1.2525,
+      "step": 3824
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.835239456754826e-05,
+      "loss": 1.2552,
+      "step": 3825
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.837026447462473e-05,
+      "loss": 1.2604,
+      "step": 3826
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.838813438170122e-05,
+      "loss": 1.2794,
+      "step": 3827
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.84060042887777e-05,
+      "loss": 1.255,
+      "step": 3828
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.842387419585419e-05,
+      "loss": 1.2466,
+      "step": 3829
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.844174410293067e-05,
+      "loss": 1.309,
+      "step": 3830
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.845961401000716e-05,
+      "loss": 1.2892,
+      "step": 3831
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.847748391708363e-05,
+      "loss": 1.3133,
+      "step": 3832
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.849535382416012e-05,
+      "loss": 1.2481,
+      "step": 3833
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.85132237312366e-05,
+      "loss": 1.2984,
+      "step": 3834
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.853109363831308e-05,
+      "loss": 1.2625,
+      "step": 3835
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.854896354538956e-05,
+      "loss": 1.2712,
+      "step": 3836
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.856683345246605e-05,
+      "loss": 1.271,
+      "step": 3837
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.858470335954253e-05,
+      "loss": 1.2673,
+      "step": 3838
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.860257326661902e-05,
+      "loss": 1.2903,
+      "step": 3839
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.86204431736955e-05,
+      "loss": 1.2808,
+      "step": 3840
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.863831308077199e-05,
+      "loss": 1.2632,
+      "step": 3841
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.865618298784847e-05,
+      "loss": 1.2511,
+      "step": 3842
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.867405289492494e-05,
+      "loss": 1.3049,
+      "step": 3843
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.869192280200144e-05,
+      "loss": 1.2748,
+      "step": 3844
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.870979270907791e-05,
+      "loss": 1.2692,
+      "step": 3845
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.87276626161544e-05,
+      "loss": 1.222,
+      "step": 3846
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.874553252323088e-05,
+      "loss": 1.2458,
+      "step": 3847
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.876340243030737e-05,
+      "loss": 1.2425,
+      "step": 3848
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.878127233738385e-05,
+      "loss": 1.2681,
+      "step": 3849
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.879914224446034e-05,
+      "loss": 1.2723,
+      "step": 3850
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.881701215153682e-05,
+      "loss": 1.2827,
+      "step": 3851
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.883488205861331e-05,
+      "loss": 1.2725,
+      "step": 3852
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.885275196568977e-05,
+      "loss": 1.2793,
+      "step": 3853
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.887062187276626e-05,
+      "loss": 1.2793,
+      "step": 3854
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.888849177984274e-05,
+      "loss": 1.2628,
+      "step": 3855
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.890636168691923e-05,
+      "loss": 1.2919,
+      "step": 3856
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.892423159399571e-05,
+      "loss": 1.2608,
+      "step": 3857
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.89421015010722e-05,
+      "loss": 1.2567,
+      "step": 3858
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.895997140814868e-05,
+      "loss": 1.2794,
+      "step": 3859
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.897784131522517e-05,
+      "loss": 1.2804,
+      "step": 3860
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.899571122230165e-05,
+      "loss": 1.3001,
+      "step": 3861
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.901358112937814e-05,
+      "loss": 1.2456,
+      "step": 3862
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.903145103645461e-05,
+      "loss": 1.2782,
+      "step": 3863
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.904932094353109e-05,
+      "loss": 1.2759,
+      "step": 3864
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.906719085060758e-05,
+      "loss": 1.2715,
+      "step": 3865
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.908506075768406e-05,
+      "loss": 1.3034,
+      "step": 3866
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.910293066476055e-05,
+      "loss": 1.2416,
+      "step": 3867
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.912080057183703e-05,
+      "loss": 1.2403,
+      "step": 3868
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.913867047891352e-05,
+      "loss": 1.2683,
+      "step": 3869
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.915654038599e-05,
+      "loss": 1.2969,
+      "step": 3870
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.917441029306649e-05,
+      "loss": 1.2364,
+      "step": 3871
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.919228020014297e-05,
+      "loss": 1.281,
+      "step": 3872
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.921015010721944e-05,
+      "loss": 1.2541,
+      "step": 3873
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.922802001429592e-05,
+      "loss": 1.247,
+      "step": 3874
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.924588992137241e-05,
+      "loss": 1.2726,
+      "step": 3875
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.926375982844889e-05,
+      "loss": 1.2401,
+      "step": 3876
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.928162973552538e-05,
+      "loss": 1.3147,
+      "step": 3877
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.929949964260186e-05,
+      "loss": 1.284,
+      "step": 3878
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.931736954967835e-05,
+      "loss": 1.2987,
+      "step": 3879
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.933523945675483e-05,
+      "loss": 1.2719,
+      "step": 3880
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.935310936383132e-05,
+      "loss": 1.3269,
+      "step": 3881
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.93709792709078e-05,
+      "loss": 1.2815,
+      "step": 3882
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.938884917798428e-05,
+      "loss": 1.2864,
+      "step": 3883
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.940671908506076e-05,
+      "loss": 1.2653,
+      "step": 3884
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.942458899213724e-05,
+      "loss": 1.2733,
+      "step": 3885
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.944245889921373e-05,
+      "loss": 1.3202,
+      "step": 3886
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.946032880629021e-05,
+      "loss": 1.2371,
+      "step": 3887
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.94781987133667e-05,
+      "loss": 1.2466,
+      "step": 3888
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.949606862044318e-05,
+      "loss": 1.2577,
+      "step": 3889
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.951393852751967e-05,
+      "loss": 1.2853,
+      "step": 3890
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.953180843459614e-05,
+      "loss": 1.2689,
+      "step": 3891
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.954967834167264e-05,
+      "loss": 1.2689,
+      "step": 3892
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.956754824874911e-05,
+      "loss": 1.2786,
+      "step": 3893
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.958541815582559e-05,
+      "loss": 1.26,
+      "step": 3894
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.960328806290207e-05,
+      "loss": 1.2791,
+      "step": 3895
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.962115796997856e-05,
+      "loss": 1.2686,
+      "step": 3896
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.963902787705504e-05,
+      "loss": 1.2287,
+      "step": 3897
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.965689778413153e-05,
+      "loss": 1.2754,
+      "step": 3898
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.9674767691208e-05,
+      "loss": 1.26,
+      "step": 3899
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.96926375982845e-05,
+      "loss": 1.2492,
+      "step": 3900
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.971050750536097e-05,
+      "loss": 1.2548,
+      "step": 3901
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.972837741243746e-05,
+      "loss": 1.2653,
+      "step": 3902
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.974624731951394e-05,
+      "loss": 1.2515,
+      "step": 3903
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.976411722659042e-05,
+      "loss": 1.2665,
+      "step": 3904
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.978198713366691e-05,
+      "loss": 1.2665,
+      "step": 3905
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.979985704074339e-05,
+      "loss": 1.2706,
+      "step": 3906
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.981772694781988e-05,
+      "loss": 1.3156,
+      "step": 3907
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.983559685489636e-05,
+      "loss": 1.2564,
+      "step": 3908
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.985346676197285e-05,
+      "loss": 1.2595,
+      "step": 3909
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.987133666904932e-05,
+      "loss": 1.2797,
+      "step": 3910
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.988920657612581e-05,
+      "loss": 1.2845,
+      "step": 3911
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.990707648320229e-05,
+      "loss": 1.3372,
+      "step": 3912
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.992494639027878e-05,
+      "loss": 1.2804,
+      "step": 3913
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.994281629735526e-05,
+      "loss": 1.2609,
+      "step": 3914
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.996068620443174e-05,
+      "loss": 1.2872,
+      "step": 3915
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.997855611150822e-05,
+      "loss": 1.2683,
+      "step": 3916
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.99964260185847e-05,
+      "loss": 1.2715,
+      "step": 3917
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.001429592566118e-05,
+      "loss": 1.2835,
+      "step": 3918
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.003216583273767e-05,
+      "loss": 1.273,
+      "step": 3919
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.005003573981415e-05,
+      "loss": 1.2585,
+      "step": 3920
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.006790564689064e-05,
+      "loss": 1.2326,
+      "step": 3921
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.008577555396712e-05,
+      "loss": 1.2639,
+      "step": 3922
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.010364546104361e-05,
+      "loss": 1.2854,
+      "step": 3923
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.012151536812009e-05,
+      "loss": 1.2374,
+      "step": 3924
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.013938527519657e-05,
+      "loss": 1.2638,
+      "step": 3925
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.015725518227306e-05,
+      "loss": 1.283,
+      "step": 3926
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.017512508934953e-05,
+      "loss": 1.2493,
+      "step": 3927
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.019299499642603e-05,
+      "loss": 1.2402,
+      "step": 3928
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.02108649035025e-05,
+      "loss": 1.2667,
+      "step": 3929
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.0228734810579e-05,
+      "loss": 1.2899,
+      "step": 3930
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.024660471765547e-05,
+      "loss": 1.2664,
+      "step": 3931
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.026447462473196e-05,
+      "loss": 1.286,
+      "step": 3932
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.028234453180844e-05,
+      "loss": 1.2799,
+      "step": 3933
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.030021443888493e-05,
+      "loss": 1.2741,
+      "step": 3934
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.03180843459614e-05,
+      "loss": 1.3127,
+      "step": 3935
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.033595425303789e-05,
+      "loss": 1.2721,
+      "step": 3936
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.035382416011436e-05,
+      "loss": 1.2537,
+      "step": 3937
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.037169406719085e-05,
+      "loss": 1.2497,
+      "step": 3938
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.038956397426733e-05,
+      "loss": 1.2852,
+      "step": 3939
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.040743388134382e-05,
+      "loss": 1.2631,
+      "step": 3940
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.04253037884203e-05,
+      "loss": 1.2746,
+      "step": 3941
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.044317369549679e-05,
+      "loss": 1.239,
+      "step": 3942
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.046104360257327e-05,
+      "loss": 1.2895,
+      "step": 3943
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.047891350964976e-05,
+      "loss": 1.2377,
+      "step": 3944
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.049678341672624e-05,
+      "loss": 1.2569,
+      "step": 3945
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.051465332380271e-05,
+      "loss": 1.2621,
+      "step": 3946
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.05325232308792e-05,
+      "loss": 1.2932,
+      "step": 3947
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.055039313795568e-05,
+      "loss": 1.2461,
+      "step": 3948
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.056826304503217e-05,
+      "loss": 1.2896,
+      "step": 3949
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.058613295210865e-05,
+      "loss": 1.2843,
+      "step": 3950
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.060400285918514e-05,
+      "loss": 1.2564,
+      "step": 3951
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.062187276626162e-05,
+      "loss": 1.2415,
+      "step": 3952
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.063974267333811e-05,
+      "loss": 1.2748,
+      "step": 3953
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.065761258041459e-05,
+      "loss": 1.3058,
+      "step": 3954
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.067548248749108e-05,
+      "loss": 1.2562,
+      "step": 3955
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.069335239456754e-05,
+      "loss": 1.2625,
+      "step": 3956
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.071122230164403e-05,
+      "loss": 1.2851,
+      "step": 3957
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.072909220872051e-05,
+      "loss": 1.3216,
+      "step": 3958
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.0746962115797e-05,
+      "loss": 1.2528,
+      "step": 3959
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.076483202287348e-05,
+      "loss": 1.2872,
+      "step": 3960
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.078270192994997e-05,
+      "loss": 1.2621,
+      "step": 3961
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.080057183702645e-05,
+      "loss": 1.2604,
+      "step": 3962
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.081844174410294e-05,
+      "loss": 1.2471,
+      "step": 3963
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.083631165117942e-05,
+      "loss": 1.2869,
+      "step": 3964
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.08541815582559e-05,
+      "loss": 1.2698,
+      "step": 3965
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.087205146533238e-05,
+      "loss": 1.2579,
+      "step": 3966
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.088992137240886e-05,
+      "loss": 1.2596,
+      "step": 3967
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.090779127948535e-05,
+      "loss": 1.2446,
+      "step": 3968
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.092566118656183e-05,
+      "loss": 1.2624,
+      "step": 3969
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.094353109363832e-05,
+      "loss": 1.2742,
+      "step": 3970
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.09614010007148e-05,
+      "loss": 1.2632,
+      "step": 3971
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.097927090779129e-05,
+      "loss": 1.296,
+      "step": 3972
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.099714081486777e-05,
+      "loss": 1.2639,
+      "step": 3973
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.101501072194426e-05,
+      "loss": 1.2752,
+      "step": 3974
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.103288062902073e-05,
+      "loss": 1.2248,
+      "step": 3975
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.105075053609721e-05,
+      "loss": 1.3124,
+      "step": 3976
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.106862044317369e-05,
+      "loss": 1.2639,
+      "step": 3977
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.108649035025018e-05,
+      "loss": 1.2563,
+      "step": 3978
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.110436025732666e-05,
+      "loss": 1.2637,
+      "step": 3979
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.112223016440315e-05,
+      "loss": 1.2293,
+      "step": 3980
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.114010007147963e-05,
+      "loss": 1.2713,
+      "step": 3981
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.115796997855612e-05,
+      "loss": 1.2656,
+      "step": 3982
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.11758398856326e-05,
+      "loss": 1.2418,
+      "step": 3983
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.119370979270909e-05,
+      "loss": 1.3261,
+      "step": 3984
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.121157969978556e-05,
+      "loss": 1.2717,
+      "step": 3985
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.122944960686205e-05,
+      "loss": 1.2894,
+      "step": 3986
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.124731951393853e-05,
+      "loss": 1.2908,
+      "step": 3987
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.126518942101501e-05,
+      "loss": 1.3104,
+      "step": 3988
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.12830593280915e-05,
+      "loss": 1.247,
+      "step": 3989
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.130092923516798e-05,
+      "loss": 1.2528,
+      "step": 3990
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.131879914224447e-05,
+      "loss": 1.2705,
+      "step": 3991
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.133666904932095e-05,
+      "loss": 1.2848,
+      "step": 3992
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.135453895639744e-05,
+      "loss": 1.2555,
+      "step": 3993
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.137240886347391e-05,
+      "loss": 1.2635,
+      "step": 3994
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.13902787705504e-05,
+      "loss": 1.2563,
+      "step": 3995
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.140814867762688e-05,
+      "loss": 1.272,
+      "step": 3996
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.142601858470336e-05,
+      "loss": 1.2757,
+      "step": 3997
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.144388849177984e-05,
+      "loss": 1.2507,
+      "step": 3998
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.146175839885633e-05,
+      "loss": 1.3009,
+      "step": 3999
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.14796283059328e-05,
+      "loss": 1.2603,
+      "step": 4000
+    },
+    {
+      "epoch": 0.64,
+      "eval_loss": 1.3113023042678833,
+      "eval_runtime": 61.4287,
+      "eval_samples_per_second": 162.79,
+      "eval_steps_per_second": 0.228,
+      "step": 4000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.14974982130093e-05,
+      "loss": 1.2962,
+      "step": 4001
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.151536812008577e-05,
+      "loss": 1.2838,
+      "step": 4002
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.153323802716227e-05,
+      "loss": 1.2812,
+      "step": 4003
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.155110793423874e-05,
+      "loss": 1.2457,
+      "step": 4004
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.156897784131523e-05,
+      "loss": 1.2869,
+      "step": 4005
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.158684774839171e-05,
+      "loss": 1.295,
+      "step": 4006
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.160471765546819e-05,
+      "loss": 1.2495,
+      "step": 4007
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.162258756254468e-05,
+      "loss": 1.2658,
+      "step": 4008
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.164045746962116e-05,
+      "loss": 1.2766,
+      "step": 4009
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.165832737669765e-05,
+      "loss": 1.2765,
+      "step": 4010
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.167619728377412e-05,
+      "loss": 1.2714,
+      "step": 4011
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.169406719085062e-05,
+      "loss": 1.2978,
+      "step": 4012
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.17119370979271e-05,
+      "loss": 1.2694,
+      "step": 4013
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.172980700500358e-05,
+      "loss": 1.2834,
+      "step": 4014
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.174767691208006e-05,
+      "loss": 1.2848,
+      "step": 4015
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.176554681915655e-05,
+      "loss": 1.2778,
+      "step": 4016
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.178341672623302e-05,
+      "loss": 1.2186,
+      "step": 4017
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.180128663330951e-05,
+      "loss": 1.2745,
+      "step": 4018
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.181915654038598e-05,
+      "loss": 1.2634,
+      "step": 4019
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.183702644746248e-05,
+      "loss": 1.2617,
+      "step": 4020
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.185489635453895e-05,
+      "loss": 1.245,
+      "step": 4021
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.187276626161544e-05,
+      "loss": 1.2559,
+      "step": 4022
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.189063616869192e-05,
+      "loss": 1.2884,
+      "step": 4023
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.190850607576841e-05,
+      "loss": 1.2783,
+      "step": 4024
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.192637598284489e-05,
+      "loss": 1.295,
+      "step": 4025
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.194424588992138e-05,
+      "loss": 1.2896,
+      "step": 4026
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.196211579699786e-05,
+      "loss": 1.2883,
+      "step": 4027
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.197998570407434e-05,
+      "loss": 1.2763,
+      "step": 4028
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.199785561115083e-05,
+      "loss": 1.2986,
+      "step": 4029
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.20157255182273e-05,
+      "loss": 1.2644,
+      "step": 4030
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.20335954253038e-05,
+      "loss": 1.2593,
+      "step": 4031
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.205146533238027e-05,
+      "loss": 1.2521,
+      "step": 4032
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.206933523945676e-05,
+      "loss": 1.3005,
+      "step": 4033
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.208720514653324e-05,
+      "loss": 1.2817,
+      "step": 4034
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.210507505360973e-05,
+      "loss": 1.2211,
+      "step": 4035
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.212294496068621e-05,
+      "loss": 1.2572,
+      "step": 4036
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.21408148677627e-05,
+      "loss": 1.2504,
+      "step": 4037
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.215868477483916e-05,
+      "loss": 1.3166,
+      "step": 4038
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.217655468191565e-05,
+      "loss": 1.2691,
+      "step": 4039
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.219442458899213e-05,
+      "loss": 1.2314,
+      "step": 4040
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.221229449606862e-05,
+      "loss": 1.3129,
+      "step": 4041
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.22301644031451e-05,
+      "loss": 1.2546,
+      "step": 4042
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.224803431022159e-05,
+      "loss": 1.3096,
+      "step": 4043
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.226590421729807e-05,
+      "loss": 1.2601,
+      "step": 4044
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.228377412437456e-05,
+      "loss": 1.2725,
+      "step": 4045
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.230164403145104e-05,
+      "loss": 1.2526,
+      "step": 4046
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.231951393852753e-05,
+      "loss": 1.2647,
+      "step": 4047
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.2337383845604e-05,
+      "loss": 1.2576,
+      "step": 4048
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.235525375268048e-05,
+      "loss": 1.2768,
+      "step": 4049
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.237312365975697e-05,
+      "loss": 1.2774,
+      "step": 4050
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.239099356683345e-05,
+      "loss": 1.2553,
+      "step": 4051
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.240886347390994e-05,
+      "loss": 1.3043,
+      "step": 4052
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.242673338098642e-05,
+      "loss": 1.2648,
+      "step": 4053
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.244460328806291e-05,
+      "loss": 1.2911,
+      "step": 4054
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.246247319513939e-05,
+      "loss": 1.2757,
+      "step": 4055
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.248034310221588e-05,
+      "loss": 1.2807,
+      "step": 4056
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.249821300929236e-05,
+      "loss": 1.2932,
+      "step": 4057
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.251608291636885e-05,
+      "loss": 1.3075,
+      "step": 4058
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.253395282344531e-05,
+      "loss": 1.2804,
+      "step": 4059
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.25518227305218e-05,
+      "loss": 1.2783,
+      "step": 4060
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.256969263759828e-05,
+      "loss": 1.2597,
+      "step": 4061
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.258756254467477e-05,
+      "loss": 1.2743,
+      "step": 4062
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.260543245175125e-05,
+      "loss": 1.2615,
+      "step": 4063
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.262330235882774e-05,
+      "loss": 1.2836,
+      "step": 4064
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.264117226590422e-05,
+      "loss": 1.2901,
+      "step": 4065
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.265904217298071e-05,
+      "loss": 1.2818,
+      "step": 4066
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.267691208005719e-05,
+      "loss": 1.2513,
+      "step": 4067
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.269478198713368e-05,
+      "loss": 1.3048,
+      "step": 4068
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.271265189421015e-05,
+      "loss": 1.2794,
+      "step": 4069
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.273052180128663e-05,
+      "loss": 1.2612,
+      "step": 4070
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.274839170836312e-05,
+      "loss": 1.2951,
+      "step": 4071
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.27662616154396e-05,
+      "loss": 1.2728,
+      "step": 4072
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.278413152251609e-05,
+      "loss": 1.2473,
+      "step": 4073
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.280200142959257e-05,
+      "loss": 1.2594,
+      "step": 4074
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.281987133666906e-05,
+      "loss": 1.2775,
+      "step": 4075
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.283774124374554e-05,
+      "loss": 1.3256,
+      "step": 4076
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.285561115082203e-05,
+      "loss": 1.2489,
+      "step": 4077
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.28734810578985e-05,
+      "loss": 1.2794,
+      "step": 4078
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.289135096497498e-05,
+      "loss": 1.254,
+      "step": 4079
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.290922087205146e-05,
+      "loss": 1.2939,
+      "step": 4080
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.292709077912795e-05,
+      "loss": 1.2826,
+      "step": 4081
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.294496068620443e-05,
+      "loss": 1.2602,
+      "step": 4082
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.296283059328092e-05,
+      "loss": 1.2479,
+      "step": 4083
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.29807005003574e-05,
+      "loss": 1.2617,
+      "step": 4084
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.299857040743389e-05,
+      "loss": 1.2715,
+      "step": 4085
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.301644031451036e-05,
+      "loss": 1.2663,
+      "step": 4086
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.303431022158686e-05,
+      "loss": 1.2831,
+      "step": 4087
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.305218012866333e-05,
+      "loss": 1.2695,
+      "step": 4088
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.307005003573981e-05,
+      "loss": 1.2659,
+      "step": 4089
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.30879199428163e-05,
+      "loss": 1.2846,
+      "step": 4090
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.310578984989278e-05,
+      "loss": 1.25,
+      "step": 4091
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.312365975696927e-05,
+      "loss": 1.28,
+      "step": 4092
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.314152966404575e-05,
+      "loss": 1.3039,
+      "step": 4093
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.315939957112224e-05,
+      "loss": 1.2871,
+      "step": 4094
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.317726947819872e-05,
+      "loss": 1.2699,
+      "step": 4095
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.31951393852752e-05,
+      "loss": 1.3212,
+      "step": 4096
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.321300929235168e-05,
+      "loss": 1.2613,
+      "step": 4097
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.323087919942817e-05,
+      "loss": 1.2647,
+      "step": 4098
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.324874910650465e-05,
+      "loss": 1.2684,
+      "step": 4099
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.326661901358113e-05,
+      "loss": 1.2647,
+      "step": 4100
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.32844889206576e-05,
+      "loss": 1.2955,
+      "step": 4101
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.33023588277341e-05,
+      "loss": 1.2717,
+      "step": 4102
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.332022873481058e-05,
+      "loss": 1.2904,
+      "step": 4103
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.333809864188707e-05,
+      "loss": 1.306,
+      "step": 4104
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.335596854896354e-05,
+      "loss": 1.2467,
+      "step": 4105
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.337383845604003e-05,
+      "loss": 1.2801,
+      "step": 4106
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.339170836311651e-05,
+      "loss": 1.3106,
+      "step": 4107
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.3409578270193e-05,
+      "loss": 1.2903,
+      "step": 4108
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.342744817726948e-05,
+      "loss": 1.3176,
+      "step": 4109
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.344531808434596e-05,
+      "loss": 1.2855,
+      "step": 4110
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.346318799142245e-05,
+      "loss": 1.2843,
+      "step": 4111
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.348105789849893e-05,
+      "loss": 1.2699,
+      "step": 4112
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.349892780557542e-05,
+      "loss": 1.2663,
+      "step": 4113
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.35167977126519e-05,
+      "loss": 1.2731,
+      "step": 4114
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.353466761972839e-05,
+      "loss": 1.2677,
+      "step": 4115
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.355253752680486e-05,
+      "loss": 1.2796,
+      "step": 4116
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.357040743388135e-05,
+      "loss": 1.2459,
+      "step": 4117
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.358827734095783e-05,
+      "loss": 1.2913,
+      "step": 4118
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.360614724803432e-05,
+      "loss": 1.23,
+      "step": 4119
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.362401715511079e-05,
+      "loss": 1.2699,
+      "step": 4120
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.364188706218728e-05,
+      "loss": 1.2604,
+      "step": 4121
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.365975696926375e-05,
+      "loss": 1.2652,
+      "step": 4122
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.367762687634025e-05,
+      "loss": 1.2871,
+      "step": 4123
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.369549678341672e-05,
+      "loss": 1.3,
+      "step": 4124
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.371336669049321e-05,
+      "loss": 1.2771,
+      "step": 4125
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.373123659756969e-05,
+      "loss": 1.2879,
+      "step": 4126
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.374910650464618e-05,
+      "loss": 1.3209,
+      "step": 4127
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.376697641172266e-05,
+      "loss": 1.3131,
+      "step": 4128
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.378484631879915e-05,
+      "loss": 1.2698,
+      "step": 4129
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.380271622587563e-05,
+      "loss": 1.2563,
+      "step": 4130
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.38205861329521e-05,
+      "loss": 1.3129,
+      "step": 4131
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.38384560400286e-05,
+      "loss": 1.3277,
+      "step": 4132
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.385632594710507e-05,
+      "loss": 1.3083,
+      "step": 4133
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.387419585418156e-05,
+      "loss": 1.2606,
+      "step": 4134
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.389206576125804e-05,
+      "loss": 1.253,
+      "step": 4135
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.390993566833453e-05,
+      "loss": 1.3012,
+      "step": 4136
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.392780557541101e-05,
+      "loss": 1.2704,
+      "step": 4137
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.39456754824875e-05,
+      "loss": 1.2964,
+      "step": 4138
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.396354538956398e-05,
+      "loss": 1.2688,
+      "step": 4139
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.398141529664047e-05,
+      "loss": 1.2396,
+      "step": 4140
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.399928520371693e-05,
+      "loss": 1.281,
+      "step": 4141
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.401715511079342e-05,
+      "loss": 1.2874,
+      "step": 4142
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.40350250178699e-05,
+      "loss": 1.2543,
+      "step": 4143
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.405289492494639e-05,
+      "loss": 1.2636,
+      "step": 4144
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.407076483202287e-05,
+      "loss": 1.2412,
+      "step": 4145
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.408863473909936e-05,
+      "loss": 1.2609,
+      "step": 4146
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.410650464617584e-05,
+      "loss": 1.2664,
+      "step": 4147
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.412437455325233e-05,
+      "loss": 1.2507,
+      "step": 4148
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.414224446032881e-05,
+      "loss": 1.2758,
+      "step": 4149
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.41601143674053e-05,
+      "loss": 1.2643,
+      "step": 4150
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.417798427448178e-05,
+      "loss": 1.2656,
+      "step": 4151
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.419585418155825e-05,
+      "loss": 1.2902,
+      "step": 4152
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.421372408863474e-05,
+      "loss": 1.2783,
+      "step": 4153
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.423159399571122e-05,
+      "loss": 1.2857,
+      "step": 4154
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.424946390278771e-05,
+      "loss": 1.2774,
+      "step": 4155
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.426733380986419e-05,
+      "loss": 1.2688,
+      "step": 4156
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.428520371694068e-05,
+      "loss": 1.2415,
+      "step": 4157
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.430307362401716e-05,
+      "loss": 1.3057,
+      "step": 4158
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.432094353109365e-05,
+      "loss": 1.2995,
+      "step": 4159
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.433881343817013e-05,
+      "loss": 1.3096,
+      "step": 4160
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.43566833452466e-05,
+      "loss": 1.2868,
+      "step": 4161
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.437455325232308e-05,
+      "loss": 1.2466,
+      "step": 4162
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.439242315939957e-05,
+      "loss": 1.2584,
+      "step": 4163
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.441029306647605e-05,
+      "loss": 1.2546,
+      "step": 4164
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.442816297355254e-05,
+      "loss": 1.2581,
+      "step": 4165
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.444603288062902e-05,
+      "loss": 1.2868,
+      "step": 4166
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.446390278770551e-05,
+      "loss": 1.2745,
+      "step": 4167
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.448177269478199e-05,
+      "loss": 1.2669,
+      "step": 4168
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.449964260185848e-05,
+      "loss": 1.2881,
+      "step": 4169
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.451751250893497e-05,
+      "loss": 1.2751,
+      "step": 4170
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.453538241601145e-05,
+      "loss": 1.2589,
+      "step": 4171
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.455325232308792e-05,
+      "loss": 1.2993,
+      "step": 4172
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.45711222301644e-05,
+      "loss": 1.3197,
+      "step": 4173
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.458899213724089e-05,
+      "loss": 1.2916,
+      "step": 4174
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.460686204431737e-05,
+      "loss": 1.2496,
+      "step": 4175
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.462473195139386e-05,
+      "loss": 1.2779,
+      "step": 4176
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.464260185847034e-05,
+      "loss": 1.265,
+      "step": 4177
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.466047176554683e-05,
+      "loss": 1.2749,
+      "step": 4178
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.46783416726233e-05,
+      "loss": 1.2878,
+      "step": 4179
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.46962115796998e-05,
+      "loss": 1.2789,
+      "step": 4180
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.471408148677627e-05,
+      "loss": 1.2718,
+      "step": 4181
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.473195139385275e-05,
+      "loss": 1.2554,
+      "step": 4182
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.474982130092923e-05,
+      "loss": 1.2794,
+      "step": 4183
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.476769120800572e-05,
+      "loss": 1.2555,
+      "step": 4184
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.47855611150822e-05,
+      "loss": 1.3089,
+      "step": 4185
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.480343102215869e-05,
+      "loss": 1.2522,
+      "step": 4186
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.482130092923517e-05,
+      "loss": 1.2866,
+      "step": 4187
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.483917083631166e-05,
+      "loss": 1.2824,
+      "step": 4188
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.485704074338815e-05,
+      "loss": 1.2821,
+      "step": 4189
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.487491065046462e-05,
+      "loss": 1.2697,
+      "step": 4190
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.489278055754112e-05,
+      "loss": 1.2511,
+      "step": 4191
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.491065046461758e-05,
+      "loss": 1.3126,
+      "step": 4192
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.492852037169407e-05,
+      "loss": 1.2684,
+      "step": 4193
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.494639027877055e-05,
+      "loss": 1.2984,
+      "step": 4194
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.496426018584704e-05,
+      "loss": 1.2804,
+      "step": 4195
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.498213009292352e-05,
+      "loss": 1.2358,
+      "step": 4196
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.500000000000001e-05,
+      "loss": 1.2804,
+      "step": 4197
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.501786990707648e-05,
+      "loss": 1.2832,
+      "step": 4198
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.503573981415298e-05,
+      "loss": 1.2884,
+      "step": 4199
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.505360972122945e-05,
+      "loss": 1.2965,
+      "step": 4200
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.507147962830594e-05,
+      "loss": 1.253,
+      "step": 4201
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.508934953538242e-05,
+      "loss": 1.3067,
+      "step": 4202
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.51072194424589e-05,
+      "loss": 1.268,
+      "step": 4203
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.512508934953538e-05,
+      "loss": 1.3046,
+      "step": 4204
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.514295925661187e-05,
+      "loss": 1.2948,
+      "step": 4205
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.516082916368834e-05,
+      "loss": 1.2931,
+      "step": 4206
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.517869907076484e-05,
+      "loss": 1.2389,
+      "step": 4207
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.519656897784133e-05,
+      "loss": 1.2504,
+      "step": 4208
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.52144388849178e-05,
+      "loss": 1.2615,
+      "step": 4209
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.52323087919943e-05,
+      "loss": 1.2744,
+      "step": 4210
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.525017869907077e-05,
+      "loss": 1.2905,
+      "step": 4211
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.526804860614726e-05,
+      "loss": 1.2619,
+      "step": 4212
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.528591851322373e-05,
+      "loss": 1.2914,
+      "step": 4213
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.530378842030022e-05,
+      "loss": 1.2707,
+      "step": 4214
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.53216583273767e-05,
+      "loss": 1.2429,
+      "step": 4215
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.533952823445319e-05,
+      "loss": 1.2887,
+      "step": 4216
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.535739814152966e-05,
+      "loss": 1.2702,
+      "step": 4217
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.537526804860615e-05,
+      "loss": 1.2479,
+      "step": 4218
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.539313795568263e-05,
+      "loss": 1.275,
+      "step": 4219
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.541100786275912e-05,
+      "loss": 1.2779,
+      "step": 4220
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.54288777698356e-05,
+      "loss": 1.2858,
+      "step": 4221
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.544674767691209e-05,
+      "loss": 1.2664,
+      "step": 4222
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.546461758398856e-05,
+      "loss": 1.2578,
+      "step": 4223
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.548248749106505e-05,
+      "loss": 1.2551,
+      "step": 4224
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.550035739814152e-05,
+      "loss": 1.2757,
+      "step": 4225
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.551822730521801e-05,
+      "loss": 1.3029,
+      "step": 4226
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.55360972122945e-05,
+      "loss": 1.2541,
+      "step": 4227
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.555396711937098e-05,
+      "loss": 1.2476,
+      "step": 4228
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.557183702644747e-05,
+      "loss": 1.2781,
+      "step": 4229
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.558970693352395e-05,
+      "loss": 1.2495,
+      "step": 4230
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.560757684060044e-05,
+      "loss": 1.2617,
+      "step": 4231
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.562544674767692e-05,
+      "loss": 1.2873,
+      "step": 4232
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.564331665475341e-05,
+      "loss": 1.2611,
+      "step": 4233
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.566118656182987e-05,
+      "loss": 1.2639,
+      "step": 4234
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.567905646890637e-05,
+      "loss": 1.306,
+      "step": 4235
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.569692637598284e-05,
+      "loss": 1.278,
+      "step": 4236
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.571479628305933e-05,
+      "loss": 1.2526,
+      "step": 4237
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.573266619013581e-05,
+      "loss": 1.3071,
+      "step": 4238
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.57505360972123e-05,
+      "loss": 1.2829,
+      "step": 4239
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.576840600428878e-05,
+      "loss": 1.2802,
+      "step": 4240
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.578627591136527e-05,
+      "loss": 1.2734,
+      "step": 4241
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.580414581844175e-05,
+      "loss": 1.2772,
+      "step": 4242
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.582201572551824e-05,
+      "loss": 1.3001,
+      "step": 4243
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.58398856325947e-05,
+      "loss": 1.2728,
+      "step": 4244
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.58577555396712e-05,
+      "loss": 1.2675,
+      "step": 4245
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.587562544674768e-05,
+      "loss": 1.2904,
+      "step": 4246
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.589349535382416e-05,
+      "loss": 1.2584,
+      "step": 4247
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.591136526090065e-05,
+      "loss": 1.2561,
+      "step": 4248
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.592923516797713e-05,
+      "loss": 1.2974,
+      "step": 4249
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.594710507505362e-05,
+      "loss": 1.2767,
+      "step": 4250
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.59649749821301e-05,
+      "loss": 1.2718,
+      "step": 4251
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.598284488920659e-05,
+      "loss": 1.2518,
+      "step": 4252
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.600071479628307e-05,
+      "loss": 1.263,
+      "step": 4253
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.601858470335954e-05,
+      "loss": 1.2464,
+      "step": 4254
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.603645461043602e-05,
+      "loss": 1.2408,
+      "step": 4255
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.605432451751251e-05,
+      "loss": 1.2611,
+      "step": 4256
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.607219442458899e-05,
+      "loss": 1.283,
+      "step": 4257
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.609006433166548e-05,
+      "loss": 1.2241,
+      "step": 4258
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.610793423874196e-05,
+      "loss": 1.2802,
+      "step": 4259
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.612580414581845e-05,
+      "loss": 1.2725,
+      "step": 4260
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.614367405289493e-05,
+      "loss": 1.2743,
+      "step": 4261
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.616154395997142e-05,
+      "loss": 1.2602,
+      "step": 4262
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.61794138670479e-05,
+      "loss": 1.2901,
+      "step": 4263
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.619728377412437e-05,
+      "loss": 1.2632,
+      "step": 4264
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.621515368120086e-05,
+      "loss": 1.2835,
+      "step": 4265
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.623302358827734e-05,
+      "loss": 1.3159,
+      "step": 4266
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.625089349535383e-05,
+      "loss": 1.2656,
+      "step": 4267
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.626876340243031e-05,
+      "loss": 1.2861,
+      "step": 4268
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.62866333095068e-05,
+      "loss": 1.2601,
+      "step": 4269
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.630450321658328e-05,
+      "loss": 1.2418,
+      "step": 4270
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.632237312365977e-05,
+      "loss": 1.278,
+      "step": 4271
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.634024303073625e-05,
+      "loss": 1.2682,
+      "step": 4272
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.635811293781274e-05,
+      "loss": 1.2691,
+      "step": 4273
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.637598284488921e-05,
+      "loss": 1.27,
+      "step": 4274
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.639385275196569e-05,
+      "loss": 1.294,
+      "step": 4275
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.641172265904217e-05,
+      "loss": 1.2673,
+      "step": 4276
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.642959256611866e-05,
+      "loss": 1.2615,
+      "step": 4277
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.644746247319514e-05,
+      "loss": 1.2916,
+      "step": 4278
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.646533238027163e-05,
+      "loss": 1.2887,
+      "step": 4279
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.64832022873481e-05,
+      "loss": 1.2879,
+      "step": 4280
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.65010721944246e-05,
+      "loss": 1.2696,
+      "step": 4281
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.651894210150107e-05,
+      "loss": 1.2642,
+      "step": 4282
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.653681200857757e-05,
+      "loss": 1.2756,
+      "step": 4283
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.655468191565404e-05,
+      "loss": 1.2888,
+      "step": 4284
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.657255182273052e-05,
+      "loss": 1.2887,
+      "step": 4285
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.659042172980701e-05,
+      "loss": 1.2717,
+      "step": 4286
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.660829163688349e-05,
+      "loss": 1.2504,
+      "step": 4287
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.662616154395998e-05,
+      "loss": 1.2993,
+      "step": 4288
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.664403145103646e-05,
+      "loss": 1.2823,
+      "step": 4289
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.666190135811295e-05,
+      "loss": 1.2936,
+      "step": 4290
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.667977126518943e-05,
+      "loss": 1.2822,
+      "step": 4291
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.669764117226592e-05,
+      "loss": 1.3069,
+      "step": 4292
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.67155110793424e-05,
+      "loss": 1.2729,
+      "step": 4293
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.673338098641888e-05,
+      "loss": 1.2797,
+      "step": 4294
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.675125089349535e-05,
+      "loss": 1.2922,
+      "step": 4295
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.676912080057184e-05,
+      "loss": 1.278,
+      "step": 4296
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.678699070764832e-05,
+      "loss": 1.3001,
+      "step": 4297
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.680486061472481e-05,
+      "loss": 1.2499,
+      "step": 4298
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.682273052180129e-05,
+      "loss": 1.2835,
+      "step": 4299
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.684060042887778e-05,
+      "loss": 1.284,
+      "step": 4300
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.685847033595425e-05,
+      "loss": 1.3,
+      "step": 4301
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.687634024303074e-05,
+      "loss": 1.2803,
+      "step": 4302
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.689421015010722e-05,
+      "loss": 1.2928,
+      "step": 4303
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.691208005718371e-05,
+      "loss": 1.3193,
+      "step": 4304
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.692994996426019e-05,
+      "loss": 1.2882,
+      "step": 4305
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.694781987133667e-05,
+      "loss": 1.2913,
+      "step": 4306
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.696568977841316e-05,
+      "loss": 1.2874,
+      "step": 4307
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.698355968548964e-05,
+      "loss": 1.3069,
+      "step": 4308
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.700142959256613e-05,
+      "loss": 1.2705,
+      "step": 4309
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.70192994996426e-05,
+      "loss": 1.3256,
+      "step": 4310
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.70371694067191e-05,
+      "loss": 1.3221,
+      "step": 4311
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.705503931379557e-05,
+      "loss": 1.2948,
+      "step": 4312
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.707290922087206e-05,
+      "loss": 1.3086,
+      "step": 4313
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.709077912794854e-05,
+      "loss": 1.301,
+      "step": 4314
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.710864903502503e-05,
+      "loss": 1.2828,
+      "step": 4315
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.71265189421015e-05,
+      "loss": 1.2693,
+      "step": 4316
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.714438884917799e-05,
+      "loss": 1.3108,
+      "step": 4317
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.716225875625446e-05,
+      "loss": 1.2965,
+      "step": 4318
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.718012866333096e-05,
+      "loss": 1.2694,
+      "step": 4319
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.719799857040743e-05,
+      "loss": 1.309,
+      "step": 4320
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.721586847748392e-05,
+      "loss": 1.2653,
+      "step": 4321
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.72337383845604e-05,
+      "loss": 1.3223,
+      "step": 4322
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.725160829163689e-05,
+      "loss": 1.2543,
+      "step": 4323
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.726947819871337e-05,
+      "loss": 1.2676,
+      "step": 4324
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.728734810578986e-05,
+      "loss": 1.2857,
+      "step": 4325
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.730521801286634e-05,
+      "loss": 1.2722,
+      "step": 4326
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.732308791994282e-05,
+      "loss": 1.3129,
+      "step": 4327
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.73409578270193e-05,
+      "loss": 1.32,
+      "step": 4328
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.735882773409578e-05,
+      "loss": 1.2747,
+      "step": 4329
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.737669764117227e-05,
+      "loss": 1.2912,
+      "step": 4330
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.739456754824875e-05,
+      "loss": 1.2728,
+      "step": 4331
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.741243745532524e-05,
+      "loss": 1.2856,
+      "step": 4332
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.743030736240172e-05,
+      "loss": 1.2485,
+      "step": 4333
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.744817726947821e-05,
+      "loss": 1.32,
+      "step": 4334
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.746604717655469e-05,
+      "loss": 1.3208,
+      "step": 4335
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.748391708363117e-05,
+      "loss": 1.2513,
+      "step": 4336
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.750178699070764e-05,
+      "loss": 1.3198,
+      "step": 4337
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.751965689778413e-05,
+      "loss": 1.2638,
+      "step": 4338
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.753752680486061e-05,
+      "loss": 1.2861,
+      "step": 4339
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.75553967119371e-05,
+      "loss": 1.2476,
+      "step": 4340
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.757326661901358e-05,
+      "loss": 1.3045,
+      "step": 4341
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.759113652609007e-05,
+      "loss": 1.264,
+      "step": 4342
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.760900643316655e-05,
+      "loss": 1.3102,
+      "step": 4343
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.762687634024304e-05,
+      "loss": 1.2926,
+      "step": 4344
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.764474624731952e-05,
+      "loss": 1.3099,
+      "step": 4345
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.766261615439601e-05,
+      "loss": 1.2872,
+      "step": 4346
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.768048606147249e-05,
+      "loss": 1.3081,
+      "step": 4347
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.769835596854896e-05,
+      "loss": 1.307,
+      "step": 4348
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.771622587562545e-05,
+      "loss": 1.2883,
+      "step": 4349
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.773409578270193e-05,
+      "loss": 1.3016,
+      "step": 4350
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.775196568977842e-05,
+      "loss": 1.2737,
+      "step": 4351
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.77698355968549e-05,
+      "loss": 1.2419,
+      "step": 4352
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.778770550393139e-05,
+      "loss": 1.2429,
+      "step": 4353
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.780557541100787e-05,
+      "loss": 1.3379,
+      "step": 4354
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.782344531808436e-05,
+      "loss": 1.2953,
+      "step": 4355
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.784131522516084e-05,
+      "loss": 1.2898,
+      "step": 4356
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.785918513223731e-05,
+      "loss": 1.3095,
+      "step": 4357
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.787705503931379e-05,
+      "loss": 1.2898,
+      "step": 4358
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.789492494639028e-05,
+      "loss": 1.2953,
+      "step": 4359
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.791279485346676e-05,
+      "loss": 1.2958,
+      "step": 4360
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.793066476054325e-05,
+      "loss": 1.3146,
+      "step": 4361
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.794853466761973e-05,
+      "loss": 1.2994,
+      "step": 4362
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.796640457469622e-05,
+      "loss": 1.3144,
+      "step": 4363
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.79842744817727e-05,
+      "loss": 1.2707,
+      "step": 4364
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.800214438884919e-05,
+      "loss": 1.26,
+      "step": 4365
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.802001429592566e-05,
+      "loss": 1.287,
+      "step": 4366
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.803788420300214e-05,
+      "loss": 1.3035,
+      "step": 4367
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.805575411007863e-05,
+      "loss": 1.3074,
+      "step": 4368
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.807362401715511e-05,
+      "loss": 1.2927,
+      "step": 4369
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.80914939242316e-05,
+      "loss": 1.2645,
+      "step": 4370
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.810936383130808e-05,
+      "loss": 1.2937,
+      "step": 4371
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.812723373838457e-05,
+      "loss": 1.271,
+      "step": 4372
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.814510364546105e-05,
+      "loss": 1.2864,
+      "step": 4373
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.816297355253754e-05,
+      "loss": 1.2859,
+      "step": 4374
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.818084345961402e-05,
+      "loss": 1.2554,
+      "step": 4375
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.81987133666905e-05,
+      "loss": 1.3164,
+      "step": 4376
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.821658327376698e-05,
+      "loss": 1.2906,
+      "step": 4377
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.823445318084346e-05,
+      "loss": 1.288,
+      "step": 4378
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.825232308791994e-05,
+      "loss": 1.2759,
+      "step": 4379
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.827019299499643e-05,
+      "loss": 1.3067,
+      "step": 4380
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.828806290207291e-05,
+      "loss": 1.2701,
+      "step": 4381
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.83059328091494e-05,
+      "loss": 1.2394,
+      "step": 4382
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.832380271622588e-05,
+      "loss": 1.252,
+      "step": 4383
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.834167262330237e-05,
+      "loss": 1.2712,
+      "step": 4384
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.835954253037884e-05,
+      "loss": 1.2781,
+      "step": 4385
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.837741243745533e-05,
+      "loss": 1.2785,
+      "step": 4386
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.839528234453181e-05,
+      "loss": 1.2871,
+      "step": 4387
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.841315225160829e-05,
+      "loss": 1.2642,
+      "step": 4388
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.843102215868478e-05,
+      "loss": 1.2499,
+      "step": 4389
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.844889206576126e-05,
+      "loss": 1.2601,
+      "step": 4390
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.846676197283775e-05,
+      "loss": 1.25,
+      "step": 4391
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.848463187991423e-05,
+      "loss": 1.3031,
+      "step": 4392
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.850250178699072e-05,
+      "loss": 1.2469,
+      "step": 4393
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.85203716940672e-05,
+      "loss": 1.3114,
+      "step": 4394
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.853824160114369e-05,
+      "loss": 1.3051,
+      "step": 4395
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.855611150822016e-05,
+      "loss": 1.3041,
+      "step": 4396
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.857398141529665e-05,
+      "loss": 1.2823,
+      "step": 4397
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.859185132237312e-05,
+      "loss": 1.3261,
+      "step": 4398
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.860972122944961e-05,
+      "loss": 1.2772,
+      "step": 4399
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.862759113652609e-05,
+      "loss": 1.2843,
+      "step": 4400
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.864546104360258e-05,
+      "loss": 1.2652,
+      "step": 4401
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.866333095067905e-05,
+      "loss": 1.2856,
+      "step": 4402
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.868120085775555e-05,
+      "loss": 1.2641,
+      "step": 4403
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.869907076483202e-05,
+      "loss": 1.2971,
+      "step": 4404
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.871694067190851e-05,
+      "loss": 1.3023,
+      "step": 4405
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.873481057898499e-05,
+      "loss": 1.27,
+      "step": 4406
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.875268048606148e-05,
+      "loss": 1.2977,
+      "step": 4407
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.877055039313796e-05,
+      "loss": 1.2748,
+      "step": 4408
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.878842030021444e-05,
+      "loss": 1.3089,
+      "step": 4409
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.880629020729093e-05,
+      "loss": 1.2596,
+      "step": 4410
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.88241601143674e-05,
+      "loss": 1.2703,
+      "step": 4411
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.88420300214439e-05,
+      "loss": 1.2455,
+      "step": 4412
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.885989992852037e-05,
+      "loss": 1.2393,
+      "step": 4413
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.887776983559686e-05,
+      "loss": 1.2671,
+      "step": 4414
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.889563974267334e-05,
+      "loss": 1.289,
+      "step": 4415
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.891350964974983e-05,
+      "loss": 1.2611,
+      "step": 4416
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.893137955682631e-05,
+      "loss": 1.282,
+      "step": 4417
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.89492494639028e-05,
+      "loss": 1.2874,
+      "step": 4418
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.896711937097927e-05,
+      "loss": 1.2878,
+      "step": 4419
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.898498927805576e-05,
+      "loss": 1.314,
+      "step": 4420
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.900285918513223e-05,
+      "loss": 1.2735,
+      "step": 4421
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.902072909220872e-05,
+      "loss": 1.3086,
+      "step": 4422
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.90385989992852e-05,
+      "loss": 1.2666,
+      "step": 4423
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.905646890636169e-05,
+      "loss": 1.2858,
+      "step": 4424
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.907433881343817e-05,
+      "loss": 1.2667,
+      "step": 4425
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.909220872051466e-05,
+      "loss": 1.2988,
+      "step": 4426
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.911007862759114e-05,
+      "loss": 1.3092,
+      "step": 4427
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.912794853466763e-05,
+      "loss": 1.2951,
+      "step": 4428
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.914581844174411e-05,
+      "loss": 1.2837,
+      "step": 4429
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.916368834882058e-05,
+      "loss": 1.2962,
+      "step": 4430
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.918155825589708e-05,
+      "loss": 1.2632,
+      "step": 4431
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.919942816297355e-05,
+      "loss": 1.2362,
+      "step": 4432
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.921729807005004e-05,
+      "loss": 1.2998,
+      "step": 4433
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.923516797712652e-05,
+      "loss": 1.2846,
+      "step": 4434
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.925303788420301e-05,
+      "loss": 1.2799,
+      "step": 4435
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.927090779127949e-05,
+      "loss": 1.2735,
+      "step": 4436
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.928877769835598e-05,
+      "loss": 1.2348,
+      "step": 4437
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.930664760543246e-05,
+      "loss": 1.258,
+      "step": 4438
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.932451751250894e-05,
+      "loss": 1.3028,
+      "step": 4439
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.934238741958541e-05,
+      "loss": 1.2816,
+      "step": 4440
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.93602573266619e-05,
+      "loss": 1.2918,
+      "step": 4441
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.937812723373838e-05,
+      "loss": 1.3007,
+      "step": 4442
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.939599714081487e-05,
+      "loss": 1.3327,
+      "step": 4443
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.941386704789135e-05,
+      "loss": 1.2564,
+      "step": 4444
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.943173695496784e-05,
+      "loss": 1.2874,
+      "step": 4445
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.944960686204432e-05,
+      "loss": 1.2933,
+      "step": 4446
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.946747676912081e-05,
+      "loss": 1.3089,
+      "step": 4447
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.948534667619729e-05,
+      "loss": 1.2705,
+      "step": 4448
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.950321658327378e-05,
+      "loss": 1.2803,
+      "step": 4449
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.952108649035025e-05,
+      "loss": 1.2847,
+      "step": 4450
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.953895639742673e-05,
+      "loss": 1.2707,
+      "step": 4451
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.955682630450322e-05,
+      "loss": 1.2809,
+      "step": 4452
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.95746962115797e-05,
+      "loss": 1.2733,
+      "step": 4453
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.959256611865619e-05,
+      "loss": 1.2697,
+      "step": 4454
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.961043602573267e-05,
+      "loss": 1.2785,
+      "step": 4455
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.962830593280916e-05,
+      "loss": 1.2797,
+      "step": 4456
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.964617583988564e-05,
+      "loss": 1.2668,
+      "step": 4457
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.966404574696213e-05,
+      "loss": 1.2715,
+      "step": 4458
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.96819156540386e-05,
+      "loss": 1.2598,
+      "step": 4459
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.969978556111508e-05,
+      "loss": 1.2593,
+      "step": 4460
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.971765546819156e-05,
+      "loss": 1.2998,
+      "step": 4461
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.973552537526805e-05,
+      "loss": 1.2527,
+      "step": 4462
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.975339528234453e-05,
+      "loss": 1.303,
+      "step": 4463
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.977126518942102e-05,
+      "loss": 1.2427,
+      "step": 4464
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.97891350964975e-05,
+      "loss": 1.2661,
+      "step": 4465
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.980700500357399e-05,
+      "loss": 1.2677,
+      "step": 4466
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.982487491065047e-05,
+      "loss": 1.2562,
+      "step": 4467
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.984274481772696e-05,
+      "loss": 1.287,
+      "step": 4468
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.986061472480343e-05,
+      "loss": 1.3104,
+      "step": 4469
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.987848463187991e-05,
+      "loss": 1.3304,
+      "step": 4470
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.98963545389564e-05,
+      "loss": 1.2741,
+      "step": 4471
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.991422444603288e-05,
+      "loss": 1.2711,
+      "step": 4472
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.993209435310937e-05,
+      "loss": 1.3167,
+      "step": 4473
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.994996426018585e-05,
+      "loss": 1.3146,
+      "step": 4474
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.996783416726234e-05,
+      "loss": 1.2898,
+      "step": 4475
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.998570407433882e-05,
+      "loss": 1.2897,
+      "step": 4476
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.000357398141531e-05,
+      "loss": 1.2706,
+      "step": 4477
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.002144388849178e-05,
+      "loss": 1.302,
+      "step": 4478
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.003931379556828e-05,
+      "loss": 1.3087,
+      "step": 4479
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.005718370264474e-05,
+      "loss": 1.2541,
+      "step": 4480
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.007505360972123e-05,
+      "loss": 1.2668,
+      "step": 4481
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.009292351679771e-05,
+      "loss": 1.2926,
+      "step": 4482
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.01107934238742e-05,
+      "loss": 1.2646,
+      "step": 4483
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.012866333095068e-05,
+      "loss": 1.2832,
+      "step": 4484
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.014653323802717e-05,
+      "loss": 1.3054,
+      "step": 4485
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.016440314510364e-05,
+      "loss": 1.2457,
+      "step": 4486
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.018227305218014e-05,
+      "loss": 1.2951,
+      "step": 4487
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.020014295925661e-05,
+      "loss": 1.2661,
+      "step": 4488
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.02180128663331e-05,
+      "loss": 1.2959,
+      "step": 4489
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.023588277340958e-05,
+      "loss": 1.3119,
+      "step": 4490
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.025375268048606e-05,
+      "loss": 1.2926,
+      "step": 4491
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.027162258756255e-05,
+      "loss": 1.2942,
+      "step": 4492
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.028949249463903e-05,
+      "loss": 1.307,
+      "step": 4493
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.030736240171552e-05,
+      "loss": 1.2648,
+      "step": 4494
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.0325232308792e-05,
+      "loss": 1.2683,
+      "step": 4495
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.034310221586849e-05,
+      "loss": 1.2324,
+      "step": 4496
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.036097212294496e-05,
+      "loss": 1.2737,
+      "step": 4497
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.037884203002145e-05,
+      "loss": 1.2863,
+      "step": 4498
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.039671193709793e-05,
+      "loss": 1.3107,
+      "step": 4499
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.041458184417442e-05,
+      "loss": 1.3184,
+      "step": 4500
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.043245175125089e-05,
+      "loss": 1.2922,
+      "step": 4501
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.045032165832738e-05,
+      "loss": 1.2666,
+      "step": 4502
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.046819156540386e-05,
+      "loss": 1.2819,
+      "step": 4503
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.048606147248035e-05,
+      "loss": 1.2303,
+      "step": 4504
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.050393137955682e-05,
+      "loss": 1.2681,
+      "step": 4505
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.052180128663331e-05,
+      "loss": 1.3091,
+      "step": 4506
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.053967119370979e-05,
+      "loss": 1.2806,
+      "step": 4507
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.055754110078628e-05,
+      "loss": 1.2794,
+      "step": 4508
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.057541100786276e-05,
+      "loss": 1.2795,
+      "step": 4509
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.059328091493925e-05,
+      "loss": 1.2243,
+      "step": 4510
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.061115082201573e-05,
+      "loss": 1.293,
+      "step": 4511
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.06290207290922e-05,
+      "loss": 1.3137,
+      "step": 4512
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.06468906361687e-05,
+      "loss": 1.2516,
+      "step": 4513
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.066476054324517e-05,
+      "loss": 1.2725,
+      "step": 4514
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.068263045032167e-05,
+      "loss": 1.2995,
+      "step": 4515
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.070050035739814e-05,
+      "loss": 1.3021,
+      "step": 4516
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.071837026447463e-05,
+      "loss": 1.2724,
+      "step": 4517
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.073624017155111e-05,
+      "loss": 1.2899,
+      "step": 4518
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.07541100786276e-05,
+      "loss": 1.3211,
+      "step": 4519
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.077197998570408e-05,
+      "loss": 1.2402,
+      "step": 4520
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.078984989278057e-05,
+      "loss": 1.2466,
+      "step": 4521
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.080771979985703e-05,
+      "loss": 1.2932,
+      "step": 4522
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.082558970693353e-05,
+      "loss": 1.287,
+      "step": 4523
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.084345961401e-05,
+      "loss": 1.2931,
+      "step": 4524
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.08613295210865e-05,
+      "loss": 1.2831,
+      "step": 4525
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.087919942816297e-05,
+      "loss": 1.2596,
+      "step": 4526
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.089706933523946e-05,
+      "loss": 1.2709,
+      "step": 4527
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.091493924231594e-05,
+      "loss": 1.3009,
+      "step": 4528
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.093280914939243e-05,
+      "loss": 1.2838,
+      "step": 4529
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.095067905646891e-05,
+      "loss": 1.2755,
+      "step": 4530
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.09685489635454e-05,
+      "loss": 1.2553,
+      "step": 4531
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.098641887062188e-05,
+      "loss": 1.284,
+      "step": 4532
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.100428877769835e-05,
+      "loss": 1.2485,
+      "step": 4533
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.102215868477484e-05,
+      "loss": 1.325,
+      "step": 4534
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.104002859185132e-05,
+      "loss": 1.2553,
+      "step": 4535
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.105789849892781e-05,
+      "loss": 1.2773,
+      "step": 4536
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.107576840600429e-05,
+      "loss": 1.3032,
+      "step": 4537
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.109363831308078e-05,
+      "loss": 1.3239,
+      "step": 4538
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.111150822015726e-05,
+      "loss": 1.286,
+      "step": 4539
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.112937812723375e-05,
+      "loss": 1.2662,
+      "step": 4540
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.114724803431023e-05,
+      "loss": 1.3147,
+      "step": 4541
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.11651179413867e-05,
+      "loss": 1.3003,
+      "step": 4542
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.118298784846318e-05,
+      "loss": 1.3013,
+      "step": 4543
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.120085775553967e-05,
+      "loss": 1.2653,
+      "step": 4544
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.121872766261615e-05,
+      "loss": 1.3036,
+      "step": 4545
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.123659756969264e-05,
+      "loss": 1.3203,
+      "step": 4546
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.125446747676912e-05,
+      "loss": 1.2817,
+      "step": 4547
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.127233738384561e-05,
+      "loss": 1.2976,
+      "step": 4548
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.129020729092209e-05,
+      "loss": 1.3018,
+      "step": 4549
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.130807719799858e-05,
+      "loss": 1.2831,
+      "step": 4550
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.132594710507506e-05,
+      "loss": 1.2879,
+      "step": 4551
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.134381701215153e-05,
+      "loss": 1.267,
+      "step": 4552
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.136168691922802e-05,
+      "loss": 1.2808,
+      "step": 4553
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.13795568263045e-05,
+      "loss": 1.284,
+      "step": 4554
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.139742673338099e-05,
+      "loss": 1.2836,
+      "step": 4555
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.141529664045747e-05,
+      "loss": 1.2554,
+      "step": 4556
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.143316654753396e-05,
+      "loss": 1.2774,
+      "step": 4557
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.145103645461044e-05,
+      "loss": 1.3088,
+      "step": 4558
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.146890636168693e-05,
+      "loss": 1.3196,
+      "step": 4559
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.14867762687634e-05,
+      "loss": 1.3017,
+      "step": 4560
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.15046461758399e-05,
+      "loss": 1.3389,
+      "step": 4561
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.152251608291637e-05,
+      "loss": 1.2709,
+      "step": 4562
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.154038598999285e-05,
+      "loss": 1.2808,
+      "step": 4563
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.155825589706933e-05,
+      "loss": 1.2703,
+      "step": 4564
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.157612580414582e-05,
+      "loss": 1.2946,
+      "step": 4565
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.15939957112223e-05,
+      "loss": 1.3127,
+      "step": 4566
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.161186561829879e-05,
+      "loss": 1.3127,
+      "step": 4567
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.162973552537527e-05,
+      "loss": 1.2895,
+      "step": 4568
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.164760543245176e-05,
+      "loss": 1.2719,
+      "step": 4569
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.166547533952823e-05,
+      "loss": 1.2798,
+      "step": 4570
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.168334524660473e-05,
+      "loss": 1.2999,
+      "step": 4571
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.17012151536812e-05,
+      "loss": 1.2893,
+      "step": 4572
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.171908506075768e-05,
+      "loss": 1.3082,
+      "step": 4573
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.173695496783417e-05,
+      "loss": 1.2818,
+      "step": 4574
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.175482487491065e-05,
+      "loss": 1.2642,
+      "step": 4575
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.177269478198714e-05,
+      "loss": 1.281,
+      "step": 4576
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.179056468906362e-05,
+      "loss": 1.2694,
+      "step": 4577
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.180843459614011e-05,
+      "loss": 1.3108,
+      "step": 4578
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.182630450321659e-05,
+      "loss": 1.296,
+      "step": 4579
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.184417441029308e-05,
+      "loss": 1.2996,
+      "step": 4580
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.186204431736955e-05,
+      "loss": 1.3188,
+      "step": 4581
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.187991422444605e-05,
+      "loss": 1.2901,
+      "step": 4582
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.189778413152251e-05,
+      "loss": 1.2816,
+      "step": 4583
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.1915654038599e-05,
+      "loss": 1.2665,
+      "step": 4584
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.193352394567548e-05,
+      "loss": 1.2441,
+      "step": 4585
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.195139385275197e-05,
+      "loss": 1.2938,
+      "step": 4586
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.196926375982845e-05,
+      "loss": 1.266,
+      "step": 4587
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.198713366690494e-05,
+      "loss": 1.29,
+      "step": 4588
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.200500357398141e-05,
+      "loss": 1.2521,
+      "step": 4589
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.20228734810579e-05,
+      "loss": 1.2703,
+      "step": 4590
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.204074338813438e-05,
+      "loss": 1.2565,
+      "step": 4591
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.205861329521087e-05,
+      "loss": 1.2515,
+      "step": 4592
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.207648320228735e-05,
+      "loss": 1.2983,
+      "step": 4593
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.209435310936383e-05,
+      "loss": 1.2529,
+      "step": 4594
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.211222301644032e-05,
+      "loss": 1.2647,
+      "step": 4595
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.21300929235168e-05,
+      "loss": 1.2694,
+      "step": 4596
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.214796283059329e-05,
+      "loss": 1.2628,
+      "step": 4597
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.216583273766976e-05,
+      "loss": 1.2716,
+      "step": 4598
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.218370264474626e-05,
+      "loss": 1.2852,
+      "step": 4599
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.220157255182273e-05,
+      "loss": 1.2856,
+      "step": 4600
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.221944245889922e-05,
+      "loss": 1.2387,
+      "step": 4601
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.22373123659757e-05,
+      "loss": 1.2255,
+      "step": 4602
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.225518227305219e-05,
+      "loss": 1.3137,
+      "step": 4603
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.227305218012866e-05,
+      "loss": 1.2975,
+      "step": 4604
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.229092208720515e-05,
+      "loss": 1.2813,
+      "step": 4605
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.230879199428162e-05,
+      "loss": 1.3069,
+      "step": 4606
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.232666190135812e-05,
+      "loss": 1.2837,
+      "step": 4607
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.23445318084346e-05,
+      "loss": 1.2921,
+      "step": 4608
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.236240171551108e-05,
+      "loss": 1.2908,
+      "step": 4609
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.238027162258756e-05,
+      "loss": 1.2821,
+      "step": 4610
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.239814152966405e-05,
+      "loss": 1.2717,
+      "step": 4611
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.241601143674053e-05,
+      "loss": 1.2997,
+      "step": 4612
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.243388134381702e-05,
+      "loss": 1.2751,
+      "step": 4613
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.24517512508935e-05,
+      "loss": 1.2683,
+      "step": 4614
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.246962115796998e-05,
+      "loss": 1.2622,
+      "step": 4615
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.248749106504647e-05,
+      "loss": 1.2722,
+      "step": 4616
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.250536097212294e-05,
+      "loss": 1.301,
+      "step": 4617
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.252323087919944e-05,
+      "loss": 1.2888,
+      "step": 4618
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.254110078627591e-05,
+      "loss": 1.2409,
+      "step": 4619
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.25589706933524e-05,
+      "loss": 1.2797,
+      "step": 4620
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.257684060042888e-05,
+      "loss": 1.2686,
+      "step": 4621
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.259471050750537e-05,
+      "loss": 1.3041,
+      "step": 4622
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.261258041458185e-05,
+      "loss": 1.2996,
+      "step": 4623
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.263045032165833e-05,
+      "loss": 1.2793,
+      "step": 4624
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.26483202287348e-05,
+      "loss": 1.2743,
+      "step": 4625
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.26661901358113e-05,
+      "loss": 1.2804,
+      "step": 4626
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.268406004288777e-05,
+      "loss": 1.2874,
+      "step": 4627
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.270192994996426e-05,
+      "loss": 1.2822,
+      "step": 4628
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.271979985704074e-05,
+      "loss": 1.258,
+      "step": 4629
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.273766976411723e-05,
+      "loss": 1.2712,
+      "step": 4630
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.275553967119371e-05,
+      "loss": 1.3093,
+      "step": 4631
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.27734095782702e-05,
+      "loss": 1.2608,
+      "step": 4632
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.279127948534669e-05,
+      "loss": 1.2645,
+      "step": 4633
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.280914939242317e-05,
+      "loss": 1.2871,
+      "step": 4634
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.282701929949965e-05,
+      "loss": 1.2924,
+      "step": 4635
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.284488920657612e-05,
+      "loss": 1.2873,
+      "step": 4636
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.286275911365261e-05,
+      "loss": 1.2612,
+      "step": 4637
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.288062902072909e-05,
+      "loss": 1.3145,
+      "step": 4638
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.289849892780558e-05,
+      "loss": 1.2872,
+      "step": 4639
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.291636883488206e-05,
+      "loss": 1.2762,
+      "step": 4640
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.293423874195855e-05,
+      "loss": 1.2918,
+      "step": 4641
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.295210864903503e-05,
+      "loss": 1.2738,
+      "step": 4642
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.296997855611152e-05,
+      "loss": 1.2972,
+      "step": 4643
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.2987848463188e-05,
+      "loss": 1.2875,
+      "step": 4644
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.300571837026447e-05,
+      "loss": 1.3066,
+      "step": 4645
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.302358827734095e-05,
+      "loss": 1.2982,
+      "step": 4646
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.304145818441744e-05,
+      "loss": 1.2888,
+      "step": 4647
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.305932809149392e-05,
+      "loss": 1.27,
+      "step": 4648
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.307719799857041e-05,
+      "loss": 1.2845,
+      "step": 4649
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.309506790564689e-05,
+      "loss": 1.2857,
+      "step": 4650
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.311293781272338e-05,
+      "loss": 1.2622,
+      "step": 4651
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.313080771979987e-05,
+      "loss": 1.2465,
+      "step": 4652
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.314867762687635e-05,
+      "loss": 1.2501,
+      "step": 4653
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.316654753395284e-05,
+      "loss": 1.3025,
+      "step": 4654
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.31844174410293e-05,
+      "loss": 1.2761,
+      "step": 4655
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.32022873481058e-05,
+      "loss": 1.2703,
+      "step": 4656
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.322015725518227e-05,
+      "loss": 1.2768,
+      "step": 4657
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.323802716225876e-05,
+      "loss": 1.2924,
+      "step": 4658
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.325589706933524e-05,
+      "loss": 1.3003,
+      "step": 4659
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.327376697641173e-05,
+      "loss": 1.2744,
+      "step": 4660
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.329163688348821e-05,
+      "loss": 1.337,
+      "step": 4661
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.33095067905647e-05,
+      "loss": 1.2856,
+      "step": 4662
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.332737669764118e-05,
+      "loss": 1.2731,
+      "step": 4663
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.334524660471767e-05,
+      "loss": 1.2768,
+      "step": 4664
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.336311651179414e-05,
+      "loss": 1.2944,
+      "step": 4665
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.338098641887062e-05,
+      "loss": 1.2591,
+      "step": 4666
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.33988563259471e-05,
+      "loss": 1.2963,
+      "step": 4667
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.341672623302359e-05,
+      "loss": 1.2809,
+      "step": 4668
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.343459614010007e-05,
+      "loss": 1.2827,
+      "step": 4669
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.345246604717656e-05,
+      "loss": 1.2852,
+      "step": 4670
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.347033595425305e-05,
+      "loss": 1.3191,
+      "step": 4671
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.348820586132953e-05,
+      "loss": 1.2695,
+      "step": 4672
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.350607576840602e-05,
+      "loss": 1.3357,
+      "step": 4673
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.35239456754825e-05,
+      "loss": 1.2616,
+      "step": 4674
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.354181558255899e-05,
+      "loss": 1.2796,
+      "step": 4675
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.355968548963545e-05,
+      "loss": 1.3061,
+      "step": 4676
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.357755539671194e-05,
+      "loss": 1.2881,
+      "step": 4677
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.359542530378842e-05,
+      "loss": 1.3163,
+      "step": 4678
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.361329521086491e-05,
+      "loss": 1.2642,
+      "step": 4679
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.363116511794139e-05,
+      "loss": 1.2827,
+      "step": 4680
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.364903502501788e-05,
+      "loss": 1.262,
+      "step": 4681
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.366690493209436e-05,
+      "loss": 1.2742,
+      "step": 4682
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.368477483917085e-05,
+      "loss": 1.2815,
+      "step": 4683
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.370264474624732e-05,
+      "loss": 1.3031,
+      "step": 4684
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.372051465332381e-05,
+      "loss": 1.2945,
+      "step": 4685
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.373838456040028e-05,
+      "loss": 1.2626,
+      "step": 4686
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.375625446747677e-05,
+      "loss": 1.3005,
+      "step": 4687
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.377412437455325e-05,
+      "loss": 1.2849,
+      "step": 4688
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.379199428162974e-05,
+      "loss": 1.2679,
+      "step": 4689
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.380986418870623e-05,
+      "loss": 1.2685,
+      "step": 4690
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.38277340957827e-05,
+      "loss": 1.2712,
+      "step": 4691
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.38456040028592e-05,
+      "loss": 1.3009,
+      "step": 4692
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.386347390993567e-05,
+      "loss": 1.2925,
+      "step": 4693
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.388134381701217e-05,
+      "loss": 1.2437,
+      "step": 4694
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.389921372408864e-05,
+      "loss": 1.2696,
+      "step": 4695
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.391708363116512e-05,
+      "loss": 1.3117,
+      "step": 4696
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.39349535382416e-05,
+      "loss": 1.272,
+      "step": 4697
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.395282344531809e-05,
+      "loss": 1.2818,
+      "step": 4698
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.397069335239457e-05,
+      "loss": 1.256,
+      "step": 4699
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.398856325947106e-05,
+      "loss": 1.2514,
+      "step": 4700
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.400643316654753e-05,
+      "loss": 1.2968,
+      "step": 4701
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.402430307362403e-05,
+      "loss": 1.279,
+      "step": 4702
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.40421729807005e-05,
+      "loss": 1.2676,
+      "step": 4703
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.4060042887777e-05,
+      "loss": 1.262,
+      "step": 4704
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.407791279485347e-05,
+      "loss": 1.2843,
+      "step": 4705
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.409578270192996e-05,
+      "loss": 1.2985,
+      "step": 4706
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.411365260900643e-05,
+      "loss": 1.2721,
+      "step": 4707
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.413152251608292e-05,
+      "loss": 1.2868,
+      "step": 4708
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.414939242315941e-05,
+      "loss": 1.2794,
+      "step": 4709
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.416726233023589e-05,
+      "loss": 1.2853,
+      "step": 4710
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.418513223731238e-05,
+      "loss": 1.277,
+      "step": 4711
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.420300214438885e-05,
+      "loss": 1.2963,
+      "step": 4712
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.422087205146534e-05,
+      "loss": 1.2779,
+      "step": 4713
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.423874195854182e-05,
+      "loss": 1.297,
+      "step": 4714
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.425661186561831e-05,
+      "loss": 1.2909,
+      "step": 4715
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.427448177269479e-05,
+      "loss": 1.2582,
+      "step": 4716
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.429235167977127e-05,
+      "loss": 1.285,
+      "step": 4717
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.431022158684775e-05,
+      "loss": 1.2927,
+      "step": 4718
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.432809149392424e-05,
+      "loss": 1.2881,
+      "step": 4719
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.434596140100071e-05,
+      "loss": 1.2642,
+      "step": 4720
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.43638313080772e-05,
+      "loss": 1.2555,
+      "step": 4721
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.438170121515368e-05,
+      "loss": 1.2841,
+      "step": 4722
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.439957112223017e-05,
+      "loss": 1.2834,
+      "step": 4723
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.441744102930665e-05,
+      "loss": 1.2676,
+      "step": 4724
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.443531093638314e-05,
+      "loss": 1.2634,
+      "step": 4725
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.445318084345962e-05,
+      "loss": 1.2971,
+      "step": 4726
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.44710507505361e-05,
+      "loss": 1.2429,
+      "step": 4727
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.448892065761259e-05,
+      "loss": 1.2471,
+      "step": 4728
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.450679056468906e-05,
+      "loss": 1.2809,
+      "step": 4729
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.452466047176556e-05,
+      "loss": 1.2727,
+      "step": 4730
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.454253037884203e-05,
+      "loss": 1.2951,
+      "step": 4731
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.456040028591852e-05,
+      "loss": 1.2379,
+      "step": 4732
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.4578270192995e-05,
+      "loss": 1.3029,
+      "step": 4733
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.459614010007149e-05,
+      "loss": 1.2916,
+      "step": 4734
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.461401000714797e-05,
+      "loss": 1.2626,
+      "step": 4735
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.463187991422446e-05,
+      "loss": 1.2558,
+      "step": 4736
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.464974982130094e-05,
+      "loss": 1.3092,
+      "step": 4737
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.466761972837742e-05,
+      "loss": 1.3106,
+      "step": 4738
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.468548963545389e-05,
+      "loss": 1.2824,
+      "step": 4739
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.470335954253038e-05,
+      "loss": 1.3065,
+      "step": 4740
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.472122944960686e-05,
+      "loss": 1.2915,
+      "step": 4741
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.473909935668335e-05,
+      "loss": 1.2912,
+      "step": 4742
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.475696926375983e-05,
+      "loss": 1.3363,
+      "step": 4743
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.477483917083632e-05,
+      "loss": 1.3013,
+      "step": 4744
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.47927090779128e-05,
+      "loss": 1.2875,
+      "step": 4745
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.481057898498929e-05,
+      "loss": 1.2796,
+      "step": 4746
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.482844889206577e-05,
+      "loss": 1.2463,
+      "step": 4747
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.484631879914224e-05,
+      "loss": 1.3169,
+      "step": 4748
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.486418870621873e-05,
+      "loss": 1.2889,
+      "step": 4749
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.488205861329521e-05,
+      "loss": 1.2759,
+      "step": 4750
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.48999285203717e-05,
+      "loss": 1.2832,
+      "step": 4751
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.491779842744818e-05,
+      "loss": 1.2975,
+      "step": 4752
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.493566833452467e-05,
+      "loss": 1.3112,
+      "step": 4753
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.495353824160115e-05,
+      "loss": 1.2995,
+      "step": 4754
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.497140814867764e-05,
+      "loss": 1.2791,
+      "step": 4755
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.498927805575412e-05,
+      "loss": 1.2572,
+      "step": 4756
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 8.500714796283061e-05,
+      "loss": 1.2838,
+      "step": 4757
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 8.502501786990707e-05,
+      "loss": 1.2446,
+      "step": 4758
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 8.504288777698356e-05,
+      "loss": 1.2495,
+      "step": 4759
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 8.506075768406004e-05,
+      "loss": 1.3195,
+      "step": 4760
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 8.507862759113653e-05,
+      "loss": 1.2805,
+      "step": 4761
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 8.509649749821301e-05,
+      "loss": 1.2824,
+      "step": 4762
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 8.51143674052895e-05,
+      "loss": 1.2768,
+      "step": 4763
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 8.513223731236598e-05,
+      "loss": 1.2778,
+      "step": 4764
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 8.515010721944247e-05,
+      "loss": 1.2802,
+      "step": 4765
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 8.516797712651895e-05,
+      "loss": 1.2935,
+      "step": 4766
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 8.518584703359544e-05,
+      "loss": 1.3305,
+      "step": 4767
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 8.520371694067191e-05,
+      "loss": 1.3081,
+      "step": 4768
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 8.522158684774839e-05,
+      "loss": 1.2759,
+      "step": 4769
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 8.523945675482488e-05,
+      "loss": 1.2677,
+      "step": 4770
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 8.525732666190136e-05,
+      "loss": 1.3006,
+      "step": 4771
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 8.527519656897785e-05,
+      "loss": 1.2932,
+      "step": 4772
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 8.529306647605433e-05,
+      "loss": 1.2887,
+      "step": 4773
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 8.531093638313082e-05,
+      "loss": 1.2682,
+      "step": 4774
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 8.53288062902073e-05,
+      "loss": 1.2716,
+      "step": 4775
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 8.534667619728379e-05,
+      "loss": 1.2916,
+      "step": 4776
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 8.536454610436026e-05,
+      "loss": 1.2949,
+      "step": 4777
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 8.538241601143676e-05,
+      "loss": 1.3054,
+      "step": 4778
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 8.540028591851322e-05,
+      "loss": 1.2921,
+      "step": 4779
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 8.541815582558971e-05,
+      "loss": 1.3078,
+      "step": 4780
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 8.543602573266619e-05,
+      "loss": 1.2663,
+      "step": 4781
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 8.545389563974268e-05,
+      "loss": 1.3249,
+      "step": 4782
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 8.547176554681916e-05,
+      "loss": 1.2981,
+      "step": 4783
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 8.548963545389565e-05,
+      "loss": 1.2769,
+      "step": 4784
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 8.550750536097212e-05,
+      "loss": 1.309,
+      "step": 4785
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 8.552537526804862e-05,
+      "loss": 1.2889,
+      "step": 4786
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 8.554324517512509e-05,
+      "loss": 1.2727,
+      "step": 4787
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 8.556111508220158e-05,
+      "loss": 1.2847,
+      "step": 4788
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 8.557898498927806e-05,
+      "loss": 1.2745,
+      "step": 4789
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 8.559685489635454e-05,
+      "loss": 1.2957,
+      "step": 4790
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 8.561472480343103e-05,
+      "loss": 1.2738,
+      "step": 4791
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 8.563259471050751e-05,
+      "loss": 1.2604,
+      "step": 4792
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 8.5650464617584e-05,
+      "loss": 1.2791,
+      "step": 4793
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 8.566833452466048e-05,
+      "loss": 1.2435,
+      "step": 4794
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 8.568620443173697e-05,
+      "loss": 1.2813,
+      "step": 4795
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 8.570407433881344e-05,
+      "loss": 1.3077,
+      "step": 4796
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 8.572194424588993e-05,
+      "loss": 1.2646,
+      "step": 4797
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 8.573981415296641e-05,
+      "loss": 1.2644,
+      "step": 4798
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 8.575768406004289e-05,
+      "loss": 1.2882,
+      "step": 4799
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 8.577555396711937e-05,
+      "loss": 1.2981,
+      "step": 4800
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 8.579342387419586e-05,
+      "loss": 1.2677,
+      "step": 4801
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 8.581129378127234e-05,
+      "loss": 1.2996,
+      "step": 4802
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 8.582916368834883e-05,
+      "loss": 1.293,
+      "step": 4803
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 8.58470335954253e-05,
+      "loss": 1.2737,
+      "step": 4804
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 8.58649035025018e-05,
+      "loss": 1.3044,
+      "step": 4805
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 8.588277340957827e-05,
+      "loss": 1.2932,
+      "step": 4806
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 8.590064331665476e-05,
+      "loss": 1.2731,
+      "step": 4807
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 8.591851322373124e-05,
+      "loss": 1.2637,
+      "step": 4808
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 8.593638313080773e-05,
+      "loss": 1.3059,
+      "step": 4809
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 8.595425303788421e-05,
+      "loss": 1.271,
+      "step": 4810
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 8.597212294496069e-05,
+      "loss": 1.2991,
+      "step": 4811
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 8.598999285203718e-05,
+      "loss": 1.3031,
+      "step": 4812
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 8.600786275911365e-05,
+      "loss": 1.3312,
+      "step": 4813
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 8.602573266619015e-05,
+      "loss": 1.2817,
+      "step": 4814
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 8.604360257326662e-05,
+      "loss": 1.2567,
+      "step": 4815
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 8.606147248034311e-05,
+      "loss": 1.3081,
+      "step": 4816
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 8.607934238741959e-05,
+      "loss": 1.325,
+      "step": 4817
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 8.609721229449608e-05,
+      "loss": 1.2438,
+      "step": 4818
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 8.611508220157256e-05,
+      "loss": 1.335,
+      "step": 4819
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 8.613295210864904e-05,
+      "loss": 1.2675,
+      "step": 4820
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 8.615082201572551e-05,
+      "loss": 1.2675,
+      "step": 4821
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 8.6168691922802e-05,
+      "loss": 1.2694,
+      "step": 4822
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 8.618656182987848e-05,
+      "loss": 1.2792,
+      "step": 4823
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 8.620443173695497e-05,
+      "loss": 1.3137,
+      "step": 4824
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 8.622230164403145e-05,
+      "loss": 1.281,
+      "step": 4825
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 8.624017155110794e-05,
+      "loss": 1.3288,
+      "step": 4826
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 8.625804145818442e-05,
+      "loss": 1.2742,
+      "step": 4827
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 8.627591136526091e-05,
+      "loss": 1.324,
+      "step": 4828
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 8.629378127233739e-05,
+      "loss": 1.296,
+      "step": 4829
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 8.631165117941387e-05,
+      "loss": 1.275,
+      "step": 4830
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 8.632952108649036e-05,
+      "loss": 1.2571,
+      "step": 4831
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 8.634739099356683e-05,
+      "loss": 1.257,
+      "step": 4832
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 8.636526090064332e-05,
+      "loss": 1.2625,
+      "step": 4833
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 8.63831308077198e-05,
+      "loss": 1.2764,
+      "step": 4834
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 8.640100071479629e-05,
+      "loss": 1.3068,
+      "step": 4835
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 8.641887062187277e-05,
+      "loss": 1.2887,
+      "step": 4836
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 8.643674052894926e-05,
+      "loss": 1.2486,
+      "step": 4837
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 8.645461043602574e-05,
+      "loss": 1.2442,
+      "step": 4838
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 8.647248034310223e-05,
+      "loss": 1.2673,
+      "step": 4839
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 8.64903502501787e-05,
+      "loss": 1.283,
+      "step": 4840
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 8.650822015725518e-05,
+      "loss": 1.2646,
+      "step": 4841
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 8.652609006433166e-05,
+      "loss": 1.3015,
+      "step": 4842
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 8.654395997140815e-05,
+      "loss": 1.2211,
+      "step": 4843
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 8.656182987848463e-05,
+      "loss": 1.2829,
+      "step": 4844
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 8.657969978556112e-05,
+      "loss": 1.2851,
+      "step": 4845
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 8.65975696926376e-05,
+      "loss": 1.2812,
+      "step": 4846
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 8.661543959971409e-05,
+      "loss": 1.2818,
+      "step": 4847
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 8.663330950679057e-05,
+      "loss": 1.2748,
+      "step": 4848
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 8.665117941386706e-05,
+      "loss": 1.2905,
+      "step": 4849
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 8.666904932094354e-05,
+      "loss": 1.277,
+      "step": 4850
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 8.668691922802001e-05,
+      "loss": 1.2931,
+      "step": 4851
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 8.67047891350965e-05,
+      "loss": 1.2502,
+      "step": 4852
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 8.672265904217298e-05,
+      "loss": 1.2704,
+      "step": 4853
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 8.674052894924947e-05,
+      "loss": 1.2749,
+      "step": 4854
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 8.675839885632595e-05,
+      "loss": 1.2681,
+      "step": 4855
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 8.677626876340244e-05,
+      "loss": 1.2427,
+      "step": 4856
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 8.679413867047892e-05,
+      "loss": 1.3172,
+      "step": 4857
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 8.681200857755541e-05,
+      "loss": 1.2456,
+      "step": 4858
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 8.682987848463189e-05,
+      "loss": 1.2494,
+      "step": 4859
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 8.684774839170838e-05,
+      "loss": 1.295,
+      "step": 4860
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 8.686561829878484e-05,
+      "loss": 1.3001,
+      "step": 4861
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 8.688348820586133e-05,
+      "loss": 1.2996,
+      "step": 4862
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 8.690135811293781e-05,
+      "loss": 1.2862,
+      "step": 4863
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 8.69192280200143e-05,
+      "loss": 1.3137,
+      "step": 4864
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 8.693709792709078e-05,
+      "loss": 1.2829,
+      "step": 4865
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 8.695496783416727e-05,
+      "loss": 1.2739,
+      "step": 4866
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 8.697283774124375e-05,
+      "loss": 1.3098,
+      "step": 4867
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 8.699070764832024e-05,
+      "loss": 1.2771,
+      "step": 4868
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 8.700857755539671e-05,
+      "loss": 1.3232,
+      "step": 4869
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 8.70264474624732e-05,
+      "loss": 1.2647,
+      "step": 4870
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 8.704431736954968e-05,
+      "loss": 1.2624,
+      "step": 4871
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 8.706218727662616e-05,
+      "loss": 1.2826,
+      "step": 4872
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 8.708005718370265e-05,
+      "loss": 1.3146,
+      "step": 4873
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 8.709792709077913e-05,
+      "loss": 1.3017,
+      "step": 4874
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 8.711579699785562e-05,
+      "loss": 1.2924,
+      "step": 4875
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 8.71336669049321e-05,
+      "loss": 1.3177,
+      "step": 4876
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 8.715153681200859e-05,
+      "loss": 1.3208,
+      "step": 4877
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 8.716940671908507e-05,
+      "loss": 1.3073,
+      "step": 4878
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 8.718727662616156e-05,
+      "loss": 1.302,
+      "step": 4879
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 8.720514653323803e-05,
+      "loss": 1.2897,
+      "step": 4880
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.722301644031452e-05,
+      "loss": 1.2694,
+      "step": 4881
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.724088634739099e-05,
+      "loss": 1.2793,
+      "step": 4882
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.725875625446748e-05,
+      "loss": 1.2688,
+      "step": 4883
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.727662616154396e-05,
+      "loss": 1.3237,
+      "step": 4884
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.729449606862045e-05,
+      "loss": 1.2802,
+      "step": 4885
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.731236597569693e-05,
+      "loss": 1.2904,
+      "step": 4886
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.733023588277342e-05,
+      "loss": 1.2893,
+      "step": 4887
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.73481057898499e-05,
+      "loss": 1.2667,
+      "step": 4888
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.736597569692638e-05,
+      "loss": 1.3287,
+      "step": 4889
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.738384560400286e-05,
+      "loss": 1.3074,
+      "step": 4890
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.740171551107935e-05,
+      "loss": 1.2815,
+      "step": 4891
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.741958541815583e-05,
+      "loss": 1.2748,
+      "step": 4892
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.743745532523231e-05,
+      "loss": 1.2642,
+      "step": 4893
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.74553252323088e-05,
+      "loss": 1.2623,
+      "step": 4894
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.747319513938528e-05,
+      "loss": 1.2809,
+      "step": 4895
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.749106504646177e-05,
+      "loss": 1.2705,
+      "step": 4896
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.750893495353824e-05,
+      "loss": 1.2727,
+      "step": 4897
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.752680486061474e-05,
+      "loss": 1.2747,
+      "step": 4898
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.754467476769121e-05,
+      "loss": 1.3126,
+      "step": 4899
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.75625446747677e-05,
+      "loss": 1.2925,
+      "step": 4900
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.758041458184418e-05,
+      "loss": 1.2823,
+      "step": 4901
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.759828448892066e-05,
+      "loss": 1.2958,
+      "step": 4902
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.761615439599714e-05,
+      "loss": 1.28,
+      "step": 4903
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.763402430307363e-05,
+      "loss": 1.3209,
+      "step": 4904
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.76518942101501e-05,
+      "loss": 1.3302,
+      "step": 4905
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.76697641172266e-05,
+      "loss": 1.276,
+      "step": 4906
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.768763402430307e-05,
+      "loss": 1.2631,
+      "step": 4907
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.770550393137956e-05,
+      "loss": 1.2888,
+      "step": 4908
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.772337383845604e-05,
+      "loss": 1.2644,
+      "step": 4909
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.774124374553253e-05,
+      "loss": 1.2796,
+      "step": 4910
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.775911365260901e-05,
+      "loss": 1.2587,
+      "step": 4911
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.777698355968549e-05,
+      "loss": 1.3047,
+      "step": 4912
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.779485346676198e-05,
+      "loss": 1.3013,
+      "step": 4913
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.781272337383846e-05,
+      "loss": 1.3176,
+      "step": 4914
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.783059328091495e-05,
+      "loss": 1.2608,
+      "step": 4915
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.784846318799142e-05,
+      "loss": 1.2729,
+      "step": 4916
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.786633309506791e-05,
+      "loss": 1.2954,
+      "step": 4917
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.788420300214439e-05,
+      "loss": 1.2895,
+      "step": 4918
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.790207290922088e-05,
+      "loss": 1.2854,
+      "step": 4919
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.791994281629736e-05,
+      "loss": 1.2861,
+      "step": 4920
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.793781272337385e-05,
+      "loss": 1.2775,
+      "step": 4921
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.795568263045033e-05,
+      "loss": 1.2899,
+      "step": 4922
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.79735525375268e-05,
+      "loss": 1.2801,
+      "step": 4923
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.799142244460328e-05,
+      "loss": 1.2852,
+      "step": 4924
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.800929235167977e-05,
+      "loss": 1.2738,
+      "step": 4925
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.802716225875625e-05,
+      "loss": 1.3336,
+      "step": 4926
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.804503216583274e-05,
+      "loss": 1.264,
+      "step": 4927
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.806290207290922e-05,
+      "loss": 1.3069,
+      "step": 4928
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.808077197998571e-05,
+      "loss": 1.2894,
+      "step": 4929
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.809864188706219e-05,
+      "loss": 1.2626,
+      "step": 4930
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.811651179413868e-05,
+      "loss": 1.3065,
+      "step": 4931
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.813438170121516e-05,
+      "loss": 1.2987,
+      "step": 4932
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.815225160829163e-05,
+      "loss": 1.2908,
+      "step": 4933
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.817012151536813e-05,
+      "loss": 1.2862,
+      "step": 4934
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.81879914224446e-05,
+      "loss": 1.324,
+      "step": 4935
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.82058613295211e-05,
+      "loss": 1.308,
+      "step": 4936
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.822373123659757e-05,
+      "loss": 1.259,
+      "step": 4937
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.824160114367406e-05,
+      "loss": 1.2806,
+      "step": 4938
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.825947105075054e-05,
+      "loss": 1.2989,
+      "step": 4939
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.827734095782703e-05,
+      "loss": 1.2803,
+      "step": 4940
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.829521086490351e-05,
+      "loss": 1.2973,
+      "step": 4941
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.831308077198e-05,
+      "loss": 1.3253,
+      "step": 4942
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.833095067905646e-05,
+      "loss": 1.2658,
+      "step": 4943
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.834882058613295e-05,
+      "loss": 1.2863,
+      "step": 4944
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.836669049320943e-05,
+      "loss": 1.2888,
+      "step": 4945
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.838456040028592e-05,
+      "loss": 1.2745,
+      "step": 4946
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.84024303073624e-05,
+      "loss": 1.2874,
+      "step": 4947
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.842030021443889e-05,
+      "loss": 1.2979,
+      "step": 4948
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.843817012151537e-05,
+      "loss": 1.2977,
+      "step": 4949
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.845604002859186e-05,
+      "loss": 1.2854,
+      "step": 4950
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.847390993566834e-05,
+      "loss": 1.3357,
+      "step": 4951
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.849177984274483e-05,
+      "loss": 1.2646,
+      "step": 4952
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.85096497498213e-05,
+      "loss": 1.3092,
+      "step": 4953
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.852751965689778e-05,
+      "loss": 1.3062,
+      "step": 4954
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.854538956397427e-05,
+      "loss": 1.2857,
+      "step": 4955
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.856325947105075e-05,
+      "loss": 1.2905,
+      "step": 4956
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.858112937812724e-05,
+      "loss": 1.2959,
+      "step": 4957
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.859899928520372e-05,
+      "loss": 1.3297,
+      "step": 4958
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.861686919228021e-05,
+      "loss": 1.2777,
+      "step": 4959
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.863473909935669e-05,
+      "loss": 1.2822,
+      "step": 4960
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.865260900643318e-05,
+      "loss": 1.2782,
+      "step": 4961
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.867047891350966e-05,
+      "loss": 1.3179,
+      "step": 4962
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.868834882058615e-05,
+      "loss": 1.3036,
+      "step": 4963
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.870621872766261e-05,
+      "loss": 1.2797,
+      "step": 4964
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.87240886347391e-05,
+      "loss": 1.2964,
+      "step": 4965
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.874195854181558e-05,
+      "loss": 1.2938,
+      "step": 4966
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.875982844889207e-05,
+      "loss": 1.2963,
+      "step": 4967
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.877769835596855e-05,
+      "loss": 1.2952,
+      "step": 4968
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.879556826304504e-05,
+      "loss": 1.2807,
+      "step": 4969
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.881343817012152e-05,
+      "loss": 1.2921,
+      "step": 4970
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.8831308077198e-05,
+      "loss": 1.275,
+      "step": 4971
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.884917798427448e-05,
+      "loss": 1.3017,
+      "step": 4972
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.886704789135097e-05,
+      "loss": 1.2738,
+      "step": 4973
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.888491779842745e-05,
+      "loss": 1.3124,
+      "step": 4974
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.890278770550393e-05,
+      "loss": 1.2623,
+      "step": 4975
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.892065761258042e-05,
+      "loss": 1.251,
+      "step": 4976
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.89385275196569e-05,
+      "loss": 1.295,
+      "step": 4977
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.895639742673339e-05,
+      "loss": 1.2868,
+      "step": 4978
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.897426733380987e-05,
+      "loss": 1.2882,
+      "step": 4979
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.899213724088636e-05,
+      "loss": 1.2754,
+      "step": 4980
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.901000714796283e-05,
+      "loss": 1.2749,
+      "step": 4981
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.902787705503933e-05,
+      "loss": 1.3025,
+      "step": 4982
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.90457469621158e-05,
+      "loss": 1.3019,
+      "step": 4983
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.906361686919228e-05,
+      "loss": 1.3022,
+      "step": 4984
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.908148677626876e-05,
+      "loss": 1.2773,
+      "step": 4985
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.909935668334525e-05,
+      "loss": 1.348,
+      "step": 4986
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.911722659042173e-05,
+      "loss": 1.2706,
+      "step": 4987
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.913509649749822e-05,
+      "loss": 1.2944,
+      "step": 4988
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.91529664045747e-05,
+      "loss": 1.2542,
+      "step": 4989
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.917083631165119e-05,
+      "loss": 1.2825,
+      "step": 4990
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.918870621872766e-05,
+      "loss": 1.2882,
+      "step": 4991
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.920657612580415e-05,
+      "loss": 1.2823,
+      "step": 4992
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.922444603288063e-05,
+      "loss": 1.2841,
+      "step": 4993
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.924231593995712e-05,
+      "loss": 1.2748,
+      "step": 4994
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.92601858470336e-05,
+      "loss": 1.2851,
+      "step": 4995
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.927805575411008e-05,
+      "loss": 1.2972,
+      "step": 4996
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.929592566118657e-05,
+      "loss": 1.2677,
+      "step": 4997
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.931379556826305e-05,
+      "loss": 1.315,
+      "step": 4998
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.933166547533954e-05,
+      "loss": 1.3175,
+      "step": 4999
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.934953538241601e-05,
+      "loss": 1.2946,
+      "step": 5000
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.93674052894925e-05,
+      "loss": 1.2991,
+      "step": 5001
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.938527519656898e-05,
+      "loss": 1.2743,
+      "step": 5002
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.940314510364547e-05,
+      "loss": 1.3032,
+      "step": 5003
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.942101501072195e-05,
+      "loss": 1.3469,
+      "step": 5004
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.943888491779843e-05,
+      "loss": 1.2919,
+      "step": 5005
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.94567548248749e-05,
+      "loss": 1.2896,
+      "step": 5006
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.94746247319514e-05,
+      "loss": 1.3081,
+      "step": 5007
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.949249463902787e-05,
+      "loss": 1.2909,
+      "step": 5008
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.951036454610436e-05,
+      "loss": 1.2865,
+      "step": 5009
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.952823445318084e-05,
+      "loss": 1.2973,
+      "step": 5010
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.954610436025733e-05,
+      "loss": 1.2899,
+      "step": 5011
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.956397426733381e-05,
+      "loss": 1.2883,
+      "step": 5012
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.95818441744103e-05,
+      "loss": 1.2451,
+      "step": 5013
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.959971408148678e-05,
+      "loss": 1.305,
+      "step": 5014
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.961758398856326e-05,
+      "loss": 1.2884,
+      "step": 5015
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.963545389563975e-05,
+      "loss": 1.2754,
+      "step": 5016
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.965332380271622e-05,
+      "loss": 1.3226,
+      "step": 5017
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.967119370979272e-05,
+      "loss": 1.2999,
+      "step": 5018
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.968906361686919e-05,
+      "loss": 1.2728,
+      "step": 5019
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.970693352394568e-05,
+      "loss": 1.2673,
+      "step": 5020
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.972480343102216e-05,
+      "loss": 1.2991,
+      "step": 5021
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.974267333809865e-05,
+      "loss": 1.3226,
+      "step": 5022
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.976054324517513e-05,
+      "loss": 1.3175,
+      "step": 5023
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.977841315225162e-05,
+      "loss": 1.269,
+      "step": 5024
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.97962830593281e-05,
+      "loss": 1.2933,
+      "step": 5025
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.981415296640458e-05,
+      "loss": 1.3228,
+      "step": 5026
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.983202287348105e-05,
+      "loss": 1.3009,
+      "step": 5027
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.984989278055754e-05,
+      "loss": 1.2638,
+      "step": 5028
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.986776268763402e-05,
+      "loss": 1.2965,
+      "step": 5029
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.988563259471051e-05,
+      "loss": 1.3004,
+      "step": 5030
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.990350250178699e-05,
+      "loss": 1.2576,
+      "step": 5031
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.992137240886348e-05,
+      "loss": 1.2993,
+      "step": 5032
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.993924231593996e-05,
+      "loss": 1.2976,
+      "step": 5033
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.995711222301645e-05,
+      "loss": 1.3025,
+      "step": 5034
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.997498213009293e-05,
+      "loss": 1.3137,
+      "step": 5035
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.99928520371694e-05,
+      "loss": 1.2994,
+      "step": 5036
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.00107219442459e-05,
+      "loss": 1.2746,
+      "step": 5037
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.002859185132237e-05,
+      "loss": 1.3004,
+      "step": 5038
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.004646175839886e-05,
+      "loss": 1.3282,
+      "step": 5039
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.006433166547534e-05,
+      "loss": 1.2593,
+      "step": 5040
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.008220157255183e-05,
+      "loss": 1.2709,
+      "step": 5041
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.010007147962831e-05,
+      "loss": 1.2725,
+      "step": 5042
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.01179413867048e-05,
+      "loss": 1.2911,
+      "step": 5043
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.013581129378128e-05,
+      "loss": 1.3072,
+      "step": 5044
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.015368120085777e-05,
+      "loss": 1.2607,
+      "step": 5045
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.017155110793423e-05,
+      "loss": 1.2774,
+      "step": 5046
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.018942101501072e-05,
+      "loss": 1.2718,
+      "step": 5047
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.02072909220872e-05,
+      "loss": 1.292,
+      "step": 5048
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.022516082916369e-05,
+      "loss": 1.2934,
+      "step": 5049
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.024303073624017e-05,
+      "loss": 1.3003,
+      "step": 5050
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.026090064331666e-05,
+      "loss": 1.3,
+      "step": 5051
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.027877055039314e-05,
+      "loss": 1.2381,
+      "step": 5052
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.029664045746963e-05,
+      "loss": 1.3139,
+      "step": 5053
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.03145103645461e-05,
+      "loss": 1.2958,
+      "step": 5054
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.03323802716226e-05,
+      "loss": 1.304,
+      "step": 5055
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.035025017869907e-05,
+      "loss": 1.3013,
+      "step": 5056
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.036812008577555e-05,
+      "loss": 1.3184,
+      "step": 5057
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.038598999285204e-05,
+      "loss": 1.2951,
+      "step": 5058
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.040385989992852e-05,
+      "loss": 1.3044,
+      "step": 5059
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.042172980700501e-05,
+      "loss": 1.2885,
+      "step": 5060
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.043959971408149e-05,
+      "loss": 1.3182,
+      "step": 5061
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.045746962115798e-05,
+      "loss": 1.2744,
+      "step": 5062
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.047533952823446e-05,
+      "loss": 1.2717,
+      "step": 5063
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.049320943531095e-05,
+      "loss": 1.2794,
+      "step": 5064
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.051107934238742e-05,
+      "loss": 1.2939,
+      "step": 5065
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.052894924946392e-05,
+      "loss": 1.3115,
+      "step": 5066
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.054681915654038e-05,
+      "loss": 1.299,
+      "step": 5067
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.056468906361687e-05,
+      "loss": 1.3189,
+      "step": 5068
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.058255897069335e-05,
+      "loss": 1.2803,
+      "step": 5069
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.060042887776984e-05,
+      "loss": 1.2973,
+      "step": 5070
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.061829878484632e-05,
+      "loss": 1.2854,
+      "step": 5071
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.063616869192281e-05,
+      "loss": 1.3094,
+      "step": 5072
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.065403859899928e-05,
+      "loss": 1.2565,
+      "step": 5073
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.067190850607578e-05,
+      "loss": 1.2901,
+      "step": 5074
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.068977841315225e-05,
+      "loss": 1.306,
+      "step": 5075
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.070764832022874e-05,
+      "loss": 1.2775,
+      "step": 5076
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.072551822730522e-05,
+      "loss": 1.3243,
+      "step": 5077
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.07433881343817e-05,
+      "loss": 1.2886,
+      "step": 5078
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.076125804145819e-05,
+      "loss": 1.3387,
+      "step": 5079
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.077912794853467e-05,
+      "loss": 1.2827,
+      "step": 5080
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.079699785561116e-05,
+      "loss": 1.2559,
+      "step": 5081
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.081486776268764e-05,
+      "loss": 1.3049,
+      "step": 5082
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.083273766976413e-05,
+      "loss": 1.3033,
+      "step": 5083
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.08506075768406e-05,
+      "loss": 1.3047,
+      "step": 5084
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.08684774839171e-05,
+      "loss": 1.328,
+      "step": 5085
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.088634739099357e-05,
+      "loss": 1.2503,
+      "step": 5086
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.090421729807005e-05,
+      "loss": 1.2945,
+      "step": 5087
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.092208720514653e-05,
+      "loss": 1.2877,
+      "step": 5088
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.093995711222302e-05,
+      "loss": 1.2717,
+      "step": 5089
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.09578270192995e-05,
+      "loss": 1.2743,
+      "step": 5090
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.097569692637599e-05,
+      "loss": 1.2964,
+      "step": 5091
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.099356683345246e-05,
+      "loss": 1.3095,
+      "step": 5092
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.101143674052895e-05,
+      "loss": 1.299,
+      "step": 5093
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.102930664760543e-05,
+      "loss": 1.3095,
+      "step": 5094
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.104717655468192e-05,
+      "loss": 1.2926,
+      "step": 5095
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.106504646175841e-05,
+      "loss": 1.2844,
+      "step": 5096
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.108291636883489e-05,
+      "loss": 1.281,
+      "step": 5097
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.110078627591137e-05,
+      "loss": 1.2812,
+      "step": 5098
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.111865618298785e-05,
+      "loss": 1.284,
+      "step": 5099
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.113652609006434e-05,
+      "loss": 1.3049,
+      "step": 5100
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.115439599714081e-05,
+      "loss": 1.2791,
+      "step": 5101
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.11722659042173e-05,
+      "loss": 1.2364,
+      "step": 5102
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.119013581129378e-05,
+      "loss": 1.2543,
+      "step": 5103
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.120800571837027e-05,
+      "loss": 1.2785,
+      "step": 5104
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.122587562544675e-05,
+      "loss": 1.3187,
+      "step": 5105
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.124374553252324e-05,
+      "loss": 1.3076,
+      "step": 5106
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.126161543959972e-05,
+      "loss": 1.2547,
+      "step": 5107
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.12794853466762e-05,
+      "loss": 1.2672,
+      "step": 5108
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.129735525375267e-05,
+      "loss": 1.2916,
+      "step": 5109
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.131522516082917e-05,
+      "loss": 1.2694,
+      "step": 5110
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.133309506790564e-05,
+      "loss": 1.2827,
+      "step": 5111
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.135096497498213e-05,
+      "loss": 1.2632,
+      "step": 5112
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.136883488205861e-05,
+      "loss": 1.2706,
+      "step": 5113
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.13867047891351e-05,
+      "loss": 1.2787,
+      "step": 5114
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.14045746962116e-05,
+      "loss": 1.3078,
+      "step": 5115
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.142244460328807e-05,
+      "loss": 1.2862,
+      "step": 5116
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.144031451036456e-05,
+      "loss": 1.27,
+      "step": 5117
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.145818441744103e-05,
+      "loss": 1.2994,
+      "step": 5118
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.147605432451752e-05,
+      "loss": 1.3082,
+      "step": 5119
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.1493924231594e-05,
+      "loss": 1.2783,
+      "step": 5120
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.151179413867048e-05,
+      "loss": 1.2845,
+      "step": 5121
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.152966404574696e-05,
+      "loss": 1.3593,
+      "step": 5122
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.154753395282345e-05,
+      "loss": 1.2864,
+      "step": 5123
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.156540385989993e-05,
+      "loss": 1.2821,
+      "step": 5124
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.158327376697642e-05,
+      "loss": 1.2679,
+      "step": 5125
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.16011436740529e-05,
+      "loss": 1.3327,
+      "step": 5126
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.161901358112939e-05,
+      "loss": 1.2845,
+      "step": 5127
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.163688348820585e-05,
+      "loss": 1.2414,
+      "step": 5128
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.165475339528234e-05,
+      "loss": 1.278,
+      "step": 5129
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.167262330235882e-05,
+      "loss": 1.318,
+      "step": 5130
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.169049320943531e-05,
+      "loss": 1.3,
+      "step": 5131
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.170836311651179e-05,
+      "loss": 1.2969,
+      "step": 5132
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.172623302358828e-05,
+      "loss": 1.3012,
+      "step": 5133
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.174410293066477e-05,
+      "loss": 1.3215,
+      "step": 5134
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.176197283774125e-05,
+      "loss": 1.3149,
+      "step": 5135
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.177984274481774e-05,
+      "loss": 1.2802,
+      "step": 5136
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.179771265189422e-05,
+      "loss": 1.2981,
+      "step": 5137
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.181558255897071e-05,
+      "loss": 1.3145,
+      "step": 5138
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.183345246604717e-05,
+      "loss": 1.2829,
+      "step": 5139
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.185132237312366e-05,
+      "loss": 1.2753,
+      "step": 5140
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.186919228020014e-05,
+      "loss": 1.2939,
+      "step": 5141
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.188706218727663e-05,
+      "loss": 1.2731,
+      "step": 5142
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.190493209435311e-05,
+      "loss": 1.264,
+      "step": 5143
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.19228020014296e-05,
+      "loss": 1.2453,
+      "step": 5144
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.194067190850608e-05,
+      "loss": 1.275,
+      "step": 5145
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.195854181558257e-05,
+      "loss": 1.3029,
+      "step": 5146
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.197641172265905e-05,
+      "loss": 1.3025,
+      "step": 5147
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.199428162973554e-05,
+      "loss": 1.2632,
+      "step": 5148
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.2012151536812e-05,
+      "loss": 1.2924,
+      "step": 5149
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.203002144388849e-05,
+      "loss": 1.3028,
+      "step": 5150
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.204789135096497e-05,
+      "loss": 1.2928,
+      "step": 5151
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.206576125804146e-05,
+      "loss": 1.275,
+      "step": 5152
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.208363116511795e-05,
+      "loss": 1.2965,
+      "step": 5153
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.210150107219443e-05,
+      "loss": 1.2817,
+      "step": 5154
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.211937097927092e-05,
+      "loss": 1.2949,
+      "step": 5155
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.21372408863474e-05,
+      "loss": 1.3205,
+      "step": 5156
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.215511079342389e-05,
+      "loss": 1.3323,
+      "step": 5157
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.217298070050037e-05,
+      "loss": 1.3087,
+      "step": 5158
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.219085060757684e-05,
+      "loss": 1.2965,
+      "step": 5159
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.220872051465332e-05,
+      "loss": 1.3002,
+      "step": 5160
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.222659042172981e-05,
+      "loss": 1.2786,
+      "step": 5161
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.224446032880629e-05,
+      "loss": 1.2852,
+      "step": 5162
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.226233023588278e-05,
+      "loss": 1.2775,
+      "step": 5163
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.228020014295926e-05,
+      "loss": 1.2849,
+      "step": 5164
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.229807005003575e-05,
+      "loss": 1.2614,
+      "step": 5165
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.231593995711223e-05,
+      "loss": 1.3181,
+      "step": 5166
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.233380986418872e-05,
+      "loss": 1.2942,
+      "step": 5167
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.23516797712652e-05,
+      "loss": 1.2692,
+      "step": 5168
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.236954967834169e-05,
+      "loss": 1.2879,
+      "step": 5169
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.238741958541815e-05,
+      "loss": 1.2775,
+      "step": 5170
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.240528949249464e-05,
+      "loss": 1.2996,
+      "step": 5171
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.242315939957113e-05,
+      "loss": 1.3059,
+      "step": 5172
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.244102930664761e-05,
+      "loss": 1.3126,
+      "step": 5173
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.24588992137241e-05,
+      "loss": 1.249,
+      "step": 5174
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.247676912080058e-05,
+      "loss": 1.2969,
+      "step": 5175
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.249463902787707e-05,
+      "loss": 1.2978,
+      "step": 5176
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.251250893495355e-05,
+      "loss": 1.3026,
+      "step": 5177
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.253037884203004e-05,
+      "loss": 1.284,
+      "step": 5178
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.254824874910651e-05,
+      "loss": 1.2895,
+      "step": 5179
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.256611865618299e-05,
+      "loss": 1.2999,
+      "step": 5180
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.258398856325947e-05,
+      "loss": 1.2829,
+      "step": 5181
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.260185847033596e-05,
+      "loss": 1.2776,
+      "step": 5182
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.261972837741244e-05,
+      "loss": 1.2919,
+      "step": 5183
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.263759828448893e-05,
+      "loss": 1.271,
+      "step": 5184
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.26554681915654e-05,
+      "loss": 1.3163,
+      "step": 5185
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.26733380986419e-05,
+      "loss": 1.2744,
+      "step": 5186
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.269120800571837e-05,
+      "loss": 1.2812,
+      "step": 5187
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.270907791279486e-05,
+      "loss": 1.292,
+      "step": 5188
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.272694781987134e-05,
+      "loss": 1.3057,
+      "step": 5189
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.274481772694782e-05,
+      "loss": 1.2703,
+      "step": 5190
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.276268763402431e-05,
+      "loss": 1.3018,
+      "step": 5191
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.278055754110079e-05,
+      "loss": 1.273,
+      "step": 5192
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.279842744817728e-05,
+      "loss": 1.2714,
+      "step": 5193
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.281629735525376e-05,
+      "loss": 1.2641,
+      "step": 5194
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.283416726233025e-05,
+      "loss": 1.3205,
+      "step": 5195
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.285203716940672e-05,
+      "loss": 1.2778,
+      "step": 5196
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.286990707648322e-05,
+      "loss": 1.2855,
+      "step": 5197
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.288777698355969e-05,
+      "loss": 1.2523,
+      "step": 5198
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.290564689063618e-05,
+      "loss": 1.2764,
+      "step": 5199
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.292351679771265e-05,
+      "loss": 1.289,
+      "step": 5200
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.294138670478914e-05,
+      "loss": 1.2665,
+      "step": 5201
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.295925661186562e-05,
+      "loss": 1.3439,
+      "step": 5202
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.29771265189421e-05,
+      "loss": 1.2636,
+      "step": 5203
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.299499642601858e-05,
+      "loss": 1.2916,
+      "step": 5204
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.301286633309508e-05,
+      "loss": 1.2942,
+      "step": 5205
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.303073624017155e-05,
+      "loss": 1.2713,
+      "step": 5206
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.304860614724804e-05,
+      "loss": 1.3115,
+      "step": 5207
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.306647605432452e-05,
+      "loss": 1.2989,
+      "step": 5208
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.308434596140101e-05,
+      "loss": 1.2758,
+      "step": 5209
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.310221586847749e-05,
+      "loss": 1.2174,
+      "step": 5210
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.312008577555397e-05,
+      "loss": 1.2604,
+      "step": 5211
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.313795568263046e-05,
+      "loss": 1.3386,
+      "step": 5212
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.315582558970694e-05,
+      "loss": 1.3219,
+      "step": 5213
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.317369549678343e-05,
+      "loss": 1.2763,
+      "step": 5214
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.31915654038599e-05,
+      "loss": 1.333,
+      "step": 5215
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.32094353109364e-05,
+      "loss": 1.3088,
+      "step": 5216
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.322730521801287e-05,
+      "loss": 1.2813,
+      "step": 5217
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.324517512508936e-05,
+      "loss": 1.2742,
+      "step": 5218
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.326304503216584e-05,
+      "loss": 1.3243,
+      "step": 5219
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.328091493924233e-05,
+      "loss": 1.3311,
+      "step": 5220
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.32987848463188e-05,
+      "loss": 1.2709,
+      "step": 5221
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.331665475339529e-05,
+      "loss": 1.3186,
+      "step": 5222
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.333452466047176e-05,
+      "loss": 1.2838,
+      "step": 5223
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.335239456754825e-05,
+      "loss": 1.2916,
+      "step": 5224
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.337026447462473e-05,
+      "loss": 1.303,
+      "step": 5225
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.338813438170122e-05,
+      "loss": 1.2718,
+      "step": 5226
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.34060042887777e-05,
+      "loss": 1.2994,
+      "step": 5227
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.342387419585419e-05,
+      "loss": 1.2841,
+      "step": 5228
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.344174410293067e-05,
+      "loss": 1.266,
+      "step": 5229
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.345961401000716e-05,
+      "loss": 1.2884,
+      "step": 5230
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.347748391708364e-05,
+      "loss": 1.3059,
+      "step": 5231
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.349535382416011e-05,
+      "loss": 1.2729,
+      "step": 5232
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.35132237312366e-05,
+      "loss": 1.2818,
+      "step": 5233
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.353109363831308e-05,
+      "loss": 1.2862,
+      "step": 5234
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.354896354538957e-05,
+      "loss": 1.2878,
+      "step": 5235
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.356683345246605e-05,
+      "loss": 1.3278,
+      "step": 5236
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.358470335954254e-05,
+      "loss": 1.2959,
+      "step": 5237
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.360257326661902e-05,
+      "loss": 1.2889,
+      "step": 5238
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.362044317369551e-05,
+      "loss": 1.3131,
+      "step": 5239
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.363831308077199e-05,
+      "loss": 1.3053,
+      "step": 5240
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.365618298784848e-05,
+      "loss": 1.2878,
+      "step": 5241
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.367405289492494e-05,
+      "loss": 1.259,
+      "step": 5242
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.369192280200143e-05,
+      "loss": 1.2943,
+      "step": 5243
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.370979270907791e-05,
+      "loss": 1.3027,
+      "step": 5244
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.37276626161544e-05,
+      "loss": 1.2751,
+      "step": 5245
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.374553252323088e-05,
+      "loss": 1.277,
+      "step": 5246
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.376340243030737e-05,
+      "loss": 1.2835,
+      "step": 5247
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.378127233738385e-05,
+      "loss": 1.2716,
+      "step": 5248
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.379914224446034e-05,
+      "loss": 1.2773,
+      "step": 5249
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.381701215153682e-05,
+      "loss": 1.2688,
+      "step": 5250
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.383488205861331e-05,
+      "loss": 1.3063,
+      "step": 5251
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.385275196568978e-05,
+      "loss": 1.2749,
+      "step": 5252
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.387062187276626e-05,
+      "loss": 1.3192,
+      "step": 5253
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.388849177984275e-05,
+      "loss": 1.2956,
+      "step": 5254
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.390636168691923e-05,
+      "loss": 1.3096,
+      "step": 5255
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.392423159399572e-05,
+      "loss": 1.293,
+      "step": 5256
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.39421015010722e-05,
+      "loss": 1.3197,
+      "step": 5257
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.395997140814869e-05,
+      "loss": 1.2765,
+      "step": 5258
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.397784131522517e-05,
+      "loss": 1.2867,
+      "step": 5259
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.399571122230166e-05,
+      "loss": 1.2602,
+      "step": 5260
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.401358112937814e-05,
+      "loss": 1.2438,
+      "step": 5261
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.403145103645461e-05,
+      "loss": 1.2804,
+      "step": 5262
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.404932094353109e-05,
+      "loss": 1.2771,
+      "step": 5263
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.406719085060758e-05,
+      "loss": 1.2786,
+      "step": 5264
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.408506075768406e-05,
+      "loss": 1.2896,
+      "step": 5265
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.410293066476055e-05,
+      "loss": 1.2747,
+      "step": 5266
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.412080057183703e-05,
+      "loss": 1.2939,
+      "step": 5267
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.413867047891352e-05,
+      "loss": 1.2947,
+      "step": 5268
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.415654038599e-05,
+      "loss": 1.2785,
+      "step": 5269
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.417441029306649e-05,
+      "loss": 1.287,
+      "step": 5270
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.419228020014296e-05,
+      "loss": 1.3131,
+      "step": 5271
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.421015010721944e-05,
+      "loss": 1.3165,
+      "step": 5272
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.422802001429593e-05,
+      "loss": 1.3442,
+      "step": 5273
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.424588992137241e-05,
+      "loss": 1.2898,
+      "step": 5274
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.42637598284489e-05,
+      "loss": 1.3149,
+      "step": 5275
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.428162973552538e-05,
+      "loss": 1.308,
+      "step": 5276
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.429949964260187e-05,
+      "loss": 1.2909,
+      "step": 5277
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.431736954967835e-05,
+      "loss": 1.2881,
+      "step": 5278
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.433523945675484e-05,
+      "loss": 1.2866,
+      "step": 5279
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.435310936383131e-05,
+      "loss": 1.2991,
+      "step": 5280
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.43709792709078e-05,
+      "loss": 1.2851,
+      "step": 5281
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.438884917798428e-05,
+      "loss": 1.2945,
+      "step": 5282
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.440671908506076e-05,
+      "loss": 1.3026,
+      "step": 5283
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.442458899213724e-05,
+      "loss": 1.2732,
+      "step": 5284
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.444245889921373e-05,
+      "loss": 1.3041,
+      "step": 5285
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.44603288062902e-05,
+      "loss": 1.2895,
+      "step": 5286
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.44781987133667e-05,
+      "loss": 1.2683,
+      "step": 5287
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.449606862044317e-05,
+      "loss": 1.2803,
+      "step": 5288
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.451393852751967e-05,
+      "loss": 1.2797,
+      "step": 5289
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.453180843459614e-05,
+      "loss": 1.3248,
+      "step": 5290
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.454967834167263e-05,
+      "loss": 1.3028,
+      "step": 5291
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.456754824874911e-05,
+      "loss": 1.3114,
+      "step": 5292
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.458541815582559e-05,
+      "loss": 1.2771,
+      "step": 5293
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.460328806290208e-05,
+      "loss": 1.2939,
+      "step": 5294
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.462115796997856e-05,
+      "loss": 1.2982,
+      "step": 5295
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.463902787705505e-05,
+      "loss": 1.2737,
+      "step": 5296
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.465689778413153e-05,
+      "loss": 1.2993,
+      "step": 5297
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.467476769120802e-05,
+      "loss": 1.311,
+      "step": 5298
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.46926375982845e-05,
+      "loss": 1.2703,
+      "step": 5299
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.471050750536098e-05,
+      "loss": 1.2756,
+      "step": 5300
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.472837741243746e-05,
+      "loss": 1.2949,
+      "step": 5301
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.474624731951395e-05,
+      "loss": 1.3019,
+      "step": 5302
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.476411722659042e-05,
+      "loss": 1.2636,
+      "step": 5303
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.478198713366691e-05,
+      "loss": 1.2501,
+      "step": 5304
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.479985704074339e-05,
+      "loss": 1.2804,
+      "step": 5305
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.481772694781988e-05,
+      "loss": 1.3078,
+      "step": 5306
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.483559685489635e-05,
+      "loss": 1.3188,
+      "step": 5307
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.485346676197284e-05,
+      "loss": 1.2839,
+      "step": 5308
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.487133666904932e-05,
+      "loss": 1.3009,
+      "step": 5309
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.488920657612581e-05,
+      "loss": 1.316,
+      "step": 5310
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.490707648320229e-05,
+      "loss": 1.2959,
+      "step": 5311
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.492494639027878e-05,
+      "loss": 1.2854,
+      "step": 5312
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.494281629735526e-05,
+      "loss": 1.3126,
+      "step": 5313
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.496068620443174e-05,
+      "loss": 1.3005,
+      "step": 5314
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.497855611150823e-05,
+      "loss": 1.3091,
+      "step": 5315
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.49964260185847e-05,
+      "loss": 1.2976,
+      "step": 5316
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.50142959256612e-05,
+      "loss": 1.3276,
+      "step": 5317
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.503216583273767e-05,
+      "loss": 1.2571,
+      "step": 5318
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.505003573981416e-05,
+      "loss": 1.33,
+      "step": 5319
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.506790564689064e-05,
+      "loss": 1.3069,
+      "step": 5320
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.508577555396713e-05,
+      "loss": 1.2878,
+      "step": 5321
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.510364546104361e-05,
+      "loss": 1.2911,
+      "step": 5322
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.51215153681201e-05,
+      "loss": 1.28,
+      "step": 5323
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.513938527519656e-05,
+      "loss": 1.2977,
+      "step": 5324
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.515725518227306e-05,
+      "loss": 1.3192,
+      "step": 5325
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.517512508934953e-05,
+      "loss": 1.2568,
+      "step": 5326
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.519299499642602e-05,
+      "loss": 1.2736,
+      "step": 5327
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.52108649035025e-05,
+      "loss": 1.2796,
+      "step": 5328
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.522873481057899e-05,
+      "loss": 1.2961,
+      "step": 5329
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.524660471765547e-05,
+      "loss": 1.2719,
+      "step": 5330
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.526447462473196e-05,
+      "loss": 1.3071,
+      "step": 5331
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.528234453180844e-05,
+      "loss": 1.3029,
+      "step": 5332
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.530021443888493e-05,
+      "loss": 1.2897,
+      "step": 5333
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.53180843459614e-05,
+      "loss": 1.2611,
+      "step": 5334
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.533595425303788e-05,
+      "loss": 1.2881,
+      "step": 5335
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.535382416011437e-05,
+      "loss": 1.252,
+      "step": 5336
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.537169406719085e-05,
+      "loss": 1.2935,
+      "step": 5337
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.538956397426734e-05,
+      "loss": 1.2997,
+      "step": 5338
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.540743388134382e-05,
+      "loss": 1.2856,
+      "step": 5339
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.542530378842031e-05,
+      "loss": 1.2795,
+      "step": 5340
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.544317369549679e-05,
+      "loss": 1.281,
+      "step": 5341
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.546104360257328e-05,
+      "loss": 1.3231,
+      "step": 5342
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.547891350964976e-05,
+      "loss": 1.2864,
+      "step": 5343
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.549678341672623e-05,
+      "loss": 1.2832,
+      "step": 5344
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.551465332380271e-05,
+      "loss": 1.2565,
+      "step": 5345
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.55325232308792e-05,
+      "loss": 1.2835,
+      "step": 5346
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.555039313795568e-05,
+      "loss": 1.3178,
+      "step": 5347
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.556826304503217e-05,
+      "loss": 1.3051,
+      "step": 5348
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.558613295210865e-05,
+      "loss": 1.2924,
+      "step": 5349
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.560400285918514e-05,
+      "loss": 1.2764,
+      "step": 5350
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.562187276626162e-05,
+      "loss": 1.2726,
+      "step": 5351
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.563974267333811e-05,
+      "loss": 1.3062,
+      "step": 5352
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.565761258041459e-05,
+      "loss": 1.2724,
+      "step": 5353
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.567548248749108e-05,
+      "loss": 1.2873,
+      "step": 5354
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.569335239456755e-05,
+      "loss": 1.3071,
+      "step": 5355
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.571122230164403e-05,
+      "loss": 1.2856,
+      "step": 5356
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.572909220872052e-05,
+      "loss": 1.2759,
+      "step": 5357
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.5746962115797e-05,
+      "loss": 1.3252,
+      "step": 5358
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.576483202287349e-05,
+      "loss": 1.3209,
+      "step": 5359
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.578270192994997e-05,
+      "loss": 1.2516,
+      "step": 5360
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.580057183702646e-05,
+      "loss": 1.2956,
+      "step": 5361
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.581844174410294e-05,
+      "loss": 1.3259,
+      "step": 5362
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.583631165117943e-05,
+      "loss": 1.3064,
+      "step": 5363
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.58541815582559e-05,
+      "loss": 1.3095,
+      "step": 5364
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.587205146533238e-05,
+      "loss": 1.2748,
+      "step": 5365
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.588992137240886e-05,
+      "loss": 1.3072,
+      "step": 5366
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.590779127948535e-05,
+      "loss": 1.2958,
+      "step": 5367
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.592566118656183e-05,
+      "loss": 1.3532,
+      "step": 5368
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.594353109363832e-05,
+      "loss": 1.3033,
+      "step": 5369
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.59614010007148e-05,
+      "loss": 1.2991,
+      "step": 5370
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.597927090779129e-05,
+      "loss": 1.3008,
+      "step": 5371
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.599714081486776e-05,
+      "loss": 1.2955,
+      "step": 5372
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.601501072194426e-05,
+      "loss": 1.3188,
+      "step": 5373
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.603288062902073e-05,
+      "loss": 1.3081,
+      "step": 5374
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.605075053609721e-05,
+      "loss": 1.2868,
+      "step": 5375
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.60686204431737e-05,
+      "loss": 1.3098,
+      "step": 5376
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.608649035025018e-05,
+      "loss": 1.2557,
+      "step": 5377
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.610436025732667e-05,
+      "loss": 1.2896,
+      "step": 5378
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.612223016440315e-05,
+      "loss": 1.2826,
+      "step": 5379
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.614010007147964e-05,
+      "loss": 1.3094,
+      "step": 5380
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.615796997855612e-05,
+      "loss": 1.3023,
+      "step": 5381
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.61758398856326e-05,
+      "loss": 1.3045,
+      "step": 5382
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.619370979270908e-05,
+      "loss": 1.3177,
+      "step": 5383
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.621157969978557e-05,
+      "loss": 1.2936,
+      "step": 5384
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.622944960686205e-05,
+      "loss": 1.2689,
+      "step": 5385
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.624731951393853e-05,
+      "loss": 1.3167,
+      "step": 5386
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.626518942101501e-05,
+      "loss": 1.3215,
+      "step": 5387
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.62830593280915e-05,
+      "loss": 1.2763,
+      "step": 5388
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.630092923516798e-05,
+      "loss": 1.2883,
+      "step": 5389
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.631879914224447e-05,
+      "loss": 1.3112,
+      "step": 5390
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.633666904932094e-05,
+      "loss": 1.2779,
+      "step": 5391
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.635453895639743e-05,
+      "loss": 1.3148,
+      "step": 5392
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.637240886347391e-05,
+      "loss": 1.268,
+      "step": 5393
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.63902787705504e-05,
+      "loss": 1.3018,
+      "step": 5394
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.640814867762688e-05,
+      "loss": 1.275,
+      "step": 5395
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.642601858470336e-05,
+      "loss": 1.3114,
+      "step": 5396
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.644388849177985e-05,
+      "loss": 1.2847,
+      "step": 5397
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.646175839885633e-05,
+      "loss": 1.2925,
+      "step": 5398
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.647962830593282e-05,
+      "loss": 1.2854,
+      "step": 5399
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.64974982130093e-05,
+      "loss": 1.2925,
+      "step": 5400
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.651536812008579e-05,
+      "loss": 1.2732,
+      "step": 5401
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.653323802716226e-05,
+      "loss": 1.2737,
+      "step": 5402
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.655110793423875e-05,
+      "loss": 1.3085,
+      "step": 5403
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.656897784131523e-05,
+      "loss": 1.2861,
+      "step": 5404
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.658684774839172e-05,
+      "loss": 1.3115,
+      "step": 5405
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.660471765546819e-05,
+      "loss": 1.2885,
+      "step": 5406
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.662258756254468e-05,
+      "loss": 1.2787,
+      "step": 5407
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.664045746962115e-05,
+      "loss": 1.3204,
+      "step": 5408
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.665832737669765e-05,
+      "loss": 1.2743,
+      "step": 5409
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.667619728377412e-05,
+      "loss": 1.2727,
+      "step": 5410
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.669406719085061e-05,
+      "loss": 1.2773,
+      "step": 5411
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.671193709792709e-05,
+      "loss": 1.2656,
+      "step": 5412
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.672980700500358e-05,
+      "loss": 1.2898,
+      "step": 5413
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.674767691208006e-05,
+      "loss": 1.2635,
+      "step": 5414
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.676554681915655e-05,
+      "loss": 1.2579,
+      "step": 5415
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.678341672623303e-05,
+      "loss": 1.3173,
+      "step": 5416
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.68012866333095e-05,
+      "loss": 1.3118,
+      "step": 5417
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.6819156540386e-05,
+      "loss": 1.304,
+      "step": 5418
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.683702644746247e-05,
+      "loss": 1.3361,
+      "step": 5419
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.685489635453896e-05,
+      "loss": 1.3308,
+      "step": 5420
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.687276626161544e-05,
+      "loss": 1.2914,
+      "step": 5421
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.689063616869193e-05,
+      "loss": 1.3123,
+      "step": 5422
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.690850607576841e-05,
+      "loss": 1.32,
+      "step": 5423
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.69263759828449e-05,
+      "loss": 1.2965,
+      "step": 5424
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.694424588992138e-05,
+      "loss": 1.2859,
+      "step": 5425
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.696211579699787e-05,
+      "loss": 1.3008,
+      "step": 5426
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.697998570407433e-05,
+      "loss": 1.2934,
+      "step": 5427
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.699785561115082e-05,
+      "loss": 1.3114,
+      "step": 5428
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.70157255182273e-05,
+      "loss": 1.3433,
+      "step": 5429
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.703359542530379e-05,
+      "loss": 1.3091,
+      "step": 5430
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.705146533238027e-05,
+      "loss": 1.3059,
+      "step": 5431
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.706933523945676e-05,
+      "loss": 1.3174,
+      "step": 5432
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.708720514653324e-05,
+      "loss": 1.2808,
+      "step": 5433
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.710507505360973e-05,
+      "loss": 1.3042,
+      "step": 5434
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.712294496068621e-05,
+      "loss": 1.2767,
+      "step": 5435
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.71408148677627e-05,
+      "loss": 1.3155,
+      "step": 5436
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.715868477483918e-05,
+      "loss": 1.2852,
+      "step": 5437
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.717655468191565e-05,
+      "loss": 1.3281,
+      "step": 5438
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.719442458899214e-05,
+      "loss": 1.2816,
+      "step": 5439
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.721229449606862e-05,
+      "loss": 1.3219,
+      "step": 5440
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.723016440314511e-05,
+      "loss": 1.2746,
+      "step": 5441
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.724803431022159e-05,
+      "loss": 1.326,
+      "step": 5442
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.726590421729808e-05,
+      "loss": 1.336,
+      "step": 5443
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.728377412437456e-05,
+      "loss": 1.3113,
+      "step": 5444
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.730164403145105e-05,
+      "loss": 1.2997,
+      "step": 5445
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.731951393852753e-05,
+      "loss": 1.2961,
+      "step": 5446
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.7337383845604e-05,
+      "loss": 1.3032,
+      "step": 5447
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.735525375268048e-05,
+      "loss": 1.3336,
+      "step": 5448
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.737312365975697e-05,
+      "loss": 1.2858,
+      "step": 5449
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.739099356683345e-05,
+      "loss": 1.3032,
+      "step": 5450
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.740886347390994e-05,
+      "loss": 1.2922,
+      "step": 5451
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.742673338098642e-05,
+      "loss": 1.2588,
+      "step": 5452
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.744460328806291e-05,
+      "loss": 1.3559,
+      "step": 5453
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.746247319513939e-05,
+      "loss": 1.298,
+      "step": 5454
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.748034310221588e-05,
+      "loss": 1.2914,
+      "step": 5455
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.749821300929235e-05,
+      "loss": 1.3141,
+      "step": 5456
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.751608291636885e-05,
+      "loss": 1.2645,
+      "step": 5457
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.753395282344532e-05,
+      "loss": 1.2795,
+      "step": 5458
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.75518227305218e-05,
+      "loss": 1.2958,
+      "step": 5459
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.756969263759829e-05,
+      "loss": 1.285,
+      "step": 5460
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.758756254467477e-05,
+      "loss": 1.2786,
+      "step": 5461
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.760543245175126e-05,
+      "loss": 1.3345,
+      "step": 5462
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.762330235882774e-05,
+      "loss": 1.2977,
+      "step": 5463
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.764117226590423e-05,
+      "loss": 1.2728,
+      "step": 5464
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.76590421729807e-05,
+      "loss": 1.2844,
+      "step": 5465
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.76769120800572e-05,
+      "loss": 1.3227,
+      "step": 5466
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.769478198713367e-05,
+      "loss": 1.2949,
+      "step": 5467
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.771265189421015e-05,
+      "loss": 1.2756,
+      "step": 5468
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.773052180128663e-05,
+      "loss": 1.2936,
+      "step": 5469
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.774839170836312e-05,
+      "loss": 1.3108,
+      "step": 5470
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.77662616154396e-05,
+      "loss": 1.3311,
+      "step": 5471
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.778413152251609e-05,
+      "loss": 1.2393,
+      "step": 5472
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.780200142959257e-05,
+      "loss": 1.3147,
+      "step": 5473
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.781987133666906e-05,
+      "loss": 1.3049,
+      "step": 5474
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.783774124374553e-05,
+      "loss": 1.3111,
+      "step": 5475
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.785561115082202e-05,
+      "loss": 1.2973,
+      "step": 5476
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.78734810578985e-05,
+      "loss": 1.3328,
+      "step": 5477
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.789135096497498e-05,
+      "loss": 1.3203,
+      "step": 5478
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.790922087205147e-05,
+      "loss": 1.3065,
+      "step": 5479
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.792709077912795e-05,
+      "loss": 1.2972,
+      "step": 5480
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.794496068620444e-05,
+      "loss": 1.2781,
+      "step": 5481
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.796283059328092e-05,
+      "loss": 1.3167,
+      "step": 5482
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.798070050035741e-05,
+      "loss": 1.2827,
+      "step": 5483
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.799857040743388e-05,
+      "loss": 1.3006,
+      "step": 5484
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.801644031451038e-05,
+      "loss": 1.2886,
+      "step": 5485
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.803431022158685e-05,
+      "loss": 1.3269,
+      "step": 5486
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.805218012866334e-05,
+      "loss": 1.3068,
+      "step": 5487
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.807005003573981e-05,
+      "loss": 1.2979,
+      "step": 5488
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.80879199428163e-05,
+      "loss": 1.2953,
+      "step": 5489
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.810578984989278e-05,
+      "loss": 1.2649,
+      "step": 5490
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.812365975696927e-05,
+      "loss": 1.3063,
+      "step": 5491
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.814152966404574e-05,
+      "loss": 1.3281,
+      "step": 5492
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.815939957112224e-05,
+      "loss": 1.2716,
+      "step": 5493
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.817726947819871e-05,
+      "loss": 1.2627,
+      "step": 5494
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.81951393852752e-05,
+      "loss": 1.2962,
+      "step": 5495
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.821300929235168e-05,
+      "loss": 1.2929,
+      "step": 5496
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.823087919942817e-05,
+      "loss": 1.327,
+      "step": 5497
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.824874910650465e-05,
+      "loss": 1.2584,
+      "step": 5498
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.826661901358113e-05,
+      "loss": 1.3073,
+      "step": 5499
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.828448892065762e-05,
+      "loss": 1.314,
+      "step": 5500
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.83023588277341e-05,
+      "loss": 1.3094,
+      "step": 5501
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.832022873481059e-05,
+      "loss": 1.302,
+      "step": 5502
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.833809864188706e-05,
+      "loss": 1.3157,
+      "step": 5503
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.835596854896355e-05,
+      "loss": 1.2988,
+      "step": 5504
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.837383845604003e-05,
+      "loss": 1.2729,
+      "step": 5505
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.839170836311652e-05,
+      "loss": 1.2917,
+      "step": 5506
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.8409578270193e-05,
+      "loss": 1.2944,
+      "step": 5507
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.842744817726949e-05,
+      "loss": 1.3124,
+      "step": 5508
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.844531808434596e-05,
+      "loss": 1.2786,
+      "step": 5509
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.846318799142245e-05,
+      "loss": 1.302,
+      "step": 5510
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.848105789849892e-05,
+      "loss": 1.3128,
+      "step": 5511
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.849892780557541e-05,
+      "loss": 1.3044,
+      "step": 5512
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.851679771265189e-05,
+      "loss": 1.2545,
+      "step": 5513
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.853466761972838e-05,
+      "loss": 1.3125,
+      "step": 5514
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.855253752680486e-05,
+      "loss": 1.3121,
+      "step": 5515
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.857040743388135e-05,
+      "loss": 1.3211,
+      "step": 5516
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.858827734095783e-05,
+      "loss": 1.2706,
+      "step": 5517
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.860614724803432e-05,
+      "loss": 1.3145,
+      "step": 5518
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.86240171551108e-05,
+      "loss": 1.3063,
+      "step": 5519
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.864188706218727e-05,
+      "loss": 1.2865,
+      "step": 5520
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.865975696926377e-05,
+      "loss": 1.2849,
+      "step": 5521
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.867762687634024e-05,
+      "loss": 1.2967,
+      "step": 5522
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.869549678341673e-05,
+      "loss": 1.2796,
+      "step": 5523
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.871336669049321e-05,
+      "loss": 1.3122,
+      "step": 5524
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.87312365975697e-05,
+      "loss": 1.2792,
+      "step": 5525
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.874910650464618e-05,
+      "loss": 1.3222,
+      "step": 5526
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.876697641172267e-05,
+      "loss": 1.3002,
+      "step": 5527
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.878484631879915e-05,
+      "loss": 1.2866,
+      "step": 5528
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.880271622587564e-05,
+      "loss": 1.327,
+      "step": 5529
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.88205861329521e-05,
+      "loss": 1.285,
+      "step": 5530
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.88384560400286e-05,
+      "loss": 1.3026,
+      "step": 5531
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.885632594710507e-05,
+      "loss": 1.3274,
+      "step": 5532
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.887419585418156e-05,
+      "loss": 1.2906,
+      "step": 5533
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.889206576125804e-05,
+      "loss": 1.2974,
+      "step": 5534
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.890993566833453e-05,
+      "loss": 1.2906,
+      "step": 5535
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.892780557541101e-05,
+      "loss": 1.3341,
+      "step": 5536
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.89456754824875e-05,
+      "loss": 1.2999,
+      "step": 5537
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.896354538956398e-05,
+      "loss": 1.3142,
+      "step": 5538
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.898141529664047e-05,
+      "loss": 1.3332,
+      "step": 5539
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.899928520371694e-05,
+      "loss": 1.2994,
+      "step": 5540
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.901715511079342e-05,
+      "loss": 1.279,
+      "step": 5541
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.903502501786991e-05,
+      "loss": 1.3226,
+      "step": 5542
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.905289492494639e-05,
+      "loss": 1.2642,
+      "step": 5543
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.907076483202288e-05,
+      "loss": 1.3012,
+      "step": 5544
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.908863473909936e-05,
+      "loss": 1.2998,
+      "step": 5545
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.910650464617585e-05,
+      "loss": 1.3192,
+      "step": 5546
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.912437455325233e-05,
+      "loss": 1.3067,
+      "step": 5547
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.914224446032882e-05,
+      "loss": 1.2631,
+      "step": 5548
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.91601143674053e-05,
+      "loss": 1.305,
+      "step": 5549
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.917798427448177e-05,
+      "loss": 1.2908,
+      "step": 5550
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.919585418155825e-05,
+      "loss": 1.2689,
+      "step": 5551
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.921372408863474e-05,
+      "loss": 1.2918,
+      "step": 5552
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.923159399571122e-05,
+      "loss": 1.3411,
+      "step": 5553
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.924946390278771e-05,
+      "loss": 1.2833,
+      "step": 5554
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.926733380986419e-05,
+      "loss": 1.2831,
+      "step": 5555
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.928520371694068e-05,
+      "loss": 1.315,
+      "step": 5556
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.930307362401716e-05,
+      "loss": 1.2827,
+      "step": 5557
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.932094353109365e-05,
+      "loss": 1.2607,
+      "step": 5558
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.933881343817014e-05,
+      "loss": 1.3224,
+      "step": 5559
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.93566833452466e-05,
+      "loss": 1.2688,
+      "step": 5560
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.937455325232309e-05,
+      "loss": 1.3102,
+      "step": 5561
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.939242315939957e-05,
+      "loss": 1.2999,
+      "step": 5562
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.941029306647606e-05,
+      "loss": 1.2984,
+      "step": 5563
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.942816297355254e-05,
+      "loss": 1.3155,
+      "step": 5564
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.944603288062903e-05,
+      "loss": 1.3346,
+      "step": 5565
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.94639027877055e-05,
+      "loss": 1.2999,
+      "step": 5566
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.9481772694782e-05,
+      "loss": 1.2948,
+      "step": 5567
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.949964260185847e-05,
+      "loss": 1.2886,
+      "step": 5568
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.951751250893497e-05,
+      "loss": 1.2998,
+      "step": 5569
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.953538241601144e-05,
+      "loss": 1.3158,
+      "step": 5570
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.955325232308792e-05,
+      "loss": 1.2774,
+      "step": 5571
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.95711222301644e-05,
+      "loss": 1.3064,
+      "step": 5572
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.958899213724089e-05,
+      "loss": 1.3023,
+      "step": 5573
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.960686204431737e-05,
+      "loss": 1.2908,
+      "step": 5574
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.962473195139386e-05,
+      "loss": 1.3306,
+      "step": 5575
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.964260185847033e-05,
+      "loss": 1.2745,
+      "step": 5576
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.966047176554683e-05,
+      "loss": 1.2898,
+      "step": 5577
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.967834167262332e-05,
+      "loss": 1.2844,
+      "step": 5578
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.96962115796998e-05,
+      "loss": 1.3085,
+      "step": 5579
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.971408148677628e-05,
+      "loss": 1.2985,
+      "step": 5580
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.973195139385275e-05,
+      "loss": 1.2733,
+      "step": 5581
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.974982130092924e-05,
+      "loss": 1.2879,
+      "step": 5582
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.976769120800572e-05,
+      "loss": 1.2823,
+      "step": 5583
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.978556111508221e-05,
+      "loss": 1.2877,
+      "step": 5584
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.980343102215869e-05,
+      "loss": 1.2597,
+      "step": 5585
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.982130092923518e-05,
+      "loss": 1.2749,
+      "step": 5586
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.983917083631165e-05,
+      "loss": 1.2475,
+      "step": 5587
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.985704074338814e-05,
+      "loss": 1.2982,
+      "step": 5588
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.987491065046462e-05,
+      "loss": 1.2946,
+      "step": 5589
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.989278055754111e-05,
+      "loss": 1.3077,
+      "step": 5590
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.991065046461758e-05,
+      "loss": 1.2887,
+      "step": 5591
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.992852037169407e-05,
+      "loss": 1.3199,
+      "step": 5592
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.994639027877055e-05,
+      "loss": 1.3254,
+      "step": 5593
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.996426018584704e-05,
+      "loss": 1.3198,
+      "step": 5594
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.998213009292351e-05,
+      "loss": 1.2733,
+      "step": 5595
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 0.0001,
+      "loss": 1.3207,
+      "step": 5596
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.999999999246133e-05,
+      "loss": 1.294,
+      "step": 5597
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.999999996984528e-05,
+      "loss": 1.2641,
+      "step": 5598
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.999999993215188e-05,
+      "loss": 1.272,
+      "step": 5599
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.99999998793811e-05,
+      "loss": 1.2872,
+      "step": 5600
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.999999981153296e-05,
+      "loss": 1.2875,
+      "step": 5601
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.999999972860748e-05,
+      "loss": 1.3093,
+      "step": 5602
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.99999996306046e-05,
+      "loss": 1.3181,
+      "step": 5603
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.999999951752439e-05,
+      "loss": 1.2662,
+      "step": 5604
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.99999993893668e-05,
+      "loss": 1.2894,
+      "step": 5605
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.999999924613183e-05,
+      "loss": 1.3211,
+      "step": 5606
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.999999908781953e-05,
+      "loss": 1.2802,
+      "step": 5607
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.999999891442985e-05,
+      "loss": 1.2742,
+      "step": 5608
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.999999872596281e-05,
+      "loss": 1.2859,
+      "step": 5609
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.999999852241841e-05,
+      "loss": 1.2898,
+      "step": 5610
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.999999830379665e-05,
+      "loss": 1.321,
+      "step": 5611
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.999999807009752e-05,
+      "loss": 1.2931,
+      "step": 5612
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.999999782132103e-05,
+      "loss": 1.2982,
+      "step": 5613
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.999999755746719e-05,
+      "loss": 1.3084,
+      "step": 5614
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.999999727853597e-05,
+      "loss": 1.2965,
+      "step": 5615
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.99999969845274e-05,
+      "loss": 1.3228,
+      "step": 5616
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.999999667544144e-05,
+      "loss": 1.3159,
+      "step": 5617
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.999999635127815e-05,
+      "loss": 1.2973,
+      "step": 5618
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.999999601203748e-05,
+      "loss": 1.2857,
+      "step": 5619
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.999999565771944e-05,
+      "loss": 1.3036,
+      "step": 5620
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.999999528832406e-05,
+      "loss": 1.3179,
+      "step": 5621
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.999999490385132e-05,
+      "loss": 1.318,
+      "step": 5622
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.999999450430119e-05,
+      "loss": 1.3155,
+      "step": 5623
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.999999408967373e-05,
+      "loss": 1.2799,
+      "step": 5624
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.999999365996891e-05,
+      "loss": 1.2934,
+      "step": 5625
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.999999321518669e-05,
+      "loss": 1.3127,
+      "step": 5626
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.999999275532714e-05,
+      "loss": 1.2652,
+      "step": 5627
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.999999228039024e-05,
+      "loss": 1.2956,
+      "step": 5628
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.999999179037594e-05,
+      "loss": 1.2956,
+      "step": 5629
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.99999912852843e-05,
+      "loss": 1.3146,
+      "step": 5630
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.99999907651153e-05,
+      "loss": 1.3046,
+      "step": 5631
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.999999022986894e-05,
+      "loss": 1.2924,
+      "step": 5632
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.999998967954523e-05,
+      "loss": 1.2782,
+      "step": 5633
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.999998911414415e-05,
+      "loss": 1.2764,
+      "step": 5634
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.999998853366571e-05,
+      "loss": 1.2748,
+      "step": 5635
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.999998793810989e-05,
+      "loss": 1.2897,
+      "step": 5636
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.999998732747674e-05,
+      "loss": 1.3114,
+      "step": 5637
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.999998670176622e-05,
+      "loss": 1.3305,
+      "step": 5638
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.999998606097833e-05,
+      "loss": 1.3346,
+      "step": 5639
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.99999854051131e-05,
+      "loss": 1.291,
+      "step": 5640
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.999998473417051e-05,
+      "loss": 1.2679,
+      "step": 5641
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.999998404815054e-05,
+      "loss": 1.2864,
+      "step": 5642
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.999998334705323e-05,
+      "loss": 1.295,
+      "step": 5643
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.999998263087856e-05,
+      "loss": 1.3069,
+      "step": 5644
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.999998189962654e-05,
+      "loss": 1.2895,
+      "step": 5645
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.999998115329713e-05,
+      "loss": 1.3181,
+      "step": 5646
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.999998039189041e-05,
+      "loss": 1.289,
+      "step": 5647
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.999997961540629e-05,
+      "loss": 1.3057,
+      "step": 5648
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.999997882384483e-05,
+      "loss": 1.301,
+      "step": 5649
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.999997801720602e-05,
+      "loss": 1.2559,
+      "step": 5650
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.999997719548984e-05,
+      "loss": 1.2948,
+      "step": 5651
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.999997635869631e-05,
+      "loss": 1.2566,
+      "step": 5652
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.999997550682543e-05,
+      "loss": 1.2917,
+      "step": 5653
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.999997463987718e-05,
+      "loss": 1.2983,
+      "step": 5654
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.999997375785159e-05,
+      "loss": 1.2946,
+      "step": 5655
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.999997286074863e-05,
+      "loss": 1.3003,
+      "step": 5656
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.999997194856833e-05,
+      "loss": 1.2964,
+      "step": 5657
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.999997102131067e-05,
+      "loss": 1.2721,
+      "step": 5658
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.999997007897565e-05,
+      "loss": 1.2774,
+      "step": 5659
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.999996912156328e-05,
+      "loss": 1.3011,
+      "step": 5660
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.999996814907355e-05,
+      "loss": 1.3001,
+      "step": 5661
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.999996716150647e-05,
+      "loss": 1.3007,
+      "step": 5662
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.999996615886204e-05,
+      "loss": 1.3055,
+      "step": 5663
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.999996514114026e-05,
+      "loss": 1.3002,
+      "step": 5664
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.999996410834111e-05,
+      "loss": 1.3068,
+      "step": 5665
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.999996306046462e-05,
+      "loss": 1.3252,
+      "step": 5666
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.999996199751079e-05,
+      "loss": 1.3082,
+      "step": 5667
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.999996091947958e-05,
+      "loss": 1.2851,
+      "step": 5668
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.999995982637104e-05,
+      "loss": 1.31,
+      "step": 5669
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.999995871818515e-05,
+      "loss": 1.2905,
+      "step": 5670
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.999995759492189e-05,
+      "loss": 1.324,
+      "step": 5671
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.999995645658128e-05,
+      "loss": 1.302,
+      "step": 5672
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.999995530316335e-05,
+      "loss": 1.2837,
+      "step": 5673
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.999995413466805e-05,
+      "loss": 1.2911,
+      "step": 5674
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.99999529510954e-05,
+      "loss": 1.2812,
+      "step": 5675
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.999995175244541e-05,
+      "loss": 1.2799,
+      "step": 5676
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.999995053871807e-05,
+      "loss": 1.3035,
+      "step": 5677
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.999994930991337e-05,
+      "loss": 1.2725,
+      "step": 5678
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.999994806603132e-05,
+      "loss": 1.2713,
+      "step": 5679
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.999994680707194e-05,
+      "loss": 1.2824,
+      "step": 5680
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.99999455330352e-05,
+      "loss": 1.325,
+      "step": 5681
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.999994424392112e-05,
+      "loss": 1.2906,
+      "step": 5682
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.999994293972969e-05,
+      "loss": 1.3299,
+      "step": 5683
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.999994162046091e-05,
+      "loss": 1.2917,
+      "step": 5684
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.999994028611479e-05,
+      "loss": 1.3311,
+      "step": 5685
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.999993893669133e-05,
+      "loss": 1.3065,
+      "step": 5686
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.99999375721905e-05,
+      "loss": 1.3161,
+      "step": 5687
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.999993619261235e-05,
+      "loss": 1.3279,
+      "step": 5688
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.999993479795686e-05,
+      "loss": 1.3137,
+      "step": 5689
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.999993338822403e-05,
+      "loss": 1.2986,
+      "step": 5690
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.999993196341383e-05,
+      "loss": 1.3037,
+      "step": 5691
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.999993052352629e-05,
+      "loss": 1.3001,
+      "step": 5692
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.999992906856143e-05,
+      "loss": 1.3195,
+      "step": 5693
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.999992759851921e-05,
+      "loss": 1.3633,
+      "step": 5694
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.999992611339966e-05,
+      "loss": 1.332,
+      "step": 5695
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.999992461320275e-05,
+      "loss": 1.3103,
+      "step": 5696
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.999992309792853e-05,
+      "loss": 1.3126,
+      "step": 5697
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.999992156757695e-05,
+      "loss": 1.2951,
+      "step": 5698
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.999992002214804e-05,
+      "loss": 1.286,
+      "step": 5699
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.999991846164179e-05,
+      "loss": 1.3108,
+      "step": 5700
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.999991688605818e-05,
+      "loss": 1.3281,
+      "step": 5701
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.999991529539726e-05,
+      "loss": 1.3484,
+      "step": 5702
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.999991368965898e-05,
+      "loss": 1.2888,
+      "step": 5703
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.999991206884339e-05,
+      "loss": 1.2737,
+      "step": 5704
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.999991043295043e-05,
+      "loss": 1.2774,
+      "step": 5705
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.999990878198016e-05,
+      "loss": 1.3244,
+      "step": 5706
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.999990711593254e-05,
+      "loss": 1.3356,
+      "step": 5707
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.999990543480759e-05,
+      "loss": 1.311,
+      "step": 5708
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.99999037386053e-05,
+      "loss": 1.3122,
+      "step": 5709
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.999990202732569e-05,
+      "loss": 1.2778,
+      "step": 5710
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.999990030096873e-05,
+      "loss": 1.3046,
+      "step": 5711
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.999989855953444e-05,
+      "loss": 1.2938,
+      "step": 5712
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.999989680302282e-05,
+      "loss": 1.3338,
+      "step": 5713
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.999989503143387e-05,
+      "loss": 1.3139,
+      "step": 5714
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.99998932447676e-05,
+      "loss": 1.3141,
+      "step": 5715
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.999989144302397e-05,
+      "loss": 1.3204,
+      "step": 5716
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.999988962620303e-05,
+      "loss": 1.2831,
+      "step": 5717
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.999988779430476e-05,
+      "loss": 1.2838,
+      "step": 5718
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.999988594732916e-05,
+      "loss": 1.2951,
+      "step": 5719
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.999988408527624e-05,
+      "loss": 1.2749,
+      "step": 5720
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.999988220814596e-05,
+      "loss": 1.2856,
+      "step": 5721
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.999988031593838e-05,
+      "loss": 1.3057,
+      "step": 5722
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.999987840865346e-05,
+      "loss": 1.2675,
+      "step": 5723
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.999987648629121e-05,
+      "loss": 1.2712,
+      "step": 5724
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.999987454885165e-05,
+      "loss": 1.3161,
+      "step": 5725
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.999987259633475e-05,
+      "loss": 1.3373,
+      "step": 5726
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.999987062874054e-05,
+      "loss": 1.318,
+      "step": 5727
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.999986864606899e-05,
+      "loss": 1.2969,
+      "step": 5728
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.999986664832013e-05,
+      "loss": 1.3157,
+      "step": 5729
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.999986463549393e-05,
+      "loss": 1.2852,
+      "step": 5730
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.999986260759042e-05,
+      "loss": 1.269,
+      "step": 5731
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.999986056460959e-05,
+      "loss": 1.2766,
+      "step": 5732
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.999985850655144e-05,
+      "loss": 1.336,
+      "step": 5733
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.999985643341596e-05,
+      "loss": 1.3015,
+      "step": 5734
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.999985434520317e-05,
+      "loss": 1.2563,
+      "step": 5735
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.999985224191306e-05,
+      "loss": 1.278,
+      "step": 5736
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.999985012354562e-05,
+      "loss": 1.28,
+      "step": 5737
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.999984799010087e-05,
+      "loss": 1.2899,
+      "step": 5738
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.99998458415788e-05,
+      "loss": 1.2739,
+      "step": 5739
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.999984367797941e-05,
+      "loss": 1.3096,
+      "step": 5740
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.999984149930271e-05,
+      "loss": 1.2967,
+      "step": 5741
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.99998393055487e-05,
+      "loss": 1.312,
+      "step": 5742
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.999983709671736e-05,
+      "loss": 1.3112,
+      "step": 5743
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.999983487280873e-05,
+      "loss": 1.3262,
+      "step": 5744
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.999983263382276e-05,
+      "loss": 1.3172,
+      "step": 5745
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.999983037975949e-05,
+      "loss": 1.2904,
+      "step": 5746
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.99998281106189e-05,
+      "loss": 1.3136,
+      "step": 5747
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.999982582640101e-05,
+      "loss": 1.3264,
+      "step": 5748
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.999982352710579e-05,
+      "loss": 1.284,
+      "step": 5749
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.999982121273328e-05,
+      "loss": 1.258,
+      "step": 5750
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.999981888328346e-05,
+      "loss": 1.2797,
+      "step": 5751
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.999981653875633e-05,
+      "loss": 1.2728,
+      "step": 5752
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.999981417915189e-05,
+      "loss": 1.2595,
+      "step": 5753
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.999981180447014e-05,
+      "loss": 1.3171,
+      "step": 5754
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.999980941471108e-05,
+      "loss": 1.3035,
+      "step": 5755
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.999980700987473e-05,
+      "loss": 1.291,
+      "step": 5756
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.999980458996105e-05,
+      "loss": 1.3047,
+      "step": 5757
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.999980215497008e-05,
+      "loss": 1.3291,
+      "step": 5758
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.99997997049018e-05,
+      "loss": 1.2986,
+      "step": 5759
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.999979723975623e-05,
+      "loss": 1.3006,
+      "step": 5760
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.999979475953335e-05,
+      "loss": 1.3341,
+      "step": 5761
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.999979226423317e-05,
+      "loss": 1.2855,
+      "step": 5762
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.999978975385569e-05,
+      "loss": 1.2936,
+      "step": 5763
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.99997872284009e-05,
+      "loss": 1.2902,
+      "step": 5764
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.999978468786883e-05,
+      "loss": 1.2859,
+      "step": 5765
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.999978213225945e-05,
+      "loss": 1.2838,
+      "step": 5766
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.999977956157277e-05,
+      "loss": 1.2915,
+      "step": 5767
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.99997769758088e-05,
+      "loss": 1.3056,
+      "step": 5768
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.999977437496753e-05,
+      "loss": 1.2891,
+      "step": 5769
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.999977175904896e-05,
+      "loss": 1.2986,
+      "step": 5770
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.99997691280531e-05,
+      "loss": 1.314,
+      "step": 5771
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.999976648197995e-05,
+      "loss": 1.2986,
+      "step": 5772
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.999976382082952e-05,
+      "loss": 1.2737,
+      "step": 5773
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.999976114460176e-05,
+      "loss": 1.3266,
+      "step": 5774
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.999975845329674e-05,
+      "loss": 1.3154,
+      "step": 5775
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.999975574691442e-05,
+      "loss": 1.2543,
+      "step": 5776
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.99997530254548e-05,
+      "loss": 1.2535,
+      "step": 5777
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.999975028891792e-05,
+      "loss": 1.2894,
+      "step": 5778
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.999974753730375e-05,
+      "loss": 1.3064,
+      "step": 5779
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.999974477061226e-05,
+      "loss": 1.3144,
+      "step": 5780
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.999974198884351e-05,
+      "loss": 1.2983,
+      "step": 5781
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.999973919199746e-05,
+      "loss": 1.3027,
+      "step": 5782
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.999973638007414e-05,
+      "loss": 1.2968,
+      "step": 5783
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.999973355307352e-05,
+      "loss": 1.2888,
+      "step": 5784
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.999973071099562e-05,
+      "loss": 1.3276,
+      "step": 5785
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.999972785384045e-05,
+      "loss": 1.2782,
+      "step": 5786
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.9999724981608e-05,
+      "loss": 1.3089,
+      "step": 5787
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.999972209429826e-05,
+      "loss": 1.3159,
+      "step": 5788
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.999971919191124e-05,
+      "loss": 1.3043,
+      "step": 5789
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.999971627444694e-05,
+      "loss": 1.2829,
+      "step": 5790
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.999971334190536e-05,
+      "loss": 1.3062,
+      "step": 5791
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.999971039428652e-05,
+      "loss": 1.3187,
+      "step": 5792
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.999970743159039e-05,
+      "loss": 1.3026,
+      "step": 5793
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.999970445381699e-05,
+      "loss": 1.3075,
+      "step": 5794
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.999970146096631e-05,
+      "loss": 1.31,
+      "step": 5795
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.999969845303836e-05,
+      "loss": 1.2758,
+      "step": 5796
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.999969543003314e-05,
+      "loss": 1.2936,
+      "step": 5797
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.999969239195065e-05,
+      "loss": 1.3225,
+      "step": 5798
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.999968933879089e-05,
+      "loss": 1.2936,
+      "step": 5799
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.999968627055386e-05,
+      "loss": 1.3562,
+      "step": 5800
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.999968318723956e-05,
+      "loss": 1.3014,
+      "step": 5801
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.999968008884798e-05,
+      "loss": 1.3119,
+      "step": 5802
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.999967697537916e-05,
+      "loss": 1.2925,
+      "step": 5803
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.999967384683304e-05,
+      "loss": 1.2942,
+      "step": 5804
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.999967070320968e-05,
+      "loss": 1.3113,
+      "step": 5805
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.999966754450905e-05,
+      "loss": 1.3255,
+      "step": 5806
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.999966437073116e-05,
+      "loss": 1.291,
+      "step": 5807
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.9999661181876e-05,
+      "loss": 1.2844,
+      "step": 5808
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.999965797794359e-05,
+      "loss": 1.3315,
+      "step": 5809
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.999965475893391e-05,
+      "loss": 1.2746,
+      "step": 5810
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.999965152484696e-05,
+      "loss": 1.2853,
+      "step": 5811
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.999964827568277e-05,
+      "loss": 1.2817,
+      "step": 5812
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.999964501144133e-05,
+      "loss": 1.2894,
+      "step": 5813
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.999964173212262e-05,
+      "loss": 1.2928,
+      "step": 5814
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.999963843772665e-05,
+      "loss": 1.3206,
+      "step": 5815
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.999963512825343e-05,
+      "loss": 1.3038,
+      "step": 5816
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.999963180370296e-05,
+      "loss": 1.283,
+      "step": 5817
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.999962846407524e-05,
+      "loss": 1.2848,
+      "step": 5818
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.999962510937026e-05,
+      "loss": 1.2819,
+      "step": 5819
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.999962173958804e-05,
+      "loss": 1.2897,
+      "step": 5820
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.999961835472858e-05,
+      "loss": 1.3252,
+      "step": 5821
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.999961495479186e-05,
+      "loss": 1.298,
+      "step": 5822
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.999961153977789e-05,
+      "loss": 1.3137,
+      "step": 5823
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.999960810968667e-05,
+      "loss": 1.3033,
+      "step": 5824
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.99996046645182e-05,
+      "loss": 1.319,
+      "step": 5825
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.999960120427251e-05,
+      "loss": 1.2741,
+      "step": 5826
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.999959772894955e-05,
+      "loss": 1.3102,
+      "step": 5827
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.999959423854936e-05,
+      "loss": 1.3043,
+      "step": 5828
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.999959073307195e-05,
+      "loss": 1.3108,
+      "step": 5829
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.999958721251727e-05,
+      "loss": 1.3103,
+      "step": 5830
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.999958367688536e-05,
+      "loss": 1.3147,
+      "step": 5831
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.999958012617623e-05,
+      "loss": 1.3397,
+      "step": 5832
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.999957656038985e-05,
+      "loss": 1.3386,
+      "step": 5833
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.999957297952623e-05,
+      "loss": 1.293,
+      "step": 5834
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.999956938358537e-05,
+      "loss": 1.2941,
+      "step": 5835
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.999956577256728e-05,
+      "loss": 1.3278,
+      "step": 5836
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.999956214647197e-05,
+      "loss": 1.3062,
+      "step": 5837
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.999955850529941e-05,
+      "loss": 1.3098,
+      "step": 5838
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.999955484904964e-05,
+      "loss": 1.303,
+      "step": 5839
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.999955117772263e-05,
+      "loss": 1.3,
+      "step": 5840
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.99995474913184e-05,
+      "loss": 1.3121,
+      "step": 5841
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.999954378983693e-05,
+      "loss": 1.2816,
+      "step": 5842
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.999954007327826e-05,
+      "loss": 1.3189,
+      "step": 5843
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.999953634164233e-05,
+      "loss": 1.3322,
+      "step": 5844
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.999953259492921e-05,
+      "loss": 1.3216,
+      "step": 5845
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.999952883313884e-05,
+      "loss": 1.2767,
+      "step": 5846
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.999952505627126e-05,
+      "loss": 1.333,
+      "step": 5847
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.999952126432645e-05,
+      "loss": 1.2841,
+      "step": 5848
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.999951745730444e-05,
+      "loss": 1.2969,
+      "step": 5849
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.999951363520521e-05,
+      "loss": 1.2971,
+      "step": 5850
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.999950979802875e-05,
+      "loss": 1.3265,
+      "step": 5851
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.999950594577508e-05,
+      "loss": 1.3365,
+      "step": 5852
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.99995020784442e-05,
+      "loss": 1.2749,
+      "step": 5853
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.99994981960361e-05,
+      "loss": 1.2848,
+      "step": 5854
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.99994942985508e-05,
+      "loss": 1.2809,
+      "step": 5855
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.999949038598828e-05,
+      "loss": 1.3113,
+      "step": 5856
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.999948645834855e-05,
+      "loss": 1.2791,
+      "step": 5857
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.999948251563162e-05,
+      "loss": 1.3042,
+      "step": 5858
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.999947855783747e-05,
+      "loss": 1.2678,
+      "step": 5859
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.999947458496612e-05,
+      "loss": 1.2885,
+      "step": 5860
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.999947059701756e-05,
+      "loss": 1.3032,
+      "step": 5861
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.999946659399181e-05,
+      "loss": 1.3378,
+      "step": 5862
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.999946257588885e-05,
+      "loss": 1.3027,
+      "step": 5863
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.999945854270869e-05,
+      "loss": 1.3171,
+      "step": 5864
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.999945449445131e-05,
+      "loss": 1.2747,
+      "step": 5865
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.999945043111676e-05,
+      "loss": 1.333,
+      "step": 5866
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.9999446352705e-05,
+      "loss": 1.3292,
+      "step": 5867
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.999944225921604e-05,
+      "loss": 1.2942,
+      "step": 5868
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.99994381506499e-05,
+      "loss": 1.3069,
+      "step": 5869
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.999943402700656e-05,
+      "loss": 1.301,
+      "step": 5870
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.999942988828602e-05,
+      "loss": 1.3268,
+      "step": 5871
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.99994257344883e-05,
+      "loss": 1.2998,
+      "step": 5872
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.999942156561338e-05,
+      "loss": 1.2854,
+      "step": 5873
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.999941738166127e-05,
+      "loss": 1.3143,
+      "step": 5874
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.999941318263198e-05,
+      "loss": 1.3085,
+      "step": 5875
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.999940896852551e-05,
+      "loss": 1.2961,
+      "step": 5876
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.999940473934184e-05,
+      "loss": 1.2826,
+      "step": 5877
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.999940049508099e-05,
+      "loss": 1.3165,
+      "step": 5878
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.999939623574296e-05,
+      "loss": 1.2853,
+      "step": 5879
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.999939196132775e-05,
+      "loss": 1.2344,
+      "step": 5880
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.999938767183535e-05,
+      "loss": 1.2966,
+      "step": 5881
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.999938336726579e-05,
+      "loss": 1.2869,
+      "step": 5882
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.999937904761905e-05,
+      "loss": 1.3161,
+      "step": 5883
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.999937471289511e-05,
+      "loss": 1.2652,
+      "step": 5884
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.999937036309401e-05,
+      "loss": 1.3001,
+      "step": 5885
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.999936599821575e-05,
+      "loss": 1.2747,
+      "step": 5886
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.99993616182603e-05,
+      "loss": 1.3071,
+      "step": 5887
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.999935722322769e-05,
+      "loss": 1.2779,
+      "step": 5888
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.99993528131179e-05,
+      "loss": 1.332,
+      "step": 5889
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.999934838793095e-05,
+      "loss": 1.3112,
+      "step": 5890
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.999934394766683e-05,
+      "loss": 1.2939,
+      "step": 5891
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.999933949232554e-05,
+      "loss": 1.3407,
+      "step": 5892
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.999933502190711e-05,
+      "loss": 1.2668,
+      "step": 5893
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.999933053641149e-05,
+      "loss": 1.3232,
+      "step": 5894
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.999932603583872e-05,
+      "loss": 1.3235,
+      "step": 5895
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.999932152018879e-05,
+      "loss": 1.3182,
+      "step": 5896
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.99993169894617e-05,
+      "loss": 1.3176,
+      "step": 5897
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.999931244365744e-05,
+      "loss": 1.3364,
+      "step": 5898
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.999930788277605e-05,
+      "loss": 1.3,
+      "step": 5899
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.999930330681748e-05,
+      "loss": 1.2978,
+      "step": 5900
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.999929871578178e-05,
+      "loss": 1.2739,
+      "step": 5901
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.99992941096689e-05,
+      "loss": 1.3052,
+      "step": 5902
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.99992894884789e-05,
+      "loss": 1.3246,
+      "step": 5903
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.999928485221173e-05,
+      "loss": 1.2941,
+      "step": 5904
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.999928020086742e-05,
+      "loss": 1.3605,
+      "step": 5905
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.999927553444597e-05,
+      "loss": 1.3249,
+      "step": 5906
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.999927085294736e-05,
+      "loss": 1.2719,
+      "step": 5907
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.999926615637162e-05,
+      "loss": 1.3072,
+      "step": 5908
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.999926144471874e-05,
+      "loss": 1.2991,
+      "step": 5909
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.999925671798871e-05,
+      "loss": 1.2662,
+      "step": 5910
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.999925197618154e-05,
+      "loss": 1.2893,
+      "step": 5911
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.999924721929723e-05,
+      "loss": 1.3006,
+      "step": 5912
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.99992424473358e-05,
+      "loss": 1.2718,
+      "step": 5913
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.999923766029722e-05,
+      "loss": 1.3168,
+      "step": 5914
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.999923285818151e-05,
+      "loss": 1.303,
+      "step": 5915
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.999922804098867e-05,
+      "loss": 1.3119,
+      "step": 5916
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.999922320871871e-05,
+      "loss": 1.2972,
+      "step": 5917
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.99992183613716e-05,
+      "loss": 1.2989,
+      "step": 5918
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.999921349894736e-05,
+      "loss": 1.3145,
+      "step": 5919
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.999920862144602e-05,
+      "loss": 1.3156,
+      "step": 5920
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.999920372886755e-05,
+      "loss": 1.2834,
+      "step": 5921
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.999919882121194e-05,
+      "loss": 1.343,
+      "step": 5922
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.999919389847921e-05,
+      "loss": 1.3368,
+      "step": 5923
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.999918896066938e-05,
+      "loss": 1.3174,
+      "step": 5924
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.999918400778242e-05,
+      "loss": 1.3385,
+      "step": 5925
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.999917903981833e-05,
+      "loss": 1.2865,
+      "step": 5926
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.999917405677715e-05,
+      "loss": 1.2954,
+      "step": 5927
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.999916905865883e-05,
+      "loss": 1.2908,
+      "step": 5928
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.999916404546341e-05,
+      "loss": 1.2772,
+      "step": 5929
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.999915901719088e-05,
+      "loss": 1.2893,
+      "step": 5930
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.999915397384123e-05,
+      "loss": 1.2702,
+      "step": 5931
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.999914891541448e-05,
+      "loss": 1.2947,
+      "step": 5932
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.999914384191062e-05,
+      "loss": 1.3014,
+      "step": 5933
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.999913875332967e-05,
+      "loss": 1.2605,
+      "step": 5934
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.99991336496716e-05,
+      "loss": 1.3116,
+      "step": 5935
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.999912853093643e-05,
+      "loss": 1.2574,
+      "step": 5936
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.999912339712416e-05,
+      "loss": 1.272,
+      "step": 5937
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.99991182482348e-05,
+      "loss": 1.295,
+      "step": 5938
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.999911308426834e-05,
+      "loss": 1.2827,
+      "step": 5939
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.999910790522477e-05,
+      "loss": 1.3126,
+      "step": 5940
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.999910271110412e-05,
+      "loss": 1.327,
+      "step": 5941
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.999909750190637e-05,
+      "loss": 1.2866,
+      "step": 5942
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.999909227763153e-05,
+      "loss": 1.3051,
+      "step": 5943
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.999908703827961e-05,
+      "loss": 1.3099,
+      "step": 5944
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.99990817838506e-05,
+      "loss": 1.3153,
+      "step": 5945
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.999907651434449e-05,
+      "loss": 1.3022,
+      "step": 5946
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.999907122976131e-05,
+      "loss": 1.2668,
+      "step": 5947
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.999906593010105e-05,
+      "loss": 1.2821,
+      "step": 5948
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.999906061536368e-05,
+      "loss": 1.28,
+      "step": 5949
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.999905528554925e-05,
+      "loss": 1.2776,
+      "step": 5950
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.999904994065775e-05,
+      "loss": 1.306,
+      "step": 5951
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.999904458068916e-05,
+      "loss": 1.2776,
+      "step": 5952
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.99990392056435e-05,
+      "loss": 1.2905,
+      "step": 5953
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.999903381552075e-05,
+      "loss": 1.2531,
+      "step": 5954
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.999902841032096e-05,
+      "loss": 1.2674,
+      "step": 5955
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.999902299004408e-05,
+      "loss": 1.2705,
+      "step": 5956
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.999901755469014e-05,
+      "loss": 1.3021,
+      "step": 5957
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.999901210425913e-05,
+      "loss": 1.2681,
+      "step": 5958
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.999900663875103e-05,
+      "loss": 1.3035,
+      "step": 5959
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.999900115816589e-05,
+      "loss": 1.3206,
+      "step": 5960
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.99989956625037e-05,
+      "loss": 1.2823,
+      "step": 5961
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.999899015176444e-05,
+      "loss": 1.3197,
+      "step": 5962
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.999898462594812e-05,
+      "loss": 1.3006,
+      "step": 5963
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.999897908505473e-05,
+      "loss": 1.3134,
+      "step": 5964
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.999897352908429e-05,
+      "loss": 1.3276,
+      "step": 5965
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.99989679580368e-05,
+      "loss": 1.342,
+      "step": 5966
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.999896237191226e-05,
+      "loss": 1.3074,
+      "step": 5967
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.999895677071067e-05,
+      "loss": 1.3229,
+      "step": 5968
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.999895115443203e-05,
+      "loss": 1.2829,
+      "step": 5969
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.999894552307635e-05,
+      "loss": 1.3007,
+      "step": 5970
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.999893987664362e-05,
+      "loss": 1.2944,
+      "step": 5971
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.999893421513384e-05,
+      "loss": 1.2686,
+      "step": 5972
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.999892853854702e-05,
+      "loss": 1.3143,
+      "step": 5973
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.999892284688317e-05,
+      "loss": 1.3137,
+      "step": 5974
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.999891714014227e-05,
+      "loss": 1.2593,
+      "step": 5975
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.999891141832434e-05,
+      "loss": 1.3097,
+      "step": 5976
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.999890568142936e-05,
+      "loss": 1.3138,
+      "step": 5977
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.999889992945737e-05,
+      "loss": 1.3321,
+      "step": 5978
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.999889416240833e-05,
+      "loss": 1.3033,
+      "step": 5979
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.999888838028226e-05,
+      "loss": 1.3173,
+      "step": 5980
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.999888258307918e-05,
+      "loss": 1.3094,
+      "step": 5981
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.999887677079907e-05,
+      "loss": 1.2882,
+      "step": 5982
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.999887094344192e-05,
+      "loss": 1.3024,
+      "step": 5983
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.999886510100776e-05,
+      "loss": 1.2901,
+      "step": 5984
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.999885924349657e-05,
+      "loss": 1.3045,
+      "step": 5985
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.999885337090838e-05,
+      "loss": 1.2848,
+      "step": 5986
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.999884748324316e-05,
+      "loss": 1.2777,
+      "step": 5987
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.99988415805009e-05,
+      "loss": 1.3065,
+      "step": 5988
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.999883566268166e-05,
+      "loss": 1.3082,
+      "step": 5989
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.99988297297854e-05,
+      "loss": 1.3109,
+      "step": 5990
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.999882378181212e-05,
+      "loss": 1.291,
+      "step": 5991
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.999881781876184e-05,
+      "loss": 1.3019,
+      "step": 5992
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.999881184063455e-05,
+      "loss": 1.2696,
+      "step": 5993
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.999880584743025e-05,
+      "loss": 1.3277,
+      "step": 5994
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.999879983914896e-05,
+      "loss": 1.3011,
+      "step": 5995
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.999879381579066e-05,
+      "loss": 1.3188,
+      "step": 5996
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.999878777735537e-05,
+      "loss": 1.3035,
+      "step": 5997
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.999878172384307e-05,
+      "loss": 1.2663,
+      "step": 5998
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.999877565525377e-05,
+      "loss": 1.2928,
+      "step": 5999
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.999876957158749e-05,
+      "loss": 1.2948,
+      "step": 6000
+    },
+    {
+      "epoch": 0.97,
+      "eval_loss": 1.3237231969833374,
+      "eval_runtime": 61.5644,
+      "eval_samples_per_second": 162.432,
+      "eval_steps_per_second": 0.227,
+      "step": 6000
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.99987634728442e-05,
+      "loss": 1.2833,
+      "step": 6001
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.999875735902394e-05,
+      "loss": 1.2633,
+      "step": 6002
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.999875123012669e-05,
+      "loss": 1.3416,
+      "step": 6003
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.999874508615243e-05,
+      "loss": 1.3081,
+      "step": 6004
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.999873892710122e-05,
+      "loss": 1.2954,
+      "step": 6005
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.9998732752973e-05,
+      "loss": 1.3047,
+      "step": 6006
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.999872656376781e-05,
+      "loss": 1.2796,
+      "step": 6007
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.999872035948563e-05,
+      "loss": 1.3074,
+      "step": 6008
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.99987141401265e-05,
+      "loss": 1.3074,
+      "step": 6009
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.999870790569037e-05,
+      "loss": 1.295,
+      "step": 6010
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.999870165617727e-05,
+      "loss": 1.3438,
+      "step": 6011
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.99986953915872e-05,
+      "loss": 1.3499,
+      "step": 6012
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.999868911192017e-05,
+      "loss": 1.3325,
+      "step": 6013
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.999868281717615e-05,
+      "loss": 1.301,
+      "step": 6014
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.999867650735517e-05,
+      "loss": 1.3025,
+      "step": 6015
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.999867018245723e-05,
+      "loss": 1.3071,
+      "step": 6016
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.999866384248233e-05,
+      "loss": 1.322,
+      "step": 6017
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.999865748743048e-05,
+      "loss": 1.3052,
+      "step": 6018
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.999865111730166e-05,
+      "loss": 1.3062,
+      "step": 6019
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.999864473209588e-05,
+      "loss": 1.3058,
+      "step": 6020
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.999863833181315e-05,
+      "loss": 1.3295,
+      "step": 6021
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.999863191645346e-05,
+      "loss": 1.3066,
+      "step": 6022
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.999862548601684e-05,
+      "loss": 1.3083,
+      "step": 6023
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.999861904050327e-05,
+      "loss": 1.3232,
+      "step": 6024
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.999861257991273e-05,
+      "loss": 1.2654,
+      "step": 6025
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.999860610424526e-05,
+      "loss": 1.2858,
+      "step": 6026
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.999859961350085e-05,
+      "loss": 1.3095,
+      "step": 6027
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.999859310767948e-05,
+      "loss": 1.334,
+      "step": 6028
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.999858658678118e-05,
+      "loss": 1.3141,
+      "step": 6029
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.999858005080596e-05,
+      "loss": 1.3082,
+      "step": 6030
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.999857349975379e-05,
+      "loss": 1.2955,
+      "step": 6031
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.999856693362468e-05,
+      "loss": 1.3373,
+      "step": 6032
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.999856035241865e-05,
+      "loss": 1.2743,
+      "step": 6033
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.999855375613569e-05,
+      "loss": 1.2887,
+      "step": 6034
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.99985471447758e-05,
+      "loss": 1.2819,
+      "step": 6035
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.999854051833899e-05,
+      "loss": 1.2908,
+      "step": 6036
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.999853387682525e-05,
+      "loss": 1.245,
+      "step": 6037
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.999852722023459e-05,
+      "loss": 1.2652,
+      "step": 6038
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.999852054856702e-05,
+      "loss": 1.2684,
+      "step": 6039
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.999851386182252e-05,
+      "loss": 1.2806,
+      "step": 6040
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.999850716000112e-05,
+      "loss": 1.3017,
+      "step": 6041
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.99985004431028e-05,
+      "loss": 1.2764,
+      "step": 6042
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.999849371112755e-05,
+      "loss": 1.2278,
+      "step": 6043
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.999848696407541e-05,
+      "loss": 1.3058,
+      "step": 6044
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.999848020194635e-05,
+      "loss": 1.2917,
+      "step": 6045
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.99984734247404e-05,
+      "loss": 1.3012,
+      "step": 6046
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.999846663245756e-05,
+      "loss": 1.2699,
+      "step": 6047
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.999845982509778e-05,
+      "loss": 1.3207,
+      "step": 6048
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.999845300266113e-05,
+      "loss": 1.3203,
+      "step": 6049
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.999844616514758e-05,
+      "loss": 1.2886,
+      "step": 6050
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.999843931255712e-05,
+      "loss": 1.2715,
+      "step": 6051
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.99984324448898e-05,
+      "loss": 1.2644,
+      "step": 6052
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.999842556214555e-05,
+      "loss": 1.2977,
+      "step": 6053
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.999841866432444e-05,
+      "loss": 1.3133,
+      "step": 6054
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.999841175142644e-05,
+      "loss": 1.2951,
+      "step": 6055
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.999840482345154e-05,
+      "loss": 1.317,
+      "step": 6056
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.999839788039977e-05,
+      "loss": 1.2849,
+      "step": 6057
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.999839092227112e-05,
+      "loss": 1.2925,
+      "step": 6058
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.999838394906558e-05,
+      "loss": 1.3066,
+      "step": 6059
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.999837696078319e-05,
+      "loss": 1.311,
+      "step": 6060
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.999836995742388e-05,
+      "loss": 1.2655,
+      "step": 6061
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.999836293898774e-05,
+      "loss": 1.2725,
+      "step": 6062
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.999835590547472e-05,
+      "loss": 1.2938,
+      "step": 6063
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.999834885688483e-05,
+      "loss": 1.2913,
+      "step": 6064
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.999834179321808e-05,
+      "loss": 1.3075,
+      "step": 6065
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.999833471447447e-05,
+      "loss": 1.3209,
+      "step": 6066
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.999832762065398e-05,
+      "loss": 1.325,
+      "step": 6067
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.999832051175663e-05,
+      "loss": 1.2856,
+      "step": 6068
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.999831338778245e-05,
+      "loss": 1.2738,
+      "step": 6069
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.999830624873139e-05,
+      "loss": 1.3301,
+      "step": 6070
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.99982990946035e-05,
+      "loss": 1.2926,
+      "step": 6071
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.999829192539873e-05,
+      "loss": 1.2516,
+      "step": 6072
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.999828474111715e-05,
+      "loss": 1.3011,
+      "step": 6073
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.999827754175869e-05,
+      "loss": 1.292,
+      "step": 6074
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.99982703273234e-05,
+      "loss": 1.2827,
+      "step": 6075
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.999826309781126e-05,
+      "loss": 1.3093,
+      "step": 6076
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.999825585322229e-05,
+      "loss": 1.3097,
+      "step": 6077
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.99982485935565e-05,
+      "loss": 1.3095,
+      "step": 6078
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.999824131881384e-05,
+      "loss": 1.297,
+      "step": 6079
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.999823402899436e-05,
+      "loss": 1.2813,
+      "step": 6080
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.999822672409806e-05,
+      "loss": 1.2722,
+      "step": 6081
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.999821940412493e-05,
+      "loss": 1.3001,
+      "step": 6082
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.999821206907496e-05,
+      "loss": 1.2911,
+      "step": 6083
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.999820471894817e-05,
+      "loss": 1.2953,
+      "step": 6084
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.999819735374457e-05,
+      "loss": 1.3046,
+      "step": 6085
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.999818997346412e-05,
+      "loss": 1.3055,
+      "step": 6086
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.999818257810689e-05,
+      "loss": 1.301,
+      "step": 6087
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.999817516767282e-05,
+      "loss": 1.302,
+      "step": 6088
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.999816774216196e-05,
+      "loss": 1.2589,
+      "step": 6089
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.999816030157427e-05,
+      "loss": 1.2689,
+      "step": 6090
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.999815284590977e-05,
+      "loss": 1.3407,
+      "step": 6091
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.999814537516847e-05,
+      "loss": 1.3105,
+      "step": 6092
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.999813788935037e-05,
+      "loss": 1.3171,
+      "step": 6093
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.999813038845547e-05,
+      "loss": 1.295,
+      "step": 6094
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.999812287248377e-05,
+      "loss": 1.2858,
+      "step": 6095
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.999811534143526e-05,
+      "loss": 1.2699,
+      "step": 6096
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.999810779530996e-05,
+      "loss": 1.3114,
+      "step": 6097
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.999810023410787e-05,
+      "loss": 1.3433,
+      "step": 6098
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.9998092657829e-05,
+      "loss": 1.2787,
+      "step": 6099
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.999808506647333e-05,
+      "loss": 1.3224,
+      "step": 6100
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.999807746004088e-05,
+      "loss": 1.3179,
+      "step": 6101
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.999806983853163e-05,
+      "loss": 1.3066,
+      "step": 6102
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.999806220194563e-05,
+      "loss": 1.2537,
+      "step": 6103
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.999805455028283e-05,
+      "loss": 1.305,
+      "step": 6104
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.999804688354324e-05,
+      "loss": 1.2901,
+      "step": 6105
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.999803920172691e-05,
+      "loss": 1.2964,
+      "step": 6106
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.999803150483378e-05,
+      "loss": 1.3185,
+      "step": 6107
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.999802379286389e-05,
+      "loss": 1.2781,
+      "step": 6108
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.999801606581724e-05,
+      "loss": 1.2887,
+      "step": 6109
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.999800832369381e-05,
+      "loss": 1.3286,
+      "step": 6110
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.999800056649363e-05,
+      "loss": 1.3113,
+      "step": 6111
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.99979927942167e-05,
+      "loss": 1.3233,
+      "step": 6112
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.999798500686298e-05,
+      "loss": 1.3178,
+      "step": 6113
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.999797720443252e-05,
+      "loss": 1.3278,
+      "step": 6114
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.999796938692532e-05,
+      "loss": 1.3149,
+      "step": 6115
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.999796155434136e-05,
+      "loss": 1.3194,
+      "step": 6116
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.999795370668063e-05,
+      "loss": 1.3017,
+      "step": 6117
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.999794584394319e-05,
+      "loss": 1.2795,
+      "step": 6118
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.999793796612899e-05,
+      "loss": 1.3109,
+      "step": 6119
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.999793007323803e-05,
+      "loss": 1.3056,
+      "step": 6120
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.999792216527036e-05,
+      "loss": 1.3083,
+      "step": 6121
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.999791424222595e-05,
+      "loss": 1.2932,
+      "step": 6122
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.999790630410479e-05,
+      "loss": 1.3045,
+      "step": 6123
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.99978983509069e-05,
+      "loss": 1.3071,
+      "step": 6124
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.99978903826323e-05,
+      "loss": 1.3147,
+      "step": 6125
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.999788239928095e-05,
+      "loss": 1.291,
+      "step": 6126
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.999787440085288e-05,
+      "loss": 1.2958,
+      "step": 6127
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.999786638734809e-05,
+      "loss": 1.316,
+      "step": 6128
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.999785835876658e-05,
+      "loss": 1.3401,
+      "step": 6129
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.999785031510835e-05,
+      "loss": 1.2938,
+      "step": 6130
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.999784225637342e-05,
+      "loss": 1.2873,
+      "step": 6131
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.999783418256177e-05,
+      "loss": 1.2941,
+      "step": 6132
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.99978260936734e-05,
+      "loss": 1.3711,
+      "step": 6133
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.999781798970833e-05,
+      "loss": 1.3569,
+      "step": 6134
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.999780987066655e-05,
+      "loss": 1.3186,
+      "step": 6135
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.999780173654808e-05,
+      "loss": 1.286,
+      "step": 6136
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.999779358735291e-05,
+      "loss": 1.3019,
+      "step": 6137
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.999778542308102e-05,
+      "loss": 1.3164,
+      "step": 6138
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.999777724373245e-05,
+      "loss": 1.3487,
+      "step": 6139
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.999776904930719e-05,
+      "loss": 1.2921,
+      "step": 6140
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.999776083980523e-05,
+      "loss": 1.2684,
+      "step": 6141
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.99977526152266e-05,
+      "loss": 1.3211,
+      "step": 6142
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.999774437557126e-05,
+      "loss": 1.2934,
+      "step": 6143
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.999773612083925e-05,
+      "loss": 1.3196,
+      "step": 6144
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.999772785103056e-05,
+      "loss": 1.2997,
+      "step": 6145
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.999771956614519e-05,
+      "loss": 1.2959,
+      "step": 6146
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.999771126618314e-05,
+      "loss": 1.2895,
+      "step": 6147
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.999770295114441e-05,
+      "loss": 1.2743,
+      "step": 6148
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.999769462102903e-05,
+      "loss": 1.2584,
+      "step": 6149
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.999768627583697e-05,
+      "loss": 1.3327,
+      "step": 6150
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.999767791556824e-05,
+      "loss": 1.301,
+      "step": 6151
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.999766954022285e-05,
+      "loss": 1.3144,
+      "step": 6152
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.99976611498008e-05,
+      "loss": 1.3057,
+      "step": 6153
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.99976527443021e-05,
+      "loss": 1.3195,
+      "step": 6154
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.999764432372674e-05,
+      "loss": 1.3161,
+      "step": 6155
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.999763588807472e-05,
+      "loss": 1.3287,
+      "step": 6156
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.999762743734607e-05,
+      "loss": 1.3374,
+      "step": 6157
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.999761897154075e-05,
+      "loss": 1.3017,
+      "step": 6158
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.999761049065878e-05,
+      "loss": 1.2921,
+      "step": 6159
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.99976019947002e-05,
+      "loss": 1.3122,
+      "step": 6160
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.999759348366495e-05,
+      "loss": 1.2636,
+      "step": 6161
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.999758495755307e-05,
+      "loss": 1.2868,
+      "step": 6162
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.999757641636456e-05,
+      "loss": 1.3113,
+      "step": 6163
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.999756786009942e-05,
+      "loss": 1.3114,
+      "step": 6164
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.999755928875765e-05,
+      "loss": 1.2668,
+      "step": 6165
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.999755070233925e-05,
+      "loss": 1.3134,
+      "step": 6166
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.999754210084422e-05,
+      "loss": 1.2973,
+      "step": 6167
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.999753348427257e-05,
+      "loss": 1.3043,
+      "step": 6168
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.99975248526243e-05,
+      "loss": 1.3107,
+      "step": 6169
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.999751620589941e-05,
+      "loss": 1.3102,
+      "step": 6170
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.999750754409791e-05,
+      "loss": 1.2699,
+      "step": 6171
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.99974988672198e-05,
+      "loss": 1.2909,
+      "step": 6172
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.999749017526508e-05,
+      "loss": 1.2949,
+      "step": 6173
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.999748146823376e-05,
+      "loss": 1.3023,
+      "step": 6174
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.999747274612583e-05,
+      "loss": 1.2941,
+      "step": 6175
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.99974640089413e-05,
+      "loss": 1.2803,
+      "step": 6176
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.999745525668016e-05,
+      "loss": 1.284,
+      "step": 6177
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.999744648934244e-05,
+      "loss": 1.2931,
+      "step": 6178
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.999743770692811e-05,
+      "loss": 1.314,
+      "step": 6179
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.999742890943722e-05,
+      "loss": 1.2942,
+      "step": 6180
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.999742009686972e-05,
+      "loss": 1.3013,
+      "step": 6181
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.999741126922564e-05,
+      "loss": 1.2661,
+      "step": 6182
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.999740242650497e-05,
+      "loss": 1.3464,
+      "step": 6183
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.999739356870772e-05,
+      "loss": 1.3059,
+      "step": 6184
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.99973846958339e-05,
+      "loss": 1.2743,
+      "step": 6185
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.999737580788351e-05,
+      "loss": 1.3078,
+      "step": 6186
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.999736690485652e-05,
+      "loss": 1.3009,
+      "step": 6187
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.999735798675299e-05,
+      "loss": 1.29,
+      "step": 6188
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.99973490535729e-05,
+      "loss": 1.2686,
+      "step": 6189
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.999734010531622e-05,
+      "loss": 1.3268,
+      "step": 6190
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.9997331141983e-05,
+      "loss": 1.3098,
+      "step": 6191
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.99973221635732e-05,
+      "loss": 1.2789,
+      "step": 6192
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.999731317008685e-05,
+      "loss": 1.3277,
+      "step": 6193
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.999730416152397e-05,
+      "loss": 1.3284,
+      "step": 6194
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.999729513788452e-05,
+      "loss": 1.2535,
+      "step": 6195
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.999728609916852e-05,
+      "loss": 1.3307,
+      "step": 6196
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.999727704537598e-05,
+      "loss": 1.3357,
+      "step": 6197
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.999726797650691e-05,
+      "loss": 1.3267,
+      "step": 6198
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.999725889256128e-05,
+      "loss": 1.2754,
+      "step": 6199
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.999724979353911e-05,
+      "loss": 1.3029,
+      "step": 6200
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.999724067944043e-05,
+      "loss": 1.2944,
+      "step": 6201
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.99972315502652e-05,
+      "loss": 1.2924,
+      "step": 6202
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.999722240601346e-05,
+      "loss": 1.3194,
+      "step": 6203
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.999721324668517e-05,
+      "loss": 1.3283,
+      "step": 6204
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.999720407228037e-05,
+      "loss": 1.3144,
+      "step": 6205
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.999719488279906e-05,
+      "loss": 1.3213,
+      "step": 6206
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.999718567824121e-05,
+      "loss": 1.3119,
+      "step": 6207
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.999717645860687e-05,
+      "loss": 1.3003,
+      "step": 6208
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.999716722389601e-05,
+      "loss": 1.329,
+      "step": 6209
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.999715797410862e-05,
+      "loss": 1.2702,
+      "step": 6210
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.999714870924475e-05,
+      "loss": 1.3145,
+      "step": 6211
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.999713942930436e-05,
+      "loss": 1.299,
+      "step": 6212
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.999713013428749e-05,
+      "loss": 1.2802,
+      "step": 6213
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.99971208241941e-05,
+      "loss": 1.2878,
+      "step": 6214
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.999711149902422e-05,
+      "loss": 1.3014,
+      "step": 6215
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.999710215877785e-05,
+      "loss": 1.2958,
+      "step": 6216
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.9997092803455e-05,
+      "loss": 1.2861,
+      "step": 6217
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.999708343305566e-05,
+      "loss": 1.3345,
+      "step": 6218
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.999707404757983e-05,
+      "loss": 1.2864,
+      "step": 6219
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.999706464702752e-05,
+      "loss": 1.3096,
+      "step": 6220
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.999705523139873e-05,
+      "loss": 1.3272,
+      "step": 6221
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.999704580069346e-05,
+      "loss": 1.3053,
+      "step": 6222
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.999703635491173e-05,
+      "loss": 1.3005,
+      "step": 6223
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.999702689405352e-05,
+      "loss": 1.2986,
+      "step": 6224
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.999701741811885e-05,
+      "loss": 1.2877,
+      "step": 6225
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.999700792710773e-05,
+      "loss": 1.3141,
+      "step": 6226
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.999699842102013e-05,
+      "loss": 1.3295,
+      "step": 6227
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.999698889985607e-05,
+      "loss": 1.2994,
+      "step": 6228
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.999697936361556e-05,
+      "loss": 1.2982,
+      "step": 6229
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.99969698122986e-05,
+      "loss": 1.3059,
+      "step": 6230
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.999696024590519e-05,
+      "loss": 1.2662,
+      "step": 6231
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.999695066443533e-05,
+      "loss": 1.286,
+      "step": 6232
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.999694106788904e-05,
+      "loss": 1.3323,
+      "step": 6233
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.99969314562663e-05,
+      "loss": 1.291,
+      "step": 6234
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.99969218295671e-05,
+      "loss": 1.2947,
+      "step": 6235
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.99969121877915e-05,
+      "loss": 1.3186,
+      "step": 6236
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.999690253093945e-05,
+      "loss": 1.286,
+      "step": 6237
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.999689285901096e-05,
+      "loss": 1.2954,
+      "step": 6238
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.999688317200606e-05,
+      "loss": 1.2758,
+      "step": 6239
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.999687346992475e-05,
+      "loss": 1.3087,
+      "step": 6240
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.9996863752767e-05,
+      "loss": 1.2625,
+      "step": 6241
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.999685402053282e-05,
+      "loss": 1.2855,
+      "step": 6242
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.999684427322224e-05,
+      "loss": 1.2946,
+      "step": 6243
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.999683451083525e-05,
+      "loss": 1.2862,
+      "step": 6244
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.999682473337187e-05,
+      "loss": 1.2931,
+      "step": 6245
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.999681494083206e-05,
+      "loss": 1.2988,
+      "step": 6246
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.999680513321586e-05,
+      "loss": 1.3077,
+      "step": 6247
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.999679531052324e-05,
+      "loss": 1.2964,
+      "step": 6248
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.999678547275424e-05,
+      "loss": 1.2696,
+      "step": 6249
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.999677561990883e-05,
+      "loss": 1.3066,
+      "step": 6250
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.999676575198704e-05,
+      "loss": 1.2824,
+      "step": 6251
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.999675586898887e-05,
+      "loss": 1.2702,
+      "step": 6252
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.999674597091431e-05,
+      "loss": 1.2682,
+      "step": 6253
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.999673605776336e-05,
+      "loss": 1.2512,
+      "step": 6254
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.999672612953604e-05,
+      "loss": 1.2886,
+      "step": 6255
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.999671618623234e-05,
+      "loss": 1.2996,
+      "step": 6256
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.999670622785226e-05,
+      "loss": 1.2864,
+      "step": 6257
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.999669625439582e-05,
+      "loss": 1.2781,
+      "step": 6258
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.999668626586302e-05,
+      "loss": 1.3022,
+      "step": 6259
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.999667626225385e-05,
+      "loss": 1.3173,
+      "step": 6260
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.999666624356831e-05,
+      "loss": 1.2996,
+      "step": 6261
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.999665620980643e-05,
+      "loss": 1.2896,
+      "step": 6262
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.999664616096818e-05,
+      "loss": 1.326,
+      "step": 6263
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.999663609705358e-05,
+      "loss": 1.2749,
+      "step": 6264
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.999662601806263e-05,
+      "loss": 1.3075,
+      "step": 6265
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.999661592399534e-05,
+      "loss": 1.2972,
+      "step": 6266
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.999660581485169e-05,
+      "loss": 1.2966,
+      "step": 6267
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.999659569063172e-05,
+      "loss": 1.2897,
+      "step": 6268
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.999658555133542e-05,
+      "loss": 1.2743,
+      "step": 6269
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.999657539696277e-05,
+      "loss": 1.3019,
+      "step": 6270
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.999656522751378e-05,
+      "loss": 1.328,
+      "step": 6271
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.999655504298848e-05,
+      "loss": 1.2732,
+      "step": 6272
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.999654484338685e-05,
+      "loss": 1.294,
+      "step": 6273
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.999653462870891e-05,
+      "loss": 1.2982,
+      "step": 6274
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.999652439895464e-05,
+      "loss": 1.2753,
+      "step": 6275
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.999651415412406e-05,
+      "loss": 1.2904,
+      "step": 6276
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.999650389421715e-05,
+      "loss": 1.2721,
+      "step": 6277
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.999649361923396e-05,
+      "loss": 1.37,
+      "step": 6278
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.999648332917444e-05,
+      "loss": 1.3477,
+      "step": 6279
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.999647302403862e-05,
+      "loss": 1.3486,
+      "step": 6280
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.999646270382652e-05,
+      "loss": 1.2818,
+      "step": 6281
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.999645236853809e-05,
+      "loss": 1.3353,
+      "step": 6282
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.99964420181734e-05,
+      "loss": 1.3072,
+      "step": 6283
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.999643165273241e-05,
+      "loss": 1.3302,
+      "step": 6284
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.999642127221512e-05,
+      "loss": 1.3346,
+      "step": 6285
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.999641087662155e-05,
+      "loss": 1.3207,
+      "step": 6286
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.999640046595171e-05,
+      "loss": 1.5271,
+      "step": 6287
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.99963900402056e-05,
+      "loss": 1.296,
+      "step": 6288
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.99963795993832e-05,
+      "loss": 1.2875,
+      "step": 6289
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.999636914348452e-05,
+      "loss": 1.3154,
+      "step": 6290
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.999635867250958e-05,
+      "loss": 1.3326,
+      "step": 6291
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.999634818645838e-05,
+      "loss": 1.2819,
+      "step": 6292
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.999633768533092e-05,
+      "loss": 1.3086,
+      "step": 6293
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.99963271691272e-05,
+      "loss": 1.3099,
+      "step": 6294
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.99963166378472e-05,
+      "loss": 1.3088,
+      "step": 6295
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.999630609149098e-05,
+      "loss": 1.3426,
+      "step": 6296
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.99962955300585e-05,
+      "loss": 1.2796,
+      "step": 6297
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.999628495354978e-05,
+      "loss": 1.3235,
+      "step": 6298
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.99962743619648e-05,
+      "loss": 1.313,
+      "step": 6299
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.99962637553036e-05,
+      "loss": 1.3028,
+      "step": 6300
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.999625313356615e-05,
+      "loss": 1.3031,
+      "step": 6301
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.999624249675245e-05,
+      "loss": 1.3045,
+      "step": 6302
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.999623184486256e-05,
+      "loss": 1.2848,
+      "step": 6303
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.999622117789641e-05,
+      "loss": 1.302,
+      "step": 6304
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.999621049585406e-05,
+      "loss": 1.3121,
+      "step": 6305
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.999619979873547e-05,
+      "loss": 1.2862,
+      "step": 6306
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.999618908654066e-05,
+      "loss": 1.307,
+      "step": 6307
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.999617835926966e-05,
+      "loss": 1.3276,
+      "step": 6308
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.999616761692243e-05,
+      "loss": 1.297,
+      "step": 6309
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.9996156859499e-05,
+      "loss": 1.2572,
+      "step": 6310
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.999614608699936e-05,
+      "loss": 1.269,
+      "step": 6311
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.999613529942353e-05,
+      "loss": 1.2937,
+      "step": 6312
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.999612449677149e-05,
+      "loss": 1.3422,
+      "step": 6313
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.999611367904327e-05,
+      "loss": 1.321,
+      "step": 6314
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.999610284623883e-05,
+      "loss": 1.3022,
+      "step": 6315
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.999609199835823e-05,
+      "loss": 1.2767,
+      "step": 6316
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.999608113540144e-05,
+      "loss": 1.3005,
+      "step": 6317
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.999607025736845e-05,
+      "loss": 1.3017,
+      "step": 6318
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.999605936425931e-05,
+      "loss": 1.3141,
+      "step": 6319
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.999604845607397e-05,
+      "loss": 1.3186,
+      "step": 6320
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.999603753281248e-05,
+      "loss": 1.3005,
+      "step": 6321
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.999602659447481e-05,
+      "loss": 1.3186,
+      "step": 6322
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.999601564106098e-05,
+      "loss": 1.3005,
+      "step": 6323
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.999600467257099e-05,
+      "loss": 1.2839,
+      "step": 6324
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.999599368900483e-05,
+      "loss": 1.2978,
+      "step": 6325
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.999598269036252e-05,
+      "loss": 1.2648,
+      "step": 6326
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.999597167664406e-05,
+      "loss": 1.2762,
+      "step": 6327
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.999596064784946e-05,
+      "loss": 1.2888,
+      "step": 6328
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.99959496039787e-05,
+      "loss": 1.2715,
+      "step": 6329
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.999593854503179e-05,
+      "loss": 1.3171,
+      "step": 6330
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.999592747100876e-05,
+      "loss": 1.235,
+      "step": 6331
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.99959163819096e-05,
+      "loss": 1.2876,
+      "step": 6332
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.99959052777343e-05,
+      "loss": 1.2943,
+      "step": 6333
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.999589415848287e-05,
+      "loss": 1.2936,
+      "step": 6334
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.99958830241553e-05,
+      "loss": 1.3063,
+      "step": 6335
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.999587187475163e-05,
+      "loss": 1.2917,
+      "step": 6336
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.999586071027183e-05,
+      "loss": 1.255,
+      "step": 6337
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.999584953071592e-05,
+      "loss": 1.2492,
+      "step": 6338
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.99958383360839e-05,
+      "loss": 1.3153,
+      "step": 6339
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.999582712637578e-05,
+      "loss": 1.2663,
+      "step": 6340
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.999581590159154e-05,
+      "loss": 1.2893,
+      "step": 6341
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.99958046617312e-05,
+      "loss": 1.3044,
+      "step": 6342
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.999579340679477e-05,
+      "loss": 1.3184,
+      "step": 6343
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.999578213678224e-05,
+      "loss": 1.3235,
+      "step": 6344
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.99957708516936e-05,
+      "loss": 1.3373,
+      "step": 6345
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.99957595515289e-05,
+      "loss": 1.2933,
+      "step": 6346
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.999574823628811e-05,
+      "loss": 1.3097,
+      "step": 6347
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.999573690597124e-05,
+      "loss": 1.3185,
+      "step": 6348
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.999572556057827e-05,
+      "loss": 1.3386,
+      "step": 6349
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.999571420010924e-05,
+      "loss": 1.3201,
+      "step": 6350
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.999570282456417e-05,
+      "loss": 1.3175,
+      "step": 6351
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.999569143394299e-05,
+      "loss": 1.2796,
+      "step": 6352
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.999568002824576e-05,
+      "loss": 1.2962,
+      "step": 6353
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.999566860747247e-05,
+      "loss": 1.3211,
+      "step": 6354
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.999565717162313e-05,
+      "loss": 1.2646,
+      "step": 6355
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.999564572069773e-05,
+      "loss": 1.3111,
+      "step": 6356
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.999563425469627e-05,
+      "loss": 1.3148,
+      "step": 6357
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.999562277361877e-05,
+      "loss": 1.2945,
+      "step": 6358
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.999561127746524e-05,
+      "loss": 1.3247,
+      "step": 6359
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.999559976623565e-05,
+      "loss": 1.3431,
+      "step": 6360
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.999558823993004e-05,
+      "loss": 1.2822,
+      "step": 6361
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.999557669854839e-05,
+      "loss": 1.3382,
+      "step": 6362
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.999556514209071e-05,
+      "loss": 1.3376,
+      "step": 6363
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.9995553570557e-05,
+      "loss": 1.303,
+      "step": 6364
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.999554198394727e-05,
+      "loss": 1.3017,
+      "step": 6365
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.999553038226153e-05,
+      "loss": 1.2985,
+      "step": 6366
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.999551876549976e-05,
+      "loss": 1.2707,
+      "step": 6367
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.999550713366199e-05,
+      "loss": 1.2807,
+      "step": 6368
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.999549548674821e-05,
+      "loss": 1.2889,
+      "step": 6369
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.999548382475844e-05,
+      "loss": 1.2912,
+      "step": 6370
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.999547214769263e-05,
+      "loss": 1.3021,
+      "step": 6371
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.999546045555084e-05,
+      "loss": 1.25,
+      "step": 6372
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.999544874833307e-05,
+      "loss": 1.2834,
+      "step": 6373
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.999543702603928e-05,
+      "loss": 1.2624,
+      "step": 6374
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.999542528866954e-05,
+      "loss": 1.3064,
+      "step": 6375
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.999541353622379e-05,
+      "loss": 1.2591,
+      "step": 6376
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.999540176870207e-05,
+      "loss": 1.2958,
+      "step": 6377
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.999538998610436e-05,
+      "loss": 1.2758,
+      "step": 6378
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.99953781884307e-05,
+      "loss": 1.2853,
+      "step": 6379
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.999536637568105e-05,
+      "loss": 1.2876,
+      "step": 6380
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.999535454785545e-05,
+      "loss": 1.3098,
+      "step": 6381
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.999534270495387e-05,
+      "loss": 1.3112,
+      "step": 6382
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.999533084697634e-05,
+      "loss": 1.2906,
+      "step": 6383
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.999531897392286e-05,
+      "loss": 1.2929,
+      "step": 6384
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.999530708579342e-05,
+      "loss": 1.3041,
+      "step": 6385
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.999529518258803e-05,
+      "loss": 1.3078,
+      "step": 6386
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.999528326430671e-05,
+      "loss": 1.2922,
+      "step": 6387
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.999527133094944e-05,
+      "loss": 1.3249,
+      "step": 6388
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.999525938251623e-05,
+      "loss": 1.3125,
+      "step": 6389
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.999524741900708e-05,
+      "loss": 1.3031,
+      "step": 6390
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.999523544042202e-05,
+      "loss": 1.3041,
+      "step": 6391
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.999522344676102e-05,
+      "loss": 1.3114,
+      "step": 6392
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.99952114380241e-05,
+      "loss": 1.2907,
+      "step": 6393
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.999519941421127e-05,
+      "loss": 1.286,
+      "step": 6394
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.999518737532251e-05,
+      "loss": 1.3279,
+      "step": 6395
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.999517532135783e-05,
+      "loss": 1.2601,
+      "step": 6396
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.999516325231726e-05,
+      "loss": 1.2321,
+      "step": 6397
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.999515116820077e-05,
+      "loss": 1.2593,
+      "step": 6398
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.999513906900839e-05,
+      "loss": 1.2889,
+      "step": 6399
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.999512695474012e-05,
+      "loss": 1.2741,
+      "step": 6400
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.999511482539594e-05,
+      "loss": 1.2654,
+      "step": 6401
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.999510268097588e-05,
+      "loss": 1.3045,
+      "step": 6402
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.999509052147993e-05,
+      "loss": 1.3756,
+      "step": 6403
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.99950783469081e-05,
+      "loss": 1.3721,
+      "step": 6404
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.99950661572604e-05,
+      "loss": 1.366,
+      "step": 6405
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.999505395253681e-05,
+      "loss": 1.3805,
+      "step": 6406
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.999504173273735e-05,
+      "loss": 1.3702,
+      "step": 6407
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.999502949786202e-05,
+      "loss": 1.6505,
+      "step": 6408
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.999501724791083e-05,
+      "loss": 1.4332,
+      "step": 6409
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.999500498288378e-05,
+      "loss": 1.3794,
+      "step": 6410
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.999499270278086e-05,
+      "loss": 1.4305,
+      "step": 6411
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.999498040760211e-05,
+      "loss": 1.5643,
+      "step": 6412
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.999496809734749e-05,
+      "loss": 1.4611,
+      "step": 6413
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.999495577201702e-05,
+      "loss": 1.4257,
+      "step": 6414
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.999494343161074e-05,
+      "loss": 1.3763,
+      "step": 6415
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.99949310761286e-05,
+      "loss": 1.3826,
+      "step": 6416
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.999491870557062e-05,
+      "loss": 1.3624,
+      "step": 6417
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.999490631993682e-05,
+      "loss": 1.395,
+      "step": 6418
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.999489391922718e-05,
+      "loss": 1.374,
+      "step": 6419
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.999488150344174e-05,
+      "loss": 1.4396,
+      "step": 6420
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.999486907258047e-05,
+      "loss": 1.3758,
+      "step": 6421
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.999485662664336e-05,
+      "loss": 1.401,
+      "step": 6422
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.999484416563047e-05,
+      "loss": 1.4183,
+      "step": 6423
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.999483168954176e-05,
+      "loss": 1.3814,
+      "step": 6424
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.999481919837723e-05,
+      "loss": 1.3879,
+      "step": 6425
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.999480669213692e-05,
+      "loss": 1.3639,
+      "step": 6426
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.99947941708208e-05,
+      "loss": 1.3933,
+      "step": 6427
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.99947816344289e-05,
+      "loss": 1.3446,
+      "step": 6428
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.999476908296121e-05,
+      "loss": 1.381,
+      "step": 6429
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.999475651641773e-05,
+      "loss": 1.3979,
+      "step": 6430
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.999474393479847e-05,
+      "loss": 1.3947,
+      "step": 6431
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.999473133810342e-05,
+      "loss": 1.425,
+      "step": 6432
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.999471872633259e-05,
+      "loss": 1.3653,
+      "step": 6433
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.9994706099486e-05,
+      "loss": 1.3997,
+      "step": 6434
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.999469345756365e-05,
+      "loss": 1.4406,
+      "step": 6435
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.999468080056553e-05,
+      "loss": 1.3602,
+      "step": 6436
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.999466812849165e-05,
+      "loss": 1.408,
+      "step": 6437
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.999465544134202e-05,
+      "loss": 1.3559,
+      "step": 6438
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.999464273911663e-05,
+      "loss": 1.3658,
+      "step": 6439
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.999463002181551e-05,
+      "loss": 1.3502,
+      "step": 6440
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.999461728943862e-05,
+      "loss": 1.3772,
+      "step": 6441
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.999460454198602e-05,
+      "loss": 1.3647,
+      "step": 6442
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.999459177945766e-05,
+      "loss": 1.3749,
+      "step": 6443
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.999457900185357e-05,
+      "loss": 1.3901,
+      "step": 6444
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.999456620917376e-05,
+      "loss": 1.3971,
+      "step": 6445
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.999455340141822e-05,
+      "loss": 1.4175,
+      "step": 6446
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.999454057858696e-05,
+      "loss": 1.4309,
+      "step": 6447
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.999452774067999e-05,
+      "loss": 1.444,
+      "step": 6448
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.999451488769729e-05,
+      "loss": 1.3918,
+      "step": 6449
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.999450201963889e-05,
+      "loss": 1.4114,
+      "step": 6450
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.99944891365048e-05,
+      "loss": 1.397,
+      "step": 6451
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.9994476238295e-05,
+      "loss": 1.4307,
+      "step": 6452
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.999446332500948e-05,
+      "loss": 1.3765,
+      "step": 6453
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.999445039664829e-05,
+      "loss": 1.4144,
+      "step": 6454
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.99944374532114e-05,
+      "loss": 1.4267,
+      "step": 6455
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.999442449469882e-05,
+      "loss": 1.4228,
+      "step": 6456
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.999441152111058e-05,
+      "loss": 1.4168,
+      "step": 6457
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.999439853244665e-05,
+      "loss": 1.3957,
+      "step": 6458
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.999438552870704e-05,
+      "loss": 1.4153,
+      "step": 6459
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.999437250989177e-05,
+      "loss": 1.3352,
+      "step": 6460
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.999435947600082e-05,
+      "loss": 1.3956,
+      "step": 6461
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.99943464270342e-05,
+      "loss": 1.383,
+      "step": 6462
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.999433336299195e-05,
+      "loss": 1.3779,
+      "step": 6463
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.999432028387403e-05,
+      "loss": 1.3463,
+      "step": 6464
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.999430718968046e-05,
+      "loss": 1.3939,
+      "step": 6465
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.999429408041125e-05,
+      "loss": 1.3891,
+      "step": 6466
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.999428095606637e-05,
+      "loss": 1.3635,
+      "step": 6467
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.999426781664589e-05,
+      "loss": 1.3785,
+      "step": 6468
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.999425466214976e-05,
+      "loss": 1.3694,
+      "step": 6469
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.999424149257799e-05,
+      "loss": 1.3698,
+      "step": 6470
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.99942283079306e-05,
+      "loss": 1.3746,
+      "step": 6471
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.999421510820759e-05,
+      "loss": 1.3496,
+      "step": 6472
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.999420189340895e-05,
+      "loss": 1.3709,
+      "step": 6473
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.999418866353471e-05,
+      "loss": 1.4181,
+      "step": 6474
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.999417541858487e-05,
+      "loss": 1.3744,
+      "step": 6475
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.99941621585594e-05,
+      "loss": 1.3547,
+      "step": 6476
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.999414888345833e-05,
+      "loss": 1.3531,
+      "step": 6477
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.999413559328167e-05,
+      "loss": 1.382,
+      "step": 6478
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.99941222880294e-05,
+      "loss": 1.3809,
+      "step": 6479
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.999410896770157e-05,
+      "loss": 1.3607,
+      "step": 6480
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.999409563229812e-05,
+      "loss": 1.3944,
+      "step": 6481
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.99940822818191e-05,
+      "loss": 1.3793,
+      "step": 6482
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.99940689162645e-05,
+      "loss": 1.3895,
+      "step": 6483
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.999405553563432e-05,
+      "loss": 1.4041,
+      "step": 6484
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.999404213992858e-05,
+      "loss": 1.365,
+      "step": 6485
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.999402872914727e-05,
+      "loss": 1.3629,
+      "step": 6486
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.99940153032904e-05,
+      "loss": 1.3896,
+      "step": 6487
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.999400186235797e-05,
+      "loss": 1.4068,
+      "step": 6488
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.999398840634997e-05,
+      "loss": 1.3894,
+      "step": 6489
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.999397493526645e-05,
+      "loss": 1.3741,
+      "step": 6490
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.999396144910736e-05,
+      "loss": 1.3631,
+      "step": 6491
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.999394794787273e-05,
+      "loss": 1.3629,
+      "step": 6492
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.999393443156258e-05,
+      "loss": 1.3864,
+      "step": 6493
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.999392090017688e-05,
+      "loss": 1.403,
+      "step": 6494
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.999390735371564e-05,
+      "loss": 1.355,
+      "step": 6495
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.999389379217889e-05,
+      "loss": 1.3738,
+      "step": 6496
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.999388021556662e-05,
+      "loss": 1.3611,
+      "step": 6497
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.999386662387882e-05,
+      "loss": 1.3786,
+      "step": 6498
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.999385301711551e-05,
+      "loss": 1.3476,
+      "step": 6499
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.999383939527669e-05,
+      "loss": 1.3903,
+      "step": 6500
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.999382575836238e-05,
+      "loss": 1.351,
+      "step": 6501
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.999381210637255e-05,
+      "loss": 1.397,
+      "step": 6502
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.999379843930723e-05,
+      "loss": 1.3777,
+      "step": 6503
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.999378475716641e-05,
+      "loss": 1.381,
+      "step": 6504
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.999377105995012e-05,
+      "loss": 1.3916,
+      "step": 6505
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.999375734765832e-05,
+      "loss": 1.3631,
+      "step": 6506
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.999374362029105e-05,
+      "loss": 1.3758,
+      "step": 6507
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.99937298778483e-05,
+      "loss": 1.3751,
+      "step": 6508
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.999371612033008e-05,
+      "loss": 1.3807,
+      "step": 6509
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.99937023477364e-05,
+      "loss": 1.3659,
+      "step": 6510
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.999368856006724e-05,
+      "loss": 1.3786,
+      "step": 6511
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.999367475732264e-05,
+      "loss": 1.3388,
+      "step": 6512
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.999366093950258e-05,
+      "loss": 1.3767,
+      "step": 6513
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.999364710660704e-05,
+      "loss": 1.3725,
+      "step": 6514
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.999363325863608e-05,
+      "loss": 1.3774,
+      "step": 6515
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.999361939558967e-05,
+      "loss": 1.3593,
+      "step": 6516
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.999360551746782e-05,
+      "loss": 1.3821,
+      "step": 6517
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.999359162427053e-05,
+      "loss": 1.4056,
+      "step": 6518
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.999357771599782e-05,
+      "loss": 1.3832,
+      "step": 6519
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.999356379264968e-05,
+      "loss": 1.3667,
+      "step": 6520
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.999354985422612e-05,
+      "loss": 1.371,
+      "step": 6521
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.999353590072713e-05,
+      "loss": 1.4035,
+      "step": 6522
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.999352193215275e-05,
+      "loss": 1.3923,
+      "step": 6523
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.999350794850292e-05,
+      "loss": 1.3744,
+      "step": 6524
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.999349394977773e-05,
+      "loss": 1.3849,
+      "step": 6525
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.999347993597713e-05,
+      "loss": 1.344,
+      "step": 6526
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.99934659071011e-05,
+      "loss": 1.3603,
+      "step": 6527
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.99934518631497e-05,
+      "loss": 1.3759,
+      "step": 6528
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.999343780412291e-05,
+      "loss": 1.3895,
+      "step": 6529
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.999342373002075e-05,
+      "loss": 1.3738,
+      "step": 6530
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.999340964084318e-05,
+      "loss": 1.3757,
+      "step": 6531
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.999339553659025e-05,
+      "loss": 1.3871,
+      "step": 6532
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.999338141726193e-05,
+      "loss": 1.3998,
+      "step": 6533
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.999336728285826e-05,
+      "loss": 1.4083,
+      "step": 6534
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.999335313337923e-05,
+      "loss": 1.4092,
+      "step": 6535
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.999333896882484e-05,
+      "loss": 1.3706,
+      "step": 6536
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.99933247891951e-05,
+      "loss": 1.4024,
+      "step": 6537
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.999331059449e-05,
+      "loss": 1.3674,
+      "step": 6538
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.999329638470956e-05,
+      "loss": 1.3496,
+      "step": 6539
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.999328215985378e-05,
+      "loss": 1.4101,
+      "step": 6540
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.999326791992265e-05,
+      "loss": 1.3449,
+      "step": 6541
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.999325366491618e-05,
+      "loss": 1.3853,
+      "step": 6542
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.99932393948344e-05,
+      "loss": 1.3623,
+      "step": 6543
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.99932251096773e-05,
+      "loss": 1.3267,
+      "step": 6544
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.999321080944485e-05,
+      "loss": 1.3718,
+      "step": 6545
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.999319649413713e-05,
+      "loss": 1.3466,
+      "step": 6546
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.999318216375407e-05,
+      "loss": 1.3976,
+      "step": 6547
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.999316781829568e-05,
+      "loss": 1.3674,
+      "step": 6548
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.999315345776202e-05,
+      "loss": 1.3594,
+      "step": 6549
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.999313908215305e-05,
+      "loss": 1.348,
+      "step": 6550
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.99931246914688e-05,
+      "loss": 1.3594,
+      "step": 6551
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.999311028570924e-05,
+      "loss": 1.3484,
+      "step": 6552
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.999309586487439e-05,
+      "loss": 1.3987,
+      "step": 6553
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.999308142896428e-05,
+      "loss": 1.3467,
+      "step": 6554
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.99930669779789e-05,
+      "loss": 1.3753,
+      "step": 6555
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.999305251191821e-05,
+      "loss": 1.3859,
+      "step": 6556
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.999303803078229e-05,
+      "loss": 1.3733,
+      "step": 6557
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.999302353457109e-05,
+      "loss": 1.3687,
+      "step": 6558
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.999300902328463e-05,
+      "loss": 1.3756,
+      "step": 6559
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.999299449692292e-05,
+      "loss": 1.39,
+      "step": 6560
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.999297995548595e-05,
+      "loss": 1.3564,
+      "step": 6561
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.999296539897374e-05,
+      "loss": 1.3343,
+      "step": 6562
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.99929508273863e-05,
+      "loss": 1.3591,
+      "step": 6563
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.99929362407236e-05,
+      "loss": 1.3497,
+      "step": 6564
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.999292163898566e-05,
+      "loss": 1.3699,
+      "step": 6565
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.999290702217253e-05,
+      "loss": 1.3683,
+      "step": 6566
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.999289239028413e-05,
+      "loss": 1.3431,
+      "step": 6567
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.999287774332053e-05,
+      "loss": 1.351,
+      "step": 6568
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.999286308128172e-05,
+      "loss": 1.3399,
+      "step": 6569
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.99928484041677e-05,
+      "loss": 1.321,
+      "step": 6570
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.999283371197847e-05,
+      "loss": 1.3431,
+      "step": 6571
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.999281900471405e-05,
+      "loss": 1.332,
+      "step": 6572
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.999280428237441e-05,
+      "loss": 1.3619,
+      "step": 6573
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.999278954495959e-05,
+      "loss": 1.3323,
+      "step": 6574
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.999277479246956e-05,
+      "loss": 1.3725,
+      "step": 6575
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.999276002490437e-05,
+      "loss": 1.3765,
+      "step": 6576
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.999274524226399e-05,
+      "loss": 1.3698,
+      "step": 6577
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.999273044454845e-05,
+      "loss": 1.3749,
+      "step": 6578
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.999271563175771e-05,
+      "loss": 1.3639,
+      "step": 6579
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.999270080389182e-05,
+      "loss": 1.3525,
+      "step": 6580
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.999268596095078e-05,
+      "loss": 1.3895,
+      "step": 6581
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.999267110293455e-05,
+      "loss": 1.358,
+      "step": 6582
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.99926562298432e-05,
+      "loss": 1.3569,
+      "step": 6583
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.999264134167668e-05,
+      "loss": 1.3512,
+      "step": 6584
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.999262643843502e-05,
+      "loss": 1.3576,
+      "step": 6585
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.999261152011824e-05,
+      "loss": 1.3502,
+      "step": 6586
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.99925965867263e-05,
+      "loss": 1.3817,
+      "step": 6587
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.999258163825923e-05,
+      "loss": 1.3792,
+      "step": 6588
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.999256667471704e-05,
+      "loss": 1.3898,
+      "step": 6589
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.999255169609973e-05,
+      "loss": 1.3873,
+      "step": 6590
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.99925367024073e-05,
+      "loss": 1.3541,
+      "step": 6591
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.999252169363975e-05,
+      "loss": 1.3509,
+      "step": 6592
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.999250666979712e-05,
+      "loss": 1.3668,
+      "step": 6593
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.999249163087936e-05,
+      "loss": 1.3529,
+      "step": 6594
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.99924765768865e-05,
+      "loss": 1.3593,
+      "step": 6595
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.999246150781856e-05,
+      "loss": 1.3813,
+      "step": 6596
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.999244642367552e-05,
+      "loss": 1.3661,
+      "step": 6597
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.99924313244574e-05,
+      "loss": 1.345,
+      "step": 6598
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.99924162101642e-05,
+      "loss": 1.3799,
+      "step": 6599
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.999240108079592e-05,
+      "loss": 1.3238,
+      "step": 6600
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.999238593635257e-05,
+      "loss": 1.385,
+      "step": 6601
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.999237077683415e-05,
+      "loss": 1.3805,
+      "step": 6602
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.999235560224067e-05,
+      "loss": 1.3864,
+      "step": 6603
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.999234041257213e-05,
+      "loss": 1.3541,
+      "step": 6604
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.999232520782854e-05,
+      "loss": 1.3837,
+      "step": 6605
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.999230998800988e-05,
+      "loss": 1.3333,
+      "step": 6606
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.99922947531162e-05,
+      "loss": 1.3956,
+      "step": 6607
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.999227950314747e-05,
+      "loss": 1.3284,
+      "step": 6608
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.999226423810373e-05,
+      "loss": 1.3706,
+      "step": 6609
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.999224895798494e-05,
+      "loss": 1.3667,
+      "step": 6610
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.999223366279112e-05,
+      "loss": 1.3212,
+      "step": 6611
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.999221835252228e-05,
+      "loss": 1.3979,
+      "step": 6612
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.999220302717842e-05,
+      "loss": 1.3691,
+      "step": 6613
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.999218768675955e-05,
+      "loss": 1.3849,
+      "step": 6614
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.999217233126567e-05,
+      "loss": 1.3962,
+      "step": 6615
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.999215696069679e-05,
+      "loss": 1.3774,
+      "step": 6616
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.999214157505292e-05,
+      "loss": 1.362,
+      "step": 6617
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.999212617433406e-05,
+      "loss": 1.3659,
+      "step": 6618
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.99921107585402e-05,
+      "loss": 1.3472,
+      "step": 6619
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.999209532767135e-05,
+      "loss": 1.3369,
+      "step": 6620
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.999207988172754e-05,
+      "loss": 1.3756,
+      "step": 6621
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.999206442070873e-05,
+      "loss": 1.3718,
+      "step": 6622
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.999204894461496e-05,
+      "loss": 1.3702,
+      "step": 6623
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.999203345344623e-05,
+      "loss": 1.3685,
+      "step": 6624
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.999201794720255e-05,
+      "loss": 1.3423,
+      "step": 6625
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.999200242588389e-05,
+      "loss": 1.3664,
+      "step": 6626
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.999198688949028e-05,
+      "loss": 1.3766,
+      "step": 6627
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.999197133802173e-05,
+      "loss": 1.3782,
+      "step": 6628
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.999195577147824e-05,
+      "loss": 1.367,
+      "step": 6629
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.999194018985983e-05,
+      "loss": 1.383,
+      "step": 6630
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.999192459316646e-05,
+      "loss": 1.3698,
+      "step": 6631
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.999190898139817e-05,
+      "loss": 1.3773,
+      "step": 6632
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.999189335455495e-05,
+      "loss": 1.37,
+      "step": 6633
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.999187771263682e-05,
+      "loss": 1.3202,
+      "step": 6634
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.999186205564377e-05,
+      "loss": 1.3518,
+      "step": 6635
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.999184638357581e-05,
+      "loss": 1.3793,
+      "step": 6636
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.999183069643296e-05,
+      "loss": 1.389,
+      "step": 6637
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.999181499421521e-05,
+      "loss": 1.3602,
+      "step": 6638
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.999179927692255e-05,
+      "loss": 1.3755,
+      "step": 6639
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.9991783544555e-05,
+      "loss": 1.3576,
+      "step": 6640
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.999176779711256e-05,
+      "loss": 1.3561,
+      "step": 6641
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.999175203459527e-05,
+      "loss": 1.3596,
+      "step": 6642
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.999173625700308e-05,
+      "loss": 1.4002,
+      "step": 6643
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.999172046433603e-05,
+      "loss": 1.3736,
+      "step": 6644
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.999170465659409e-05,
+      "loss": 1.358,
+      "step": 6645
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.99916888337773e-05,
+      "loss": 1.3229,
+      "step": 6646
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.999167299588566e-05,
+      "loss": 1.3676,
+      "step": 6647
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.999165714291915e-05,
+      "loss": 1.3429,
+      "step": 6648
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.999164127487783e-05,
+      "loss": 1.3743,
+      "step": 6649
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.999162539176164e-05,
+      "loss": 1.3722,
+      "step": 6650
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.999160949357062e-05,
+      "loss": 1.3823,
+      "step": 6651
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.999159358030475e-05,
+      "loss": 1.3428,
+      "step": 6652
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.999157765196408e-05,
+      "loss": 1.3545,
+      "step": 6653
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.999156170854856e-05,
+      "loss": 1.3674,
+      "step": 6654
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.999154575005824e-05,
+      "loss": 1.3374,
+      "step": 6655
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.999152977649309e-05,
+      "loss": 1.3595,
+      "step": 6656
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.999151378785314e-05,
+      "loss": 1.3686,
+      "step": 6657
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.999149778413838e-05,
+      "loss": 1.3857,
+      "step": 6658
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.999148176534883e-05,
+      "loss": 1.3494,
+      "step": 6659
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.999146573148448e-05,
+      "loss": 1.3713,
+      "step": 6660
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.999144968254535e-05,
+      "loss": 1.3728,
+      "step": 6661
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.999143361853143e-05,
+      "loss": 1.3478,
+      "step": 6662
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.999141753944272e-05,
+      "loss": 1.3384,
+      "step": 6663
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.999140144527923e-05,
+      "loss": 1.3569,
+      "step": 6664
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.999138533604099e-05,
+      "loss": 1.3763,
+      "step": 6665
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.999136921172798e-05,
+      "loss": 1.3727,
+      "step": 6666
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.99913530723402e-05,
+      "loss": 1.3832,
+      "step": 6667
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.999133691787767e-05,
+      "loss": 1.3867,
+      "step": 6668
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.999132074834039e-05,
+      "loss": 1.3523,
+      "step": 6669
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.999130456372836e-05,
+      "loss": 1.3831,
+      "step": 6670
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.999128836404159e-05,
+      "loss": 1.3703,
+      "step": 6671
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.999127214928008e-05,
+      "loss": 1.3783,
+      "step": 6672
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.999125591944385e-05,
+      "loss": 1.3647,
+      "step": 6673
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.999123967453288e-05,
+      "loss": 1.3848,
+      "step": 6674
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.99912234145472e-05,
+      "loss": 1.3468,
+      "step": 6675
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.99912071394868e-05,
+      "loss": 1.3418,
+      "step": 6676
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.999119084935169e-05,
+      "loss": 1.3265,
+      "step": 6677
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.999117454414186e-05,
+      "loss": 1.388,
+      "step": 6678
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.999115822385734e-05,
+      "loss": 1.348,
+      "step": 6679
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.999114188849811e-05,
+      "loss": 1.3588,
+      "step": 6680
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.999112553806421e-05,
+      "loss": 1.3476,
+      "step": 6681
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.999110917255561e-05,
+      "loss": 1.3683,
+      "step": 6682
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.999109279197233e-05,
+      "loss": 1.3577,
+      "step": 6683
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.999107639631438e-05,
+      "loss": 1.3447,
+      "step": 6684
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.999105998558174e-05,
+      "loss": 1.3665,
+      "step": 6685
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.999104355977444e-05,
+      "loss": 1.3475,
+      "step": 6686
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.999102711889247e-05,
+      "loss": 1.3824,
+      "step": 6687
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.999101066293586e-05,
+      "loss": 1.352,
+      "step": 6688
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.999099419190459e-05,
+      "loss": 1.3682,
+      "step": 6689
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.999097770579868e-05,
+      "loss": 1.3607,
+      "step": 6690
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.999096120461813e-05,
+      "loss": 1.3241,
+      "step": 6691
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.999094468836292e-05,
+      "loss": 1.3416,
+      "step": 6692
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.999092815703309e-05,
+      "loss": 1.3623,
+      "step": 6693
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.999091161062863e-05,
+      "loss": 1.3298,
+      "step": 6694
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.999089504914955e-05,
+      "loss": 1.3593,
+      "step": 6695
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.999087847259585e-05,
+      "loss": 1.3511,
+      "step": 6696
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.999086188096754e-05,
+      "loss": 1.3334,
+      "step": 6697
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.999084527426463e-05,
+      "loss": 1.3514,
+      "step": 6698
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.99908286524871e-05,
+      "loss": 1.4023,
+      "step": 6699
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.999081201563499e-05,
+      "loss": 1.3547,
+      "step": 6700
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.999079536370828e-05,
+      "loss": 1.3439,
+      "step": 6701
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.999077869670697e-05,
+      "loss": 1.3608,
+      "step": 6702
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.999076201463109e-05,
+      "loss": 1.3661,
+      "step": 6703
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.999074531748062e-05,
+      "loss": 1.327,
+      "step": 6704
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.99907286052556e-05,
+      "loss": 1.3626,
+      "step": 6705
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.9990711877956e-05,
+      "loss": 1.3257,
+      "step": 6706
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.999069513558184e-05,
+      "loss": 1.3727,
+      "step": 6707
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.999067837813312e-05,
+      "loss": 1.3422,
+      "step": 6708
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.999066160560984e-05,
+      "loss": 1.3591,
+      "step": 6709
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.999064481801202e-05,
+      "loss": 1.3691,
+      "step": 6710
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.999062801533967e-05,
+      "loss": 1.3576,
+      "step": 6711
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.999061119759276e-05,
+      "loss": 1.3757,
+      "step": 6712
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.999059436477135e-05,
+      "loss": 1.3131,
+      "step": 6713
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.999057751687537e-05,
+      "loss": 1.3238,
+      "step": 6714
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.999056065390489e-05,
+      "loss": 1.361,
+      "step": 6715
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.999054377585991e-05,
+      "loss": 1.3644,
+      "step": 6716
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.99905268827404e-05,
+      "loss": 1.3521,
+      "step": 6717
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.999050997454638e-05,
+      "loss": 1.3306,
+      "step": 6718
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.999049305127786e-05,
+      "loss": 1.3652,
+      "step": 6719
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.999047611293485e-05,
+      "loss": 1.3555,
+      "step": 6720
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.999045915951734e-05,
+      "loss": 1.3482,
+      "step": 6721
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.999044219102536e-05,
+      "loss": 1.3332,
+      "step": 6722
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.999042520745889e-05,
+      "loss": 1.3312,
+      "step": 6723
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.999040820881794e-05,
+      "loss": 1.3265,
+      "step": 6724
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.999039119510251e-05,
+      "loss": 1.3251,
+      "step": 6725
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.999037416631264e-05,
+      "loss": 1.356,
+      "step": 6726
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.99903571224483e-05,
+      "loss": 1.3614,
+      "step": 6727
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.99903400635095e-05,
+      "loss": 1.3696,
+      "step": 6728
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.999032298949624e-05,
+      "loss": 1.3755,
+      "step": 6729
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.999030590040854e-05,
+      "loss": 1.346,
+      "step": 6730
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.999028879624641e-05,
+      "loss": 1.3377,
+      "step": 6731
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.999027167700985e-05,
+      "loss": 1.3386,
+      "step": 6732
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.999025454269885e-05,
+      "loss": 1.3485,
+      "step": 6733
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.999023739331342e-05,
+      "loss": 1.4009,
+      "step": 6734
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.999022022885358e-05,
+      "loss": 1.3815,
+      "step": 6735
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.999020304931933e-05,
+      "loss": 1.3355,
+      "step": 6736
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.999018585471065e-05,
+      "loss": 1.3341,
+      "step": 6737
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.999016864502759e-05,
+      "loss": 1.3283,
+      "step": 6738
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.99901514202701e-05,
+      "loss": 1.3346,
+      "step": 6739
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.999013418043825e-05,
+      "loss": 1.3485,
+      "step": 6740
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.9990116925532e-05,
+      "loss": 1.3427,
+      "step": 6741
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.999009965555137e-05,
+      "loss": 1.3419,
+      "step": 6742
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.999008237049635e-05,
+      "loss": 1.3444,
+      "step": 6743
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.999006507036696e-05,
+      "loss": 1.3616,
+      "step": 6744
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.999004775516322e-05,
+      "loss": 1.3303,
+      "step": 6745
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.99900304248851e-05,
+      "loss": 1.3694,
+      "step": 6746
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.999001307953262e-05,
+      "loss": 1.3271,
+      "step": 6747
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.998999571910581e-05,
+      "loss": 1.3804,
+      "step": 6748
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.998997834360464e-05,
+      "loss": 1.3836,
+      "step": 6749
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.998996095302913e-05,
+      "loss": 1.3797,
+      "step": 6750
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.998994354737929e-05,
+      "loss": 1.3523,
+      "step": 6751
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.998992612665511e-05,
+      "loss": 1.3392,
+      "step": 6752
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.99899086908566e-05,
+      "loss": 1.3497,
+      "step": 6753
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.998989123998379e-05,
+      "loss": 1.3537,
+      "step": 6754
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.998987377403665e-05,
+      "loss": 1.3518,
+      "step": 6755
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.998985629301521e-05,
+      "loss": 1.3287,
+      "step": 6756
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.998983879691945e-05,
+      "loss": 1.3358,
+      "step": 6757
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.998982128574939e-05,
+      "loss": 1.3183,
+      "step": 6758
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.998980375950505e-05,
+      "loss": 1.3558,
+      "step": 6759
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.998978621818641e-05,
+      "loss": 1.373,
+      "step": 6760
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.99897686617935e-05,
+      "loss": 1.337,
+      "step": 6761
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.998975109032631e-05,
+      "loss": 1.3203,
+      "step": 6762
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.998973350378485e-05,
+      "loss": 1.3836,
+      "step": 6763
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.99897159021691e-05,
+      "loss": 1.3374,
+      "step": 6764
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.99896982854791e-05,
+      "loss": 1.3637,
+      "step": 6765
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.998968065371486e-05,
+      "loss": 1.3453,
+      "step": 6766
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.998966300687635e-05,
+      "loss": 1.3557,
+      "step": 6767
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.99896453449636e-05,
+      "loss": 1.3548,
+      "step": 6768
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.99896276679766e-05,
+      "loss": 1.3342,
+      "step": 6769
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.998960997591538e-05,
+      "loss": 1.3568,
+      "step": 6770
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.998959226877991e-05,
+      "loss": 1.3514,
+      "step": 6771
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.998957454657023e-05,
+      "loss": 1.3702,
+      "step": 6772
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.998955680928634e-05,
+      "loss": 1.3193,
+      "step": 6773
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.998953905692821e-05,
+      "loss": 1.3771,
+      "step": 6774
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.998952128949588e-05,
+      "loss": 1.3412,
+      "step": 6775
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.998950350698937e-05,
+      "loss": 1.3548,
+      "step": 6776
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.998948570940864e-05,
+      "loss": 1.3533,
+      "step": 6777
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.998946789675373e-05,
+      "loss": 1.3248,
+      "step": 6778
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.998945006902462e-05,
+      "loss": 1.3789,
+      "step": 6779
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.998943222622132e-05,
+      "loss": 1.3268,
+      "step": 6780
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.998941436834386e-05,
+      "loss": 1.3901,
+      "step": 6781
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.998939649539223e-05,
+      "loss": 1.3972,
+      "step": 6782
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.998937860736643e-05,
+      "loss": 1.3175,
+      "step": 6783
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.998936070426646e-05,
+      "loss": 1.3096,
+      "step": 6784
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.998934278609234e-05,
+      "loss": 1.3434,
+      "step": 6785
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.998932485284407e-05,
+      "loss": 1.3392,
+      "step": 6786
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.998930690452166e-05,
+      "loss": 1.345,
+      "step": 6787
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.998928894112512e-05,
+      "loss": 1.3629,
+      "step": 6788
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.998927096265442e-05,
+      "loss": 1.3348,
+      "step": 6789
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.998925296910962e-05,
+      "loss": 1.3371,
+      "step": 6790
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.998923496049069e-05,
+      "loss": 1.3438,
+      "step": 6791
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.998921693679764e-05,
+      "loss": 1.3363,
+      "step": 6792
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.998919889803047e-05,
+      "loss": 1.3209,
+      "step": 6793
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.99891808441892e-05,
+      "loss": 1.3521,
+      "step": 6794
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.998916277527383e-05,
+      "loss": 1.3159,
+      "step": 6795
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.998914469128438e-05,
+      "loss": 1.3397,
+      "step": 6796
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.998912659222082e-05,
+      "loss": 1.3069,
+      "step": 6797
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.998910847808319e-05,
+      "loss": 1.3395,
+      "step": 6798
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.998909034887146e-05,
+      "loss": 1.3178,
+      "step": 6799
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.998907220458566e-05,
+      "loss": 1.3338,
+      "step": 6800
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.998905404522581e-05,
+      "loss": 1.3611,
+      "step": 6801
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.998903587079189e-05,
+      "loss": 1.3483,
+      "step": 6802
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.998901768128391e-05,
+      "loss": 1.3577,
+      "step": 6803
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.998899947670189e-05,
+      "loss": 1.3188,
+      "step": 6804
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.998898125704581e-05,
+      "loss": 1.3333,
+      "step": 6805
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.998896302231568e-05,
+      "loss": 1.3043,
+      "step": 6806
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.998894477251155e-05,
+      "loss": 1.3504,
+      "step": 6807
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.998892650763336e-05,
+      "loss": 1.3286,
+      "step": 6808
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.998890822768114e-05,
+      "loss": 1.3414,
+      "step": 6809
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.998888993265493e-05,
+      "loss": 1.3467,
+      "step": 6810
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.99888716225547e-05,
+      "loss": 1.3313,
+      "step": 6811
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.998885329738045e-05,
+      "loss": 1.3755,
+      "step": 6812
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.998883495713221e-05,
+      "loss": 1.3479,
+      "step": 6813
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.998881660180995e-05,
+      "loss": 1.3328,
+      "step": 6814
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.998879823141371e-05,
+      "loss": 1.3377,
+      "step": 6815
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.99887798459435e-05,
+      "loss": 1.3723,
+      "step": 6816
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.99887614453993e-05,
+      "loss": 1.3442,
+      "step": 6817
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.998874302978113e-05,
+      "loss": 1.3312,
+      "step": 6818
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.998872459908899e-05,
+      "loss": 1.3501,
+      "step": 6819
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.998870615332287e-05,
+      "loss": 1.3311,
+      "step": 6820
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.998868769248281e-05,
+      "loss": 1.3555,
+      "step": 6821
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.99886692165688e-05,
+      "loss": 1.3427,
+      "step": 6822
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.998865072558085e-05,
+      "loss": 1.3174,
+      "step": 6823
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.998863221951894e-05,
+      "loss": 1.3207,
+      "step": 6824
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.99886136983831e-05,
+      "loss": 1.3296,
+      "step": 6825
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.998859516217333e-05,
+      "loss": 1.3593,
+      "step": 6826
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.998857661088966e-05,
+      "loss": 1.355,
+      "step": 6827
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.998855804453203e-05,
+      "loss": 1.3697,
+      "step": 6828
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.998853946310051e-05,
+      "loss": 1.3277,
+      "step": 6829
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.998852086659509e-05,
+      "loss": 1.3477,
+      "step": 6830
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.998850225501575e-05,
+      "loss": 1.3472,
+      "step": 6831
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.998848362836253e-05,
+      "loss": 1.3463,
+      "step": 6832
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.998846498663541e-05,
+      "loss": 1.3645,
+      "step": 6833
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.99884463298344e-05,
+      "loss": 1.3655,
+      "step": 6834
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.998842765795952e-05,
+      "loss": 1.3683,
+      "step": 6835
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.998840897101076e-05,
+      "loss": 1.3182,
+      "step": 6836
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.998839026898814e-05,
+      "loss": 1.3588,
+      "step": 6837
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.998837155189165e-05,
+      "loss": 1.3414,
+      "step": 6838
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.998835281972129e-05,
+      "loss": 1.2925,
+      "step": 6839
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.998833407247713e-05,
+      "loss": 1.3479,
+      "step": 6840
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.998831531015906e-05,
+      "loss": 1.3588,
+      "step": 6841
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.998829653276719e-05,
+      "loss": 1.334,
+      "step": 6842
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.998827774030148e-05,
+      "loss": 1.3538,
+      "step": 6843
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.998825893276193e-05,
+      "loss": 1.3423,
+      "step": 6844
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.998824011014857e-05,
+      "loss": 1.3503,
+      "step": 6845
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.998822127246139e-05,
+      "loss": 1.3378,
+      "step": 6846
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.998820241970039e-05,
+      "loss": 1.3228,
+      "step": 6847
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.99881835518656e-05,
+      "loss": 1.329,
+      "step": 6848
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.9988164668957e-05,
+      "loss": 1.3451,
+      "step": 6849
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.99881457709746e-05,
+      "loss": 1.3394,
+      "step": 6850
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.998812685791842e-05,
+      "loss": 1.3453,
+      "step": 6851
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.998810792978846e-05,
+      "loss": 1.3187,
+      "step": 6852
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.998808898658472e-05,
+      "loss": 1.3012,
+      "step": 6853
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.99880700283072e-05,
+      "loss": 1.3619,
+      "step": 6854
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.998805105495592e-05,
+      "loss": 1.3106,
+      "step": 6855
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.998803206653087e-05,
+      "loss": 1.3312,
+      "step": 6856
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.99880130630321e-05,
+      "loss": 1.3119,
+      "step": 6857
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.998799404445954e-05,
+      "loss": 1.3451,
+      "step": 6858
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.998797501081325e-05,
+      "loss": 1.3395,
+      "step": 6859
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.998795596209323e-05,
+      "loss": 1.3508,
+      "step": 6860
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.998793689829948e-05,
+      "loss": 1.334,
+      "step": 6861
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.9987917819432e-05,
+      "loss": 1.298,
+      "step": 6862
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.99878987254908e-05,
+      "loss": 1.3586,
+      "step": 6863
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.998787961647589e-05,
+      "loss": 1.3671,
+      "step": 6864
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.998786049238727e-05,
+      "loss": 1.3646,
+      "step": 6865
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.998784135322494e-05,
+      "loss": 1.3262,
+      "step": 6866
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.998782219898892e-05,
+      "loss": 1.3266,
+      "step": 6867
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.99878030296792e-05,
+      "loss": 1.3106,
+      "step": 6868
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.998778384529582e-05,
+      "loss": 1.3315,
+      "step": 6869
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.998776464583873e-05,
+      "loss": 1.3324,
+      "step": 6870
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.998774543130798e-05,
+      "loss": 1.3655,
+      "step": 6871
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.998772620170355e-05,
+      "loss": 1.333,
+      "step": 6872
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.998770695702549e-05,
+      "loss": 1.3749,
+      "step": 6873
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.998768769727375e-05,
+      "loss": 1.3485,
+      "step": 6874
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.998766842244835e-05,
+      "loss": 1.3192,
+      "step": 6875
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.998764913254932e-05,
+      "loss": 1.3871,
+      "step": 6876
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.998762982757665e-05,
+      "loss": 1.3311,
+      "step": 6877
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.998761050753034e-05,
+      "loss": 1.3453,
+      "step": 6878
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.998759117241043e-05,
+      "loss": 1.3617,
+      "step": 6879
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.998757182221687e-05,
+      "loss": 1.3194,
+      "step": 6880
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.99875524569497e-05,
+      "loss": 1.318,
+      "step": 6881
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.998753307660891e-05,
+      "loss": 1.3192,
+      "step": 6882
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.998751368119455e-05,
+      "loss": 1.3064,
+      "step": 6883
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.998749427070655e-05,
+      "loss": 1.297,
+      "step": 6884
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.998747484514499e-05,
+      "loss": 1.3164,
+      "step": 6885
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.998745540450983e-05,
+      "loss": 1.3271,
+      "step": 6886
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.998743594880108e-05,
+      "loss": 1.3229,
+      "step": 6887
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.998741647801877e-05,
+      "loss": 1.3338,
+      "step": 6888
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.998739699216289e-05,
+      "loss": 1.3374,
+      "step": 6889
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.998737749123345e-05,
+      "loss": 1.3483,
+      "step": 6890
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.998735797523045e-05,
+      "loss": 1.3497,
+      "step": 6891
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.998733844415389e-05,
+      "loss": 1.3509,
+      "step": 6892
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.998731889800379e-05,
+      "loss": 1.3596,
+      "step": 6893
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.998729933678015e-05,
+      "loss": 1.3543,
+      "step": 6894
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.998727976048299e-05,
+      "loss": 1.3183,
+      "step": 6895
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.998726016911229e-05,
+      "loss": 1.3083,
+      "step": 6896
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.998724056266806e-05,
+      "loss": 1.3668,
+      "step": 6897
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.998722094115033e-05,
+      "loss": 1.297,
+      "step": 6898
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.998720130455909e-05,
+      "loss": 1.3327,
+      "step": 6899
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.998718165289434e-05,
+      "loss": 1.3582,
+      "step": 6900
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.998716198615609e-05,
+      "loss": 1.3173,
+      "step": 6901
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.998714230434435e-05,
+      "loss": 1.3302,
+      "step": 6902
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.998712260745913e-05,
+      "loss": 1.3301,
+      "step": 6903
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.998710289550042e-05,
+      "loss": 1.3641,
+      "step": 6904
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.998708316846824e-05,
+      "loss": 1.3202,
+      "step": 6905
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.99870634263626e-05,
+      "loss": 1.3127,
+      "step": 6906
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.998704366918348e-05,
+      "loss": 1.3293,
+      "step": 6907
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.99870238969309e-05,
+      "loss": 1.3283,
+      "step": 6908
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.99870041096049e-05,
+      "loss": 1.3827,
+      "step": 6909
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.998698430720543e-05,
+      "loss": 1.3466,
+      "step": 6910
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.998696448973253e-05,
+      "loss": 1.3098,
+      "step": 6911
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.998694465718622e-05,
+      "loss": 1.3555,
+      "step": 6912
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.998692480956645e-05,
+      "loss": 1.3986,
+      "step": 6913
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.998690494687327e-05,
+      "loss": 1.4071,
+      "step": 6914
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.998688506910667e-05,
+      "loss": 1.3524,
+      "step": 6915
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.998686517626668e-05,
+      "loss": 1.3519,
+      "step": 6916
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.998684526835326e-05,
+      "loss": 1.3573,
+      "step": 6917
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.998682534536647e-05,
+      "loss": 1.3733,
+      "step": 6918
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.998680540730628e-05,
+      "loss": 1.33,
+      "step": 6919
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.99867854541727e-05,
+      "loss": 1.3321,
+      "step": 6920
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.998676548596575e-05,
+      "loss": 1.3521,
+      "step": 6921
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.998674550268544e-05,
+      "loss": 1.3587,
+      "step": 6922
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.998672550433173e-05,
+      "loss": 1.3813,
+      "step": 6923
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.998670549090467e-05,
+      "loss": 1.3348,
+      "step": 6924
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.998668546240427e-05,
+      "loss": 1.3528,
+      "step": 6925
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.998666541883051e-05,
+      "loss": 1.3266,
+      "step": 6926
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.998664536018341e-05,
+      "loss": 1.3407,
+      "step": 6927
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.998662528646299e-05,
+      "loss": 1.3032,
+      "step": 6928
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.998660519766923e-05,
+      "loss": 1.3401,
+      "step": 6929
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.998658509380213e-05,
+      "loss": 1.3383,
+      "step": 6930
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.998656497486174e-05,
+      "loss": 1.3383,
+      "step": 6931
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.9986544840848e-05,
+      "loss": 1.3441,
+      "step": 6932
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.998652469176099e-05,
+      "loss": 1.3336,
+      "step": 6933
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.998650452760066e-05,
+      "loss": 1.3543,
+      "step": 6934
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.998648434836704e-05,
+      "loss": 1.3612,
+      "step": 6935
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.998646415406015e-05,
+      "loss": 1.3267,
+      "step": 6936
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.998644394467996e-05,
+      "loss": 1.3147,
+      "step": 6937
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.998642372022648e-05,
+      "loss": 1.3351,
+      "step": 6938
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.998640348069977e-05,
+      "loss": 1.3494,
+      "step": 6939
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.998638322609977e-05,
+      "loss": 1.3293,
+      "step": 6940
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.998636295642651e-05,
+      "loss": 1.336,
+      "step": 6941
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.998634267168001e-05,
+      "loss": 1.3164,
+      "step": 6942
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.998632237186026e-05,
+      "loss": 1.332,
+      "step": 6943
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.998630205696728e-05,
+      "loss": 1.3166,
+      "step": 6944
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.998628172700106e-05,
+      "loss": 1.3405,
+      "step": 6945
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.998626138196161e-05,
+      "loss": 1.3642,
+      "step": 6946
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.998624102184895e-05,
+      "loss": 1.332,
+      "step": 6947
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.998622064666306e-05,
+      "loss": 1.3505,
+      "step": 6948
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.998620025640398e-05,
+      "loss": 1.311,
+      "step": 6949
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.998617985107169e-05,
+      "loss": 1.3508,
+      "step": 6950
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.998615943066621e-05,
+      "loss": 1.3483,
+      "step": 6951
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.998613899518751e-05,
+      "loss": 1.3408,
+      "step": 6952
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.998611854463567e-05,
+      "loss": 1.3017,
+      "step": 6953
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.998609807901063e-05,
+      "loss": 1.3173,
+      "step": 6954
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.998607759831242e-05,
+      "loss": 1.3619,
+      "step": 6955
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.998605710254105e-05,
+      "loss": 1.3509,
+      "step": 6956
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.998603659169653e-05,
+      "loss": 1.3233,
+      "step": 6957
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.998601606577884e-05,
+      "loss": 1.3625,
+      "step": 6958
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.9985995524788e-05,
+      "loss": 1.3624,
+      "step": 6959
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.998597496872403e-05,
+      "loss": 1.3432,
+      "step": 6960
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.998595439758692e-05,
+      "loss": 1.333,
+      "step": 6961
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.998593381137668e-05,
+      "loss": 1.3165,
+      "step": 6962
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.998591321009335e-05,
+      "loss": 1.3473,
+      "step": 6963
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.998589259373688e-05,
+      "loss": 1.3631,
+      "step": 6964
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.998587196230729e-05,
+      "loss": 1.3801,
+      "step": 6965
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.99858513158046e-05,
+      "loss": 1.368,
+      "step": 6966
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.998583065422883e-05,
+      "loss": 1.3448,
+      "step": 6967
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.998580997757996e-05,
+      "loss": 1.3487,
+      "step": 6968
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.9985789285858e-05,
+      "loss": 1.3204,
+      "step": 6969
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.998576857906298e-05,
+      "loss": 1.327,
+      "step": 6970
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.998574785719487e-05,
+      "loss": 1.3287,
+      "step": 6971
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.998572712025371e-05,
+      "loss": 1.3668,
+      "step": 6972
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.998570636823948e-05,
+      "loss": 1.3463,
+      "step": 6973
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.99856856011522e-05,
+      "loss": 1.3214,
+      "step": 6974
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.998566481899186e-05,
+      "loss": 1.3368,
+      "step": 6975
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.99856440217585e-05,
+      "loss": 1.328,
+      "step": 6976
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.99856232094521e-05,
+      "loss": 1.3283,
+      "step": 6977
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.998560238207266e-05,
+      "loss": 1.2964,
+      "step": 6978
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.998558153962021e-05,
+      "loss": 1.3264,
+      "step": 6979
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.998556068209475e-05,
+      "loss": 1.3181,
+      "step": 6980
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.998553980949628e-05,
+      "loss": 1.3436,
+      "step": 6981
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.99855189218248e-05,
+      "loss": 1.3563,
+      "step": 6982
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.998549801908032e-05,
+      "loss": 1.3356,
+      "step": 6983
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.998547710126285e-05,
+      "loss": 1.3294,
+      "step": 6984
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.99854561683724e-05,
+      "loss": 1.3511,
+      "step": 6985
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.998543522040897e-05,
+      "loss": 1.311,
+      "step": 6986
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.998541425737257e-05,
+      "loss": 1.3003,
+      "step": 6987
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.998539327926321e-05,
+      "loss": 1.3136,
+      "step": 6988
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.998537228608088e-05,
+      "loss": 1.3545,
+      "step": 6989
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.99853512778256e-05,
+      "loss": 1.3399,
+      "step": 6990
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.998533025449739e-05,
+      "loss": 1.311,
+      "step": 6991
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.998530921609624e-05,
+      "loss": 1.3607,
+      "step": 6992
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.998528816262214e-05,
+      "loss": 1.3388,
+      "step": 6993
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.998526709407511e-05,
+      "loss": 1.3344,
+      "step": 6994
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.998524601045518e-05,
+      "loss": 1.3225,
+      "step": 6995
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.998522491176232e-05,
+      "loss": 1.3066,
+      "step": 6996
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.998520379799656e-05,
+      "loss": 1.308,
+      "step": 6997
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.998518266915789e-05,
+      "loss": 1.3568,
+      "step": 6998
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.998516152524633e-05,
+      "loss": 1.3078,
+      "step": 6999
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.998514036626188e-05,
+      "loss": 1.3239,
+      "step": 7000
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.998511919220454e-05,
+      "loss": 1.3223,
+      "step": 7001
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.998509800307434e-05,
+      "loss": 1.3486,
+      "step": 7002
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.998507679887127e-05,
+      "loss": 1.314,
+      "step": 7003
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.998505557959533e-05,
+      "loss": 1.3437,
+      "step": 7004
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.998503434524654e-05,
+      "loss": 1.3544,
+      "step": 7005
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.998501309582489e-05,
+      "loss": 1.3219,
+      "step": 7006
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.998499183133039e-05,
+      "loss": 1.3635,
+      "step": 7007
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.998497055176306e-05,
+      "loss": 1.2903,
+      "step": 7008
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.998494925712291e-05,
+      "loss": 1.3545,
+      "step": 7009
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.998492794740992e-05,
+      "loss": 1.3216,
+      "step": 7010
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.998490662262411e-05,
+      "loss": 1.3728,
+      "step": 7011
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.998488528276551e-05,
+      "loss": 1.3375,
+      "step": 7012
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.998486392783408e-05,
+      "loss": 1.357,
+      "step": 7013
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.998484255782987e-05,
+      "loss": 1.3355,
+      "step": 7014
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.998482117275285e-05,
+      "loss": 1.323,
+      "step": 7015
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.998479977260305e-05,
+      "loss": 1.3138,
+      "step": 7016
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.998477835738047e-05,
+      "loss": 1.3231,
+      "step": 7017
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.998475692708513e-05,
+      "loss": 1.3426,
+      "step": 7018
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.9984735481717e-05,
+      "loss": 1.3455,
+      "step": 7019
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.998471402127614e-05,
+      "loss": 1.3444,
+      "step": 7020
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.99846925457625e-05,
+      "loss": 1.3629,
+      "step": 7021
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.998467105517613e-05,
+      "loss": 1.3225,
+      "step": 7022
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.998464954951701e-05,
+      "loss": 1.3314,
+      "step": 7023
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.998462802878516e-05,
+      "loss": 1.324,
+      "step": 7024
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.998460649298058e-05,
+      "loss": 1.3541,
+      "step": 7025
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.998458494210327e-05,
+      "loss": 1.3165,
+      "step": 7026
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.998456337615325e-05,
+      "loss": 1.3378,
+      "step": 7027
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.998454179513053e-05,
+      "loss": 1.316,
+      "step": 7028
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.99845201990351e-05,
+      "loss": 1.3422,
+      "step": 7029
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.998449858786697e-05,
+      "loss": 1.333,
+      "step": 7030
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.998447696162616e-05,
+      "loss": 1.3261,
+      "step": 7031
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.998445532031266e-05,
+      "loss": 1.3326,
+      "step": 7032
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.998443366392649e-05,
+      "loss": 1.3557,
+      "step": 7033
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.998441199246766e-05,
+      "loss": 1.3477,
+      "step": 7034
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.998439030593615e-05,
+      "loss": 1.3155,
+      "step": 7035
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.9984368604332e-05,
+      "loss": 1.3501,
+      "step": 7036
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.99843468876552e-05,
+      "loss": 1.3066,
+      "step": 7037
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.998432515590575e-05,
+      "loss": 1.3194,
+      "step": 7038
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.998430340908365e-05,
+      "loss": 1.358,
+      "step": 7039
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.998428164718894e-05,
+      "loss": 1.3517,
+      "step": 7040
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.99842598702216e-05,
+      "loss": 1.3325,
+      "step": 7041
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.998423807818164e-05,
+      "loss": 1.3473,
+      "step": 7042
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.998421627106907e-05,
+      "loss": 1.2848,
+      "step": 7043
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.99841944488839e-05,
+      "loss": 1.3223,
+      "step": 7044
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.998417261162612e-05,
+      "loss": 1.3278,
+      "step": 7045
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.998415075929576e-05,
+      "loss": 1.3235,
+      "step": 7046
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.998412889189282e-05,
+      "loss": 1.3191,
+      "step": 7047
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.998410700941729e-05,
+      "loss": 1.3357,
+      "step": 7048
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.998408511186921e-05,
+      "loss": 1.3227,
+      "step": 7049
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.998406319924855e-05,
+      "loss": 1.3505,
+      "step": 7050
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.998404127155534e-05,
+      "loss": 1.4077,
+      "step": 7051
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.998401932878958e-05,
+      "loss": 1.3227,
+      "step": 7052
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.998399737095127e-05,
+      "loss": 1.3205,
+      "step": 7053
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.998397539804043e-05,
+      "loss": 1.3417,
+      "step": 7054
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.998395341005705e-05,
+      "loss": 1.3544,
+      "step": 7055
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.998393140700114e-05,
+      "loss": 1.3218,
+      "step": 7056
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.998390938887273e-05,
+      "loss": 1.3501,
+      "step": 7057
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.998388735567179e-05,
+      "loss": 1.353,
+      "step": 7058
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.998386530739836e-05,
+      "loss": 1.3132,
+      "step": 7059
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.998384324405243e-05,
+      "loss": 1.3573,
+      "step": 7060
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.998382116563401e-05,
+      "loss": 1.3155,
+      "step": 7061
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.99837990721431e-05,
+      "loss": 1.3519,
+      "step": 7062
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.99837769635797e-05,
+      "loss": 1.3579,
+      "step": 7063
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.998375483994384e-05,
+      "loss": 1.3428,
+      "step": 7064
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.998373270123553e-05,
+      "loss": 1.3184,
+      "step": 7065
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.998371054745474e-05,
+      "loss": 1.3438,
+      "step": 7066
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.998368837860151e-05,
+      "loss": 1.3465,
+      "step": 7067
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.998366619467584e-05,
+      "loss": 1.3754,
+      "step": 7068
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.998364399567772e-05,
+      "loss": 1.3113,
+      "step": 7069
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.998362178160718e-05,
+      "loss": 1.3176,
+      "step": 7070
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.998359955246422e-05,
+      "loss": 1.3402,
+      "step": 7071
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.998357730824884e-05,
+      "loss": 1.3234,
+      "step": 7072
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.998355504896102e-05,
+      "loss": 1.3466,
+      "step": 7073
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.998353277460083e-05,
+      "loss": 1.3295,
+      "step": 7074
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.998351048516822e-05,
+      "loss": 1.3439,
+      "step": 7075
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.998348818066323e-05,
+      "loss": 1.3809,
+      "step": 7076
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.998346586108585e-05,
+      "loss": 1.3564,
+      "step": 7077
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.99834435264361e-05,
+      "loss": 1.3735,
+      "step": 7078
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.998342117671397e-05,
+      "loss": 1.3539,
+      "step": 7079
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.998339881191949e-05,
+      "loss": 1.3789,
+      "step": 7080
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.998337643205264e-05,
+      "loss": 1.3406,
+      "step": 7081
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.998335403711345e-05,
+      "loss": 1.353,
+      "step": 7082
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.998333162710191e-05,
+      "loss": 1.3049,
+      "step": 7083
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.998330920201804e-05,
+      "loss": 1.3302,
+      "step": 7084
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.998328676186184e-05,
+      "loss": 1.3071,
+      "step": 7085
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.998326430663332e-05,
+      "loss": 1.3275,
+      "step": 7086
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.998324183633246e-05,
+      "loss": 1.3361,
+      "step": 7087
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.99832193509593e-05,
+      "loss": 1.3006,
+      "step": 7088
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.998319685051384e-05,
+      "loss": 1.2832,
+      "step": 7089
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.998317433499609e-05,
+      "loss": 1.3365,
+      "step": 7090
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.998315180440605e-05,
+      "loss": 1.3623,
+      "step": 7091
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.998312925874372e-05,
+      "loss": 1.3179,
+      "step": 7092
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.998310669800912e-05,
+      "loss": 1.3489,
+      "step": 7093
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.998308412220225e-05,
+      "loss": 1.3351,
+      "step": 7094
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.998306153132311e-05,
+      "loss": 1.3275,
+      "step": 7095
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.998303892537174e-05,
+      "loss": 1.3518,
+      "step": 7096
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.998301630434808e-05,
+      "loss": 1.3398,
+      "step": 7097
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.99829936682522e-05,
+      "loss": 1.3435,
+      "step": 7098
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.998297101708409e-05,
+      "loss": 1.3451,
+      "step": 7099
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.998294835084374e-05,
+      "loss": 1.331,
+      "step": 7100
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.998292566953119e-05,
+      "loss": 1.3485,
+      "step": 7101
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.99829029731464e-05,
+      "loss": 1.3162,
+      "step": 7102
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.998288026168942e-05,
+      "loss": 1.3493,
+      "step": 7103
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.998285753516024e-05,
+      "loss": 1.3328,
+      "step": 7104
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.998283479355886e-05,
+      "loss": 1.3172,
+      "step": 7105
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.998281203688529e-05,
+      "loss": 1.3214,
+      "step": 7106
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.998278926513954e-05,
+      "loss": 1.3291,
+      "step": 7107
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.998276647832161e-05,
+      "loss": 1.3659,
+      "step": 7108
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.998274367643152e-05,
+      "loss": 1.3258,
+      "step": 7109
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.998272085946926e-05,
+      "loss": 1.3118,
+      "step": 7110
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.998269802743488e-05,
+      "loss": 1.3561,
+      "step": 7111
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.998267518032832e-05,
+      "loss": 1.3482,
+      "step": 7112
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.998265231814963e-05,
+      "loss": 1.3304,
+      "step": 7113
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.998262944089882e-05,
+      "loss": 1.3295,
+      "step": 7114
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.998260654857587e-05,
+      "loss": 1.3321,
+      "step": 7115
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.99825836411808e-05,
+      "loss": 1.3426,
+      "step": 7116
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.998256071871364e-05,
+      "loss": 1.3375,
+      "step": 7117
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.998253778117436e-05,
+      "loss": 1.3426,
+      "step": 7118
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.998251482856298e-05,
+      "loss": 1.3506,
+      "step": 7119
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.998249186087952e-05,
+      "loss": 1.3465,
+      "step": 7120
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.998246887812397e-05,
+      "loss": 1.3127,
+      "step": 7121
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.998244588029635e-05,
+      "loss": 1.3303,
+      "step": 7122
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.998242286739665e-05,
+      "loss": 1.3552,
+      "step": 7123
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.99823998394249e-05,
+      "loss": 1.3557,
+      "step": 7124
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.998237679638108e-05,
+      "loss": 1.3093,
+      "step": 7125
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.998235373826521e-05,
+      "loss": 1.3438,
+      "step": 7126
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.998233066507729e-05,
+      "loss": 1.3286,
+      "step": 7127
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.998230757681737e-05,
+      "loss": 1.3178,
+      "step": 7128
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.998228447348538e-05,
+      "loss": 1.323,
+      "step": 7129
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.99822613550814e-05,
+      "loss": 1.3363,
+      "step": 7130
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.99822382216054e-05,
+      "loss": 1.3143,
+      "step": 7131
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.998221507305738e-05,
+      "loss": 1.3246,
+      "step": 7132
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.998219190943738e-05,
+      "loss": 1.3466,
+      "step": 7133
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.998216873074538e-05,
+      "loss": 1.3126,
+      "step": 7134
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.998214553698137e-05,
+      "loss": 1.3335,
+      "step": 7135
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.998212232814541e-05,
+      "loss": 1.3055,
+      "step": 7136
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.998209910423746e-05,
+      "loss": 1.3172,
+      "step": 7137
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.998207586525755e-05,
+      "loss": 1.3221,
+      "step": 7138
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.998205261120568e-05,
+      "loss": 1.3115,
+      "step": 7139
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.998202934208187e-05,
+      "loss": 1.337,
+      "step": 7140
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.99820060578861e-05,
+      "loss": 1.3271,
+      "step": 7141
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.998198275861841e-05,
+      "loss": 1.3427,
+      "step": 7142
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.998195944427878e-05,
+      "loss": 1.3288,
+      "step": 7143
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.998193611486723e-05,
+      "loss": 1.3068,
+      "step": 7144
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.998191277038376e-05,
+      "loss": 1.366,
+      "step": 7145
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.998188941082838e-05,
+      "loss": 1.3663,
+      "step": 7146
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.99818660362011e-05,
+      "loss": 1.2819,
+      "step": 7147
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.998184264650192e-05,
+      "loss": 1.342,
+      "step": 7148
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.998181924173085e-05,
+      "loss": 1.3562,
+      "step": 7149
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.998179582188791e-05,
+      "loss": 1.3103,
+      "step": 7150
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.99817723869731e-05,
+      "loss": 1.3246,
+      "step": 7151
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.998174893698642e-05,
+      "loss": 1.3157,
+      "step": 7152
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.998172547192787e-05,
+      "loss": 1.3445,
+      "step": 7153
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.998170199179747e-05,
+      "loss": 1.309,
+      "step": 7154
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.998167849659523e-05,
+      "loss": 1.3399,
+      "step": 7155
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.998165498632115e-05,
+      "loss": 1.3141,
+      "step": 7156
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.998163146097523e-05,
+      "loss": 1.3373,
+      "step": 7157
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.99816079205575e-05,
+      "loss": 1.3303,
+      "step": 7158
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.998158436506795e-05,
+      "loss": 1.3414,
+      "step": 7159
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.998156079450658e-05,
+      "loss": 1.3156,
+      "step": 7160
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.998153720887342e-05,
+      "loss": 1.3321,
+      "step": 7161
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.998151360816845e-05,
+      "loss": 1.3039,
+      "step": 7162
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.99814899923917e-05,
+      "loss": 1.3236,
+      "step": 7163
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.998146636154318e-05,
+      "loss": 1.3065,
+      "step": 7164
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.998144271562286e-05,
+      "loss": 1.3315,
+      "step": 7165
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.998141905463079e-05,
+      "loss": 1.3163,
+      "step": 7166
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.998139537856695e-05,
+      "loss": 1.3377,
+      "step": 7167
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.998137168743135e-05,
+      "loss": 1.343,
+      "step": 7168
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.998134798122403e-05,
+      "loss": 1.3566,
+      "step": 7169
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.998132425994495e-05,
+      "loss": 1.3179,
+      "step": 7170
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.998130052359416e-05,
+      "loss": 1.305,
+      "step": 7171
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.998127677217163e-05,
+      "loss": 1.3562,
+      "step": 7172
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.998125300567738e-05,
+      "loss": 1.3187,
+      "step": 7173
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.998122922411143e-05,
+      "loss": 1.3519,
+      "step": 7174
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.998120542747377e-05,
+      "loss": 1.3663,
+      "step": 7175
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.998118161576442e-05,
+      "loss": 1.3318,
+      "step": 7176
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.998115778898337e-05,
+      "loss": 1.3304,
+      "step": 7177
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.998113394713064e-05,
+      "loss": 1.3511,
+      "step": 7178
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.998111009020624e-05,
+      "loss": 1.3135,
+      "step": 7179
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.998108621821019e-05,
+      "loss": 1.3258,
+      "step": 7180
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.998106233114246e-05,
+      "loss": 1.3336,
+      "step": 7181
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.998103842900308e-05,
+      "loss": 1.3684,
+      "step": 7182
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.998101451179205e-05,
+      "loss": 1.3398,
+      "step": 7183
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.998099057950938e-05,
+      "loss": 1.355,
+      "step": 7184
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.99809666321551e-05,
+      "loss": 1.3309,
+      "step": 7185
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.998094266972917e-05,
+      "loss": 1.3551,
+      "step": 7186
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.998091869223165e-05,
+      "loss": 1.3327,
+      "step": 7187
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.99808946996625e-05,
+      "loss": 1.3064,
+      "step": 7188
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.998087069202175e-05,
+      "loss": 1.3136,
+      "step": 7189
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.998084666930943e-05,
+      "loss": 1.3424,
+      "step": 7190
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.998082263152548e-05,
+      "loss": 1.3079,
+      "step": 7191
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.998079857866998e-05,
+      "loss": 1.2913,
+      "step": 7192
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.998077451074291e-05,
+      "loss": 1.3576,
+      "step": 7193
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.998075042774427e-05,
+      "loss": 1.3055,
+      "step": 7194
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.998072632967406e-05,
+      "loss": 1.3058,
+      "step": 7195
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.998070221653229e-05,
+      "loss": 1.3202,
+      "step": 7196
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.998067808831899e-05,
+      "loss": 1.3138,
+      "step": 7197
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.998065394503415e-05,
+      "loss": 1.3097,
+      "step": 7198
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.998062978667779e-05,
+      "loss": 1.327,
+      "step": 7199
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.99806056132499e-05,
+      "loss": 1.3239,
+      "step": 7200
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.99805814247505e-05,
+      "loss": 1.3204,
+      "step": 7201
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.998055722117959e-05,
+      "loss": 1.3565,
+      "step": 7202
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.998053300253718e-05,
+      "loss": 1.3068,
+      "step": 7203
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.998050876882328e-05,
+      "loss": 1.3772,
+      "step": 7204
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.998048452003788e-05,
+      "loss": 1.2991,
+      "step": 7205
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.998046025618102e-05,
+      "loss": 1.326,
+      "step": 7206
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.998043597725267e-05,
+      "loss": 1.3445,
+      "step": 7207
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.998041168325288e-05,
+      "loss": 1.3244,
+      "step": 7208
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.99803873741816e-05,
+      "loss": 1.3263,
+      "step": 7209
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.99803630500389e-05,
+      "loss": 1.3191,
+      "step": 7210
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.998033871082476e-05,
+      "loss": 1.3167,
+      "step": 7211
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.998031435653916e-05,
+      "loss": 1.321,
+      "step": 7212
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.998028998718214e-05,
+      "loss": 1.3287,
+      "step": 7213
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.998026560275371e-05,
+      "loss": 1.3377,
+      "step": 7214
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.998024120325386e-05,
+      "loss": 1.3355,
+      "step": 7215
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.998021678868263e-05,
+      "loss": 1.3105,
+      "step": 7216
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.998019235903997e-05,
+      "loss": 1.3034,
+      "step": 7217
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.998016791432595e-05,
+      "loss": 1.3534,
+      "step": 7218
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.998014345454053e-05,
+      "loss": 1.3089,
+      "step": 7219
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.998011897968372e-05,
+      "loss": 1.3415,
+      "step": 7220
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.998009448975557e-05,
+      "loss": 1.3056,
+      "step": 7221
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.998006998475603e-05,
+      "loss": 1.3418,
+      "step": 7222
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.998004546468516e-05,
+      "loss": 1.3588,
+      "step": 7223
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.998002092954293e-05,
+      "loss": 1.3455,
+      "step": 7224
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.997999637932937e-05,
+      "loss": 1.3349,
+      "step": 7225
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.997997181404447e-05,
+      "loss": 1.2981,
+      "step": 7226
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.997994723368826e-05,
+      "loss": 1.3354,
+      "step": 7227
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.997992263826073e-05,
+      "loss": 1.3628,
+      "step": 7228
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.997989802776188e-05,
+      "loss": 1.3768,
+      "step": 7229
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.997987340219173e-05,
+      "loss": 1.3235,
+      "step": 7230
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.997984876155029e-05,
+      "loss": 1.3369,
+      "step": 7231
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.997982410583757e-05,
+      "loss": 1.3165,
+      "step": 7232
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.997979943505356e-05,
+      "loss": 1.3412,
+      "step": 7233
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.997977474919828e-05,
+      "loss": 1.3382,
+      "step": 7234
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.997975004827174e-05,
+      "loss": 1.3262,
+      "step": 7235
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.997972533227394e-05,
+      "loss": 1.3536,
+      "step": 7236
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.99797006012049e-05,
+      "loss": 1.3305,
+      "step": 7237
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.997967585506461e-05,
+      "loss": 1.3311,
+      "step": 7238
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.997965109385308e-05,
+      "loss": 1.3268,
+      "step": 7239
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.997962631757033e-05,
+      "loss": 1.3689,
+      "step": 7240
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.997960152621636e-05,
+      "loss": 1.3259,
+      "step": 7241
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.997957671979119e-05,
+      "loss": 1.3457,
+      "step": 7242
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.997955189829479e-05,
+      "loss": 1.3542,
+      "step": 7243
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.99795270617272e-05,
+      "loss": 1.3322,
+      "step": 7244
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.997950221008843e-05,
+      "loss": 1.3493,
+      "step": 7245
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.997947734337847e-05,
+      "loss": 1.321,
+      "step": 7246
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.997945246159735e-05,
+      "loss": 1.3261,
+      "step": 7247
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.997942756474504e-05,
+      "loss": 1.3209,
+      "step": 7248
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.997940265282158e-05,
+      "loss": 1.3607,
+      "step": 7249
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.997937772582698e-05,
+      "loss": 1.3648,
+      "step": 7250
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.997935278376123e-05,
+      "loss": 1.3783,
+      "step": 7251
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.997932782662434e-05,
+      "loss": 1.3451,
+      "step": 7252
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.99793028544163e-05,
+      "loss": 1.3568,
+      "step": 7253
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.997927786713716e-05,
+      "loss": 1.3028,
+      "step": 7254
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.997925286478692e-05,
+      "loss": 1.2961,
+      "step": 7255
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.997922784736553e-05,
+      "loss": 1.3408,
+      "step": 7256
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.997920281487308e-05,
+      "loss": 1.3497,
+      "step": 7257
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.997917776730951e-05,
+      "loss": 1.3521,
+      "step": 7258
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.997915270467489e-05,
+      "loss": 1.3039,
+      "step": 7259
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.997912762696916e-05,
+      "loss": 1.3639,
+      "step": 7260
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.997910253419237e-05,
+      "loss": 1.3183,
+      "step": 7261
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.997907742634452e-05,
+      "loss": 1.3438,
+      "step": 7262
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.997905230342561e-05,
+      "loss": 1.3537,
+      "step": 7263
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.997902716543569e-05,
+      "loss": 1.3415,
+      "step": 7264
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.99790020123747e-05,
+      "loss": 1.2935,
+      "step": 7265
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.997897684424267e-05,
+      "loss": 1.3433,
+      "step": 7266
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.997895166103962e-05,
+      "loss": 1.293,
+      "step": 7267
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.997892646276556e-05,
+      "loss": 1.2963,
+      "step": 7268
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.997890124942048e-05,
+      "loss": 1.3132,
+      "step": 7269
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.997887602100442e-05,
+      "loss": 1.3068,
+      "step": 7270
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.997885077751736e-05,
+      "loss": 1.3292,
+      "step": 7271
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.997882551895931e-05,
+      "loss": 1.3348,
+      "step": 7272
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.997880024533028e-05,
+      "loss": 1.3512,
+      "step": 7273
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.997877495663028e-05,
+      "loss": 1.3249,
+      "step": 7274
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.997874965285933e-05,
+      "loss": 1.3321,
+      "step": 7275
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.997872433401742e-05,
+      "loss": 1.3572,
+      "step": 7276
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.997869900010455e-05,
+      "loss": 1.3501,
+      "step": 7277
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.997867365112075e-05,
+      "loss": 1.3188,
+      "step": 7278
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.997864828706601e-05,
+      "loss": 1.3009,
+      "step": 7279
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.997862290794035e-05,
+      "loss": 1.315,
+      "step": 7280
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.997859751374378e-05,
+      "loss": 1.3386,
+      "step": 7281
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.99785721044763e-05,
+      "loss": 1.3024,
+      "step": 7282
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.997854668013791e-05,
+      "loss": 1.351,
+      "step": 7283
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.997852124072862e-05,
+      "loss": 1.3414,
+      "step": 7284
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.997849578624845e-05,
+      "loss": 1.3229,
+      "step": 7285
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.997847031669741e-05,
+      "loss": 1.3214,
+      "step": 7286
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.99784448320755e-05,
+      "loss": 1.3335,
+      "step": 7287
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.997841933238271e-05,
+      "loss": 1.3407,
+      "step": 7288
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.997839381761907e-05,
+      "loss": 1.319,
+      "step": 7289
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.997836828778458e-05,
+      "loss": 1.3391,
+      "step": 7290
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.997834274287925e-05,
+      "loss": 1.3527,
+      "step": 7291
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.997831718290311e-05,
+      "loss": 1.3131,
+      "step": 7292
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.997829160785612e-05,
+      "loss": 1.2855,
+      "step": 7293
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.997826601773833e-05,
+      "loss": 1.3625,
+      "step": 7294
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.99782404125497e-05,
+      "loss": 1.3032,
+      "step": 7295
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.99782147922903e-05,
+      "loss": 1.3459,
+      "step": 7296
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.99781891569601e-05,
+      "loss": 1.3365,
+      "step": 7297
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.99781635065591e-05,
+      "loss": 1.3454,
+      "step": 7298
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.997813784108734e-05,
+      "loss": 1.3385,
+      "step": 7299
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.99781121605448e-05,
+      "loss": 1.3108,
+      "step": 7300
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.99780864649315e-05,
+      "loss": 1.3678,
+      "step": 7301
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.997806075424744e-05,
+      "loss": 1.3455,
+      "step": 7302
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.997803502849264e-05,
+      "loss": 1.3648,
+      "step": 7303
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.997800928766708e-05,
+      "loss": 1.333,
+      "step": 7304
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 9.997798353177082e-05,
+      "loss": 1.3357,
+      "step": 7305
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 9.997795776080381e-05,
+      "loss": 1.3501,
+      "step": 7306
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 9.997793197476609e-05,
+      "loss": 1.343,
+      "step": 7307
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 9.997790617365766e-05,
+      "loss": 1.3349,
+      "step": 7308
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 9.997788035747852e-05,
+      "loss": 1.2923,
+      "step": 7309
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 9.99778545262287e-05,
+      "loss": 1.3495,
+      "step": 7310
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 9.99778286799082e-05,
+      "loss": 1.3161,
+      "step": 7311
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 9.997780281851702e-05,
+      "loss": 1.3294,
+      "step": 7312
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 9.997777694205517e-05,
+      "loss": 1.2971,
+      "step": 7313
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 9.997775105052264e-05,
+      "loss": 1.3267,
+      "step": 7314
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 9.997772514391947e-05,
+      "loss": 1.3394,
+      "step": 7315
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 9.997769922224566e-05,
+      "loss": 1.3117,
+      "step": 7316
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 9.99776732855012e-05,
+      "loss": 1.3487,
+      "step": 7317
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 9.997764733368613e-05,
+      "loss": 1.3512,
+      "step": 7318
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 9.997762136680043e-05,
+      "loss": 1.325,
+      "step": 7319
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 9.99775953848441e-05,
+      "loss": 1.3225,
+      "step": 7320
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 9.997756938781718e-05,
+      "loss": 1.3354,
+      "step": 7321
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 9.997754337571963e-05,
+      "loss": 1.3204,
+      "step": 7322
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 9.997751734855151e-05,
+      "loss": 1.3144,
+      "step": 7323
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 9.997749130631281e-05,
+      "loss": 1.3349,
+      "step": 7324
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 9.997746524900353e-05,
+      "loss": 1.2994,
+      "step": 7325
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 9.997743917662368e-05,
+      "loss": 1.288,
+      "step": 7326
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 9.997741308917328e-05,
+      "loss": 1.3175,
+      "step": 7327
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 9.997738698665231e-05,
+      "loss": 1.3157,
+      "step": 7328
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 9.99773608690608e-05,
+      "loss": 1.3363,
+      "step": 7329
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 9.997733473639876e-05,
+      "loss": 1.3374,
+      "step": 7330
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 9.997730858866618e-05,
+      "loss": 1.3106,
+      "step": 7331
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 9.99772824258631e-05,
+      "loss": 1.3437,
+      "step": 7332
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 9.99772562479895e-05,
+      "loss": 1.3208,
+      "step": 7333
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 9.997723005504539e-05,
+      "loss": 1.346,
+      "step": 7334
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 9.997720384703079e-05,
+      "loss": 1.2608,
+      "step": 7335
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 9.99771776239457e-05,
+      "loss": 1.2986,
+      "step": 7336
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 9.997715138579012e-05,
+      "loss": 1.3036,
+      "step": 7337
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 9.997712513256407e-05,
+      "loss": 1.2967,
+      "step": 7338
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 9.997709886426754e-05,
+      "loss": 1.2914,
+      "step": 7339
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 9.997707258090058e-05,
+      "loss": 1.308,
+      "step": 7340
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 9.997704628246315e-05,
+      "loss": 1.3221,
+      "step": 7341
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 9.99770199689553e-05,
+      "loss": 1.3021,
+      "step": 7342
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 9.9976993640377e-05,
+      "loss": 1.3159,
+      "step": 7343
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 9.997696729672828e-05,
+      "loss": 1.3014,
+      "step": 7344
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 9.997694093800913e-05,
+      "loss": 1.3267,
+      "step": 7345
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 9.997691456421958e-05,
+      "loss": 1.3145,
+      "step": 7346
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 9.997688817535963e-05,
+      "loss": 1.3331,
+      "step": 7347
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 9.997686177142929e-05,
+      "loss": 1.3297,
+      "step": 7348
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 9.997683535242854e-05,
+      "loss": 1.2956,
+      "step": 7349
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 9.997680891835743e-05,
+      "loss": 1.3128,
+      "step": 7350
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 9.997678246921596e-05,
+      "loss": 1.3524,
+      "step": 7351
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 9.997675600500412e-05,
+      "loss": 1.331,
+      "step": 7352
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 9.997672952572193e-05,
+      "loss": 1.3535,
+      "step": 7353
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 9.997670303136937e-05,
+      "loss": 1.3149,
+      "step": 7354
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 9.99766765219465e-05,
+      "loss": 1.3346,
+      "step": 7355
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 9.997664999745329e-05,
+      "loss": 1.3429,
+      "step": 7356
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 9.997662345788976e-05,
+      "loss": 1.3715,
+      "step": 7357
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 9.997659690325591e-05,
+      "loss": 1.3263,
+      "step": 7358
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 9.997657033355175e-05,
+      "loss": 1.3011,
+      "step": 7359
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 9.99765437487773e-05,
+      "loss": 1.3432,
+      "step": 7360
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 9.997651714893257e-05,
+      "loss": 1.322,
+      "step": 7361
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 9.997649053401755e-05,
+      "loss": 1.3515,
+      "step": 7362
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 9.997646390403225e-05,
+      "loss": 1.3291,
+      "step": 7363
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 9.997643725897667e-05,
+      "loss": 1.3121,
+      "step": 7364
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 9.997641059885085e-05,
+      "loss": 1.2911,
+      "step": 7365
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 9.997638392365478e-05,
+      "loss": 1.3409,
+      "step": 7366
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 9.997635723338848e-05,
+      "loss": 1.3191,
+      "step": 7367
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 9.997633052805193e-05,
+      "loss": 1.3342,
+      "step": 7368
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 9.997630380764516e-05,
+      "loss": 1.3347,
+      "step": 7369
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 9.997627707216817e-05,
+      "loss": 1.3221,
+      "step": 7370
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 9.997625032162097e-05,
+      "loss": 1.2892,
+      "step": 7371
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 9.997622355600356e-05,
+      "loss": 1.3569,
+      "step": 7372
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 9.997619677531598e-05,
+      "loss": 1.3348,
+      "step": 7373
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 9.997616997955819e-05,
+      "loss": 1.3016,
+      "step": 7374
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 9.997614316873023e-05,
+      "loss": 1.3141,
+      "step": 7375
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 9.99761163428321e-05,
+      "loss": 1.3248,
+      "step": 7376
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 9.997608950186381e-05,
+      "loss": 1.2865,
+      "step": 7377
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 9.997606264582537e-05,
+      "loss": 1.3213,
+      "step": 7378
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 9.997603577471679e-05,
+      "loss": 1.3205,
+      "step": 7379
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 9.997600888853806e-05,
+      "loss": 1.3462,
+      "step": 7380
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 9.997598198728922e-05,
+      "loss": 1.3533,
+      "step": 7381
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 9.997595507097024e-05,
+      "loss": 1.3297,
+      "step": 7382
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 9.997592813958116e-05,
+      "loss": 1.3542,
+      "step": 7383
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 9.997590119312196e-05,
+      "loss": 1.3292,
+      "step": 7384
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 9.997587423159268e-05,
+      "loss": 1.3253,
+      "step": 7385
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 9.997584725499331e-05,
+      "loss": 1.32,
+      "step": 7386
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 9.997582026332386e-05,
+      "loss": 1.3514,
+      "step": 7387
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 9.997579325658432e-05,
+      "loss": 1.3159,
+      "step": 7388
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 9.997576623477473e-05,
+      "loss": 1.3301,
+      "step": 7389
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 9.997573919789509e-05,
+      "loss": 1.3372,
+      "step": 7390
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 9.99757121459454e-05,
+      "loss": 1.3204,
+      "step": 7391
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 9.997568507892567e-05,
+      "loss": 1.3235,
+      "step": 7392
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 9.997565799683591e-05,
+      "loss": 1.3284,
+      "step": 7393
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 9.997563089967612e-05,
+      "loss": 1.3264,
+      "step": 7394
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 9.997560378744631e-05,
+      "loss": 1.323,
+      "step": 7395
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 9.99755766601465e-05,
+      "loss": 1.3166,
+      "step": 7396
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 9.99755495177767e-05,
+      "loss": 1.3387,
+      "step": 7397
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 9.99755223603369e-05,
+      "loss": 1.3479,
+      "step": 7398
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 9.997549518782714e-05,
+      "loss": 1.3069,
+      "step": 7399
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 9.997546800024738e-05,
+      "loss": 1.3576,
+      "step": 7400
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 9.997544079759766e-05,
+      "loss": 1.3557,
+      "step": 7401
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 9.997541357987799e-05,
+      "loss": 1.3231,
+      "step": 7402
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 9.997538634708836e-05,
+      "loss": 1.3177,
+      "step": 7403
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 9.997535909922879e-05,
+      "loss": 1.2959,
+      "step": 7404
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 9.997533183629929e-05,
+      "loss": 1.3102,
+      "step": 7405
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 9.997530455829986e-05,
+      "loss": 1.3287,
+      "step": 7406
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 9.997527726523053e-05,
+      "loss": 1.3312,
+      "step": 7407
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 9.997524995709128e-05,
+      "loss": 1.3317,
+      "step": 7408
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 9.997522263388213e-05,
+      "loss": 1.3174,
+      "step": 7409
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 9.99751952956031e-05,
+      "loss": 1.2901,
+      "step": 7410
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 9.997516794225417e-05,
+      "loss": 1.3547,
+      "step": 7411
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 9.997514057383537e-05,
+      "loss": 1.3138,
+      "step": 7412
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 9.997511319034671e-05,
+      "loss": 1.3424,
+      "step": 7413
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 9.997508579178819e-05,
+      "loss": 1.3537,
+      "step": 7414
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 9.997505837815981e-05,
+      "loss": 1.3348,
+      "step": 7415
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 9.997503094946158e-05,
+      "loss": 1.2971,
+      "step": 7416
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 9.997500350569354e-05,
+      "loss": 1.3592,
+      "step": 7417
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 9.997497604685566e-05,
+      "loss": 1.3007,
+      "step": 7418
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 9.997494857294797e-05,
+      "loss": 1.3618,
+      "step": 7419
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 9.997492108397048e-05,
+      "loss": 1.343,
+      "step": 7420
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 9.997489357992317e-05,
+      "loss": 1.3143,
+      "step": 7421
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 9.997486606080607e-05,
+      "loss": 1.3483,
+      "step": 7422
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 9.99748385266192e-05,
+      "loss": 1.3515,
+      "step": 7423
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 9.997481097736254e-05,
+      "loss": 1.3614,
+      "step": 7424
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 9.997478341303612e-05,
+      "loss": 1.3417,
+      "step": 7425
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 9.997475583363993e-05,
+      "loss": 1.3208,
+      "step": 7426
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 9.9974728239174e-05,
+      "loss": 1.3165,
+      "step": 7427
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 9.997470062963832e-05,
+      "loss": 1.3553,
+      "step": 7428
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 9.997467300503292e-05,
+      "loss": 1.3528,
+      "step": 7429
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.997464536535778e-05,
+      "loss": 1.3006,
+      "step": 7430
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.997461771061293e-05,
+      "loss": 1.2973,
+      "step": 7431
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.997459004079837e-05,
+      "loss": 1.3187,
+      "step": 7432
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.997456235591409e-05,
+      "loss": 1.3246,
+      "step": 7433
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.997453465596015e-05,
+      "loss": 1.3241,
+      "step": 7434
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.997450694093651e-05,
+      "loss": 1.3585,
+      "step": 7435
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.997447921084318e-05,
+      "loss": 1.3186,
+      "step": 7436
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.99744514656802e-05,
+      "loss": 1.3116,
+      "step": 7437
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.997442370544756e-05,
+      "loss": 1.3521,
+      "step": 7438
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.997439593014527e-05,
+      "loss": 1.3132,
+      "step": 7439
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.997436813977332e-05,
+      "loss": 1.3383,
+      "step": 7440
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.997434033433175e-05,
+      "loss": 1.3322,
+      "step": 7441
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.997431251382057e-05,
+      "loss": 1.3792,
+      "step": 7442
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.997428467823975e-05,
+      "loss": 1.3301,
+      "step": 7443
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.997425682758932e-05,
+      "loss": 1.3383,
+      "step": 7444
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.99742289618693e-05,
+      "loss": 1.3173,
+      "step": 7445
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.997420108107969e-05,
+      "loss": 1.3211,
+      "step": 7446
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.997417318522049e-05,
+      "loss": 1.3327,
+      "step": 7447
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.997414527429172e-05,
+      "loss": 1.3388,
+      "step": 7448
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.997411734829338e-05,
+      "loss": 1.347,
+      "step": 7449
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.997408940722548e-05,
+      "loss": 1.3181,
+      "step": 7450
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.997406145108804e-05,
+      "loss": 1.3331,
+      "step": 7451
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.997403347988103e-05,
+      "loss": 1.3233,
+      "step": 7452
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.997400549360452e-05,
+      "loss": 1.3209,
+      "step": 7453
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.997397749225847e-05,
+      "loss": 1.298,
+      "step": 7454
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.99739494758429e-05,
+      "loss": 1.3343,
+      "step": 7455
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.997392144435783e-05,
+      "loss": 1.324,
+      "step": 7456
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.997389339780326e-05,
+      "loss": 1.2939,
+      "step": 7457
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.997386533617919e-05,
+      "loss": 1.3288,
+      "step": 7458
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.997383725948566e-05,
+      "loss": 1.3553,
+      "step": 7459
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.997380916772263e-05,
+      "loss": 1.3373,
+      "step": 7460
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.997378106089015e-05,
+      "loss": 1.3355,
+      "step": 7461
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.99737529389882e-05,
+      "loss": 1.3092,
+      "step": 7462
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.997372480201683e-05,
+      "loss": 1.3396,
+      "step": 7463
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.997369664997599e-05,
+      "loss": 1.3335,
+      "step": 7464
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.997366848286573e-05,
+      "loss": 1.3304,
+      "step": 7465
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.997364030068605e-05,
+      "loss": 1.3316,
+      "step": 7466
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.997361210343695e-05,
+      "loss": 1.3005,
+      "step": 7467
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.997358389111844e-05,
+      "loss": 1.3328,
+      "step": 7468
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.997355566373055e-05,
+      "loss": 1.3279,
+      "step": 7469
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.997352742127325e-05,
+      "loss": 1.3549,
+      "step": 7470
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.997349916374658e-05,
+      "loss": 1.3197,
+      "step": 7471
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.997347089115054e-05,
+      "loss": 1.3807,
+      "step": 7472
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.997344260348513e-05,
+      "loss": 1.3001,
+      "step": 7473
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.997341430075036e-05,
+      "loss": 1.3249,
+      "step": 7474
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.997338598294626e-05,
+      "loss": 1.3253,
+      "step": 7475
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.99733576500728e-05,
+      "loss": 1.3218,
+      "step": 7476
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.997332930213004e-05,
+      "loss": 1.3163,
+      "step": 7477
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.997330093911794e-05,
+      "loss": 1.3076,
+      "step": 7478
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.997327256103653e-05,
+      "loss": 1.3153,
+      "step": 7479
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.997324416788582e-05,
+      "loss": 1.3437,
+      "step": 7480
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.99732157596658e-05,
+      "loss": 1.3134,
+      "step": 7481
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.997318733637651e-05,
+      "loss": 1.3281,
+      "step": 7482
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.997315889801795e-05,
+      "loss": 1.3475,
+      "step": 7483
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.99731304445901e-05,
+      "loss": 1.3212,
+      "step": 7484
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.9973101976093e-05,
+      "loss": 1.3088,
+      "step": 7485
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.997307349252665e-05,
+      "loss": 1.3362,
+      "step": 7486
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.997304499389105e-05,
+      "loss": 1.3219,
+      "step": 7487
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.997301648018622e-05,
+      "loss": 1.3201,
+      "step": 7488
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.997298795141215e-05,
+      "loss": 1.322,
+      "step": 7489
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.997295940756888e-05,
+      "loss": 1.3133,
+      "step": 7490
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.99729308486564e-05,
+      "loss": 1.3839,
+      "step": 7491
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.99729022746747e-05,
+      "loss": 1.3584,
+      "step": 7492
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.997287368562382e-05,
+      "loss": 1.3619,
+      "step": 7493
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.997284508150377e-05,
+      "loss": 1.326,
+      "step": 7494
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.997281646231453e-05,
+      "loss": 1.3497,
+      "step": 7495
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.997278782805613e-05,
+      "loss": 1.3271,
+      "step": 7496
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.997275917872857e-05,
+      "loss": 1.3237,
+      "step": 7497
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.997273051433184e-05,
+      "loss": 1.3052,
+      "step": 7498
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.9972701834866e-05,
+      "loss": 1.3176,
+      "step": 7499
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.997267314033103e-05,
+      "loss": 1.3252,
+      "step": 7500
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.997264443072692e-05,
+      "loss": 1.3194,
+      "step": 7501
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.99726157060537e-05,
+      "loss": 1.3116,
+      "step": 7502
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.997258696631138e-05,
+      "loss": 1.327,
+      "step": 7503
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.997255821149996e-05,
+      "loss": 1.3521,
+      "step": 7504
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.997252944161945e-05,
+      "loss": 1.3354,
+      "step": 7505
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.997250065666986e-05,
+      "loss": 1.3243,
+      "step": 7506
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.997247185665119e-05,
+      "loss": 1.3283,
+      "step": 7507
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.997244304156348e-05,
+      "loss": 1.2887,
+      "step": 7508
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.997241421140671e-05,
+      "loss": 1.3472,
+      "step": 7509
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.997238536618087e-05,
+      "loss": 1.2905,
+      "step": 7510
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.997235650588603e-05,
+      "loss": 1.3241,
+      "step": 7511
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.997232763052214e-05,
+      "loss": 1.3093,
+      "step": 7512
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.997229874008924e-05,
+      "loss": 1.3166,
+      "step": 7513
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.997226983458733e-05,
+      "loss": 1.3159,
+      "step": 7514
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.997224091401641e-05,
+      "loss": 1.3511,
+      "step": 7515
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.99722119783765e-05,
+      "loss": 1.3063,
+      "step": 7516
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.997218302766762e-05,
+      "loss": 1.324,
+      "step": 7517
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.997215406188976e-05,
+      "loss": 1.3192,
+      "step": 7518
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.997212508104292e-05,
+      "loss": 1.3051,
+      "step": 7519
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.997209608512713e-05,
+      "loss": 1.3055,
+      "step": 7520
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.99720670741424e-05,
+      "loss": 1.3177,
+      "step": 7521
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.997203804808873e-05,
+      "loss": 1.3328,
+      "step": 7522
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.997200900696611e-05,
+      "loss": 1.3122,
+      "step": 7523
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.997197995077457e-05,
+      "loss": 1.2817,
+      "step": 7524
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.997195087951413e-05,
+      "loss": 1.3266,
+      "step": 7525
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.997192179318477e-05,
+      "loss": 1.3375,
+      "step": 7526
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.997189269178654e-05,
+      "loss": 1.3015,
+      "step": 7527
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.99718635753194e-05,
+      "loss": 1.352,
+      "step": 7528
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.997183444378338e-05,
+      "loss": 1.2884,
+      "step": 7529
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.99718052971785e-05,
+      "loss": 1.3413,
+      "step": 7530
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.997177613550475e-05,
+      "loss": 1.2999,
+      "step": 7531
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.997174695876216e-05,
+      "loss": 1.2876,
+      "step": 7532
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.997171776695072e-05,
+      "loss": 1.3127,
+      "step": 7533
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.997168856007045e-05,
+      "loss": 1.305,
+      "step": 7534
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.997165933812134e-05,
+      "loss": 1.326,
+      "step": 7535
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.997163010110342e-05,
+      "loss": 1.289,
+      "step": 7536
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.99716008490167e-05,
+      "loss": 1.35,
+      "step": 7537
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.997157158186117e-05,
+      "loss": 1.3315,
+      "step": 7538
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.997154229963685e-05,
+      "loss": 1.3167,
+      "step": 7539
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.997151300234375e-05,
+      "loss": 1.3016,
+      "step": 7540
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.997148368998189e-05,
+      "loss": 1.3175,
+      "step": 7541
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.997145436255124e-05,
+      "loss": 1.3115,
+      "step": 7542
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.997142502005185e-05,
+      "loss": 1.311,
+      "step": 7543
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.99713956624837e-05,
+      "loss": 1.3217,
+      "step": 7544
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.997136628984682e-05,
+      "loss": 1.3315,
+      "step": 7545
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.997133690214122e-05,
+      "loss": 1.3183,
+      "step": 7546
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.99713074993669e-05,
+      "loss": 1.3438,
+      "step": 7547
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.997127808152387e-05,
+      "loss": 1.3334,
+      "step": 7548
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.997124864861213e-05,
+      "loss": 1.3555,
+      "step": 7549
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.99712192006317e-05,
+      "loss": 1.3687,
+      "step": 7550
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.997118973758257e-05,
+      "loss": 1.3158,
+      "step": 7551
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.997116025946478e-05,
+      "loss": 1.3328,
+      "step": 7552
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.997113076627833e-05,
+      "loss": 1.3195,
+      "step": 7553
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.99711012580232e-05,
+      "loss": 1.347,
+      "step": 7554
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.997107173469943e-05,
+      "loss": 1.3023,
+      "step": 7555
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.997104219630705e-05,
+      "loss": 1.3309,
+      "step": 7556
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.997101264284601e-05,
+      "loss": 1.3435,
+      "step": 7557
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.997098307431634e-05,
+      "loss": 1.3336,
+      "step": 7558
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.997095349071808e-05,
+      "loss": 1.3099,
+      "step": 7559
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.997092389205121e-05,
+      "loss": 1.3207,
+      "step": 7560
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.997089427831571e-05,
+      "loss": 1.3331,
+      "step": 7561
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.997086464951166e-05,
+      "loss": 1.3006,
+      "step": 7562
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.997083500563903e-05,
+      "loss": 1.34,
+      "step": 7563
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.997080534669782e-05,
+      "loss": 1.3429,
+      "step": 7564
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.997077567268806e-05,
+      "loss": 1.2903,
+      "step": 7565
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.997074598360974e-05,
+      "loss": 1.3097,
+      "step": 7566
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.99707162794629e-05,
+      "loss": 1.3077,
+      "step": 7567
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.99706865602475e-05,
+      "loss": 1.3519,
+      "step": 7568
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.997065682596358e-05,
+      "loss": 1.3134,
+      "step": 7569
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.997062707661116e-05,
+      "loss": 1.3038,
+      "step": 7570
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.997059731219024e-05,
+      "loss": 1.3085,
+      "step": 7571
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.99705675327008e-05,
+      "loss": 1.3604,
+      "step": 7572
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.997053773814288e-05,
+      "loss": 1.3419,
+      "step": 7573
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.997050792851647e-05,
+      "loss": 1.3457,
+      "step": 7574
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.997047810382161e-05,
+      "loss": 1.3148,
+      "step": 7575
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.997044826405828e-05,
+      "loss": 1.3577,
+      "step": 7576
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.99704184092265e-05,
+      "loss": 1.3193,
+      "step": 7577
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.997038853932628e-05,
+      "loss": 1.3432,
+      "step": 7578
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.997035865435762e-05,
+      "loss": 1.3392,
+      "step": 7579
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.997032875432054e-05,
+      "loss": 1.3196,
+      "step": 7580
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.997029883921505e-05,
+      "loss": 1.33,
+      "step": 7581
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.997026890904113e-05,
+      "loss": 1.2934,
+      "step": 7582
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.997023896379883e-05,
+      "loss": 1.3493,
+      "step": 7583
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.997020900348815e-05,
+      "loss": 1.3378,
+      "step": 7584
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.997017902810906e-05,
+      "loss": 1.3127,
+      "step": 7585
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.997014903766163e-05,
+      "loss": 1.3439,
+      "step": 7586
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.997011903214581e-05,
+      "loss": 1.2951,
+      "step": 7587
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.997008901156166e-05,
+      "loss": 1.3351,
+      "step": 7588
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.997005897590916e-05,
+      "loss": 1.3306,
+      "step": 7589
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.997002892518833e-05,
+      "loss": 1.3427,
+      "step": 7590
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.996999885939917e-05,
+      "loss": 1.3094,
+      "step": 7591
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.996996877854169e-05,
+      "loss": 1.3485,
+      "step": 7592
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.996993868261591e-05,
+      "loss": 1.3456,
+      "step": 7593
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.996990857162183e-05,
+      "loss": 1.3196,
+      "step": 7594
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.996987844555945e-05,
+      "loss": 1.3524,
+      "step": 7595
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.99698483044288e-05,
+      "loss": 1.3329,
+      "step": 7596
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.996981814822987e-05,
+      "loss": 1.3427,
+      "step": 7597
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.996978797696268e-05,
+      "loss": 1.3199,
+      "step": 7598
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.996975779062725e-05,
+      "loss": 1.3061,
+      "step": 7599
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.996972758922356e-05,
+      "loss": 1.3357,
+      "step": 7600
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.996969737275165e-05,
+      "loss": 1.2951,
+      "step": 7601
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.99696671412115e-05,
+      "loss": 1.3029,
+      "step": 7602
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.996963689460314e-05,
+      "loss": 1.3168,
+      "step": 7603
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.996960663292658e-05,
+      "loss": 1.351,
+      "step": 7604
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.996957635618182e-05,
+      "loss": 1.3703,
+      "step": 7605
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.996954606436886e-05,
+      "loss": 1.338,
+      "step": 7606
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.996951575748772e-05,
+      "loss": 1.3146,
+      "step": 7607
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.996948543553842e-05,
+      "loss": 1.2955,
+      "step": 7608
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.996945509852095e-05,
+      "loss": 1.3304,
+      "step": 7609
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.996942474643533e-05,
+      "loss": 1.336,
+      "step": 7610
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.996939437928157e-05,
+      "loss": 1.3322,
+      "step": 7611
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.996936399705967e-05,
+      "loss": 1.333,
+      "step": 7612
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.996933359976965e-05,
+      "loss": 1.3262,
+      "step": 7613
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.996930318741153e-05,
+      "loss": 1.3253,
+      "step": 7614
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.996927275998528e-05,
+      "loss": 1.3354,
+      "step": 7615
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.996924231749095e-05,
+      "loss": 1.2803,
+      "step": 7616
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.996921185992852e-05,
+      "loss": 1.3609,
+      "step": 7617
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.9969181387298e-05,
+      "loss": 1.3322,
+      "step": 7618
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.996915089959942e-05,
+      "loss": 1.3322,
+      "step": 7619
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.996912039683277e-05,
+      "loss": 1.3151,
+      "step": 7620
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.996908987899809e-05,
+      "loss": 1.2973,
+      "step": 7621
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.996905934609534e-05,
+      "loss": 1.3397,
+      "step": 7622
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.996902879812459e-05,
+      "loss": 1.3346,
+      "step": 7623
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.99689982350858e-05,
+      "loss": 1.3268,
+      "step": 7624
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.996896765697899e-05,
+      "loss": 1.2999,
+      "step": 7625
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.996893706380419e-05,
+      "loss": 1.3254,
+      "step": 7626
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.996890645556137e-05,
+      "loss": 1.3221,
+      "step": 7627
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.996887583225059e-05,
+      "loss": 1.2986,
+      "step": 7628
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.99688451938718e-05,
+      "loss": 1.3229,
+      "step": 7629
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.996881454042506e-05,
+      "loss": 1.3355,
+      "step": 7630
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.996878387191036e-05,
+      "loss": 1.2824,
+      "step": 7631
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.996875318832771e-05,
+      "loss": 1.325,
+      "step": 7632
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.996872248967712e-05,
+      "loss": 1.3024,
+      "step": 7633
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.99686917759586e-05,
+      "loss": 1.3169,
+      "step": 7634
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.996866104717215e-05,
+      "loss": 1.3387,
+      "step": 7635
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.99686303033178e-05,
+      "loss": 1.3003,
+      "step": 7636
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.996859954439554e-05,
+      "loss": 1.2918,
+      "step": 7637
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.996856877040538e-05,
+      "loss": 1.3454,
+      "step": 7638
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.996853798134733e-05,
+      "loss": 1.3443,
+      "step": 7639
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.996850717722142e-05,
+      "loss": 1.3371,
+      "step": 7640
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.996847635802764e-05,
+      "loss": 1.3381,
+      "step": 7641
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.9968445523766e-05,
+      "loss": 1.3103,
+      "step": 7642
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.99684146744365e-05,
+      "loss": 1.3528,
+      "step": 7643
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.996838381003917e-05,
+      "loss": 1.328,
+      "step": 7644
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.996835293057402e-05,
+      "loss": 1.3104,
+      "step": 7645
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.996832203604105e-05,
+      "loss": 1.3721,
+      "step": 7646
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.996829112644026e-05,
+      "loss": 1.3092,
+      "step": 7647
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.996826020177166e-05,
+      "loss": 1.3573,
+      "step": 7648
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.996822926203529e-05,
+      "loss": 1.3133,
+      "step": 7649
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.996819830723113e-05,
+      "loss": 1.3146,
+      "step": 7650
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.996816733735918e-05,
+      "loss": 1.3182,
+      "step": 7651
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.996813635241947e-05,
+      "loss": 1.3318,
+      "step": 7652
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.996810535241202e-05,
+      "loss": 1.3467,
+      "step": 7653
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.996807433733681e-05,
+      "loss": 1.3217,
+      "step": 7654
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.996804330719387e-05,
+      "loss": 1.3178,
+      "step": 7655
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.99680122619832e-05,
+      "loss": 1.3318,
+      "step": 7656
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.996798120170483e-05,
+      "loss": 1.3217,
+      "step": 7657
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.996795012635873e-05,
+      "loss": 1.3396,
+      "step": 7658
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.996791903594493e-05,
+      "loss": 1.3378,
+      "step": 7659
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.996788793046345e-05,
+      "loss": 1.3346,
+      "step": 7660
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.996785680991429e-05,
+      "loss": 1.327,
+      "step": 7661
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.996782567429747e-05,
+      "loss": 1.3476,
+      "step": 7662
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.996779452361299e-05,
+      "loss": 1.3374,
+      "step": 7663
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.996776335786085e-05,
+      "loss": 1.3266,
+      "step": 7664
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.996773217704106e-05,
+      "loss": 1.372,
+      "step": 7665
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.996770098115363e-05,
+      "loss": 1.3188,
+      "step": 7666
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.996766977019859e-05,
+      "loss": 1.3022,
+      "step": 7667
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.996763854417594e-05,
+      "loss": 1.321,
+      "step": 7668
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.996760730308567e-05,
+      "loss": 1.2956,
+      "step": 7669
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.996757604692781e-05,
+      "loss": 1.3622,
+      "step": 7670
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.996754477570239e-05,
+      "loss": 1.3445,
+      "step": 7671
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.996751348940936e-05,
+      "loss": 1.2955,
+      "step": 7672
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.996748218804877e-05,
+      "loss": 1.327,
+      "step": 7673
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.996745087162062e-05,
+      "loss": 1.3191,
+      "step": 7674
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.996741954012494e-05,
+      "loss": 1.3444,
+      "step": 7675
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.99673881935617e-05,
+      "loss": 1.3625,
+      "step": 7676
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.996735683193095e-05,
+      "loss": 1.3476,
+      "step": 7677
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.996732545523267e-05,
+      "loss": 1.3236,
+      "step": 7678
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.996729406346688e-05,
+      "loss": 1.3337,
+      "step": 7679
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.996726265663358e-05,
+      "loss": 1.3035,
+      "step": 7680
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.99672312347328e-05,
+      "loss": 1.3475,
+      "step": 7681
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.996719979776454e-05,
+      "loss": 1.3052,
+      "step": 7682
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.996716834572881e-05,
+      "loss": 1.3081,
+      "step": 7683
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.996713687862561e-05,
+      "loss": 1.2729,
+      "step": 7684
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.996710539645495e-05,
+      "loss": 1.3311,
+      "step": 7685
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.996707389921686e-05,
+      "loss": 1.3107,
+      "step": 7686
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.996704238691132e-05,
+      "loss": 1.3352,
+      "step": 7687
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.996701085953837e-05,
+      "loss": 1.336,
+      "step": 7688
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.9966979317098e-05,
+      "loss": 1.3244,
+      "step": 7689
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.996694775959022e-05,
+      "loss": 1.3236,
+      "step": 7690
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.996691618701506e-05,
+      "loss": 1.3048,
+      "step": 7691
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.99668845993725e-05,
+      "loss": 1.3054,
+      "step": 7692
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.996685299666255e-05,
+      "loss": 1.3608,
+      "step": 7693
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.996682137888525e-05,
+      "loss": 1.2921,
+      "step": 7694
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.996678974604059e-05,
+      "loss": 1.334,
+      "step": 7695
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.996675809812857e-05,
+      "loss": 1.339,
+      "step": 7696
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.996672643514921e-05,
+      "loss": 1.3333,
+      "step": 7697
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.996669475710254e-05,
+      "loss": 1.3419,
+      "step": 7698
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.996666306398855e-05,
+      "loss": 1.3359,
+      "step": 7699
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.996663135580722e-05,
+      "loss": 1.3171,
+      "step": 7700
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.996659963255861e-05,
+      "loss": 1.3417,
+      "step": 7701
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.99665678942427e-05,
+      "loss": 1.3193,
+      "step": 7702
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.996653614085952e-05,
+      "loss": 1.3063,
+      "step": 7703
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.996650437240907e-05,
+      "loss": 1.2871,
+      "step": 7704
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.996647258889133e-05,
+      "loss": 1.3086,
+      "step": 7705
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.996644079030637e-05,
+      "loss": 1.3095,
+      "step": 7706
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.996640897665415e-05,
+      "loss": 1.3058,
+      "step": 7707
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.996637714793469e-05,
+      "loss": 1.3232,
+      "step": 7708
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.996634530414801e-05,
+      "loss": 1.2944,
+      "step": 7709
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.996631344529412e-05,
+      "loss": 1.3541,
+      "step": 7710
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.996628157137303e-05,
+      "loss": 1.2766,
+      "step": 7711
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.996624968238474e-05,
+      "loss": 1.3464,
+      "step": 7712
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.996621777832926e-05,
+      "loss": 1.2922,
+      "step": 7713
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.99661858592066e-05,
+      "loss": 1.3087,
+      "step": 7714
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.996615392501679e-05,
+      "loss": 1.3314,
+      "step": 7715
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.996612197575981e-05,
+      "loss": 1.3138,
+      "step": 7716
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.996609001143568e-05,
+      "loss": 1.3199,
+      "step": 7717
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.996605803204442e-05,
+      "loss": 1.3424,
+      "step": 7718
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.996602603758603e-05,
+      "loss": 1.3028,
+      "step": 7719
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.996599402806052e-05,
+      "loss": 1.3629,
+      "step": 7720
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.996596200346789e-05,
+      "loss": 1.3219,
+      "step": 7721
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.996592996380819e-05,
+      "loss": 1.3259,
+      "step": 7722
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.996589790908138e-05,
+      "loss": 1.3094,
+      "step": 7723
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.996586583928748e-05,
+      "loss": 1.2851,
+      "step": 7724
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.996583375442652e-05,
+      "loss": 1.3496,
+      "step": 7725
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.996580165449852e-05,
+      "loss": 1.3318,
+      "step": 7726
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.996576953950345e-05,
+      "loss": 1.2975,
+      "step": 7727
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.996573740944133e-05,
+      "loss": 1.3021,
+      "step": 7728
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.99657052643122e-05,
+      "loss": 1.3333,
+      "step": 7729
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.996567310411602e-05,
+      "loss": 1.3217,
+      "step": 7730
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.996564092885286e-05,
+      "loss": 1.3156,
+      "step": 7731
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.996560873852268e-05,
+      "loss": 1.3433,
+      "step": 7732
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.99655765331255e-05,
+      "loss": 1.3434,
+      "step": 7733
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.996554431266135e-05,
+      "loss": 1.3416,
+      "step": 7734
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.996551207713023e-05,
+      "loss": 1.316,
+      "step": 7735
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.996547982653213e-05,
+      "loss": 1.3055,
+      "step": 7736
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.996544756086709e-05,
+      "loss": 1.3238,
+      "step": 7737
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.99654152801351e-05,
+      "loss": 1.3115,
+      "step": 7738
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.996538298433618e-05,
+      "loss": 1.2853,
+      "step": 7739
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.996535067347033e-05,
+      "loss": 1.3117,
+      "step": 7740
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.996531834753758e-05,
+      "loss": 1.3115,
+      "step": 7741
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.99652860065379e-05,
+      "loss": 1.3272,
+      "step": 7742
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.996525365047135e-05,
+      "loss": 1.3043,
+      "step": 7743
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.99652212793379e-05,
+      "loss": 1.3279,
+      "step": 7744
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.996518889313758e-05,
+      "loss": 1.3265,
+      "step": 7745
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.996515649187039e-05,
+      "loss": 1.3138,
+      "step": 7746
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.996512407553634e-05,
+      "loss": 1.3308,
+      "step": 7747
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.996509164413546e-05,
+      "loss": 1.3134,
+      "step": 7748
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.996505919766772e-05,
+      "loss": 1.3515,
+      "step": 7749
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.996502673613317e-05,
+      "loss": 1.3436,
+      "step": 7750
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.996499425953178e-05,
+      "loss": 1.3137,
+      "step": 7751
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.996496176786361e-05,
+      "loss": 1.3842,
+      "step": 7752
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.996492926112864e-05,
+      "loss": 1.338,
+      "step": 7753
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.996489673932687e-05,
+      "loss": 1.3233,
+      "step": 7754
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.996486420245831e-05,
+      "loss": 1.3318,
+      "step": 7755
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.996483165052302e-05,
+      "loss": 1.3207,
+      "step": 7756
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.996479908352094e-05,
+      "loss": 1.2943,
+      "step": 7757
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.996476650145212e-05,
+      "loss": 1.307,
+      "step": 7758
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.996473390431657e-05,
+      "loss": 1.3273,
+      "step": 7759
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.996470129211429e-05,
+      "loss": 1.333,
+      "step": 7760
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.996466866484529e-05,
+      "loss": 1.3319,
+      "step": 7761
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.996463602250956e-05,
+      "loss": 1.3203,
+      "step": 7762
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.996460336510715e-05,
+      "loss": 1.3307,
+      "step": 7763
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.996457069263805e-05,
+      "loss": 1.3307,
+      "step": 7764
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.996453800510225e-05,
+      "loss": 1.3339,
+      "step": 7765
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.996450530249981e-05,
+      "loss": 1.3576,
+      "step": 7766
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.99644725848307e-05,
+      "loss": 1.3506,
+      "step": 7767
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.996443985209494e-05,
+      "loss": 1.3087,
+      "step": 7768
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.996440710429253e-05,
+      "loss": 1.3175,
+      "step": 7769
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.996437434142351e-05,
+      "loss": 1.3115,
+      "step": 7770
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.996434156348785e-05,
+      "loss": 1.3067,
+      "step": 7771
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.996430877048559e-05,
+      "loss": 1.3768,
+      "step": 7772
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.996427596241673e-05,
+      "loss": 1.2799,
+      "step": 7773
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.996424313928127e-05,
+      "loss": 1.3051,
+      "step": 7774
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.996421030107922e-05,
+      "loss": 1.3139,
+      "step": 7775
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.996417744781063e-05,
+      "loss": 1.3313,
+      "step": 7776
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.996414457947546e-05,
+      "loss": 1.288,
+      "step": 7777
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.996411169607374e-05,
+      "loss": 1.3371,
+      "step": 7778
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.996407879760548e-05,
+      "loss": 1.3325,
+      "step": 7779
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.996404588407069e-05,
+      "loss": 1.3436,
+      "step": 7780
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.996401295546937e-05,
+      "loss": 1.3148,
+      "step": 7781
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.996398001180156e-05,
+      "loss": 1.3172,
+      "step": 7782
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.996394705306723e-05,
+      "loss": 1.3211,
+      "step": 7783
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.996391407926641e-05,
+      "loss": 1.3131,
+      "step": 7784
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.996388109039912e-05,
+      "loss": 1.3474,
+      "step": 7785
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.996384808646534e-05,
+      "loss": 1.3278,
+      "step": 7786
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.99638150674651e-05,
+      "loss": 1.3373,
+      "step": 7787
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.996378203339843e-05,
+      "loss": 1.3576,
+      "step": 7788
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.99637489842653e-05,
+      "loss": 1.3208,
+      "step": 7789
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.996371592006575e-05,
+      "loss": 1.3218,
+      "step": 7790
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.996368284079975e-05,
+      "loss": 1.3097,
+      "step": 7791
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.996364974646737e-05,
+      "loss": 1.3239,
+      "step": 7792
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.996361663706857e-05,
+      "loss": 1.2997,
+      "step": 7793
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.99635835126034e-05,
+      "loss": 1.3213,
+      "step": 7794
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.996355037307183e-05,
+      "loss": 1.3233,
+      "step": 7795
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.996351721847388e-05,
+      "loss": 1.3012,
+      "step": 7796
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.996348404880958e-05,
+      "loss": 1.3182,
+      "step": 7797
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.996345086407893e-05,
+      "loss": 1.3414,
+      "step": 7798
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.996341766428195e-05,
+      "loss": 1.3249,
+      "step": 7799
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.996338444941861e-05,
+      "loss": 1.339,
+      "step": 7800
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.996335121948896e-05,
+      "loss": 1.2931,
+      "step": 7801
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.9963317974493e-05,
+      "loss": 1.3321,
+      "step": 7802
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.996328471443073e-05,
+      "loss": 1.3262,
+      "step": 7803
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.996325143930219e-05,
+      "loss": 1.3478,
+      "step": 7804
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.996321814910734e-05,
+      "loss": 1.3185,
+      "step": 7805
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.996318484384623e-05,
+      "loss": 1.3477,
+      "step": 7806
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.996315152351888e-05,
+      "loss": 1.3461,
+      "step": 7807
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.996311818812525e-05,
+      "loss": 1.3257,
+      "step": 7808
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.99630848376654e-05,
+      "loss": 1.3157,
+      "step": 7809
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.996305147213929e-05,
+      "loss": 1.3619,
+      "step": 7810
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.996301809154698e-05,
+      "loss": 1.2985,
+      "step": 7811
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.996298469588846e-05,
+      "loss": 1.307,
+      "step": 7812
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.996295128516371e-05,
+      "loss": 1.332,
+      "step": 7813
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.996291785937279e-05,
+      "loss": 1.3192,
+      "step": 7814
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.99628844185157e-05,
+      "loss": 1.3502,
+      "step": 7815
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.996285096259241e-05,
+      "loss": 1.2886,
+      "step": 7816
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.996281749160299e-05,
+      "loss": 1.3265,
+      "step": 7817
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.996278400554739e-05,
+      "loss": 1.3204,
+      "step": 7818
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.996275050442567e-05,
+      "loss": 1.3235,
+      "step": 7819
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.996271698823783e-05,
+      "loss": 1.3384,
+      "step": 7820
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.996268345698384e-05,
+      "loss": 1.3108,
+      "step": 7821
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.996264991066375e-05,
+      "loss": 1.3274,
+      "step": 7822
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.996261634927757e-05,
+      "loss": 1.2679,
+      "step": 7823
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.996258277282528e-05,
+      "loss": 1.2834,
+      "step": 7824
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.996254918130693e-05,
+      "loss": 1.2992,
+      "step": 7825
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.99625155747225e-05,
+      "loss": 1.3415,
+      "step": 7826
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.9962481953072e-05,
+      "loss": 1.3468,
+      "step": 7827
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.996244831635548e-05,
+      "loss": 1.3433,
+      "step": 7828
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.99624146645729e-05,
+      "loss": 1.3073,
+      "step": 7829
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.99623809977243e-05,
+      "loss": 1.2961,
+      "step": 7830
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.996234731580967e-05,
+      "loss": 1.3153,
+      "step": 7831
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.996231361882903e-05,
+      "loss": 1.3004,
+      "step": 7832
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.996227990678239e-05,
+      "loss": 1.3144,
+      "step": 7833
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.996224617966978e-05,
+      "loss": 1.2985,
+      "step": 7834
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.996221243749117e-05,
+      "loss": 1.3358,
+      "step": 7835
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.99621786802466e-05,
+      "loss": 1.3301,
+      "step": 7836
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.996214490793608e-05,
+      "loss": 1.2855,
+      "step": 7837
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.99621111205596e-05,
+      "loss": 1.3081,
+      "step": 7838
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.996207731811718e-05,
+      "loss": 1.3253,
+      "step": 7839
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.996204350060883e-05,
+      "loss": 1.2998,
+      "step": 7840
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.996200966803458e-05,
+      "loss": 1.3165,
+      "step": 7841
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.99619758203944e-05,
+      "loss": 1.3152,
+      "step": 7842
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.996194195768834e-05,
+      "loss": 1.3215,
+      "step": 7843
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.996190807991638e-05,
+      "loss": 1.2811,
+      "step": 7844
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.996187418707856e-05,
+      "loss": 1.2972,
+      "step": 7845
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.996184027917487e-05,
+      "loss": 1.2994,
+      "step": 7846
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.996180635620532e-05,
+      "loss": 1.309,
+      "step": 7847
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.996177241816991e-05,
+      "loss": 1.2975,
+      "step": 7848
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.996173846506869e-05,
+      "loss": 1.3116,
+      "step": 7849
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.996170449690163e-05,
+      "loss": 1.3044,
+      "step": 7850
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.996167051366874e-05,
+      "loss": 1.3161,
+      "step": 7851
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.996163651537008e-05,
+      "loss": 1.3045,
+      "step": 7852
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.996160250200559e-05,
+      "loss": 1.2863,
+      "step": 7853
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.996156847357534e-05,
+      "loss": 1.3195,
+      "step": 7854
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.99615344300793e-05,
+      "loss": 1.3524,
+      "step": 7855
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.996150037151751e-05,
+      "loss": 1.3184,
+      "step": 7856
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.996146629788996e-05,
+      "loss": 1.3174,
+      "step": 7857
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.996143220919667e-05,
+      "loss": 1.2885,
+      "step": 7858
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.996139810543763e-05,
+      "loss": 1.3535,
+      "step": 7859
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.996136398661289e-05,
+      "loss": 1.3358,
+      "step": 7860
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.996132985272242e-05,
+      "loss": 1.3366,
+      "step": 7861
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.996129570376626e-05,
+      "loss": 1.2895,
+      "step": 7862
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.99612615397444e-05,
+      "loss": 1.3401,
+      "step": 7863
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.996122736065685e-05,
+      "loss": 1.3136,
+      "step": 7864
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.996119316650365e-05,
+      "loss": 1.319,
+      "step": 7865
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.996115895728477e-05,
+      "loss": 1.332,
+      "step": 7866
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.996112473300026e-05,
+      "loss": 1.3093,
+      "step": 7867
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.996109049365009e-05,
+      "loss": 1.3456,
+      "step": 7868
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.99610562392343e-05,
+      "loss": 1.3301,
+      "step": 7869
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.996102196975288e-05,
+      "loss": 1.3498,
+      "step": 7870
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.996098768520585e-05,
+      "loss": 1.3223,
+      "step": 7871
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.996095338559324e-05,
+      "loss": 1.3416,
+      "step": 7872
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.996091907091502e-05,
+      "loss": 1.3148,
+      "step": 7873
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.996088474117125e-05,
+      "loss": 1.3219,
+      "step": 7874
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.996085039636189e-05,
+      "loss": 1.3326,
+      "step": 7875
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.996081603648697e-05,
+      "loss": 1.3136,
+      "step": 7876
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.99607816615465e-05,
+      "loss": 1.3229,
+      "step": 7877
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.99607472715405e-05,
+      "loss": 1.3659,
+      "step": 7878
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.996071286646898e-05,
+      "loss": 1.3205,
+      "step": 7879
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.996067844633194e-05,
+      "loss": 1.3066,
+      "step": 7880
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.996064401112939e-05,
+      "loss": 1.3175,
+      "step": 7881
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.996060956086134e-05,
+      "loss": 1.3327,
+      "step": 7882
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.996057509552781e-05,
+      "loss": 1.3236,
+      "step": 7883
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.99605406151288e-05,
+      "loss": 1.3586,
+      "step": 7884
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.996050611966434e-05,
+      "loss": 1.3045,
+      "step": 7885
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.996047160913442e-05,
+      "loss": 1.3074,
+      "step": 7886
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.996043708353905e-05,
+      "loss": 1.3362,
+      "step": 7887
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.996040254287825e-05,
+      "loss": 1.3091,
+      "step": 7888
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.996036798715202e-05,
+      "loss": 1.327,
+      "step": 7889
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.99603334163604e-05,
+      "loss": 1.3458,
+      "step": 7890
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.996029883050335e-05,
+      "loss": 1.2992,
+      "step": 7891
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.996026422958094e-05,
+      "loss": 1.2804,
+      "step": 7892
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.996022961359312e-05,
+      "loss": 1.3414,
+      "step": 7893
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.996019498253995e-05,
+      "loss": 1.327,
+      "step": 7894
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.99601603364214e-05,
+      "loss": 1.2874,
+      "step": 7895
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.996012567523751e-05,
+      "loss": 1.3272,
+      "step": 7896
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.996009099898829e-05,
+      "loss": 1.3095,
+      "step": 7897
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.996005630767373e-05,
+      "loss": 1.2941,
+      "step": 7898
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.996002160129385e-05,
+      "loss": 1.3202,
+      "step": 7899
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.995998687984867e-05,
+      "loss": 1.3157,
+      "step": 7900
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.99599521433382e-05,
+      "loss": 1.3411,
+      "step": 7901
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.995991739176242e-05,
+      "loss": 1.304,
+      "step": 7902
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.995988262512138e-05,
+      "loss": 1.3237,
+      "step": 7903
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.995984784341507e-05,
+      "loss": 1.2978,
+      "step": 7904
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.99598130466435e-05,
+      "loss": 1.3191,
+      "step": 7905
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.995977823480669e-05,
+      "loss": 1.3154,
+      "step": 7906
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.995974340790465e-05,
+      "loss": 1.3003,
+      "step": 7907
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.995970856593738e-05,
+      "loss": 1.3242,
+      "step": 7908
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.99596737089049e-05,
+      "loss": 1.3057,
+      "step": 7909
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.995963883680721e-05,
+      "loss": 1.3333,
+      "step": 7910
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.995960394964435e-05,
+      "loss": 1.3424,
+      "step": 7911
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.995956904741628e-05,
+      "loss": 1.3579,
+      "step": 7912
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.995953413012305e-05,
+      "loss": 1.306,
+      "step": 7913
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.995949919776466e-05,
+      "loss": 1.3417,
+      "step": 7914
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.99594642503411e-05,
+      "loss": 1.3237,
+      "step": 7915
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.995942928785243e-05,
+      "loss": 1.3505,
+      "step": 7916
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.995939431029862e-05,
+      "loss": 1.3475,
+      "step": 7917
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.995935931767969e-05,
+      "loss": 1.3565,
+      "step": 7918
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.995932430999564e-05,
+      "loss": 1.3003,
+      "step": 7919
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.995928928724651e-05,
+      "loss": 1.3106,
+      "step": 7920
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.995925424943228e-05,
+      "loss": 1.3107,
+      "step": 7921
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.9959219196553e-05,
+      "loss": 1.3461,
+      "step": 7922
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.995918412860862e-05,
+      "loss": 1.3321,
+      "step": 7923
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.995914904559921e-05,
+      "loss": 1.2987,
+      "step": 7924
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.995911394752474e-05,
+      "loss": 1.3427,
+      "step": 7925
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.995907883438525e-05,
+      "loss": 1.3328,
+      "step": 7926
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.995904370618072e-05,
+      "loss": 1.3429,
+      "step": 7927
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.995900856291119e-05,
+      "loss": 1.3046,
+      "step": 7928
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.995897340457665e-05,
+      "loss": 1.3227,
+      "step": 7929
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.995893823117712e-05,
+      "loss": 1.3345,
+      "step": 7930
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.995890304271261e-05,
+      "loss": 1.3247,
+      "step": 7931
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.995886783918312e-05,
+      "loss": 1.3238,
+      "step": 7932
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.995883262058868e-05,
+      "loss": 1.3098,
+      "step": 7933
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.995879738692929e-05,
+      "loss": 1.316,
+      "step": 7934
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.995876213820497e-05,
+      "loss": 1.3201,
+      "step": 7935
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.99587268744157e-05,
+      "loss": 1.3159,
+      "step": 7936
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.995869159556152e-05,
+      "loss": 1.3027,
+      "step": 7937
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.995865630164245e-05,
+      "loss": 1.3227,
+      "step": 7938
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.995862099265847e-05,
+      "loss": 1.3135,
+      "step": 7939
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.99585856686096e-05,
+      "loss": 1.3569,
+      "step": 7940
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.995855032949586e-05,
+      "loss": 1.3302,
+      "step": 7941
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.995851497531727e-05,
+      "loss": 1.3386,
+      "step": 7942
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.99584796060738e-05,
+      "loss": 1.3079,
+      "step": 7943
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.99584442217655e-05,
+      "loss": 1.2929,
+      "step": 7944
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.995840882239238e-05,
+      "loss": 1.317,
+      "step": 7945
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.995837340795443e-05,
+      "loss": 1.2829,
+      "step": 7946
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.995833797845167e-05,
+      "loss": 1.3236,
+      "step": 7947
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.99583025338841e-05,
+      "loss": 1.3239,
+      "step": 7948
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.995826707425175e-05,
+      "loss": 1.3143,
+      "step": 7949
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.995823159955462e-05,
+      "loss": 1.3333,
+      "step": 7950
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.99581961097927e-05,
+      "loss": 1.339,
+      "step": 7951
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.995816060496606e-05,
+      "loss": 1.3345,
+      "step": 7952
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.995812508507464e-05,
+      "loss": 1.3287,
+      "step": 7953
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.995808955011852e-05,
+      "loss": 1.3537,
+      "step": 7954
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.995805400009764e-05,
+      "loss": 1.3248,
+      "step": 7955
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.995801843501207e-05,
+      "loss": 1.3221,
+      "step": 7956
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.995798285486178e-05,
+      "loss": 1.2948,
+      "step": 7957
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.99579472596468e-05,
+      "loss": 1.3317,
+      "step": 7958
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.995791164936714e-05,
+      "loss": 1.3242,
+      "step": 7959
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.995787602402282e-05,
+      "loss": 1.3343,
+      "step": 7960
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.995784038361382e-05,
+      "loss": 1.3572,
+      "step": 7961
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.995780472814018e-05,
+      "loss": 1.3297,
+      "step": 7962
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.995776905760191e-05,
+      "loss": 1.3124,
+      "step": 7963
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.995773337199898e-05,
+      "loss": 1.3441,
+      "step": 7964
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.995769767133145e-05,
+      "loss": 1.3086,
+      "step": 7965
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.995766195559932e-05,
+      "loss": 1.3125,
+      "step": 7966
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.99576262248026e-05,
+      "loss": 1.3283,
+      "step": 7967
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.995759047894127e-05,
+      "loss": 1.3102,
+      "step": 7968
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.995755471801538e-05,
+      "loss": 1.3361,
+      "step": 7969
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.995751894202494e-05,
+      "loss": 1.3132,
+      "step": 7970
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.995748315096992e-05,
+      "loss": 1.3149,
+      "step": 7971
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.995744734485038e-05,
+      "loss": 1.3074,
+      "step": 7972
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.995741152366629e-05,
+      "loss": 1.2947,
+      "step": 7973
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.995737568741769e-05,
+      "loss": 1.3242,
+      "step": 7974
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.995733983610457e-05,
+      "loss": 1.3192,
+      "step": 7975
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.995730396972696e-05,
+      "loss": 1.3289,
+      "step": 7976
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.995726808828485e-05,
+      "loss": 1.32,
+      "step": 7977
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.995723219177829e-05,
+      "loss": 1.3171,
+      "step": 7978
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.995719628020724e-05,
+      "loss": 1.3256,
+      "step": 7979
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.995716035357173e-05,
+      "loss": 1.3004,
+      "step": 7980
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.995712441187179e-05,
+      "loss": 1.3266,
+      "step": 7981
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.995708845510741e-05,
+      "loss": 1.3345,
+      "step": 7982
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.99570524832786e-05,
+      "loss": 1.2764,
+      "step": 7983
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.995701649638539e-05,
+      "loss": 1.332,
+      "step": 7984
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.995698049442777e-05,
+      "loss": 1.3086,
+      "step": 7985
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.995694447740575e-05,
+      "loss": 1.3256,
+      "step": 7986
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.995690844531938e-05,
+      "loss": 1.3225,
+      "step": 7987
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.995687239816861e-05,
+      "loss": 1.2834,
+      "step": 7988
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.99568363359535e-05,
+      "loss": 1.2857,
+      "step": 7989
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.995680025867403e-05,
+      "loss": 1.3323,
+      "step": 7990
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.995676416633025e-05,
+      "loss": 1.3494,
+      "step": 7991
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.995672805892211e-05,
+      "loss": 1.3309,
+      "step": 7992
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.995669193644968e-05,
+      "loss": 1.3327,
+      "step": 7993
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.995665579891293e-05,
+      "loss": 1.2842,
+      "step": 7994
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.995661964631188e-05,
+      "loss": 1.2956,
+      "step": 7995
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.995658347864657e-05,
+      "loss": 1.3267,
+      "step": 7996
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.995654729591698e-05,
+      "loss": 1.3508,
+      "step": 7997
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.995651109812312e-05,
+      "loss": 1.3048,
+      "step": 7998
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.995647488526503e-05,
+      "loss": 1.3182,
+      "step": 7999
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.995643865734268e-05,
+      "loss": 1.3175,
+      "step": 8000
+    },
+    {
+      "epoch": 1.29,
+      "eval_loss": 1.3409901857376099,
+      "eval_runtime": 61.4422,
+      "eval_samples_per_second": 162.755,
+      "eval_steps_per_second": 0.228,
+      "step": 8000
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.995640241435613e-05,
+      "loss": 1.3259,
+      "step": 8001
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.995636615630535e-05,
+      "loss": 1.3206,
+      "step": 8002
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.995632988319036e-05,
+      "loss": 1.343,
+      "step": 8003
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.995629359501118e-05,
+      "loss": 1.2809,
+      "step": 8004
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.995625729176781e-05,
+      "loss": 1.3098,
+      "step": 8005
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.995622097346026e-05,
+      "loss": 1.3021,
+      "step": 8006
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.995618464008857e-05,
+      "loss": 1.3154,
+      "step": 8007
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.995614829165272e-05,
+      "loss": 1.3311,
+      "step": 8008
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.995611192815273e-05,
+      "loss": 1.318,
+      "step": 8009
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.995607554958862e-05,
+      "loss": 1.3063,
+      "step": 8010
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.995603915596036e-05,
+      "loss": 1.3081,
+      "step": 8011
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.995600274726802e-05,
+      "loss": 1.3276,
+      "step": 8012
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.995596632351158e-05,
+      "loss": 1.3205,
+      "step": 8013
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.995592988469106e-05,
+      "loss": 1.3192,
+      "step": 8014
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.995589343080646e-05,
+      "loss": 1.3272,
+      "step": 8015
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.99558569618578e-05,
+      "loss": 1.2961,
+      "step": 8016
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.995582047784509e-05,
+      "loss": 1.3247,
+      "step": 8017
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.995578397876833e-05,
+      "loss": 1.294,
+      "step": 8018
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.995574746462755e-05,
+      "loss": 1.2988,
+      "step": 8019
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.995571093542273e-05,
+      "loss": 1.3108,
+      "step": 8020
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.995567439115392e-05,
+      "loss": 1.3522,
+      "step": 8021
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.995563783182112e-05,
+      "loss": 1.3289,
+      "step": 8022
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.995560125742432e-05,
+      "loss": 1.3185,
+      "step": 8023
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.995556466796354e-05,
+      "loss": 1.3904,
+      "step": 8024
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.995552806343882e-05,
+      "loss": 1.3218,
+      "step": 8025
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.995549144385012e-05,
+      "loss": 1.3093,
+      "step": 8026
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.99554548091975e-05,
+      "loss": 1.3185,
+      "step": 8027
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.995541815948093e-05,
+      "loss": 1.3569,
+      "step": 8028
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.995538149470045e-05,
+      "loss": 1.2874,
+      "step": 8029
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.995534481485607e-05,
+      "loss": 1.3265,
+      "step": 8030
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.995530811994778e-05,
+      "loss": 1.3169,
+      "step": 8031
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.99552714099756e-05,
+      "loss": 1.3232,
+      "step": 8032
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.995523468493957e-05,
+      "loss": 1.3563,
+      "step": 8033
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.995519794483965e-05,
+      "loss": 1.3016,
+      "step": 8034
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.995516118967588e-05,
+      "loss": 1.3246,
+      "step": 8035
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.995512441944827e-05,
+      "loss": 1.3209,
+      "step": 8036
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.995508763415684e-05,
+      "loss": 1.3001,
+      "step": 8037
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.995505083380158e-05,
+      "loss": 1.2991,
+      "step": 8038
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.995501401838251e-05,
+      "loss": 1.3103,
+      "step": 8039
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.995497718789965e-05,
+      "loss": 1.3011,
+      "step": 8040
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.9954940342353e-05,
+      "loss": 1.3259,
+      "step": 8041
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.995490348174257e-05,
+      "loss": 1.3717,
+      "step": 8042
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.995486660606836e-05,
+      "loss": 1.3234,
+      "step": 8043
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.995482971533042e-05,
+      "loss": 1.3222,
+      "step": 8044
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.995479280952873e-05,
+      "loss": 1.3079,
+      "step": 8045
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.995475588866332e-05,
+      "loss": 1.355,
+      "step": 8046
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.995471895273418e-05,
+      "loss": 1.3005,
+      "step": 8047
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.995468200174133e-05,
+      "loss": 1.3088,
+      "step": 8048
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.995464503568479e-05,
+      "loss": 1.3535,
+      "step": 8049
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.995460805456456e-05,
+      "loss": 1.3243,
+      "step": 8050
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.995457105838064e-05,
+      "loss": 1.3065,
+      "step": 8051
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.995453404713307e-05,
+      "loss": 1.3481,
+      "step": 8052
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.995449702082185e-05,
+      "loss": 1.3711,
+      "step": 8053
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.995445997944697e-05,
+      "loss": 1.3123,
+      "step": 8054
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.995442292300849e-05,
+      "loss": 1.3521,
+      "step": 8055
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.995438585150636e-05,
+      "loss": 1.3407,
+      "step": 8056
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.995434876494063e-05,
+      "loss": 1.3421,
+      "step": 8057
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.99543116633113e-05,
+      "loss": 1.3009,
+      "step": 8058
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.99542745466184e-05,
+      "loss": 1.3098,
+      "step": 8059
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.995423741486191e-05,
+      "loss": 1.3477,
+      "step": 8060
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.995420026804188e-05,
+      "loss": 1.3262,
+      "step": 8061
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.995416310615827e-05,
+      "loss": 1.3209,
+      "step": 8062
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.995412592921114e-05,
+      "loss": 1.3192,
+      "step": 8063
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.995408873720046e-05,
+      "loss": 1.2906,
+      "step": 8064
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.995405153012627e-05,
+      "loss": 1.3073,
+      "step": 8065
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.995401430798856e-05,
+      "loss": 1.3266,
+      "step": 8066
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.995397707078738e-05,
+      "loss": 1.3722,
+      "step": 8067
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.99539398185227e-05,
+      "loss": 1.3223,
+      "step": 8068
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.995390255119454e-05,
+      "loss": 1.3247,
+      "step": 8069
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.995386526880292e-05,
+      "loss": 1.3071,
+      "step": 8070
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.995382797134787e-05,
+      "loss": 1.3449,
+      "step": 8071
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.995379065882935e-05,
+      "loss": 1.3061,
+      "step": 8072
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.995375333124741e-05,
+      "loss": 1.3126,
+      "step": 8073
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.995371598860206e-05,
+      "loss": 1.3265,
+      "step": 8074
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.995367863089331e-05,
+      "loss": 1.322,
+      "step": 8075
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.995364125812114e-05,
+      "loss": 1.3288,
+      "step": 8076
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.99536038702856e-05,
+      "loss": 1.3153,
+      "step": 8077
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.99535664673867e-05,
+      "loss": 1.3048,
+      "step": 8078
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.995352904942442e-05,
+      "loss": 1.3529,
+      "step": 8079
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.995349161639879e-05,
+      "loss": 1.3426,
+      "step": 8080
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.995345416830984e-05,
+      "loss": 1.3349,
+      "step": 8081
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.995341670515755e-05,
+      "loss": 1.3118,
+      "step": 8082
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.995337922694195e-05,
+      "loss": 1.3319,
+      "step": 8083
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.995334173366302e-05,
+      "loss": 1.3274,
+      "step": 8084
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.995330422532082e-05,
+      "loss": 1.3336,
+      "step": 8085
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.995326670191534e-05,
+      "loss": 1.3375,
+      "step": 8086
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.995322916344658e-05,
+      "loss": 1.324,
+      "step": 8087
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.995319160991456e-05,
+      "loss": 1.327,
+      "step": 8088
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.99531540413193e-05,
+      "loss": 1.322,
+      "step": 8089
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.995311645766079e-05,
+      "loss": 1.2864,
+      "step": 8090
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.995307885893907e-05,
+      "loss": 1.3301,
+      "step": 8091
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.995304124515413e-05,
+      "loss": 1.3007,
+      "step": 8092
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.995300361630597e-05,
+      "loss": 1.2967,
+      "step": 8093
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.995296597239464e-05,
+      "loss": 1.3067,
+      "step": 8094
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.99529283134201e-05,
+      "loss": 1.3323,
+      "step": 8095
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.995289063938243e-05,
+      "loss": 1.3365,
+      "step": 8096
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.995285295028158e-05,
+      "loss": 1.347,
+      "step": 8097
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.99528152461176e-05,
+      "loss": 1.328,
+      "step": 8098
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.995277752689047e-05,
+      "loss": 1.342,
+      "step": 8099
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.995273979260022e-05,
+      "loss": 1.2964,
+      "step": 8100
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.995270204324687e-05,
+      "loss": 1.3332,
+      "step": 8101
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.995266427883041e-05,
+      "loss": 1.3299,
+      "step": 8102
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.995262649935086e-05,
+      "loss": 1.3453,
+      "step": 8103
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.995258870480823e-05,
+      "loss": 1.3548,
+      "step": 8104
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.995255089520254e-05,
+      "loss": 1.3173,
+      "step": 8105
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.995251307053378e-05,
+      "loss": 1.3391,
+      "step": 8106
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.995247523080198e-05,
+      "loss": 1.3072,
+      "step": 8107
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.995243737600716e-05,
+      "loss": 1.3252,
+      "step": 8108
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.995239950614932e-05,
+      "loss": 1.278,
+      "step": 8109
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.995236162122847e-05,
+      "loss": 1.5056,
+      "step": 8110
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.995232372124462e-05,
+      "loss": 1.2753,
+      "step": 8111
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.995228580619778e-05,
+      "loss": 1.3027,
+      "step": 8112
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.995224787608795e-05,
+      "loss": 1.3209,
+      "step": 8113
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.995220993091518e-05,
+      "loss": 1.2908,
+      "step": 8114
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.995217197067944e-05,
+      "loss": 1.3381,
+      "step": 8115
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.995213399538077e-05,
+      "loss": 1.3337,
+      "step": 8116
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.995209600501917e-05,
+      "loss": 1.31,
+      "step": 8117
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.995205799959465e-05,
+      "loss": 1.3411,
+      "step": 8118
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.995201997910722e-05,
+      "loss": 1.2908,
+      "step": 8119
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.995198194355691e-05,
+      "loss": 1.3369,
+      "step": 8120
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.995194389294368e-05,
+      "loss": 1.3432,
+      "step": 8121
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.995190582726761e-05,
+      "loss": 1.3309,
+      "step": 8122
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.995186774652868e-05,
+      "loss": 1.3018,
+      "step": 8123
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.99518296507269e-05,
+      "loss": 1.3045,
+      "step": 8124
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.995179153986227e-05,
+      "loss": 1.3388,
+      "step": 8125
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.995175341393483e-05,
+      "loss": 1.3321,
+      "step": 8126
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.995171527294456e-05,
+      "loss": 1.3106,
+      "step": 8127
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.995167711689148e-05,
+      "loss": 1.3307,
+      "step": 8128
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.995163894577562e-05,
+      "loss": 1.3328,
+      "step": 8129
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.995160075959698e-05,
+      "loss": 1.3437,
+      "step": 8130
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.995156255835558e-05,
+      "loss": 1.3338,
+      "step": 8131
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.995152434205142e-05,
+      "loss": 1.3297,
+      "step": 8132
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.99514861106845e-05,
+      "loss": 1.2938,
+      "step": 8133
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.995144786425486e-05,
+      "loss": 1.3072,
+      "step": 8134
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.99514096027625e-05,
+      "loss": 1.3286,
+      "step": 8135
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.995137132620742e-05,
+      "loss": 1.3155,
+      "step": 8136
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.995133303458963e-05,
+      "loss": 1.3365,
+      "step": 8137
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.995129472790917e-05,
+      "loss": 1.2873,
+      "step": 8138
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.995125640616603e-05,
+      "loss": 1.3294,
+      "step": 8139
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.995121806936024e-05,
+      "loss": 1.3231,
+      "step": 8140
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.995117971749176e-05,
+      "loss": 1.3234,
+      "step": 8141
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.995114135056066e-05,
+      "loss": 1.3122,
+      "step": 8142
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.995110296856693e-05,
+      "loss": 1.3331,
+      "step": 8143
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.995106457151058e-05,
+      "loss": 1.2775,
+      "step": 8144
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.995102615939162e-05,
+      "loss": 1.3264,
+      "step": 8145
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.995098773221007e-05,
+      "loss": 1.3104,
+      "step": 8146
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.995094928996593e-05,
+      "loss": 1.3135,
+      "step": 8147
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.995091083265923e-05,
+      "loss": 1.3075,
+      "step": 8148
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.995087236028997e-05,
+      "loss": 1.3322,
+      "step": 8149
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.995083387285815e-05,
+      "loss": 1.3279,
+      "step": 8150
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.995079537036379e-05,
+      "loss": 1.3226,
+      "step": 8151
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.99507568528069e-05,
+      "loss": 1.3428,
+      "step": 8152
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.995071832018751e-05,
+      "loss": 1.3376,
+      "step": 8153
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.995067977250562e-05,
+      "loss": 1.321,
+      "step": 8154
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.995064120976123e-05,
+      "loss": 1.3375,
+      "step": 8155
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.995060263195437e-05,
+      "loss": 1.3381,
+      "step": 8156
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.995056403908503e-05,
+      "loss": 1.3192,
+      "step": 8157
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.995052543115325e-05,
+      "loss": 1.3291,
+      "step": 8158
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.9950486808159e-05,
+      "loss": 1.3386,
+      "step": 8159
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.995044817010232e-05,
+      "loss": 1.2836,
+      "step": 8160
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.995040951698325e-05,
+      "loss": 1.3054,
+      "step": 8161
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.995037084880174e-05,
+      "loss": 1.3318,
+      "step": 8162
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.995033216555785e-05,
+      "loss": 1.3486,
+      "step": 8163
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.995029346725157e-05,
+      "loss": 1.3426,
+      "step": 8164
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.995025475388291e-05,
+      "loss": 1.323,
+      "step": 8165
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.99502160254519e-05,
+      "loss": 1.2989,
+      "step": 8166
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.995017728195853e-05,
+      "loss": 1.3258,
+      "step": 8167
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.995013852340282e-05,
+      "loss": 1.297,
+      "step": 8168
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.995009974978477e-05,
+      "loss": 1.3166,
+      "step": 8169
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.995006096110443e-05,
+      "loss": 1.3858,
+      "step": 8170
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.995002215736178e-05,
+      "loss": 1.3099,
+      "step": 8171
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.994998333855682e-05,
+      "loss": 1.3516,
+      "step": 8172
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.99499445046896e-05,
+      "loss": 1.2722,
+      "step": 8173
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.994990565576009e-05,
+      "loss": 1.3143,
+      "step": 8174
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.994986679176834e-05,
+      "loss": 1.3135,
+      "step": 8175
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.994982791271434e-05,
+      "loss": 1.3005,
+      "step": 8176
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.99497890185981e-05,
+      "loss": 1.3019,
+      "step": 8177
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.994975010941963e-05,
+      "loss": 1.2988,
+      "step": 8178
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.994971118517897e-05,
+      "loss": 1.3324,
+      "step": 8179
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.99496722458761e-05,
+      "loss": 1.2988,
+      "step": 8180
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.994963329151104e-05,
+      "loss": 1.3152,
+      "step": 8181
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.994959432208382e-05,
+      "loss": 1.3095,
+      "step": 8182
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.994955533759442e-05,
+      "loss": 1.3253,
+      "step": 8183
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.994951633804288e-05,
+      "loss": 1.325,
+      "step": 8184
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.994947732342918e-05,
+      "loss": 1.3072,
+      "step": 8185
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.994943829375337e-05,
+      "loss": 1.3137,
+      "step": 8186
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.994939924901545e-05,
+      "loss": 1.3376,
+      "step": 8187
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.994936018921541e-05,
+      "loss": 1.3007,
+      "step": 8188
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.994932111435328e-05,
+      "loss": 1.3152,
+      "step": 8189
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.994928202442907e-05,
+      "loss": 1.3271,
+      "step": 8190
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.994924291944278e-05,
+      "loss": 1.3427,
+      "step": 8191
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.994920379939445e-05,
+      "loss": 1.3281,
+      "step": 8192
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.994916466428408e-05,
+      "loss": 1.3142,
+      "step": 8193
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.994912551411167e-05,
+      "loss": 1.3431,
+      "step": 8194
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.994908634887721e-05,
+      "loss": 1.2896,
+      "step": 8195
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.994904716858078e-05,
+      "loss": 1.3552,
+      "step": 8196
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.994900797322233e-05,
+      "loss": 1.3339,
+      "step": 8197
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.994896876280188e-05,
+      "loss": 1.2878,
+      "step": 8198
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.994892953731948e-05,
+      "loss": 1.3254,
+      "step": 8199
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.994889029677512e-05,
+      "loss": 1.2977,
+      "step": 8200
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.994885104116878e-05,
+      "loss": 1.3345,
+      "step": 8201
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.994881177050052e-05,
+      "loss": 1.3031,
+      "step": 8202
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.994877248477032e-05,
+      "loss": 1.2736,
+      "step": 8203
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.994873318397822e-05,
+      "loss": 1.3218,
+      "step": 8204
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.994869386812423e-05,
+      "loss": 1.2925,
+      "step": 8205
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.994865453720831e-05,
+      "loss": 1.3181,
+      "step": 8206
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.994861519123053e-05,
+      "loss": 1.3105,
+      "step": 8207
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.994857583019089e-05,
+      "loss": 1.27,
+      "step": 8208
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.994853645408937e-05,
+      "loss": 1.3066,
+      "step": 8209
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.994849706292603e-05,
+      "loss": 1.2921,
+      "step": 8210
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.994845765670084e-05,
+      "loss": 1.3205,
+      "step": 8211
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.994841823541383e-05,
+      "loss": 1.3135,
+      "step": 8212
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.994837879906502e-05,
+      "loss": 1.3564,
+      "step": 8213
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.994833934765441e-05,
+      "loss": 1.3369,
+      "step": 8214
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.994829988118202e-05,
+      "loss": 1.3089,
+      "step": 8215
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.994826039964785e-05,
+      "loss": 1.3113,
+      "step": 8216
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.994822090305191e-05,
+      "loss": 1.3381,
+      "step": 8217
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.994818139139424e-05,
+      "loss": 1.3115,
+      "step": 8218
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.994814186467483e-05,
+      "loss": 1.3265,
+      "step": 8219
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.994810232289367e-05,
+      "loss": 1.3081,
+      "step": 8220
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.994806276605082e-05,
+      "loss": 1.3031,
+      "step": 8221
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.994802319414627e-05,
+      "loss": 1.2758,
+      "step": 8222
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.994798360718002e-05,
+      "loss": 1.2945,
+      "step": 8223
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.99479440051521e-05,
+      "loss": 1.2795,
+      "step": 8224
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.99479043880625e-05,
+      "loss": 1.3238,
+      "step": 8225
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.994786475591126e-05,
+      "loss": 1.3023,
+      "step": 8226
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.994782510869836e-05,
+      "loss": 1.3445,
+      "step": 8227
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.994778544642385e-05,
+      "loss": 1.3242,
+      "step": 8228
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.994774576908772e-05,
+      "loss": 1.2666,
+      "step": 8229
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.994770607668998e-05,
+      "loss": 1.3168,
+      "step": 8230
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.994766636923064e-05,
+      "loss": 1.3205,
+      "step": 8231
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.994762664670971e-05,
+      "loss": 1.3303,
+      "step": 8232
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.994758690912724e-05,
+      "loss": 1.304,
+      "step": 8233
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.994754715648318e-05,
+      "loss": 1.2983,
+      "step": 8234
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.99475073887776e-05,
+      "loss": 1.3234,
+      "step": 8235
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.994746760601047e-05,
+      "loss": 1.3223,
+      "step": 8236
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.994742780818182e-05,
+      "loss": 1.3188,
+      "step": 8237
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.994738799529167e-05,
+      "loss": 1.3122,
+      "step": 8238
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.994734816734e-05,
+      "loss": 1.3273,
+      "step": 8239
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.994730832432686e-05,
+      "loss": 1.3103,
+      "step": 8240
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.994726846625224e-05,
+      "loss": 1.2983,
+      "step": 8241
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.994722859311615e-05,
+      "loss": 1.317,
+      "step": 8242
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.994718870491863e-05,
+      "loss": 1.3437,
+      "step": 8243
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.994714880165966e-05,
+      "loss": 1.3187,
+      "step": 8244
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.994710888333927e-05,
+      "loss": 1.3047,
+      "step": 8245
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.994706894995746e-05,
+      "loss": 1.2904,
+      "step": 8246
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.994702900151425e-05,
+      "loss": 1.3634,
+      "step": 8247
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.994698903800965e-05,
+      "loss": 1.2729,
+      "step": 8248
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.994694905944366e-05,
+      "loss": 1.3187,
+      "step": 8249
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.994690906581632e-05,
+      "loss": 1.3293,
+      "step": 8250
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.994686905712762e-05,
+      "loss": 1.2863,
+      "step": 8251
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.994682903337759e-05,
+      "loss": 1.3506,
+      "step": 8252
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.994678899456622e-05,
+      "loss": 1.3214,
+      "step": 8253
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.994674894069354e-05,
+      "loss": 1.3019,
+      "step": 8254
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.994670887175954e-05,
+      "loss": 1.3278,
+      "step": 8255
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.994666878776426e-05,
+      "loss": 1.3205,
+      "step": 8256
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.994662868870768e-05,
+      "loss": 1.3333,
+      "step": 8257
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.994658857458984e-05,
+      "loss": 1.3457,
+      "step": 8258
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.994654844541076e-05,
+      "loss": 1.3282,
+      "step": 8259
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.994650830117042e-05,
+      "loss": 1.3529,
+      "step": 8260
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.994646814186884e-05,
+      "loss": 1.3298,
+      "step": 8261
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.994642796750605e-05,
+      "loss": 1.2905,
+      "step": 8262
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.994638777808205e-05,
+      "loss": 1.299,
+      "step": 8263
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.994634757359684e-05,
+      "loss": 1.2631,
+      "step": 8264
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.994630735405046e-05,
+      "loss": 1.3133,
+      "step": 8265
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.99462671194429e-05,
+      "loss": 1.3087,
+      "step": 8266
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.99462268697742e-05,
+      "loss": 1.3641,
+      "step": 8267
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.994618660504433e-05,
+      "loss": 1.3059,
+      "step": 8268
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.994614632525334e-05,
+      "loss": 1.3517,
+      "step": 8269
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.994610603040121e-05,
+      "loss": 1.3154,
+      "step": 8270
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.994606572048797e-05,
+      "loss": 1.3039,
+      "step": 8271
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.994602539551365e-05,
+      "loss": 1.3584,
+      "step": 8272
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.994598505547822e-05,
+      "loss": 1.3215,
+      "step": 8273
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.994594470038172e-05,
+      "loss": 1.3622,
+      "step": 8274
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.994590433022417e-05,
+      "loss": 1.355,
+      "step": 8275
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.994586394500556e-05,
+      "loss": 1.3053,
+      "step": 8276
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.99458235447259e-05,
+      "loss": 1.3553,
+      "step": 8277
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.994578312938523e-05,
+      "loss": 1.3315,
+      "step": 8278
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.994574269898354e-05,
+      "loss": 1.3098,
+      "step": 8279
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.994570225352084e-05,
+      "loss": 1.353,
+      "step": 8280
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.994566179299716e-05,
+      "loss": 1.3365,
+      "step": 8281
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.99456213174125e-05,
+      "loss": 1.2967,
+      "step": 8282
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.994558082676688e-05,
+      "loss": 1.3045,
+      "step": 8283
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.994554032106029e-05,
+      "loss": 1.2839,
+      "step": 8284
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.994549980029278e-05,
+      "loss": 1.3392,
+      "step": 8285
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.994545926446433e-05,
+      "loss": 1.3312,
+      "step": 8286
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.994541871357497e-05,
+      "loss": 1.2787,
+      "step": 8287
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.99453781476247e-05,
+      "loss": 1.321,
+      "step": 8288
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.994533756661353e-05,
+      "loss": 1.3173,
+      "step": 8289
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.99452969705415e-05,
+      "loss": 1.3223,
+      "step": 8290
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.994525635940859e-05,
+      "loss": 1.3228,
+      "step": 8291
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.994521573321483e-05,
+      "loss": 1.3139,
+      "step": 8292
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.994517509196022e-05,
+      "loss": 1.3122,
+      "step": 8293
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.994513443564476e-05,
+      "loss": 1.2983,
+      "step": 8294
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.994509376426852e-05,
+      "loss": 1.2998,
+      "step": 8295
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.994505307783146e-05,
+      "loss": 1.326,
+      "step": 8296
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.99450123763336e-05,
+      "loss": 1.3223,
+      "step": 8297
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.994497165977496e-05,
+      "loss": 1.3025,
+      "step": 8298
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.994493092815554e-05,
+      "loss": 1.332,
+      "step": 8299
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.994489018147537e-05,
+      "loss": 1.3067,
+      "step": 8300
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.994484941973446e-05,
+      "loss": 1.2891,
+      "step": 8301
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.994480864293282e-05,
+      "loss": 1.3376,
+      "step": 8302
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.994476785107045e-05,
+      "loss": 1.3157,
+      "step": 8303
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.99447270441474e-05,
+      "loss": 1.3102,
+      "step": 8304
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.994468622216362e-05,
+      "loss": 1.2926,
+      "step": 8305
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.994464538511916e-05,
+      "loss": 1.3145,
+      "step": 8306
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.994460453301404e-05,
+      "loss": 1.319,
+      "step": 8307
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.994456366584825e-05,
+      "loss": 1.2958,
+      "step": 8308
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.994452278362183e-05,
+      "loss": 1.3444,
+      "step": 8309
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.994448188633477e-05,
+      "loss": 1.3414,
+      "step": 8310
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.994444097398708e-05,
+      "loss": 1.2733,
+      "step": 8311
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.99444000465788e-05,
+      "loss": 1.3422,
+      "step": 8312
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.994435910410988e-05,
+      "loss": 1.2995,
+      "step": 8313
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.99443181465804e-05,
+      "loss": 1.3326,
+      "step": 8314
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.994427717399036e-05,
+      "loss": 1.3454,
+      "step": 8315
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.994423618633976e-05,
+      "loss": 1.3197,
+      "step": 8316
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.99441951836286e-05,
+      "loss": 1.3107,
+      "step": 8317
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.994415416585691e-05,
+      "loss": 1.3349,
+      "step": 8318
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.99441131330247e-05,
+      "loss": 1.3544,
+      "step": 8319
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.994407208513196e-05,
+      "loss": 1.3091,
+      "step": 8320
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.994403102217873e-05,
+      "loss": 1.2921,
+      "step": 8321
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.994398994416503e-05,
+      "loss": 1.3131,
+      "step": 8322
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.994394885109084e-05,
+      "loss": 1.3315,
+      "step": 8323
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.994390774295619e-05,
+      "loss": 1.313,
+      "step": 8324
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.994386661976109e-05,
+      "loss": 1.3497,
+      "step": 8325
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.994382548150557e-05,
+      "loss": 1.2983,
+      "step": 8326
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.994378432818961e-05,
+      "loss": 1.2939,
+      "step": 8327
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.994374315981326e-05,
+      "loss": 1.3238,
+      "step": 8328
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.994370197637648e-05,
+      "loss": 1.3293,
+      "step": 8329
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.994366077787933e-05,
+      "loss": 1.3079,
+      "step": 8330
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.99436195643218e-05,
+      "loss": 1.2952,
+      "step": 8331
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.994357833570391e-05,
+      "loss": 1.3128,
+      "step": 8332
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.994353709202568e-05,
+      "loss": 1.2979,
+      "step": 8333
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.994349583328711e-05,
+      "loss": 1.3471,
+      "step": 8334
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.994345455948821e-05,
+      "loss": 1.3155,
+      "step": 8335
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.9943413270629e-05,
+      "loss": 1.3318,
+      "step": 8336
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.994337196670948e-05,
+      "loss": 1.3238,
+      "step": 8337
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.994333064772968e-05,
+      "loss": 1.3345,
+      "step": 8338
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.994328931368961e-05,
+      "loss": 1.3058,
+      "step": 8339
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.994324796458927e-05,
+      "loss": 1.303,
+      "step": 8340
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.994320660042869e-05,
+      "loss": 1.3477,
+      "step": 8341
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.994316522120786e-05,
+      "loss": 1.3283,
+      "step": 8342
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.994312382692682e-05,
+      "loss": 1.3138,
+      "step": 8343
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.994308241758556e-05,
+      "loss": 1.3523,
+      "step": 8344
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.99430409931841e-05,
+      "loss": 1.3251,
+      "step": 8345
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.994299955372247e-05,
+      "loss": 1.2951,
+      "step": 8346
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.994295809920063e-05,
+      "loss": 1.2973,
+      "step": 8347
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.994291662961866e-05,
+      "loss": 1.346,
+      "step": 8348
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.99428751449765e-05,
+      "loss": 1.3379,
+      "step": 8349
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.994283364527423e-05,
+      "loss": 1.3518,
+      "step": 8350
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.994279213051186e-05,
+      "loss": 1.3143,
+      "step": 8351
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.994275060068934e-05,
+      "loss": 1.3369,
+      "step": 8352
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.994270905580673e-05,
+      "loss": 1.2997,
+      "step": 8353
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.994266749586403e-05,
+      "loss": 1.3327,
+      "step": 8354
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.994262592086126e-05,
+      "loss": 1.3101,
+      "step": 8355
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.994258433079843e-05,
+      "loss": 1.3406,
+      "step": 8356
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.994254272567553e-05,
+      "loss": 1.3375,
+      "step": 8357
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.994250110549263e-05,
+      "loss": 1.326,
+      "step": 8358
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.994245947024968e-05,
+      "loss": 1.336,
+      "step": 8359
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.994241781994672e-05,
+      "loss": 1.3682,
+      "step": 8360
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.994237615458377e-05,
+      "loss": 1.3346,
+      "step": 8361
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.994233447416082e-05,
+      "loss": 1.3063,
+      "step": 8362
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.99422927786779e-05,
+      "loss": 1.3296,
+      "step": 8363
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.994225106813503e-05,
+      "loss": 1.2946,
+      "step": 8364
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.99422093425322e-05,
+      "loss": 1.268,
+      "step": 8365
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.994216760186943e-05,
+      "loss": 1.288,
+      "step": 8366
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.994212584614675e-05,
+      "loss": 1.3384,
+      "step": 8367
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.994208407536414e-05,
+      "loss": 1.312,
+      "step": 8368
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.994204228952164e-05,
+      "loss": 1.3339,
+      "step": 8369
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.994200048861926e-05,
+      "loss": 1.2971,
+      "step": 8370
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.9941958672657e-05,
+      "loss": 1.3082,
+      "step": 8371
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.994191684163488e-05,
+      "loss": 1.3032,
+      "step": 8372
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.994187499555293e-05,
+      "loss": 1.3097,
+      "step": 8373
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.99418331344111e-05,
+      "loss": 1.3108,
+      "step": 8374
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.994179125820948e-05,
+      "loss": 1.3186,
+      "step": 8375
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.994174936694805e-05,
+      "loss": 1.2975,
+      "step": 8376
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.994170746062681e-05,
+      "loss": 1.2888,
+      "step": 8377
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.994166553924578e-05,
+      "loss": 1.31,
+      "step": 8378
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.994162360280498e-05,
+      "loss": 1.354,
+      "step": 8379
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.994158165130443e-05,
+      "loss": 1.3292,
+      "step": 8380
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.994153968474413e-05,
+      "loss": 1.3572,
+      "step": 8381
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.99414977031241e-05,
+      "loss": 1.3643,
+      "step": 8382
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.994145570644434e-05,
+      "loss": 1.3408,
+      "step": 8383
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.994141369470486e-05,
+      "loss": 1.3167,
+      "step": 8384
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.994137166790572e-05,
+      "loss": 1.3419,
+      "step": 8385
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.994132962604686e-05,
+      "loss": 1.3207,
+      "step": 8386
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.994128756912834e-05,
+      "loss": 1.3428,
+      "step": 8387
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.994124549715015e-05,
+      "loss": 1.3186,
+      "step": 8388
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.994120341011233e-05,
+      "loss": 1.3091,
+      "step": 8389
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.994116130801488e-05,
+      "loss": 1.3016,
+      "step": 8390
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.99411191908578e-05,
+      "loss": 1.2978,
+      "step": 8391
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.994107705864112e-05,
+      "loss": 1.3297,
+      "step": 8392
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.994103491136484e-05,
+      "loss": 1.3356,
+      "step": 8393
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.994099274902897e-05,
+      "loss": 1.302,
+      "step": 8394
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.994095057163354e-05,
+      "loss": 1.3062,
+      "step": 8395
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.994090837917855e-05,
+      "loss": 1.316,
+      "step": 8396
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.994086617166401e-05,
+      "loss": 1.3181,
+      "step": 8397
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.994082394908994e-05,
+      "loss": 1.3163,
+      "step": 8398
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.994078171145636e-05,
+      "loss": 1.2803,
+      "step": 8399
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.994073945876327e-05,
+      "loss": 1.3076,
+      "step": 8400
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.994069719101069e-05,
+      "loss": 1.3564,
+      "step": 8401
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.994065490819862e-05,
+      "loss": 1.3055,
+      "step": 8402
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.994061261032709e-05,
+      "loss": 1.3026,
+      "step": 8403
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.99405702973961e-05,
+      "loss": 1.3049,
+      "step": 8404
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.994052796940567e-05,
+      "loss": 1.3178,
+      "step": 8405
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.99404856263558e-05,
+      "loss": 1.3324,
+      "step": 8406
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.994044326824653e-05,
+      "loss": 1.3033,
+      "step": 8407
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.994040089507786e-05,
+      "loss": 1.3397,
+      "step": 8408
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.994035850684977e-05,
+      "loss": 1.3355,
+      "step": 8409
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.994031610356232e-05,
+      "loss": 1.3297,
+      "step": 8410
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.994027368521552e-05,
+      "loss": 1.3333,
+      "step": 8411
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.994023125180934e-05,
+      "loss": 1.3333,
+      "step": 8412
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.994018880334383e-05,
+      "loss": 1.242,
+      "step": 8413
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.9940146339819e-05,
+      "loss": 1.3055,
+      "step": 8414
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.994010386123485e-05,
+      "loss": 1.3229,
+      "step": 8415
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.994006136759138e-05,
+      "loss": 1.3323,
+      "step": 8416
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.994001885888863e-05,
+      "loss": 1.3186,
+      "step": 8417
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.993997633512663e-05,
+      "loss": 1.3222,
+      "step": 8418
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.993993379630536e-05,
+      "loss": 1.3151,
+      "step": 8419
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.993989124242482e-05,
+      "loss": 1.3064,
+      "step": 8420
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.993984867348505e-05,
+      "loss": 1.303,
+      "step": 8421
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.993980608948606e-05,
+      "loss": 1.3283,
+      "step": 8422
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.993976349042784e-05,
+      "loss": 1.3034,
+      "step": 8423
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.993972087631043e-05,
+      "loss": 1.3167,
+      "step": 8424
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.993967824713385e-05,
+      "loss": 1.3308,
+      "step": 8425
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.993963560289808e-05,
+      "loss": 1.3139,
+      "step": 8426
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.993959294360316e-05,
+      "loss": 1.2967,
+      "step": 8427
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.993955026924908e-05,
+      "loss": 1.3651,
+      "step": 8428
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.993950757983588e-05,
+      "loss": 1.3113,
+      "step": 8429
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.993946487536354e-05,
+      "loss": 1.2991,
+      "step": 8430
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.993942215583211e-05,
+      "loss": 1.3048,
+      "step": 8431
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.993937942124157e-05,
+      "loss": 1.2903,
+      "step": 8432
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.993933667159195e-05,
+      "loss": 1.3031,
+      "step": 8433
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.993929390688325e-05,
+      "loss": 1.2893,
+      "step": 8434
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.993925112711551e-05,
+      "loss": 1.3354,
+      "step": 8435
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.993920833228872e-05,
+      "loss": 1.3224,
+      "step": 8436
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.99391655224029e-05,
+      "loss": 1.3184,
+      "step": 8437
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.993912269745806e-05,
+      "loss": 1.3169,
+      "step": 8438
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.99390798574542e-05,
+      "loss": 1.3037,
+      "step": 8439
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.993903700239138e-05,
+      "loss": 1.3034,
+      "step": 8440
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.993899413226956e-05,
+      "loss": 1.2886,
+      "step": 8441
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.993895124708877e-05,
+      "loss": 1.3517,
+      "step": 8442
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.993890834684902e-05,
+      "loss": 1.3151,
+      "step": 8443
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.993886543155035e-05,
+      "loss": 1.327,
+      "step": 8444
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.993882250119273e-05,
+      "loss": 1.308,
+      "step": 8445
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.99387795557762e-05,
+      "loss": 1.2875,
+      "step": 8446
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.993873659530078e-05,
+      "loss": 1.3012,
+      "step": 8447
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.993869361976647e-05,
+      "loss": 1.32,
+      "step": 8448
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.993865062917326e-05,
+      "loss": 1.313,
+      "step": 8449
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.99386076235212e-05,
+      "loss": 1.318,
+      "step": 8450
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.99385646028103e-05,
+      "loss": 1.302,
+      "step": 8451
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.993852156704055e-05,
+      "loss": 1.3408,
+      "step": 8452
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.993847851621198e-05,
+      "loss": 1.3174,
+      "step": 8453
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.993843545032459e-05,
+      "loss": 1.2851,
+      "step": 8454
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.993839236937842e-05,
+      "loss": 1.2849,
+      "step": 8455
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.993834927337345e-05,
+      "loss": 1.3359,
+      "step": 8456
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.993830616230972e-05,
+      "loss": 1.3145,
+      "step": 8457
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.993826303618722e-05,
+      "loss": 1.2764,
+      "step": 8458
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.993821989500598e-05,
+      "loss": 1.3382,
+      "step": 8459
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.9938176738766e-05,
+      "loss": 1.2922,
+      "step": 8460
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.99381335674673e-05,
+      "loss": 1.3185,
+      "step": 8461
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.993809038110988e-05,
+      "loss": 1.2956,
+      "step": 8462
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.993804717969378e-05,
+      "loss": 1.3356,
+      "step": 8463
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.9938003963219e-05,
+      "loss": 1.3061,
+      "step": 8464
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.993796073168555e-05,
+      "loss": 1.2942,
+      "step": 8465
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.993791748509345e-05,
+      "loss": 1.3125,
+      "step": 8466
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.993787422344269e-05,
+      "loss": 1.3096,
+      "step": 8467
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.993783094673331e-05,
+      "loss": 1.3066,
+      "step": 8468
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.993778765496531e-05,
+      "loss": 1.3289,
+      "step": 8469
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.993774434813872e-05,
+      "loss": 1.3093,
+      "step": 8470
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.993770102625353e-05,
+      "loss": 1.344,
+      "step": 8471
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.993765768930977e-05,
+      "loss": 1.2741,
+      "step": 8472
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.993761433730744e-05,
+      "loss": 1.326,
+      "step": 8473
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.993757097024655e-05,
+      "loss": 1.2743,
+      "step": 8474
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.993752758812713e-05,
+      "loss": 1.3143,
+      "step": 8475
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.993748419094919e-05,
+      "loss": 1.3034,
+      "step": 8476
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.993744077871274e-05,
+      "loss": 1.3378,
+      "step": 8477
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.993739735141779e-05,
+      "loss": 1.3078,
+      "step": 8478
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.993735390906433e-05,
+      "loss": 1.3015,
+      "step": 8479
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.993731045165244e-05,
+      "loss": 1.3343,
+      "step": 8480
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.993726697918206e-05,
+      "loss": 1.3139,
+      "step": 8481
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.993722349165325e-05,
+      "loss": 1.3283,
+      "step": 8482
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.993717998906598e-05,
+      "loss": 1.3456,
+      "step": 8483
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.993713647142031e-05,
+      "loss": 1.3441,
+      "step": 8484
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.993709293871625e-05,
+      "loss": 1.3262,
+      "step": 8485
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.993704939095376e-05,
+      "loss": 1.3321,
+      "step": 8486
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.993700582813293e-05,
+      "loss": 1.3474,
+      "step": 8487
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.99369622502537e-05,
+      "loss": 1.3136,
+      "step": 8488
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.993691865731611e-05,
+      "loss": 1.3217,
+      "step": 8489
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.993687504932021e-05,
+      "loss": 1.316,
+      "step": 8490
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.993683142626597e-05,
+      "loss": 1.3093,
+      "step": 8491
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.993678778815341e-05,
+      "loss": 1.2916,
+      "step": 8492
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.993674413498254e-05,
+      "loss": 1.3019,
+      "step": 8493
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.99367004667534e-05,
+      "loss": 1.3172,
+      "step": 8494
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.993665678346598e-05,
+      "loss": 1.3431,
+      "step": 8495
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.993661308512028e-05,
+      "loss": 1.3405,
+      "step": 8496
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.993656937171635e-05,
+      "loss": 1.3068,
+      "step": 8497
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.993652564325417e-05,
+      "loss": 1.3046,
+      "step": 8498
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.993648189973378e-05,
+      "loss": 1.2863,
+      "step": 8499
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.993643814115518e-05,
+      "loss": 1.3528,
+      "step": 8500
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.993639436751838e-05,
+      "loss": 1.3217,
+      "step": 8501
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.99363505788234e-05,
+      "loss": 1.3176,
+      "step": 8502
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.993630677507024e-05,
+      "loss": 1.3059,
+      "step": 8503
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.993626295625893e-05,
+      "loss": 1.3116,
+      "step": 8504
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.993621912238949e-05,
+      "loss": 1.3049,
+      "step": 8505
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.993617527346188e-05,
+      "loss": 1.3036,
+      "step": 8506
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.99361314094762e-05,
+      "loss": 1.2826,
+      "step": 8507
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.993608753043239e-05,
+      "loss": 1.2949,
+      "step": 8508
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.993604363633048e-05,
+      "loss": 1.33,
+      "step": 8509
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.993599972717051e-05,
+      "loss": 1.2789,
+      "step": 8510
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.993595580295248e-05,
+      "loss": 1.3162,
+      "step": 8511
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.993591186367639e-05,
+      "loss": 1.3378,
+      "step": 8512
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.993586790934227e-05,
+      "loss": 1.3109,
+      "step": 8513
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.993582393995012e-05,
+      "loss": 1.2878,
+      "step": 8514
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.993577995549996e-05,
+      "loss": 1.3354,
+      "step": 8515
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.99357359559918e-05,
+      "loss": 1.3136,
+      "step": 8516
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.993569194142566e-05,
+      "loss": 1.316,
+      "step": 8517
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.993564791180155e-05,
+      "loss": 1.3143,
+      "step": 8518
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.993560386711946e-05,
+      "loss": 1.301,
+      "step": 8519
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.993555980737946e-05,
+      "loss": 1.3247,
+      "step": 8520
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.99355157325815e-05,
+      "loss": 1.2662,
+      "step": 8521
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.993547164272563e-05,
+      "loss": 1.3188,
+      "step": 8522
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.993542753781186e-05,
+      "loss": 1.319,
+      "step": 8523
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.99353834178402e-05,
+      "loss": 1.2553,
+      "step": 8524
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.993533928281065e-05,
+      "loss": 1.2912,
+      "step": 8525
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.993529513272324e-05,
+      "loss": 1.3327,
+      "step": 8526
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.993525096757799e-05,
+      "loss": 1.3146,
+      "step": 8527
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.993520678737489e-05,
+      "loss": 1.2872,
+      "step": 8528
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.993516259211398e-05,
+      "loss": 1.3468,
+      "step": 8529
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.993511838179524e-05,
+      "loss": 1.3085,
+      "step": 8530
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.993507415641872e-05,
+      "loss": 1.296,
+      "step": 8531
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.993502991598439e-05,
+      "loss": 1.3506,
+      "step": 8532
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.993498566049231e-05,
+      "loss": 1.2793,
+      "step": 8533
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.993494138994246e-05,
+      "loss": 1.309,
+      "step": 8534
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.993489710433485e-05,
+      "loss": 1.3119,
+      "step": 8535
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.993485280366954e-05,
+      "loss": 1.2954,
+      "step": 8536
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.99348084879465e-05,
+      "loss": 1.278,
+      "step": 8537
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.993476415716576e-05,
+      "loss": 1.3048,
+      "step": 8538
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.993471981132732e-05,
+      "loss": 1.2986,
+      "step": 8539
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.99346754504312e-05,
+      "loss": 1.3153,
+      "step": 8540
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.993463107447742e-05,
+      "loss": 1.2932,
+      "step": 8541
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.993458668346598e-05,
+      "loss": 1.2816,
+      "step": 8542
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.993454227739692e-05,
+      "loss": 1.3275,
+      "step": 8543
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.993449785627023e-05,
+      "loss": 1.2946,
+      "step": 8544
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.993445342008591e-05,
+      "loss": 1.3242,
+      "step": 8545
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.9934408968844e-05,
+      "loss": 1.3151,
+      "step": 8546
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.993436450254452e-05,
+      "loss": 1.2919,
+      "step": 8547
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.993432002118746e-05,
+      "loss": 1.3409,
+      "step": 8548
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.993427552477283e-05,
+      "loss": 1.3362,
+      "step": 8549
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.993423101330069e-05,
+      "loss": 1.3502,
+      "step": 8550
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.993418648677098e-05,
+      "loss": 1.3103,
+      "step": 8551
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.993414194518378e-05,
+      "loss": 1.3157,
+      "step": 8552
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.993409738853906e-05,
+      "loss": 1.3085,
+      "step": 8553
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.993405281683685e-05,
+      "loss": 1.3153,
+      "step": 8554
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.993400823007716e-05,
+      "loss": 1.2792,
+      "step": 8555
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.993396362826003e-05,
+      "loss": 1.3143,
+      "step": 8556
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.993391901138543e-05,
+      "loss": 1.2931,
+      "step": 8557
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.993387437945341e-05,
+      "loss": 1.2978,
+      "step": 8558
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.993382973246395e-05,
+      "loss": 1.3152,
+      "step": 8559
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.993378507041709e-05,
+      "loss": 1.2809,
+      "step": 8560
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.993374039331283e-05,
+      "loss": 1.2851,
+      "step": 8561
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.993369570115118e-05,
+      "loss": 1.2976,
+      "step": 8562
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.993365099393218e-05,
+      "loss": 1.3103,
+      "step": 8563
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.99336062716558e-05,
+      "loss": 1.3281,
+      "step": 8564
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.993356153432209e-05,
+      "loss": 1.3499,
+      "step": 8565
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.993351678193104e-05,
+      "loss": 1.2745,
+      "step": 8566
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.993347201448268e-05,
+      "loss": 1.3317,
+      "step": 8567
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.993342723197703e-05,
+      "loss": 1.2957,
+      "step": 8568
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.993338243441409e-05,
+      "loss": 1.3182,
+      "step": 8569
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.993333762179386e-05,
+      "loss": 1.3093,
+      "step": 8570
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.993329279411638e-05,
+      "loss": 1.319,
+      "step": 8571
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.993324795138166e-05,
+      "loss": 1.3266,
+      "step": 8572
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.99332030935897e-05,
+      "loss": 1.2884,
+      "step": 8573
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.99331582207405e-05,
+      "loss": 1.3139,
+      "step": 8574
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.99331133328341e-05,
+      "loss": 1.314,
+      "step": 8575
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.993306842987053e-05,
+      "loss": 1.27,
+      "step": 8576
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.993302351184975e-05,
+      "loss": 1.3435,
+      "step": 8577
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.993297857877181e-05,
+      "loss": 1.3368,
+      "step": 8578
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.993293363063674e-05,
+      "loss": 1.305,
+      "step": 8579
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.993288866744451e-05,
+      "loss": 1.3201,
+      "step": 8580
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.993284368919516e-05,
+      "loss": 1.3122,
+      "step": 8581
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.99327986958887e-05,
+      "loss": 1.3479,
+      "step": 8582
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.993275368752513e-05,
+      "loss": 1.2947,
+      "step": 8583
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.993270866410448e-05,
+      "loss": 1.3474,
+      "step": 8584
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.993266362562676e-05,
+      "loss": 1.2928,
+      "step": 8585
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.993261857209199e-05,
+      "loss": 1.328,
+      "step": 8586
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.993257350350016e-05,
+      "loss": 1.3361,
+      "step": 8587
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.99325284198513e-05,
+      "loss": 1.3469,
+      "step": 8588
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.993248332114544e-05,
+      "loss": 1.3145,
+      "step": 8589
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.993243820738256e-05,
+      "loss": 1.3025,
+      "step": 8590
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.99323930785627e-05,
+      "loss": 1.3694,
+      "step": 8591
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.993234793468586e-05,
+      "loss": 1.327,
+      "step": 8592
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.993230277575205e-05,
+      "loss": 1.3344,
+      "step": 8593
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.99322576017613e-05,
+      "loss": 1.3033,
+      "step": 8594
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.99322124127136e-05,
+      "loss": 1.3387,
+      "step": 8595
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.993216720860898e-05,
+      "loss": 1.3017,
+      "step": 8596
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.993212198944748e-05,
+      "loss": 1.303,
+      "step": 8597
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.993207675522906e-05,
+      "loss": 1.3183,
+      "step": 8598
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.993203150595376e-05,
+      "loss": 1.3233,
+      "step": 8599
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.993198624162159e-05,
+      "loss": 1.3294,
+      "step": 8600
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.993194096223257e-05,
+      "loss": 1.3198,
+      "step": 8601
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.993189566778673e-05,
+      "loss": 1.3231,
+      "step": 8602
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.993185035828403e-05,
+      "loss": 1.3035,
+      "step": 8603
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.993180503372454e-05,
+      "loss": 1.3233,
+      "step": 8604
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.993175969410824e-05,
+      "loss": 1.3102,
+      "step": 8605
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.993171433943515e-05,
+      "loss": 1.2843,
+      "step": 8606
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.99316689697053e-05,
+      "loss": 1.3269,
+      "step": 8607
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.99316235849187e-05,
+      "loss": 1.3476,
+      "step": 8608
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.993157818507534e-05,
+      "loss": 1.345,
+      "step": 8609
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.993153277017526e-05,
+      "loss": 1.322,
+      "step": 8610
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.993148734021845e-05,
+      "loss": 1.3177,
+      "step": 8611
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.993144189520493e-05,
+      "loss": 1.2703,
+      "step": 8612
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.993139643513474e-05,
+      "loss": 1.3117,
+      "step": 8613
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.993135096000786e-05,
+      "loss": 1.3303,
+      "step": 8614
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.993130546982434e-05,
+      "loss": 1.2966,
+      "step": 8615
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.993125996458415e-05,
+      "loss": 1.3749,
+      "step": 8616
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.993121444428732e-05,
+      "loss": 1.3045,
+      "step": 8617
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.993116890893388e-05,
+      "loss": 1.3225,
+      "step": 8618
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.993112335852384e-05,
+      "loss": 1.3163,
+      "step": 8619
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.99310777930572e-05,
+      "loss": 1.3207,
+      "step": 8620
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.993103221253397e-05,
+      "loss": 1.2984,
+      "step": 8621
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.993098661695419e-05,
+      "loss": 1.2997,
+      "step": 8622
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.993094100631784e-05,
+      "loss": 1.3054,
+      "step": 8623
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.993089538062496e-05,
+      "loss": 1.3163,
+      "step": 8624
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.993084973987556e-05,
+      "loss": 1.3018,
+      "step": 8625
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.993080408406965e-05,
+      "loss": 1.3044,
+      "step": 8626
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.993075841320722e-05,
+      "loss": 1.3314,
+      "step": 8627
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.993071272728833e-05,
+      "loss": 1.3104,
+      "step": 8628
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.993066702631297e-05,
+      "loss": 1.304,
+      "step": 8629
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.993062131028115e-05,
+      "loss": 1.3131,
+      "step": 8630
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.993057557919288e-05,
+      "loss": 1.3194,
+      "step": 8631
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.993052983304819e-05,
+      "loss": 1.3226,
+      "step": 8632
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.993048407184709e-05,
+      "loss": 1.2868,
+      "step": 8633
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.993043829558957e-05,
+      "loss": 1.3044,
+      "step": 8634
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.993039250427568e-05,
+      "loss": 1.2518,
+      "step": 8635
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.993034669790541e-05,
+      "loss": 1.3075,
+      "step": 8636
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.993030087647879e-05,
+      "loss": 1.3311,
+      "step": 8637
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.99302550399958e-05,
+      "loss": 1.3248,
+      "step": 8638
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.99302091884565e-05,
+      "loss": 1.2919,
+      "step": 8639
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.993016332186086e-05,
+      "loss": 1.3154,
+      "step": 8640
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.993011744020895e-05,
+      "loss": 1.3315,
+      "step": 8641
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.993007154350074e-05,
+      "loss": 1.3423,
+      "step": 8642
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.993002563173624e-05,
+      "loss": 1.3178,
+      "step": 8643
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.992997970491547e-05,
+      "loss": 1.3299,
+      "step": 8644
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.992993376303847e-05,
+      "loss": 1.3087,
+      "step": 8645
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.992988780610524e-05,
+      "loss": 1.302,
+      "step": 8646
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.992984183411576e-05,
+      "loss": 1.3263,
+      "step": 8647
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.99297958470701e-05,
+      "loss": 1.3069,
+      "step": 8648
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.992974984496823e-05,
+      "loss": 1.3321,
+      "step": 8649
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.99297038278102e-05,
+      "loss": 1.2959,
+      "step": 8650
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.992965779559599e-05,
+      "loss": 1.3342,
+      "step": 8651
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.992961174832562e-05,
+      "loss": 1.2794,
+      "step": 8652
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.992956568599913e-05,
+      "loss": 1.3188,
+      "step": 8653
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.992951960861651e-05,
+      "loss": 1.2927,
+      "step": 8654
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.992947351617778e-05,
+      "loss": 1.349,
+      "step": 8655
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.992942740868293e-05,
+      "loss": 1.3325,
+      "step": 8656
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.992938128613203e-05,
+      "loss": 1.2939,
+      "step": 8657
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.992933514852506e-05,
+      "loss": 1.2895,
+      "step": 8658
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.992928899586203e-05,
+      "loss": 1.3017,
+      "step": 8659
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.992924282814295e-05,
+      "loss": 1.3057,
+      "step": 8660
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.992919664536785e-05,
+      "loss": 1.2873,
+      "step": 8661
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.992915044753675e-05,
+      "loss": 1.3329,
+      "step": 8662
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.992910423464962e-05,
+      "loss": 1.3095,
+      "step": 8663
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.992905800670653e-05,
+      "loss": 1.3302,
+      "step": 8664
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.992901176370747e-05,
+      "loss": 1.3628,
+      "step": 8665
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.992896550565243e-05,
+      "loss": 1.2913,
+      "step": 8666
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.992891923254147e-05,
+      "loss": 1.2893,
+      "step": 8667
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.992887294437458e-05,
+      "loss": 1.296,
+      "step": 8668
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.992882664115177e-05,
+      "loss": 1.3406,
+      "step": 8669
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.992878032287305e-05,
+      "loss": 1.3051,
+      "step": 8670
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.992873398953846e-05,
+      "loss": 1.334,
+      "step": 8671
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.992868764114797e-05,
+      "loss": 1.3208,
+      "step": 8672
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.992864127770165e-05,
+      "loss": 1.3172,
+      "step": 8673
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.992859489919948e-05,
+      "loss": 1.2802,
+      "step": 8674
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.992854850564146e-05,
+      "loss": 1.2955,
+      "step": 8675
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.992850209702765e-05,
+      "loss": 1.3249,
+      "step": 8676
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.9928455673358e-05,
+      "loss": 1.324,
+      "step": 8677
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.992840923463259e-05,
+      "loss": 1.2843,
+      "step": 8678
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.99283627808514e-05,
+      "loss": 1.3241,
+      "step": 8679
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.992831631201445e-05,
+      "loss": 1.2909,
+      "step": 8680
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.992826982812173e-05,
+      "loss": 1.3358,
+      "step": 8681
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.99282233291733e-05,
+      "loss": 1.3237,
+      "step": 8682
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.992817681516915e-05,
+      "loss": 1.3254,
+      "step": 8683
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.99281302861093e-05,
+      "loss": 1.3196,
+      "step": 8684
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.992808374199374e-05,
+      "loss": 1.3225,
+      "step": 8685
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.99280371828225e-05,
+      "loss": 1.2849,
+      "step": 8686
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.992799060859562e-05,
+      "loss": 1.3423,
+      "step": 8687
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.992794401931309e-05,
+      "loss": 1.287,
+      "step": 8688
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.992789741497491e-05,
+      "loss": 1.3295,
+      "step": 8689
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.992785079558112e-05,
+      "loss": 1.3429,
+      "step": 8690
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.992780416113172e-05,
+      "loss": 1.3255,
+      "step": 8691
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.992775751162672e-05,
+      "loss": 1.3252,
+      "step": 8692
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.992771084706615e-05,
+      "loss": 1.3255,
+      "step": 8693
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.992766416745001e-05,
+      "loss": 1.3331,
+      "step": 8694
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.992761747277832e-05,
+      "loss": 1.3163,
+      "step": 8695
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.992757076305109e-05,
+      "loss": 1.3357,
+      "step": 8696
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.992752403826835e-05,
+      "loss": 1.3495,
+      "step": 8697
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.992747729843009e-05,
+      "loss": 1.3073,
+      "step": 8698
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.992743054353634e-05,
+      "loss": 1.2795,
+      "step": 8699
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.992738377358711e-05,
+      "loss": 1.2826,
+      "step": 8700
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.99273369885824e-05,
+      "loss": 1.3084,
+      "step": 8701
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.992729018852227e-05,
+      "loss": 1.328,
+      "step": 8702
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.992724337340668e-05,
+      "loss": 1.2923,
+      "step": 8703
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.992719654323567e-05,
+      "loss": 1.3055,
+      "step": 8704
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.992714969800924e-05,
+      "loss": 1.3453,
+      "step": 8705
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.992710283772744e-05,
+      "loss": 1.3388,
+      "step": 8706
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.992705596239023e-05,
+      "loss": 1.2811,
+      "step": 8707
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.992700907199768e-05,
+      "loss": 1.27,
+      "step": 8708
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.992696216654975e-05,
+      "loss": 1.2864,
+      "step": 8709
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.99269152460465e-05,
+      "loss": 1.2814,
+      "step": 8710
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.992686831048792e-05,
+      "loss": 1.2986,
+      "step": 8711
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.992682135987404e-05,
+      "loss": 1.3259,
+      "step": 8712
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.992677439420485e-05,
+      "loss": 1.338,
+      "step": 8713
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.992672741348037e-05,
+      "loss": 1.2875,
+      "step": 8714
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.992668041770064e-05,
+      "loss": 1.3542,
+      "step": 8715
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.992663340686566e-05,
+      "loss": 1.3251,
+      "step": 8716
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.992658638097542e-05,
+      "loss": 1.3444,
+      "step": 8717
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.992653934002996e-05,
+      "loss": 1.3299,
+      "step": 8718
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.992649228402929e-05,
+      "loss": 1.2829,
+      "step": 8719
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.992644521297343e-05,
+      "loss": 1.3019,
+      "step": 8720
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.992639812686237e-05,
+      "loss": 1.3237,
+      "step": 8721
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.992635102569616e-05,
+      "loss": 1.3363,
+      "step": 8722
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.99263039094748e-05,
+      "loss": 1.3114,
+      "step": 8723
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.992625677819828e-05,
+      "loss": 1.2848,
+      "step": 8724
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.992620963186665e-05,
+      "loss": 1.2964,
+      "step": 8725
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.99261624704799e-05,
+      "loss": 1.3097,
+      "step": 8726
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.992611529403805e-05,
+      "loss": 1.3141,
+      "step": 8727
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.992606810254111e-05,
+      "loss": 1.3045,
+      "step": 8728
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.992602089598911e-05,
+      "loss": 1.3161,
+      "step": 8729
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.992597367438205e-05,
+      "loss": 1.2882,
+      "step": 8730
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.992592643771996e-05,
+      "loss": 1.3306,
+      "step": 8731
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.992587918600285e-05,
+      "loss": 1.2948,
+      "step": 8732
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.99258319192307e-05,
+      "loss": 1.3168,
+      "step": 8733
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.992578463740357e-05,
+      "loss": 1.3086,
+      "step": 8734
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.992573734052146e-05,
+      "loss": 1.3609,
+      "step": 8735
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.992569002858436e-05,
+      "loss": 1.3432,
+      "step": 8736
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.992564270159232e-05,
+      "loss": 1.2848,
+      "step": 8737
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.992559535954534e-05,
+      "loss": 1.3167,
+      "step": 8738
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.992554800244343e-05,
+      "loss": 1.3185,
+      "step": 8739
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.99255006302866e-05,
+      "loss": 1.3009,
+      "step": 8740
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.99254532430749e-05,
+      "loss": 1.3493,
+      "step": 8741
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.99254058408083e-05,
+      "loss": 1.3098,
+      "step": 8742
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.992535842348682e-05,
+      "loss": 1.302,
+      "step": 8743
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.992531099111048e-05,
+      "loss": 1.3081,
+      "step": 8744
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.992526354367933e-05,
+      "loss": 1.2808,
+      "step": 8745
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.992521608119332e-05,
+      "loss": 1.301,
+      "step": 8746
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.992516860365251e-05,
+      "loss": 1.323,
+      "step": 8747
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.992512111105691e-05,
+      "loss": 1.2961,
+      "step": 8748
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.992507360340652e-05,
+      "loss": 1.3065,
+      "step": 8749
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.992502608070135e-05,
+      "loss": 1.3429,
+      "step": 8750
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.992497854294145e-05,
+      "loss": 1.312,
+      "step": 8751
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.992493099012679e-05,
+      "loss": 1.3108,
+      "step": 8752
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.992488342225741e-05,
+      "loss": 1.3484,
+      "step": 8753
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.992483583933331e-05,
+      "loss": 1.3239,
+      "step": 8754
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.992478824135451e-05,
+      "loss": 1.3088,
+      "step": 8755
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.992474062832104e-05,
+      "loss": 1.3295,
+      "step": 8756
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.992469300023291e-05,
+      "loss": 1.2817,
+      "step": 8757
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.99246453570901e-05,
+      "loss": 1.304,
+      "step": 8758
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.992459769889266e-05,
+      "loss": 1.3268,
+      "step": 8759
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.99245500256406e-05,
+      "loss": 1.3272,
+      "step": 8760
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.992450233733392e-05,
+      "loss": 1.3404,
+      "step": 8761
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.992445463397264e-05,
+      "loss": 1.2888,
+      "step": 8762
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.992440691555679e-05,
+      "loss": 1.3094,
+      "step": 8763
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.992435918208636e-05,
+      "loss": 1.3019,
+      "step": 8764
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.99243114335614e-05,
+      "loss": 1.3012,
+      "step": 8765
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.992426366998187e-05,
+      "loss": 1.3121,
+      "step": 8766
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.992421589134782e-05,
+      "loss": 1.3066,
+      "step": 8767
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.992416809765927e-05,
+      "loss": 1.3342,
+      "step": 8768
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.992412028891623e-05,
+      "loss": 1.2879,
+      "step": 8769
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.992407246511869e-05,
+      "loss": 1.2949,
+      "step": 8770
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.992402462626668e-05,
+      "loss": 1.2999,
+      "step": 8771
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.992397677236025e-05,
+      "loss": 1.316,
+      "step": 8772
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.992392890339935e-05,
+      "loss": 1.301,
+      "step": 8773
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.992388101938403e-05,
+      "loss": 1.3062,
+      "step": 8774
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.99238331203143e-05,
+      "loss": 1.335,
+      "step": 8775
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.992378520619019e-05,
+      "loss": 1.3263,
+      "step": 8776
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.992373727701167e-05,
+      "loss": 1.2882,
+      "step": 8777
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.992368933277882e-05,
+      "loss": 1.3034,
+      "step": 8778
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.99236413734916e-05,
+      "loss": 1.3204,
+      "step": 8779
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.992359339915003e-05,
+      "loss": 1.2966,
+      "step": 8780
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.992354540975416e-05,
+      "loss": 1.3205,
+      "step": 8781
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.992349740530397e-05,
+      "loss": 1.3289,
+      "step": 8782
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.992344938579949e-05,
+      "loss": 1.2935,
+      "step": 8783
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.992340135124072e-05,
+      "loss": 1.296,
+      "step": 8784
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.99233533016277e-05,
+      "loss": 1.3072,
+      "step": 8785
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.992330523696042e-05,
+      "loss": 1.2998,
+      "step": 8786
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.99232571572389e-05,
+      "loss": 1.3258,
+      "step": 8787
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.992320906246317e-05,
+      "loss": 1.2871,
+      "step": 8788
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.992316095263322e-05,
+      "loss": 1.2978,
+      "step": 8789
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.992311282774908e-05,
+      "loss": 1.288,
+      "step": 8790
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.992306468781077e-05,
+      "loss": 1.3088,
+      "step": 8791
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.99230165328183e-05,
+      "loss": 1.2896,
+      "step": 8792
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.992296836277167e-05,
+      "loss": 1.3171,
+      "step": 8793
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.992292017767091e-05,
+      "loss": 1.2908,
+      "step": 8794
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.992287197751601e-05,
+      "loss": 1.3344,
+      "step": 8795
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.992282376230703e-05,
+      "loss": 1.325,
+      "step": 8796
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.992277553204396e-05,
+      "loss": 1.2982,
+      "step": 8797
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.99227272867268e-05,
+      "loss": 1.2973,
+      "step": 8798
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.99226790263556e-05,
+      "loss": 1.298,
+      "step": 8799
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.992263075093032e-05,
+      "loss": 1.3062,
+      "step": 8800
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.992258246045103e-05,
+      "loss": 1.3091,
+      "step": 8801
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.992253415491771e-05,
+      "loss": 1.3264,
+      "step": 8802
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.992248583433039e-05,
+      "loss": 1.2909,
+      "step": 8803
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.992243749868909e-05,
+      "loss": 1.3244,
+      "step": 8804
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.992238914799381e-05,
+      "loss": 1.2745,
+      "step": 8805
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.992234078224456e-05,
+      "loss": 1.2757,
+      "step": 8806
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.992229240144138e-05,
+      "loss": 1.3121,
+      "step": 8807
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.992224400558426e-05,
+      "loss": 1.3229,
+      "step": 8808
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.992219559467323e-05,
+      "loss": 1.3104,
+      "step": 8809
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.992214716870829e-05,
+      "loss": 1.3249,
+      "step": 8810
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.992209872768947e-05,
+      "loss": 1.306,
+      "step": 8811
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.992205027161678e-05,
+      "loss": 1.3392,
+      "step": 8812
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.992200180049023e-05,
+      "loss": 1.293,
+      "step": 8813
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.992195331430982e-05,
+      "loss": 1.3355,
+      "step": 8814
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.99219048130756e-05,
+      "loss": 1.3176,
+      "step": 8815
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.992185629678757e-05,
+      "loss": 1.2889,
+      "step": 8816
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.992180776544573e-05,
+      "loss": 1.3389,
+      "step": 8817
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.992175921905012e-05,
+      "loss": 1.3352,
+      "step": 8818
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.992171065760073e-05,
+      "loss": 1.3205,
+      "step": 8819
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.992166208109757e-05,
+      "loss": 1.3003,
+      "step": 8820
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.992161348954068e-05,
+      "loss": 1.3141,
+      "step": 8821
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.992156488293007e-05,
+      "loss": 1.3268,
+      "step": 8822
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.992151626126575e-05,
+      "loss": 1.3141,
+      "step": 8823
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.992146762454772e-05,
+      "loss": 1.2976,
+      "step": 8824
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.992141897277602e-05,
+      "loss": 1.2935,
+      "step": 8825
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.992137030595065e-05,
+      "loss": 1.3111,
+      "step": 8826
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.992132162407163e-05,
+      "loss": 1.2592,
+      "step": 8827
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.992127292713897e-05,
+      "loss": 1.2704,
+      "step": 8828
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.992122421515269e-05,
+      "loss": 1.3188,
+      "step": 8829
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.992117548811279e-05,
+      "loss": 1.3079,
+      "step": 8830
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.99211267460193e-05,
+      "loss": 1.3094,
+      "step": 8831
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.992107798887223e-05,
+      "loss": 1.269,
+      "step": 8832
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.992102921667161e-05,
+      "loss": 1.3065,
+      "step": 8833
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.992098042941743e-05,
+      "loss": 1.3033,
+      "step": 8834
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.992093162710972e-05,
+      "loss": 1.3315,
+      "step": 8835
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.992088280974847e-05,
+      "loss": 1.3223,
+      "step": 8836
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.992083397733374e-05,
+      "loss": 1.3059,
+      "step": 8837
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.99207851298655e-05,
+      "loss": 1.3,
+      "step": 8838
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.99207362673438e-05,
+      "loss": 1.3198,
+      "step": 8839
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.992068738976862e-05,
+      "loss": 1.2924,
+      "step": 8840
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.992063849714e-05,
+      "loss": 1.3218,
+      "step": 8841
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.992058958945796e-05,
+      "loss": 1.3533,
+      "step": 8842
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.99205406667225e-05,
+      "loss": 1.315,
+      "step": 8843
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.992049172893362e-05,
+      "loss": 1.3094,
+      "step": 8844
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.992044277609136e-05,
+      "loss": 1.3261,
+      "step": 8845
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.992039380819573e-05,
+      "loss": 1.2797,
+      "step": 8846
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.992034482524674e-05,
+      "loss": 1.3234,
+      "step": 8847
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.992029582724442e-05,
+      "loss": 1.3117,
+      "step": 8848
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.992024681418876e-05,
+      "loss": 1.2916,
+      "step": 8849
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.992019778607977e-05,
+      "loss": 1.292,
+      "step": 8850
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.992014874291751e-05,
+      "loss": 1.3215,
+      "step": 8851
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.992009968470195e-05,
+      "loss": 1.3454,
+      "step": 8852
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.992005061143313e-05,
+      "loss": 1.291,
+      "step": 8853
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.992000152311104e-05,
+      "loss": 1.2864,
+      "step": 8854
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.991995241973572e-05,
+      "loss": 1.2826,
+      "step": 8855
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.991990330130717e-05,
+      "loss": 1.3266,
+      "step": 8856
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.991985416782541e-05,
+      "loss": 1.3451,
+      "step": 8857
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.991980501929046e-05,
+      "loss": 1.3059,
+      "step": 8858
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.991975585570233e-05,
+      "loss": 1.3506,
+      "step": 8859
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.991970667706103e-05,
+      "loss": 1.2667,
+      "step": 8860
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.991965748336658e-05,
+      "loss": 1.3203,
+      "step": 8861
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.991960827461899e-05,
+      "loss": 1.2994,
+      "step": 8862
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.99195590508183e-05,
+      "loss": 1.3389,
+      "step": 8863
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.991950981196447e-05,
+      "loss": 1.2932,
+      "step": 8864
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.991946055805757e-05,
+      "loss": 1.3304,
+      "step": 8865
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.991941128909759e-05,
+      "loss": 1.3032,
+      "step": 8866
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.991936200508454e-05,
+      "loss": 1.3018,
+      "step": 8867
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.991931270601846e-05,
+      "loss": 1.3418,
+      "step": 8868
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.991926339189933e-05,
+      "loss": 1.3307,
+      "step": 8869
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.99192140627272e-05,
+      "loss": 1.3109,
+      "step": 8870
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.991916471850204e-05,
+      "loss": 1.3231,
+      "step": 8871
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.991911535922391e-05,
+      "loss": 1.3197,
+      "step": 8872
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.991906598489282e-05,
+      "loss": 1.3469,
+      "step": 8873
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.991901659550876e-05,
+      "loss": 1.3084,
+      "step": 8874
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.991896719107174e-05,
+      "loss": 1.3412,
+      "step": 8875
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.991891777158182e-05,
+      "loss": 1.321,
+      "step": 8876
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.991886833703898e-05,
+      "loss": 1.3309,
+      "step": 8877
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.991881888744323e-05,
+      "loss": 1.3136,
+      "step": 8878
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.99187694227946e-05,
+      "loss": 1.3003,
+      "step": 8879
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.99187199430931e-05,
+      "loss": 1.3184,
+      "step": 8880
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.991867044833876e-05,
+      "loss": 1.3384,
+      "step": 8881
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.991862093853156e-05,
+      "loss": 1.3292,
+      "step": 8882
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.991857141367156e-05,
+      "loss": 1.3405,
+      "step": 8883
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.991852187375874e-05,
+      "loss": 1.3134,
+      "step": 8884
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.991847231879313e-05,
+      "loss": 1.311,
+      "step": 8885
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.991842274877475e-05,
+      "loss": 1.2998,
+      "step": 8886
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.991837316370358e-05,
+      "loss": 1.3165,
+      "step": 8887
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.991832356357967e-05,
+      "loss": 1.3014,
+      "step": 8888
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.991827394840303e-05,
+      "loss": 1.3157,
+      "step": 8889
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.991822431817368e-05,
+      "loss": 1.3293,
+      "step": 8890
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.991817467289162e-05,
+      "loss": 1.2893,
+      "step": 8891
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.991812501255687e-05,
+      "loss": 1.2936,
+      "step": 8892
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.991807533716943e-05,
+      "loss": 1.3012,
+      "step": 8893
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.991802564672936e-05,
+      "loss": 1.3164,
+      "step": 8894
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.991797594123662e-05,
+      "loss": 1.3025,
+      "step": 8895
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.991792622069128e-05,
+      "loss": 1.2963,
+      "step": 8896
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.99178764850933e-05,
+      "loss": 1.3264,
+      "step": 8897
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.991782673444273e-05,
+      "loss": 1.3418,
+      "step": 8898
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.991777696873959e-05,
+      "loss": 1.2879,
+      "step": 8899
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.991772718798386e-05,
+      "loss": 1.2815,
+      "step": 8900
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.991767739217558e-05,
+      "loss": 1.2704,
+      "step": 8901
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.991762758131476e-05,
+      "loss": 1.2882,
+      "step": 8902
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.991757775540142e-05,
+      "loss": 1.3485,
+      "step": 8903
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.991752791443557e-05,
+      "loss": 1.3333,
+      "step": 8904
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.991747805841722e-05,
+      "loss": 1.2959,
+      "step": 8905
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.991742818734639e-05,
+      "loss": 1.3093,
+      "step": 8906
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.991737830122312e-05,
+      "loss": 1.314,
+      "step": 8907
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.991732840004739e-05,
+      "loss": 1.3366,
+      "step": 8908
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.991727848381922e-05,
+      "loss": 1.3175,
+      "step": 8909
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.991722855253863e-05,
+      "loss": 1.3169,
+      "step": 8910
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.991717860620563e-05,
+      "loss": 1.315,
+      "step": 8911
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.991712864482026e-05,
+      "loss": 1.3493,
+      "step": 8912
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.991707866838249e-05,
+      "loss": 1.3177,
+      "step": 8913
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.991702867689239e-05,
+      "loss": 1.2992,
+      "step": 8914
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.991697867034993e-05,
+      "loss": 1.2859,
+      "step": 8915
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.991692864875515e-05,
+      "loss": 1.2956,
+      "step": 8916
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.991687861210805e-05,
+      "loss": 1.3473,
+      "step": 8917
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.991682856040865e-05,
+      "loss": 1.3132,
+      "step": 8918
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.991677849365697e-05,
+      "loss": 1.2707,
+      "step": 8919
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.991672841185302e-05,
+      "loss": 1.3156,
+      "step": 8920
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.991667831499682e-05,
+      "loss": 1.3252,
+      "step": 8921
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.991662820308838e-05,
+      "loss": 1.3313,
+      "step": 8922
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.991657807612773e-05,
+      "loss": 1.3294,
+      "step": 8923
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.991652793411485e-05,
+      "loss": 1.2976,
+      "step": 8924
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.99164777770498e-05,
+      "loss": 1.3166,
+      "step": 8925
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.991642760493257e-05,
+      "loss": 1.3055,
+      "step": 8926
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.991637741776316e-05,
+      "loss": 1.3335,
+      "step": 8927
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.99163272155416e-05,
+      "loss": 1.3157,
+      "step": 8928
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.991627699826793e-05,
+      "loss": 1.3461,
+      "step": 8929
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.991622676594215e-05,
+      "loss": 1.3256,
+      "step": 8930
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.991617651856425e-05,
+      "loss": 1.3292,
+      "step": 8931
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.991612625613426e-05,
+      "loss": 1.3161,
+      "step": 8932
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.99160759786522e-05,
+      "loss": 1.2957,
+      "step": 8933
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.991602568611809e-05,
+      "loss": 1.3171,
+      "step": 8934
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.991597537853194e-05,
+      "loss": 1.32,
+      "step": 8935
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.991592505589375e-05,
+      "loss": 1.3282,
+      "step": 8936
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.991587471820356e-05,
+      "loss": 1.3249,
+      "step": 8937
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.991582436546139e-05,
+      "loss": 1.3476,
+      "step": 8938
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.991577399766722e-05,
+      "loss": 1.3183,
+      "step": 8939
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.99157236148211e-05,
+      "loss": 1.3133,
+      "step": 8940
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.991567321692301e-05,
+      "loss": 1.3445,
+      "step": 8941
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.9915622803973e-05,
+      "loss": 1.309,
+      "step": 8942
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.991557237597106e-05,
+      "loss": 1.3373,
+      "step": 8943
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.991552193291722e-05,
+      "loss": 1.32,
+      "step": 8944
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.99154714748115e-05,
+      "loss": 1.2851,
+      "step": 8945
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.99154210016539e-05,
+      "loss": 1.3632,
+      "step": 8946
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.991537051344444e-05,
+      "loss": 1.3064,
+      "step": 8947
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.991532001018313e-05,
+      "loss": 1.3037,
+      "step": 8948
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.991526949187e-05,
+      "loss": 1.377,
+      "step": 8949
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.991521895850505e-05,
+      "loss": 1.3211,
+      "step": 8950
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.99151684100883e-05,
+      "loss": 1.2895,
+      "step": 8951
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.991511784661977e-05,
+      "loss": 1.3356,
+      "step": 8952
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.991506726809949e-05,
+      "loss": 1.2821,
+      "step": 8953
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.991501667452745e-05,
+      "loss": 1.3137,
+      "step": 8954
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.991496606590367e-05,
+      "loss": 1.3216,
+      "step": 8955
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.991491544222816e-05,
+      "loss": 1.3073,
+      "step": 8956
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.991486480350097e-05,
+      "loss": 1.285,
+      "step": 8957
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.991481414972206e-05,
+      "loss": 1.3009,
+      "step": 8958
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.991476348089148e-05,
+      "loss": 1.3232,
+      "step": 8959
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.991471279700925e-05,
+      "loss": 1.3275,
+      "step": 8960
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.991466209807536e-05,
+      "loss": 1.3289,
+      "step": 8961
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.991461138408985e-05,
+      "loss": 1.2698,
+      "step": 8962
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.991456065505273e-05,
+      "loss": 1.2974,
+      "step": 8963
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.9914509910964e-05,
+      "loss": 1.3033,
+      "step": 8964
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.991445915182369e-05,
+      "loss": 1.2947,
+      "step": 8965
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.991440837763182e-05,
+      "loss": 1.2882,
+      "step": 8966
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.991435758838839e-05,
+      "loss": 1.309,
+      "step": 8967
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.991430678409341e-05,
+      "loss": 1.2966,
+      "step": 8968
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.991425596474692e-05,
+      "loss": 1.2966,
+      "step": 8969
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.991420513034892e-05,
+      "loss": 1.3056,
+      "step": 8970
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.991415428089944e-05,
+      "loss": 1.3246,
+      "step": 8971
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.991410341639846e-05,
+      "loss": 1.283,
+      "step": 8972
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.991405253684604e-05,
+      "loss": 1.3225,
+      "step": 8973
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.991400164224216e-05,
+      "loss": 1.3078,
+      "step": 8974
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.991395073258687e-05,
+      "loss": 1.3248,
+      "step": 8975
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.991389980788015e-05,
+      "loss": 1.3006,
+      "step": 8976
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.991384886812202e-05,
+      "loss": 1.2727,
+      "step": 8977
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.991379791331252e-05,
+      "loss": 1.3232,
+      "step": 8978
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.991374694345163e-05,
+      "loss": 1.2879,
+      "step": 8979
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.99136959585394e-05,
+      "loss": 1.3278,
+      "step": 8980
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.991364495857585e-05,
+      "loss": 1.2966,
+      "step": 8981
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.991359394356096e-05,
+      "loss": 1.2759,
+      "step": 8982
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.991354291349476e-05,
+      "loss": 1.3303,
+      "step": 8983
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.991349186837728e-05,
+      "loss": 1.2932,
+      "step": 8984
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.99134408082085e-05,
+      "loss": 1.3154,
+      "step": 8985
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.991338973298847e-05,
+      "loss": 1.3118,
+      "step": 8986
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.99133386427172e-05,
+      "loss": 1.2938,
+      "step": 8987
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.99132875373947e-05,
+      "loss": 1.2986,
+      "step": 8988
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.991323641702097e-05,
+      "loss": 1.2998,
+      "step": 8989
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.991318528159606e-05,
+      "loss": 1.3173,
+      "step": 8990
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.991313413111995e-05,
+      "loss": 1.3161,
+      "step": 8991
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.991308296559268e-05,
+      "loss": 1.2939,
+      "step": 8992
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.991303178501425e-05,
+      "loss": 1.3062,
+      "step": 8993
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.991298058938468e-05,
+      "loss": 1.3115,
+      "step": 8994
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.9912929378704e-05,
+      "loss": 1.3041,
+      "step": 8995
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.99128781529722e-05,
+      "loss": 1.284,
+      "step": 8996
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.99128269121893e-05,
+      "loss": 1.313,
+      "step": 8997
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.991277565635535e-05,
+      "loss": 1.3371,
+      "step": 8998
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.991272438547032e-05,
+      "loss": 1.3153,
+      "step": 8999
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.991267309953425e-05,
+      "loss": 1.2984,
+      "step": 9000
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.991262179854714e-05,
+      "loss": 1.3196,
+      "step": 9001
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.991257048250904e-05,
+      "loss": 1.3347,
+      "step": 9002
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.991251915141992e-05,
+      "loss": 1.3012,
+      "step": 9003
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.991246780527981e-05,
+      "loss": 1.2972,
+      "step": 9004
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.991241644408876e-05,
+      "loss": 1.3067,
+      "step": 9005
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.991236506784674e-05,
+      "loss": 1.3145,
+      "step": 9006
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.991231367655378e-05,
+      "loss": 1.2922,
+      "step": 9007
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.99122622702099e-05,
+      "loss": 1.3259,
+      "step": 9008
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.991221084881512e-05,
+      "loss": 1.3352,
+      "step": 9009
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.991215941236944e-05,
+      "loss": 1.3391,
+      "step": 9010
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.99121079608729e-05,
+      "loss": 1.325,
+      "step": 9011
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.99120564943255e-05,
+      "loss": 1.3469,
+      "step": 9012
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.991200501272723e-05,
+      "loss": 1.3632,
+      "step": 9013
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.991195351607815e-05,
+      "loss": 1.3103,
+      "step": 9014
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.991190200437825e-05,
+      "loss": 1.3256,
+      "step": 9015
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.991185047762757e-05,
+      "loss": 1.3036,
+      "step": 9016
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.991179893582608e-05,
+      "loss": 1.3159,
+      "step": 9017
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.991174737897384e-05,
+      "loss": 1.3087,
+      "step": 9018
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.991169580707085e-05,
+      "loss": 1.3021,
+      "step": 9019
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.991164422011711e-05,
+      "loss": 1.3268,
+      "step": 9020
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.991159261811268e-05,
+      "loss": 1.2965,
+      "step": 9021
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.991154100105752e-05,
+      "loss": 1.3228,
+      "step": 9022
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.991148936895169e-05,
+      "loss": 1.3093,
+      "step": 9023
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.991143772179517e-05,
+      "loss": 1.2891,
+      "step": 9024
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.991138605958801e-05,
+      "loss": 1.2914,
+      "step": 9025
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.991133438233019e-05,
+      "loss": 1.3134,
+      "step": 9026
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.991128269002174e-05,
+      "loss": 1.3108,
+      "step": 9027
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.991123098266271e-05,
+      "loss": 1.3063,
+      "step": 9028
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.991117926025306e-05,
+      "loss": 1.3146,
+      "step": 9029
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.991112752279283e-05,
+      "loss": 1.2959,
+      "step": 9030
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.991107577028204e-05,
+      "loss": 1.3443,
+      "step": 9031
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.99110240027207e-05,
+      "loss": 1.3034,
+      "step": 9032
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.991097222010883e-05,
+      "loss": 1.2861,
+      "step": 9033
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.991092042244645e-05,
+      "loss": 1.3189,
+      "step": 9034
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.991086860973355e-05,
+      "loss": 1.3301,
+      "step": 9035
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.991081678197018e-05,
+      "loss": 1.3016,
+      "step": 9036
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.991076493915634e-05,
+      "loss": 1.3143,
+      "step": 9037
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.991071308129205e-05,
+      "loss": 1.3226,
+      "step": 9038
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.991066120837731e-05,
+      "loss": 1.3245,
+      "step": 9039
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.991060932041214e-05,
+      "loss": 1.345,
+      "step": 9040
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.991055741739658e-05,
+      "loss": 1.3296,
+      "step": 9041
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.991050549933061e-05,
+      "loss": 1.3072,
+      "step": 9042
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.991045356621427e-05,
+      "loss": 1.2979,
+      "step": 9043
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.991040161804757e-05,
+      "loss": 1.3279,
+      "step": 9044
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.991034965483053e-05,
+      "loss": 1.3178,
+      "step": 9045
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.991029767656315e-05,
+      "loss": 1.3036,
+      "step": 9046
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.991024568324547e-05,
+      "loss": 1.296,
+      "step": 9047
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.991019367487748e-05,
+      "loss": 1.2967,
+      "step": 9048
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.991014165145922e-05,
+      "loss": 1.2911,
+      "step": 9049
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.991008961299069e-05,
+      "loss": 1.3548,
+      "step": 9050
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.99100375594719e-05,
+      "loss": 1.3413,
+      "step": 9051
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.99099854909029e-05,
+      "loss": 1.2862,
+      "step": 9052
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.990993340728364e-05,
+      "loss": 1.2803,
+      "step": 9053
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.99098813086142e-05,
+      "loss": 1.2867,
+      "step": 9054
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.990982919489457e-05,
+      "loss": 1.3361,
+      "step": 9055
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.990977706612477e-05,
+      "loss": 1.3018,
+      "step": 9056
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.990972492230481e-05,
+      "loss": 1.3212,
+      "step": 9057
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.99096727634347e-05,
+      "loss": 1.2837,
+      "step": 9058
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.990962058951448e-05,
+      "loss": 1.3123,
+      "step": 9059
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.990956840054413e-05,
+      "loss": 1.3409,
+      "step": 9060
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.990951619652371e-05,
+      "loss": 1.2825,
+      "step": 9061
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.99094639774532e-05,
+      "loss": 1.3103,
+      "step": 9062
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.990941174333262e-05,
+      "loss": 1.3163,
+      "step": 9063
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.9909359494162e-05,
+      "loss": 1.3191,
+      "step": 9064
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.990930722994137e-05,
+      "loss": 1.3032,
+      "step": 9065
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.990925495067069e-05,
+      "loss": 1.2992,
+      "step": 9066
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.990920265635003e-05,
+      "loss": 1.2919,
+      "step": 9067
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.990915034697938e-05,
+      "loss": 1.2938,
+      "step": 9068
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.990909802255876e-05,
+      "loss": 1.2902,
+      "step": 9069
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.990904568308821e-05,
+      "loss": 1.326,
+      "step": 9070
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.990899332856771e-05,
+      "loss": 1.3113,
+      "step": 9071
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.990894095899728e-05,
+      "loss": 1.2979,
+      "step": 9072
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.990888857437694e-05,
+      "loss": 1.3061,
+      "step": 9073
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.990883617470674e-05,
+      "loss": 1.3128,
+      "step": 9074
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.990878375998665e-05,
+      "loss": 1.2959,
+      "step": 9075
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.99087313302167e-05,
+      "loss": 1.3204,
+      "step": 9076
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.99086788853969e-05,
+      "loss": 1.2953,
+      "step": 9077
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.99086264255273e-05,
+      "loss": 1.2949,
+      "step": 9078
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.990857395060786e-05,
+      "loss": 1.3309,
+      "step": 9079
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.990852146063865e-05,
+      "loss": 1.3148,
+      "step": 9080
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.990846895561965e-05,
+      "loss": 1.3175,
+      "step": 9081
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.99084164355509e-05,
+      "loss": 1.2879,
+      "step": 9082
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.990836390043239e-05,
+      "loss": 1.3245,
+      "step": 9083
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.990831135026415e-05,
+      "loss": 1.2882,
+      "step": 9084
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.99082587850462e-05,
+      "loss": 1.2997,
+      "step": 9085
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.990820620477856e-05,
+      "loss": 1.3165,
+      "step": 9086
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.990815360946122e-05,
+      "loss": 1.3344,
+      "step": 9087
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.990810099909423e-05,
+      "loss": 1.3091,
+      "step": 9088
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.990804837367757e-05,
+      "loss": 1.3185,
+      "step": 9089
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.990799573321129e-05,
+      "loss": 1.319,
+      "step": 9090
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.990794307769538e-05,
+      "loss": 1.3088,
+      "step": 9091
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.990789040712988e-05,
+      "loss": 1.2969,
+      "step": 9092
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.990783772151477e-05,
+      "loss": 1.3084,
+      "step": 9093
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.990778502085011e-05,
+      "loss": 1.2785,
+      "step": 9094
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.990773230513589e-05,
+      "loss": 1.3035,
+      "step": 9095
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.990767957437212e-05,
+      "loss": 1.2887,
+      "step": 9096
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.990762682855883e-05,
+      "loss": 1.3137,
+      "step": 9097
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.990757406769603e-05,
+      "loss": 1.3185,
+      "step": 9098
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.990752129178374e-05,
+      "loss": 1.3628,
+      "step": 9099
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.990746850082198e-05,
+      "loss": 1.3214,
+      "step": 9100
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.990741569481074e-05,
+      "loss": 1.2922,
+      "step": 9101
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.990736287375008e-05,
+      "loss": 1.3165,
+      "step": 9102
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.990731003763997e-05,
+      "loss": 1.3468,
+      "step": 9103
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.990725718648046e-05,
+      "loss": 1.3256,
+      "step": 9104
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.990720432027153e-05,
+      "loss": 1.3083,
+      "step": 9105
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.990715143901326e-05,
+      "loss": 1.2696,
+      "step": 9106
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.99070985427056e-05,
+      "loss": 1.3256,
+      "step": 9107
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.99070456313486e-05,
+      "loss": 1.2976,
+      "step": 9108
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.990699270494226e-05,
+      "loss": 1.2876,
+      "step": 9109
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.99069397634866e-05,
+      "loss": 1.2819,
+      "step": 9110
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.990688680698164e-05,
+      "loss": 1.2982,
+      "step": 9111
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.990683383542741e-05,
+      "loss": 1.3129,
+      "step": 9112
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.990678084882389e-05,
+      "loss": 1.3078,
+      "step": 9113
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.990672784717113e-05,
+      "loss": 1.3018,
+      "step": 9114
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.990667483046912e-05,
+      "loss": 1.2857,
+      "step": 9115
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.99066217987179e-05,
+      "loss": 1.2903,
+      "step": 9116
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.990656875191746e-05,
+      "loss": 1.3001,
+      "step": 9117
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.990651569006785e-05,
+      "loss": 1.3151,
+      "step": 9118
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.990646261316906e-05,
+      "loss": 1.3002,
+      "step": 9119
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.990640952122111e-05,
+      "loss": 1.3002,
+      "step": 9120
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.990635641422402e-05,
+      "loss": 1.3081,
+      "step": 9121
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.99063032921778e-05,
+      "loss": 1.2819,
+      "step": 9122
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.990625015508248e-05,
+      "loss": 1.3093,
+      "step": 9123
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.990619700293807e-05,
+      "loss": 1.3297,
+      "step": 9124
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.990614383574457e-05,
+      "loss": 1.3113,
+      "step": 9125
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.990609065350201e-05,
+      "loss": 1.3277,
+      "step": 9126
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.990603745621042e-05,
+      "loss": 1.3239,
+      "step": 9127
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.990598424386978e-05,
+      "loss": 1.3044,
+      "step": 9128
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.990593101648014e-05,
+      "loss": 1.3129,
+      "step": 9129
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.99058777740415e-05,
+      "loss": 1.3296,
+      "step": 9130
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.990582451655388e-05,
+      "loss": 1.2977,
+      "step": 9131
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.99057712440173e-05,
+      "loss": 1.2918,
+      "step": 9132
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.990571795643176e-05,
+      "loss": 1.3222,
+      "step": 9133
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.990566465379729e-05,
+      "loss": 1.3027,
+      "step": 9134
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.99056113361139e-05,
+      "loss": 1.3653,
+      "step": 9135
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.990555800338162e-05,
+      "loss": 1.3447,
+      "step": 9136
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.990550465560044e-05,
+      "loss": 1.3126,
+      "step": 9137
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.990545129277042e-05,
+      "loss": 1.3357,
+      "step": 9138
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.990539791489153e-05,
+      "loss": 1.3533,
+      "step": 9139
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.99053445219638e-05,
+      "loss": 1.3092,
+      "step": 9140
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.990529111398726e-05,
+      "loss": 1.3111,
+      "step": 9141
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.99052376909619e-05,
+      "loss": 1.3107,
+      "step": 9142
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.990518425288777e-05,
+      "loss": 1.3136,
+      "step": 9143
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.990513079976484e-05,
+      "loss": 1.3198,
+      "step": 9144
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.990507733159318e-05,
+      "loss": 1.3282,
+      "step": 9145
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.990502384837278e-05,
+      "loss": 1.3205,
+      "step": 9146
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.990497035010364e-05,
+      "loss": 1.3493,
+      "step": 9147
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.990491683678581e-05,
+      "loss": 1.312,
+      "step": 9148
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.990486330841928e-05,
+      "loss": 1.3048,
+      "step": 9149
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.990480976500409e-05,
+      "loss": 1.311,
+      "step": 9150
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.990475620654023e-05,
+      "loss": 1.312,
+      "step": 9151
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.990470263302771e-05,
+      "loss": 1.3227,
+      "step": 9152
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.99046490444666e-05,
+      "loss": 1.3215,
+      "step": 9153
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.990459544085684e-05,
+      "loss": 1.3012,
+      "step": 9154
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.99045418221985e-05,
+      "loss": 1.3087,
+      "step": 9155
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.99044881884916e-05,
+      "loss": 1.2892,
+      "step": 9156
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.990443453973613e-05,
+      "loss": 1.3269,
+      "step": 9157
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.99043808759321e-05,
+      "loss": 1.3094,
+      "step": 9158
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.990432719707955e-05,
+      "loss": 1.3544,
+      "step": 9159
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.990427350317848e-05,
+      "loss": 1.3194,
+      "step": 9160
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.990421979422893e-05,
+      "loss": 1.3271,
+      "step": 9161
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.990416607023087e-05,
+      "loss": 1.288,
+      "step": 9162
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.990411233118438e-05,
+      "loss": 1.315,
+      "step": 9163
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.990405857708942e-05,
+      "loss": 1.3277,
+      "step": 9164
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.990400480794604e-05,
+      "loss": 1.3139,
+      "step": 9165
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.990395102375422e-05,
+      "loss": 1.34,
+      "step": 9166
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.990389722451403e-05,
+      "loss": 1.3121,
+      "step": 9167
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.990384341022544e-05,
+      "loss": 1.2641,
+      "step": 9168
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.990378958088849e-05,
+      "loss": 1.3073,
+      "step": 9169
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.990373573650318e-05,
+      "loss": 1.2728,
+      "step": 9170
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.990368187706953e-05,
+      "loss": 1.3036,
+      "step": 9171
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.990362800258756e-05,
+      "loss": 1.3326,
+      "step": 9172
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.99035741130573e-05,
+      "loss": 1.2977,
+      "step": 9173
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.990352020847876e-05,
+      "loss": 1.3309,
+      "step": 9174
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.990346628885194e-05,
+      "loss": 1.3489,
+      "step": 9175
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.990341235417687e-05,
+      "loss": 1.2939,
+      "step": 9176
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.990335840445356e-05,
+      "loss": 1.333,
+      "step": 9177
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.990330443968204e-05,
+      "loss": 1.2852,
+      "step": 9178
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.990325045986229e-05,
+      "loss": 1.3099,
+      "step": 9179
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.990319646499437e-05,
+      "loss": 1.3233,
+      "step": 9180
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.990314245507827e-05,
+      "loss": 1.2989,
+      "step": 9181
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.990308843011402e-05,
+      "loss": 1.3178,
+      "step": 9182
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.990303439010163e-05,
+      "loss": 1.3216,
+      "step": 9183
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.990298033504111e-05,
+      "loss": 1.3366,
+      "step": 9184
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.990292626493248e-05,
+      "loss": 1.3251,
+      "step": 9185
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.990287217977577e-05,
+      "loss": 1.3132,
+      "step": 9186
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.990281807957098e-05,
+      "loss": 1.2799,
+      "step": 9187
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.990276396431812e-05,
+      "loss": 1.3348,
+      "step": 9188
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.990270983401724e-05,
+      "loss": 1.3035,
+      "step": 9189
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.990265568866833e-05,
+      "loss": 1.3091,
+      "step": 9190
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.99026015282714e-05,
+      "loss": 1.3022,
+      "step": 9191
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.990254735282648e-05,
+      "loss": 1.2906,
+      "step": 9192
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.990249316233358e-05,
+      "loss": 1.3175,
+      "step": 9193
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.990243895679273e-05,
+      "loss": 1.3149,
+      "step": 9194
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.990238473620392e-05,
+      "loss": 1.337,
+      "step": 9195
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.990233050056721e-05,
+      "loss": 1.3251,
+      "step": 9196
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.990227624988257e-05,
+      "loss": 1.3029,
+      "step": 9197
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.990222198415004e-05,
+      "loss": 1.3151,
+      "step": 9198
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.990216770336962e-05,
+      "loss": 1.2378,
+      "step": 9199
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.990211340754134e-05,
+      "loss": 1.2922,
+      "step": 9200
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.990205909666523e-05,
+      "loss": 1.3121,
+      "step": 9201
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.990200477074128e-05,
+      "loss": 1.2697,
+      "step": 9202
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.990195042976952e-05,
+      "loss": 1.3092,
+      "step": 9203
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.990189607374995e-05,
+      "loss": 1.3434,
+      "step": 9204
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.990184170268261e-05,
+      "loss": 1.2867,
+      "step": 9205
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.990178731656752e-05,
+      "loss": 1.2879,
+      "step": 9206
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.990173291540468e-05,
+      "loss": 1.2798,
+      "step": 9207
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.990167849919409e-05,
+      "loss": 1.3236,
+      "step": 9208
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.99016240679358e-05,
+      "loss": 1.2928,
+      "step": 9209
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.99015696216298e-05,
+      "loss": 1.2952,
+      "step": 9210
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.990151516027612e-05,
+      "loss": 1.3385,
+      "step": 9211
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.990146068387479e-05,
+      "loss": 1.3194,
+      "step": 9212
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.99014061924258e-05,
+      "loss": 1.3051,
+      "step": 9213
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.990135168592918e-05,
+      "loss": 1.3248,
+      "step": 9214
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.990129716438493e-05,
+      "loss": 1.3103,
+      "step": 9215
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.99012426277931e-05,
+      "loss": 1.3058,
+      "step": 9216
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.990118807615367e-05,
+      "loss": 1.2883,
+      "step": 9217
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.990113350946668e-05,
+      "loss": 1.3081,
+      "step": 9218
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.990107892773215e-05,
+      "loss": 1.3324,
+      "step": 9219
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.990102433095007e-05,
+      "loss": 1.2996,
+      "step": 9220
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.990096971912049e-05,
+      "loss": 1.3096,
+      "step": 9221
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.99009150922434e-05,
+      "loss": 1.3129,
+      "step": 9222
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.990086045031881e-05,
+      "loss": 1.3214,
+      "step": 9223
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.990080579334678e-05,
+      "loss": 1.3377,
+      "step": 9224
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.990075112132729e-05,
+      "loss": 1.3076,
+      "step": 9225
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.990069643426035e-05,
+      "loss": 1.3098,
+      "step": 9226
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.9900641732146e-05,
+      "loss": 1.3133,
+      "step": 9227
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.990058701498424e-05,
+      "loss": 1.3229,
+      "step": 9228
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.990053228277512e-05,
+      "loss": 1.3207,
+      "step": 9229
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.99004775355186e-05,
+      "loss": 1.3016,
+      "step": 9230
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.990042277321475e-05,
+      "loss": 1.3007,
+      "step": 9231
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.990036799586355e-05,
+      "loss": 1.3216,
+      "step": 9232
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.990031320346505e-05,
+      "loss": 1.3267,
+      "step": 9233
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.990025839601922e-05,
+      "loss": 1.3432,
+      "step": 9234
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.990020357352612e-05,
+      "loss": 1.3054,
+      "step": 9235
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.990014873598576e-05,
+      "loss": 1.3475,
+      "step": 9236
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.990009388339813e-05,
+      "loss": 1.3148,
+      "step": 9237
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.990003901576325e-05,
+      "loss": 1.3165,
+      "step": 9238
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.989998413308118e-05,
+      "loss": 1.302,
+      "step": 9239
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.98999292353519e-05,
+      "loss": 1.3205,
+      "step": 9240
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.989987432257541e-05,
+      "loss": 1.299,
+      "step": 9241
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.989981939475178e-05,
+      "loss": 1.3024,
+      "step": 9242
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.989976445188099e-05,
+      "loss": 1.3779,
+      "step": 9243
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.989970949396305e-05,
+      "loss": 1.3282,
+      "step": 9244
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.989965452099799e-05,
+      "loss": 1.3117,
+      "step": 9245
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.989959953298582e-05,
+      "loss": 1.2843,
+      "step": 9246
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.989954452992657e-05,
+      "loss": 1.2929,
+      "step": 9247
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.989948951182027e-05,
+      "loss": 1.3358,
+      "step": 9248
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.989943447866688e-05,
+      "loss": 1.3004,
+      "step": 9249
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.989937943046648e-05,
+      "loss": 1.3437,
+      "step": 9250
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.989932436721904e-05,
+      "loss": 1.3159,
+      "step": 9251
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.98992692889246e-05,
+      "loss": 1.3076,
+      "step": 9252
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.989921419558317e-05,
+      "loss": 1.3132,
+      "step": 9253
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.989915908719478e-05,
+      "loss": 1.2781,
+      "step": 9254
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.989910396375942e-05,
+      "loss": 1.3448,
+      "step": 9255
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.989904882527713e-05,
+      "loss": 1.2994,
+      "step": 9256
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.989899367174791e-05,
+      "loss": 1.2996,
+      "step": 9257
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.989893850317181e-05,
+      "loss": 1.3165,
+      "step": 9258
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.98988833195488e-05,
+      "loss": 1.3259,
+      "step": 9259
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.989882812087892e-05,
+      "loss": 1.3286,
+      "step": 9260
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.989877290716218e-05,
+      "loss": 1.3257,
+      "step": 9261
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.98987176783986e-05,
+      "loss": 1.3195,
+      "step": 9262
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.98986624345882e-05,
+      "loss": 1.3017,
+      "step": 9263
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.989860717573101e-05,
+      "loss": 1.2852,
+      "step": 9264
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.989855190182703e-05,
+      "loss": 1.3266,
+      "step": 9265
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.989849661287627e-05,
+      "loss": 1.2775,
+      "step": 9266
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.989844130887875e-05,
+      "loss": 1.2866,
+      "step": 9267
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.98983859898345e-05,
+      "loss": 1.3313,
+      "step": 9268
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.989833065574352e-05,
+      "loss": 1.3502,
+      "step": 9269
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.989827530660585e-05,
+      "loss": 1.3327,
+      "step": 9270
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.989821994242148e-05,
+      "loss": 1.2886,
+      "step": 9271
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.989816456319044e-05,
+      "loss": 1.2787,
+      "step": 9272
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.989810916891273e-05,
+      "loss": 1.3338,
+      "step": 9273
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.98980537595884e-05,
+      "loss": 1.3015,
+      "step": 9274
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.989799833521744e-05,
+      "loss": 1.3151,
+      "step": 9275
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.989794289579989e-05,
+      "loss": 1.3067,
+      "step": 9276
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.989788744133575e-05,
+      "loss": 1.3159,
+      "step": 9277
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.989783197182503e-05,
+      "loss": 1.3149,
+      "step": 9278
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.989777648726775e-05,
+      "loss": 1.3485,
+      "step": 9279
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.989772098766394e-05,
+      "loss": 1.3422,
+      "step": 9280
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.98976654730136e-05,
+      "loss": 1.3044,
+      "step": 9281
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.989760994331676e-05,
+      "loss": 1.3089,
+      "step": 9282
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.989755439857345e-05,
+      "loss": 1.299,
+      "step": 9283
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.989749883878364e-05,
+      "loss": 1.3246,
+      "step": 9284
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.98974432639474e-05,
+      "loss": 1.2952,
+      "step": 9285
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.989738767406471e-05,
+      "loss": 1.2756,
+      "step": 9286
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.989733206913559e-05,
+      "loss": 1.3491,
+      "step": 9287
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.989727644916007e-05,
+      "loss": 1.3099,
+      "step": 9288
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.989722081413817e-05,
+      "loss": 1.3107,
+      "step": 9289
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.98971651640699e-05,
+      "loss": 1.2894,
+      "step": 9290
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.989710949895528e-05,
+      "loss": 1.3023,
+      "step": 9291
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.989705381879432e-05,
+      "loss": 1.3357,
+      "step": 9292
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.989699812358704e-05,
+      "loss": 1.3429,
+      "step": 9293
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.989694241333345e-05,
+      "loss": 1.3181,
+      "step": 9294
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.989688668803358e-05,
+      "loss": 1.305,
+      "step": 9295
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.989683094768744e-05,
+      "loss": 1.3192,
+      "step": 9296
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.989677519229503e-05,
+      "loss": 1.3075,
+      "step": 9297
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.98967194218564e-05,
+      "loss": 1.3245,
+      "step": 9298
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.989666363637156e-05,
+      "loss": 1.3078,
+      "step": 9299
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.989660783584052e-05,
+      "loss": 1.3268,
+      "step": 9300
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.989655202026326e-05,
+      "loss": 1.2914,
+      "step": 9301
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.989649618963987e-05,
+      "loss": 1.2973,
+      "step": 9302
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.989644034397029e-05,
+      "loss": 1.346,
+      "step": 9303
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.989638448325461e-05,
+      "loss": 1.2946,
+      "step": 9304
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.98963286074928e-05,
+      "loss": 1.3103,
+      "step": 9305
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.98962727166849e-05,
+      "loss": 1.326,
+      "step": 9306
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.989621681083089e-05,
+      "loss": 1.3053,
+      "step": 9307
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.989616088993082e-05,
+      "loss": 1.2674,
+      "step": 9308
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.98961049539847e-05,
+      "loss": 1.2722,
+      "step": 9309
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.989604900299255e-05,
+      "loss": 1.3306,
+      "step": 9310
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.989599303695439e-05,
+      "loss": 1.3227,
+      "step": 9311
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.989593705587021e-05,
+      "loss": 1.3241,
+      "step": 9312
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.989588105974008e-05,
+      "loss": 1.3217,
+      "step": 9313
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.989582504856395e-05,
+      "loss": 1.323,
+      "step": 9314
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.989576902234188e-05,
+      "loss": 1.3362,
+      "step": 9315
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.989571298107388e-05,
+      "loss": 1.319,
+      "step": 9316
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.989565692475997e-05,
+      "loss": 1.3324,
+      "step": 9317
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.989560085340015e-05,
+      "loss": 1.3228,
+      "step": 9318
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.989554476699446e-05,
+      "loss": 1.3612,
+      "step": 9319
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.98954886655429e-05,
+      "loss": 1.314,
+      "step": 9320
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.989543254904548e-05,
+      "loss": 1.307,
+      "step": 9321
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.989537641750223e-05,
+      "loss": 1.3035,
+      "step": 9322
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.989532027091319e-05,
+      "loss": 1.2931,
+      "step": 9323
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.989526410927834e-05,
+      "loss": 1.3065,
+      "step": 9324
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.98952079325977e-05,
+      "loss": 1.3057,
+      "step": 9325
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.989515174087131e-05,
+      "loss": 1.3186,
+      "step": 9326
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.989509553409916e-05,
+      "loss": 1.3299,
+      "step": 9327
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.989503931228129e-05,
+      "loss": 1.2989,
+      "step": 9328
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.989498307541772e-05,
+      "loss": 1.3161,
+      "step": 9329
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.989492682350843e-05,
+      "loss": 1.2875,
+      "step": 9330
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.989487055655347e-05,
+      "loss": 1.2898,
+      "step": 9331
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.989481427455284e-05,
+      "loss": 1.2883,
+      "step": 9332
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.989475797750658e-05,
+      "loss": 1.287,
+      "step": 9333
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.989470166541468e-05,
+      "loss": 1.3158,
+      "step": 9334
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.989464533827718e-05,
+      "loss": 1.3062,
+      "step": 9335
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.989458899609409e-05,
+      "loss": 1.342,
+      "step": 9336
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.989453263886542e-05,
+      "loss": 1.2905,
+      "step": 9337
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.989447626659118e-05,
+      "loss": 1.2863,
+      "step": 9338
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.98944198792714e-05,
+      "loss": 1.289,
+      "step": 9339
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.989436347690609e-05,
+      "loss": 1.2933,
+      "step": 9340
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.989430705949528e-05,
+      "loss": 1.2951,
+      "step": 9341
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.989425062703898e-05,
+      "loss": 1.3248,
+      "step": 9342
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.98941941795372e-05,
+      "loss": 1.3134,
+      "step": 9343
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.989413771698995e-05,
+      "loss": 1.299,
+      "step": 9344
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.989408123939728e-05,
+      "loss": 1.278,
+      "step": 9345
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.989402474675917e-05,
+      "loss": 1.3144,
+      "step": 9346
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.989396823907567e-05,
+      "loss": 1.3524,
+      "step": 9347
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.989391171634676e-05,
+      "loss": 1.3216,
+      "step": 9348
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.98938551785725e-05,
+      "loss": 1.3253,
+      "step": 9349
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.989379862575288e-05,
+      "loss": 1.2955,
+      "step": 9350
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.98937420578879e-05,
+      "loss": 1.318,
+      "step": 9351
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.989368547497763e-05,
+      "loss": 1.2814,
+      "step": 9352
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.989362887702203e-05,
+      "loss": 1.3186,
+      "step": 9353
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.989357226402116e-05,
+      "loss": 1.329,
+      "step": 9354
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.989351563597501e-05,
+      "loss": 1.3322,
+      "step": 9355
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.98934589928836e-05,
+      "loss": 1.2991,
+      "step": 9356
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.989340233474698e-05,
+      "loss": 1.3358,
+      "step": 9357
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.989334566156511e-05,
+      "loss": 1.2876,
+      "step": 9358
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.989328897333806e-05,
+      "loss": 1.278,
+      "step": 9359
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.989323227006581e-05,
+      "loss": 1.3383,
+      "step": 9360
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.989317555174839e-05,
+      "loss": 1.3024,
+      "step": 9361
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.989311881838585e-05,
+      "loss": 1.3409,
+      "step": 9362
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.989306206997815e-05,
+      "loss": 1.3129,
+      "step": 9363
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.989300530652533e-05,
+      "loss": 1.2972,
+      "step": 9364
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.989294852802742e-05,
+      "loss": 1.2679,
+      "step": 9365
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.989289173448444e-05,
+      "loss": 1.3054,
+      "step": 9366
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.989283492589639e-05,
+      "loss": 1.3284,
+      "step": 9367
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.989277810226327e-05,
+      "loss": 1.3222,
+      "step": 9368
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.989272126358514e-05,
+      "loss": 1.2862,
+      "step": 9369
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.989266440986198e-05,
+      "loss": 1.3057,
+      "step": 9370
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.989260754109385e-05,
+      "loss": 1.3019,
+      "step": 9371
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.989255065728072e-05,
+      "loss": 1.3113,
+      "step": 9372
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.989249375842263e-05,
+      "loss": 1.3179,
+      "step": 9373
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.989243684451958e-05,
+      "loss": 1.3384,
+      "step": 9374
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.989237991557163e-05,
+      "loss": 1.3155,
+      "step": 9375
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.989232297157876e-05,
+      "loss": 1.2998,
+      "step": 9376
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.989226601254099e-05,
+      "loss": 1.3334,
+      "step": 9377
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.989220903845835e-05,
+      "loss": 1.307,
+      "step": 9378
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.989215204933085e-05,
+      "loss": 1.2978,
+      "step": 9379
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.989209504515851e-05,
+      "loss": 1.3431,
+      "step": 9380
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.989203802594134e-05,
+      "loss": 1.347,
+      "step": 9381
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.989198099167936e-05,
+      "loss": 1.302,
+      "step": 9382
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.98919239423726e-05,
+      "loss": 1.3012,
+      "step": 9383
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.989186687802105e-05,
+      "loss": 1.3172,
+      "step": 9384
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.989180979862477e-05,
+      "loss": 1.3126,
+      "step": 9385
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.989175270418374e-05,
+      "loss": 1.332,
+      "step": 9386
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.989169559469798e-05,
+      "loss": 1.3052,
+      "step": 9387
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.989163847016752e-05,
+      "loss": 1.2956,
+      "step": 9388
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.989158133059237e-05,
+      "loss": 1.3213,
+      "step": 9389
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.989152417597256e-05,
+      "loss": 1.3152,
+      "step": 9390
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.989146700630809e-05,
+      "loss": 1.2756,
+      "step": 9391
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.989140982159899e-05,
+      "loss": 1.2836,
+      "step": 9392
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.989135262184525e-05,
+      "loss": 1.3052,
+      "step": 9393
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.989129540704694e-05,
+      "loss": 1.306,
+      "step": 9394
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.989123817720403e-05,
+      "loss": 1.3221,
+      "step": 9395
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.989118093231656e-05,
+      "loss": 1.3058,
+      "step": 9396
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.989112367238453e-05,
+      "loss": 1.2973,
+      "step": 9397
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.989106639740798e-05,
+      "loss": 1.2878,
+      "step": 9398
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.989100910738692e-05,
+      "loss": 1.302,
+      "step": 9399
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.989095180232135e-05,
+      "loss": 1.2638,
+      "step": 9400
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.989089448221129e-05,
+      "loss": 1.2904,
+      "step": 9401
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.989083714705679e-05,
+      "loss": 1.3,
+      "step": 9402
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.989077979685783e-05,
+      "loss": 1.2662,
+      "step": 9403
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.989072243161446e-05,
+      "loss": 1.3169,
+      "step": 9404
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.989066505132667e-05,
+      "loss": 1.3224,
+      "step": 9405
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.989060765599447e-05,
+      "loss": 1.3057,
+      "step": 9406
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.98905502456179e-05,
+      "loss": 1.3033,
+      "step": 9407
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.989049282019699e-05,
+      "loss": 1.2995,
+      "step": 9408
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.989043537973173e-05,
+      "loss": 1.2895,
+      "step": 9409
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.989037792422213e-05,
+      "loss": 1.3036,
+      "step": 9410
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.989032045366823e-05,
+      "loss": 1.3485,
+      "step": 9411
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.989026296807004e-05,
+      "loss": 1.3276,
+      "step": 9412
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.98902054674276e-05,
+      "loss": 1.2595,
+      "step": 9413
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.989014795174089e-05,
+      "loss": 1.3492,
+      "step": 9414
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.989009042100995e-05,
+      "loss": 1.3127,
+      "step": 9415
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.989003287523477e-05,
+      "loss": 1.2998,
+      "step": 9416
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.988997531441539e-05,
+      "loss": 1.3067,
+      "step": 9417
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.988991773855183e-05,
+      "loss": 1.2912,
+      "step": 9418
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.98898601476441e-05,
+      "loss": 1.3077,
+      "step": 9419
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.988980254169222e-05,
+      "loss": 1.3379,
+      "step": 9420
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.988974492069622e-05,
+      "loss": 1.3237,
+      "step": 9421
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.988968728465609e-05,
+      "loss": 1.2927,
+      "step": 9422
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.988962963357186e-05,
+      "loss": 1.3038,
+      "step": 9423
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.988957196744355e-05,
+      "loss": 1.3041,
+      "step": 9424
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.988951428627118e-05,
+      "loss": 1.3138,
+      "step": 9425
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.988945659005476e-05,
+      "loss": 1.3231,
+      "step": 9426
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.988939887879431e-05,
+      "loss": 1.337,
+      "step": 9427
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.988934115248986e-05,
+      "loss": 1.3267,
+      "step": 9428
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.988928341114142e-05,
+      "loss": 1.3098,
+      "step": 9429
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.988922565474896e-05,
+      "loss": 1.3107,
+      "step": 9430
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.988916788331258e-05,
+      "loss": 1.3183,
+      "step": 9431
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.988911009683225e-05,
+      "loss": 1.3082,
+      "step": 9432
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.9889052295308e-05,
+      "loss": 1.2971,
+      "step": 9433
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.988899447873983e-05,
+      "loss": 1.3014,
+      "step": 9434
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.988893664712777e-05,
+      "loss": 1.2939,
+      "step": 9435
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.988887880047184e-05,
+      "loss": 1.3263,
+      "step": 9436
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.988882093877207e-05,
+      "loss": 1.3109,
+      "step": 9437
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.988876306202844e-05,
+      "loss": 1.3265,
+      "step": 9438
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.988870517024101e-05,
+      "loss": 1.3137,
+      "step": 9439
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.988864726340976e-05,
+      "loss": 1.32,
+      "step": 9440
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.988858934153474e-05,
+      "loss": 1.3339,
+      "step": 9441
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.988853140461595e-05,
+      "loss": 1.3247,
+      "step": 9442
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.98884734526534e-05,
+      "loss": 1.2946,
+      "step": 9443
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.988841548564712e-05,
+      "loss": 1.3186,
+      "step": 9444
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.988835750359712e-05,
+      "loss": 1.292,
+      "step": 9445
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.988829950650345e-05,
+      "loss": 1.2974,
+      "step": 9446
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.988824149436607e-05,
+      "loss": 1.2888,
+      "step": 9447
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.988818346718504e-05,
+      "loss": 1.2954,
+      "step": 9448
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.988812542496035e-05,
+      "loss": 1.3032,
+      "step": 9449
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.988806736769205e-05,
+      "loss": 1.3212,
+      "step": 9450
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.988800929538014e-05,
+      "loss": 1.3723,
+      "step": 9451
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.988795120802463e-05,
+      "loss": 1.3162,
+      "step": 9452
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.988789310562553e-05,
+      "loss": 1.3396,
+      "step": 9453
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.98878349881829e-05,
+      "loss": 1.3119,
+      "step": 9454
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.988777685569673e-05,
+      "loss": 1.34,
+      "step": 9455
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.988771870816702e-05,
+      "loss": 1.3437,
+      "step": 9456
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.98876605455938e-05,
+      "loss": 1.3662,
+      "step": 9457
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.98876023679771e-05,
+      "loss": 1.3228,
+      "step": 9458
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.988754417531694e-05,
+      "loss": 1.3341,
+      "step": 9459
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.988748596761332e-05,
+      "loss": 1.3005,
+      "step": 9460
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.988742774486626e-05,
+      "loss": 1.3346,
+      "step": 9461
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.988736950707579e-05,
+      "loss": 1.3088,
+      "step": 9462
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.98873112542419e-05,
+      "loss": 1.3318,
+      "step": 9463
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.988725298636466e-05,
+      "loss": 1.3423,
+      "step": 9464
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.988719470344404e-05,
+      "loss": 1.3326,
+      "step": 9465
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.988713640548009e-05,
+      "loss": 1.3382,
+      "step": 9466
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.988707809247277e-05,
+      "loss": 1.3251,
+      "step": 9467
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.988701976442217e-05,
+      "loss": 1.3093,
+      "step": 9468
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.988696142132828e-05,
+      "loss": 1.3197,
+      "step": 9469
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.988690306319109e-05,
+      "loss": 1.3206,
+      "step": 9470
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.988684469001066e-05,
+      "loss": 1.2901,
+      "step": 9471
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.988678630178699e-05,
+      "loss": 1.3007,
+      "step": 9472
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.988672789852008e-05,
+      "loss": 1.3108,
+      "step": 9473
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.988666948020996e-05,
+      "loss": 1.3092,
+      "step": 9474
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.988661104685665e-05,
+      "loss": 1.337,
+      "step": 9475
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.98865525984602e-05,
+      "loss": 1.3009,
+      "step": 9476
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.988649413502057e-05,
+      "loss": 1.3034,
+      "step": 9477
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.988643565653781e-05,
+      "loss": 1.3121,
+      "step": 9478
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.988637716301193e-05,
+      "loss": 1.3067,
+      "step": 9479
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.988631865444296e-05,
+      "loss": 1.2766,
+      "step": 9480
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.988626013083088e-05,
+      "loss": 1.3167,
+      "step": 9481
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.988620159217577e-05,
+      "loss": 1.3252,
+      "step": 9482
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.988614303847759e-05,
+      "loss": 1.3075,
+      "step": 9483
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.98860844697364e-05,
+      "loss": 1.3335,
+      "step": 9484
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.988602588595217e-05,
+      "loss": 1.3074,
+      "step": 9485
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.988596728712496e-05,
+      "loss": 1.2889,
+      "step": 9486
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.988590867325477e-05,
+      "loss": 1.2921,
+      "step": 9487
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.988585004434163e-05,
+      "loss": 1.2909,
+      "step": 9488
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.988579140038554e-05,
+      "loss": 1.2907,
+      "step": 9489
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.988573274138652e-05,
+      "loss": 1.3142,
+      "step": 9490
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.988567406734461e-05,
+      "loss": 1.2962,
+      "step": 9491
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.98856153782598e-05,
+      "loss": 1.3023,
+      "step": 9492
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.988555667413213e-05,
+      "loss": 1.3053,
+      "step": 9493
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.988549795496159e-05,
+      "loss": 1.2895,
+      "step": 9494
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.988543922074823e-05,
+      "loss": 1.2723,
+      "step": 9495
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.988538047149204e-05,
+      "loss": 1.3048,
+      "step": 9496
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.988532170719304e-05,
+      "loss": 1.2821,
+      "step": 9497
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.988526292785128e-05,
+      "loss": 1.2986,
+      "step": 9498
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.988520413346676e-05,
+      "loss": 1.3294,
+      "step": 9499
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.988514532403947e-05,
+      "loss": 1.3171,
+      "step": 9500
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.988508649956947e-05,
+      "loss": 1.3011,
+      "step": 9501
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.988502766005675e-05,
+      "loss": 1.2957,
+      "step": 9502
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.988496880550134e-05,
+      "loss": 1.3179,
+      "step": 9503
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.988490993590325e-05,
+      "loss": 1.3048,
+      "step": 9504
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.988485105126251e-05,
+      "loss": 1.2657,
+      "step": 9505
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.988479215157911e-05,
+      "loss": 1.2609,
+      "step": 9506
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.98847332368531e-05,
+      "loss": 1.2885,
+      "step": 9507
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.988467430708449e-05,
+      "loss": 1.298,
+      "step": 9508
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.98846153622733e-05,
+      "loss": 1.3134,
+      "step": 9509
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.988455640241952e-05,
+      "loss": 1.3299,
+      "step": 9510
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.98844974275232e-05,
+      "loss": 1.3041,
+      "step": 9511
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.988443843758434e-05,
+      "loss": 1.3167,
+      "step": 9512
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.988437943260295e-05,
+      "loss": 1.3124,
+      "step": 9513
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.988432041257909e-05,
+      "loss": 1.3369,
+      "step": 9514
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.988426137751274e-05,
+      "loss": 1.2954,
+      "step": 9515
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.988420232740393e-05,
+      "loss": 1.3354,
+      "step": 9516
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.988414326225267e-05,
+      "loss": 1.3276,
+      "step": 9517
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.9884084182059e-05,
+      "loss": 1.3421,
+      "step": 9518
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.98840250868229e-05,
+      "loss": 1.3279,
+      "step": 9519
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.988396597654442e-05,
+      "loss": 1.2907,
+      "step": 9520
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.988390685122356e-05,
+      "loss": 1.3375,
+      "step": 9521
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.988384771086034e-05,
+      "loss": 1.309,
+      "step": 9522
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.988378855545479e-05,
+      "loss": 1.3209,
+      "step": 9523
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.988372938500692e-05,
+      "loss": 1.308,
+      "step": 9524
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.988367019951674e-05,
+      "loss": 1.3109,
+      "step": 9525
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.988361099898428e-05,
+      "loss": 1.2995,
+      "step": 9526
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.988355178340955e-05,
+      "loss": 1.2728,
+      "step": 9527
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.988349255279258e-05,
+      "loss": 1.3385,
+      "step": 9528
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.988343330713337e-05,
+      "loss": 1.3094,
+      "step": 9529
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.988337404643195e-05,
+      "loss": 1.3131,
+      "step": 9530
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.988331477068835e-05,
+      "loss": 1.2995,
+      "step": 9531
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.988325547990257e-05,
+      "loss": 1.3372,
+      "step": 9532
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.98831961740746e-05,
+      "loss": 1.326,
+      "step": 9533
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.988313685320452e-05,
+      "loss": 1.3255,
+      "step": 9534
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.98830775172923e-05,
+      "loss": 1.338,
+      "step": 9535
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.988301816633799e-05,
+      "loss": 1.2971,
+      "step": 9536
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.988295880034157e-05,
+      "loss": 1.3005,
+      "step": 9537
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.988289941930309e-05,
+      "loss": 1.3404,
+      "step": 9538
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.988284002322256e-05,
+      "loss": 1.3394,
+      "step": 9539
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.988278061210001e-05,
+      "loss": 1.2668,
+      "step": 9540
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.988272118593542e-05,
+      "loss": 1.3486,
+      "step": 9541
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.988266174472884e-05,
+      "loss": 1.3248,
+      "step": 9542
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.98826022884803e-05,
+      "loss": 1.3182,
+      "step": 9543
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.988254281718977e-05,
+      "loss": 1.3286,
+      "step": 9544
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.98824833308573e-05,
+      "loss": 1.3255,
+      "step": 9545
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.988242382948292e-05,
+      "loss": 1.2825,
+      "step": 9546
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.988236431306662e-05,
+      "loss": 1.3427,
+      "step": 9547
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.988230478160843e-05,
+      "loss": 1.3188,
+      "step": 9548
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.988224523510836e-05,
+      "loss": 1.2991,
+      "step": 9549
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.988218567356645e-05,
+      "loss": 1.3257,
+      "step": 9550
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.98821260969827e-05,
+      "loss": 1.3086,
+      "step": 9551
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.988206650535713e-05,
+      "loss": 1.3396,
+      "step": 9552
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.988200689868977e-05,
+      "loss": 1.3113,
+      "step": 9553
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.98819472769806e-05,
+      "loss": 1.2816,
+      "step": 9554
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.988188764022967e-05,
+      "loss": 1.3105,
+      "step": 9555
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.988182798843701e-05,
+      "loss": 1.3148,
+      "step": 9556
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.988176832160262e-05,
+      "loss": 1.3131,
+      "step": 9557
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.988170863972652e-05,
+      "loss": 1.3214,
+      "step": 9558
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.988164894280871e-05,
+      "loss": 1.3116,
+      "step": 9559
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.988158923084925e-05,
+      "loss": 1.3221,
+      "step": 9560
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.98815295038481e-05,
+      "loss": 1.3181,
+      "step": 9561
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.988146976180534e-05,
+      "loss": 1.2548,
+      "step": 9562
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.988141000472096e-05,
+      "loss": 1.3193,
+      "step": 9563
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.988135023259495e-05,
+      "loss": 1.3078,
+      "step": 9564
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.988129044542738e-05,
+      "loss": 1.3365,
+      "step": 9565
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.988123064321824e-05,
+      "loss": 1.289,
+      "step": 9566
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.988117082596754e-05,
+      "loss": 1.3052,
+      "step": 9567
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.988111099367533e-05,
+      "loss": 1.3052,
+      "step": 9568
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.988105114634159e-05,
+      "loss": 1.3165,
+      "step": 9569
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.988099128396636e-05,
+      "loss": 1.2976,
+      "step": 9570
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.988093140654964e-05,
+      "loss": 1.3405,
+      "step": 9571
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.988087151409148e-05,
+      "loss": 1.3126,
+      "step": 9572
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.988081160659187e-05,
+      "loss": 1.3092,
+      "step": 9573
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.988075168405085e-05,
+      "loss": 1.3166,
+      "step": 9574
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.988069174646841e-05,
+      "loss": 1.2757,
+      "step": 9575
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.988063179384459e-05,
+      "loss": 1.3089,
+      "step": 9576
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.98805718261794e-05,
+      "loss": 1.3713,
+      "step": 9577
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.988051184347286e-05,
+      "loss": 1.2706,
+      "step": 9578
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.988045184572499e-05,
+      "loss": 1.3169,
+      "step": 9579
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.988039183293582e-05,
+      "loss": 1.3242,
+      "step": 9580
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.988033180510534e-05,
+      "loss": 1.3121,
+      "step": 9581
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.988027176223358e-05,
+      "loss": 1.317,
+      "step": 9582
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.988021170432056e-05,
+      "loss": 1.3099,
+      "step": 9583
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.988015163136632e-05,
+      "loss": 1.2804,
+      "step": 9584
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.988009154337084e-05,
+      "loss": 1.3029,
+      "step": 9585
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.988003144033415e-05,
+      "loss": 1.3252,
+      "step": 9586
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.987997132225628e-05,
+      "loss": 1.3338,
+      "step": 9587
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.987991118913723e-05,
+      "loss": 1.2937,
+      "step": 9588
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.987985104097704e-05,
+      "loss": 1.2933,
+      "step": 9589
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.987979087777571e-05,
+      "loss": 1.2867,
+      "step": 9590
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.987973069953328e-05,
+      "loss": 1.3078,
+      "step": 9591
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.987967050624974e-05,
+      "loss": 1.3036,
+      "step": 9592
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.987961029792514e-05,
+      "loss": 1.3561,
+      "step": 9593
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.987955007455945e-05,
+      "loss": 1.3272,
+      "step": 9594
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.987948983615274e-05,
+      "loss": 1.3357,
+      "step": 9595
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.9879429582705e-05,
+      "loss": 1.3203,
+      "step": 9596
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.987936931421625e-05,
+      "loss": 1.2916,
+      "step": 9597
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.987930903068652e-05,
+      "loss": 1.3066,
+      "step": 9598
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.987924873211582e-05,
+      "loss": 1.3141,
+      "step": 9599
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.987918841850417e-05,
+      "loss": 1.3094,
+      "step": 9600
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.987912808985158e-05,
+      "loss": 1.34,
+      "step": 9601
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.987906774615809e-05,
+      "loss": 1.3065,
+      "step": 9602
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.987900738742368e-05,
+      "loss": 1.3165,
+      "step": 9603
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.987894701364842e-05,
+      "loss": 1.3483,
+      "step": 9604
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.987888662483228e-05,
+      "loss": 1.2886,
+      "step": 9605
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.987882622097531e-05,
+      "loss": 1.2936,
+      "step": 9606
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.98787658020775e-05,
+      "loss": 1.3221,
+      "step": 9607
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.98787053681389e-05,
+      "loss": 1.3025,
+      "step": 9608
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.987864491915951e-05,
+      "loss": 1.3323,
+      "step": 9609
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.987858445513934e-05,
+      "loss": 1.3277,
+      "step": 9610
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.987852397607842e-05,
+      "loss": 1.3132,
+      "step": 9611
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.987846348197678e-05,
+      "loss": 1.3045,
+      "step": 9612
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.987840297283442e-05,
+      "loss": 1.3465,
+      "step": 9613
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.987834244865138e-05,
+      "loss": 1.3188,
+      "step": 9614
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.987828190942763e-05,
+      "loss": 1.3443,
+      "step": 9615
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.987822135516325e-05,
+      "loss": 1.3185,
+      "step": 9616
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.987816078585821e-05,
+      "loss": 1.35,
+      "step": 9617
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.987810020151256e-05,
+      "loss": 1.3169,
+      "step": 9618
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.987803960212629e-05,
+      "loss": 1.3269,
+      "step": 9619
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.987797898769944e-05,
+      "loss": 1.3073,
+      "step": 9620
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.987791835823204e-05,
+      "loss": 1.3369,
+      "step": 9621
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.987785771372407e-05,
+      "loss": 1.3078,
+      "step": 9622
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.987779705417557e-05,
+      "loss": 1.3064,
+      "step": 9623
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.987773637958655e-05,
+      "loss": 1.303,
+      "step": 9624
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.987767568995705e-05,
+      "loss": 1.3519,
+      "step": 9625
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.987761498528707e-05,
+      "loss": 1.3106,
+      "step": 9626
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.987755426557664e-05,
+      "loss": 1.3175,
+      "step": 9627
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.987749353082575e-05,
+      "loss": 1.3069,
+      "step": 9628
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.987743278103446e-05,
+      "loss": 1.3169,
+      "step": 9629
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.987737201620275e-05,
+      "loss": 1.3095,
+      "step": 9630
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.987731123633065e-05,
+      "loss": 1.3584,
+      "step": 9631
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.987725044141819e-05,
+      "loss": 1.2688,
+      "step": 9632
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.987718963146539e-05,
+      "loss": 1.3416,
+      "step": 9633
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.987712880647224e-05,
+      "loss": 1.3096,
+      "step": 9634
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.987706796643881e-05,
+      "loss": 1.3237,
+      "step": 9635
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.987700711136506e-05,
+      "loss": 1.3376,
+      "step": 9636
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.987694624125105e-05,
+      "loss": 1.2964,
+      "step": 9637
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.987688535609678e-05,
+      "loss": 1.3122,
+      "step": 9638
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.987682445590225e-05,
+      "loss": 1.3409,
+      "step": 9639
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.987676354066753e-05,
+      "loss": 1.3026,
+      "step": 9640
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.987670261039259e-05,
+      "loss": 1.2924,
+      "step": 9641
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.987664166507748e-05,
+      "loss": 1.2905,
+      "step": 9642
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.98765807047222e-05,
+      "loss": 1.3034,
+      "step": 9643
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.987651972932677e-05,
+      "loss": 1.3104,
+      "step": 9644
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.987645873889122e-05,
+      "loss": 1.3087,
+      "step": 9645
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.987639773341556e-05,
+      "loss": 1.2727,
+      "step": 9646
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.98763367128998e-05,
+      "loss": 1.3172,
+      "step": 9647
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.987627567734397e-05,
+      "loss": 1.2963,
+      "step": 9648
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.987621462674807e-05,
+      "loss": 1.3062,
+      "step": 9649
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.987615356111215e-05,
+      "loss": 1.299,
+      "step": 9650
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.987609248043621e-05,
+      "loss": 1.3312,
+      "step": 9651
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.987603138472028e-05,
+      "loss": 1.2925,
+      "step": 9652
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.987597027396437e-05,
+      "loss": 1.2989,
+      "step": 9653
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.98759091481685e-05,
+      "loss": 1.3026,
+      "step": 9654
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.987584800733266e-05,
+      "loss": 1.3133,
+      "step": 9655
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.98757868514569e-05,
+      "loss": 1.3079,
+      "step": 9656
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.987572568054125e-05,
+      "loss": 1.2839,
+      "step": 9657
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.987566449458571e-05,
+      "loss": 1.3128,
+      "step": 9658
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.98756032935903e-05,
+      "loss": 1.2953,
+      "step": 9659
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.987554207755503e-05,
+      "loss": 1.3342,
+      "step": 9660
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.987548084647993e-05,
+      "loss": 1.3312,
+      "step": 9661
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.987541960036501e-05,
+      "loss": 1.2488,
+      "step": 9662
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.987535833921029e-05,
+      "loss": 1.3103,
+      "step": 9663
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.98752970630158e-05,
+      "loss": 1.3105,
+      "step": 9664
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.987523577178155e-05,
+      "loss": 1.3053,
+      "step": 9665
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.987517446550755e-05,
+      "loss": 1.3136,
+      "step": 9666
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.987511314419384e-05,
+      "loss": 1.2988,
+      "step": 9667
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.987505180784044e-05,
+      "loss": 1.3295,
+      "step": 9668
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.987499045644733e-05,
+      "loss": 1.3054,
+      "step": 9669
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.987492909001457e-05,
+      "loss": 1.3235,
+      "step": 9670
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.987486770854216e-05,
+      "loss": 1.3425,
+      "step": 9671
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.98748063120301e-05,
+      "loss": 1.3094,
+      "step": 9672
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.987474490047843e-05,
+      "loss": 1.299,
+      "step": 9673
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.987468347388719e-05,
+      "loss": 1.2968,
+      "step": 9674
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.987462203225636e-05,
+      "loss": 1.2885,
+      "step": 9675
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.987456057558598e-05,
+      "loss": 1.3142,
+      "step": 9676
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.987449910387607e-05,
+      "loss": 1.3344,
+      "step": 9677
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.987443761712663e-05,
+      "loss": 1.3189,
+      "step": 9678
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.987437611533767e-05,
+      "loss": 1.314,
+      "step": 9679
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.987431459850927e-05,
+      "loss": 1.336,
+      "step": 9680
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.987425306664137e-05,
+      "loss": 1.3008,
+      "step": 9681
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.987419151973404e-05,
+      "loss": 1.295,
+      "step": 9682
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.98741299577873e-05,
+      "loss": 1.3189,
+      "step": 9683
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.987406838080113e-05,
+      "loss": 1.3133,
+      "step": 9684
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.987400678877558e-05,
+      "loss": 1.3194,
+      "step": 9685
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.987394518171067e-05,
+      "loss": 1.2705,
+      "step": 9686
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.987388355960639e-05,
+      "loss": 1.3516,
+      "step": 9687
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.98738219224628e-05,
+      "loss": 1.3207,
+      "step": 9688
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.987376027027986e-05,
+      "loss": 1.2944,
+      "step": 9689
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.987369860305766e-05,
+      "loss": 1.3201,
+      "step": 9690
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.987363692079617e-05,
+      "loss": 1.3162,
+      "step": 9691
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.98735752234954e-05,
+      "loss": 1.3336,
+      "step": 9692
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.987351351115542e-05,
+      "loss": 1.2996,
+      "step": 9693
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.987345178377621e-05,
+      "loss": 1.3051,
+      "step": 9694
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.987339004135778e-05,
+      "loss": 1.2716,
+      "step": 9695
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.987332828390018e-05,
+      "loss": 1.3218,
+      "step": 9696
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.987326651140341e-05,
+      "loss": 1.3022,
+      "step": 9697
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.98732047238675e-05,
+      "loss": 1.313,
+      "step": 9698
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.987314292129246e-05,
+      "loss": 1.3059,
+      "step": 9699
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.98730811036783e-05,
+      "loss": 1.334,
+      "step": 9700
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.987301927102505e-05,
+      "loss": 1.3033,
+      "step": 9701
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.987295742333274e-05,
+      "loss": 1.3227,
+      "step": 9702
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.987289556060138e-05,
+      "loss": 1.3019,
+      "step": 9703
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.987283368283096e-05,
+      "loss": 1.3372,
+      "step": 9704
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.987277179002155e-05,
+      "loss": 1.3121,
+      "step": 9705
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.987270988217313e-05,
+      "loss": 1.275,
+      "step": 9706
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.987264795928572e-05,
+      "loss": 1.2953,
+      "step": 9707
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.987258602135936e-05,
+      "loss": 1.354,
+      "step": 9708
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.987252406839406e-05,
+      "loss": 1.2942,
+      "step": 9709
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.987246210038984e-05,
+      "loss": 1.344,
+      "step": 9710
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.987240011734671e-05,
+      "loss": 1.2915,
+      "step": 9711
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.987233811926469e-05,
+      "loss": 1.3311,
+      "step": 9712
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.987227610614382e-05,
+      "loss": 1.3059,
+      "step": 9713
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.987221407798408e-05,
+      "loss": 1.3301,
+      "step": 9714
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.987215203478553e-05,
+      "loss": 1.3283,
+      "step": 9715
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.987208997654815e-05,
+      "loss": 1.3364,
+      "step": 9716
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.987202790327199e-05,
+      "loss": 1.322,
+      "step": 9717
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.987196581495704e-05,
+      "loss": 1.2989,
+      "step": 9718
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.987190371160336e-05,
+      "loss": 1.3342,
+      "step": 9719
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.987184159321093e-05,
+      "loss": 1.3008,
+      "step": 9720
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.98717794597798e-05,
+      "loss": 1.3074,
+      "step": 9721
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.987171731130993e-05,
+      "loss": 1.2735,
+      "step": 9722
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.987165514780141e-05,
+      "loss": 1.3236,
+      "step": 9723
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.987159296925423e-05,
+      "loss": 1.3574,
+      "step": 9724
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.987153077566839e-05,
+      "loss": 1.2961,
+      "step": 9725
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.987146856704395e-05,
+      "loss": 1.3478,
+      "step": 9726
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.987140634338089e-05,
+      "loss": 1.331,
+      "step": 9727
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.987134410467925e-05,
+      "loss": 1.303,
+      "step": 9728
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.987128185093904e-05,
+      "loss": 1.3002,
+      "step": 9729
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.987121958216028e-05,
+      "loss": 1.3268,
+      "step": 9730
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.987115729834299e-05,
+      "loss": 1.3145,
+      "step": 9731
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.987109499948719e-05,
+      "loss": 1.2951,
+      "step": 9732
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.98710326855929e-05,
+      "loss": 1.2773,
+      "step": 9733
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.987097035666013e-05,
+      "loss": 1.2995,
+      "step": 9734
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.987090801268891e-05,
+      "loss": 1.3389,
+      "step": 9735
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.987084565367926e-05,
+      "loss": 1.2959,
+      "step": 9736
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.987078327963117e-05,
+      "loss": 1.3206,
+      "step": 9737
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.98707208905447e-05,
+      "loss": 1.2841,
+      "step": 9738
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.987065848641986e-05,
+      "loss": 1.2896,
+      "step": 9739
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.987059606725663e-05,
+      "loss": 1.2763,
+      "step": 9740
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.98705336330551e-05,
+      "loss": 1.2955,
+      "step": 9741
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.987047118381522e-05,
+      "loss": 1.2869,
+      "step": 9742
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.987040871953703e-05,
+      "loss": 1.2912,
+      "step": 9743
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.987034624022057e-05,
+      "loss": 1.3231,
+      "step": 9744
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.987028374586584e-05,
+      "loss": 1.3331,
+      "step": 9745
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.987022123647284e-05,
+      "loss": 1.3254,
+      "step": 9746
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.987015871204164e-05,
+      "loss": 1.2631,
+      "step": 9747
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.987009617257223e-05,
+      "loss": 1.2727,
+      "step": 9748
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.987003361806462e-05,
+      "loss": 1.2759,
+      "step": 9749
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.986997104851883e-05,
+      "loss": 1.3254,
+      "step": 9750
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.986990846393489e-05,
+      "loss": 1.2912,
+      "step": 9751
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.986984586431283e-05,
+      "loss": 1.342,
+      "step": 9752
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.986978324965264e-05,
+      "loss": 1.2924,
+      "step": 9753
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.986972061995436e-05,
+      "loss": 1.2753,
+      "step": 9754
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.9869657975218e-05,
+      "loss": 1.3291,
+      "step": 9755
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.986959531544358e-05,
+      "loss": 1.3074,
+      "step": 9756
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.986953264063112e-05,
+      "loss": 1.3084,
+      "step": 9757
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.986946995078063e-05,
+      "loss": 1.2903,
+      "step": 9758
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.986940724589215e-05,
+      "loss": 1.3193,
+      "step": 9759
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.986934452596569e-05,
+      "loss": 1.325,
+      "step": 9760
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.986928179100126e-05,
+      "loss": 1.312,
+      "step": 9761
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.986921904099888e-05,
+      "loss": 1.2956,
+      "step": 9762
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.986915627595859e-05,
+      "loss": 1.3294,
+      "step": 9763
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.986909349588037e-05,
+      "loss": 1.265,
+      "step": 9764
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.986903070076428e-05,
+      "loss": 1.2796,
+      "step": 9765
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.98689678906103e-05,
+      "loss": 1.2863,
+      "step": 9766
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.986890506541848e-05,
+      "loss": 1.3225,
+      "step": 9767
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.986884222518883e-05,
+      "loss": 1.2963,
+      "step": 9768
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.986877936992136e-05,
+      "loss": 1.3428,
+      "step": 9769
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.986871649961611e-05,
+      "loss": 1.2995,
+      "step": 9770
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.986865361427307e-05,
+      "loss": 1.3397,
+      "step": 9771
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.986859071389227e-05,
+      "loss": 1.3129,
+      "step": 9772
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.986852779847375e-05,
+      "loss": 1.3176,
+      "step": 9773
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.98684648680175e-05,
+      "loss": 1.314,
+      "step": 9774
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.986840192252356e-05,
+      "loss": 1.3104,
+      "step": 9775
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.986833896199194e-05,
+      "loss": 1.2876,
+      "step": 9776
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.986827598642265e-05,
+      "loss": 1.3235,
+      "step": 9777
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.986821299581573e-05,
+      "loss": 1.2907,
+      "step": 9778
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.986814999017118e-05,
+      "loss": 1.3088,
+      "step": 9779
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.986808696948901e-05,
+      "loss": 1.3149,
+      "step": 9780
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.986802393376929e-05,
+      "loss": 1.2913,
+      "step": 9781
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.986796088301196e-05,
+      "loss": 1.3127,
+      "step": 9782
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.986789781721712e-05,
+      "loss": 1.3058,
+      "step": 9783
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.986783473638473e-05,
+      "loss": 1.3035,
+      "step": 9784
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.986777164051484e-05,
+      "loss": 1.2767,
+      "step": 9785
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.986770852960746e-05,
+      "loss": 1.2918,
+      "step": 9786
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.98676454036626e-05,
+      "loss": 1.3101,
+      "step": 9787
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.98675822626803e-05,
+      "loss": 1.3475,
+      "step": 9788
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.986751910666057e-05,
+      "loss": 1.3218,
+      "step": 9789
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.986745593560341e-05,
+      "loss": 1.2907,
+      "step": 9790
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.986739274950887e-05,
+      "loss": 1.29,
+      "step": 9791
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.986732954837695e-05,
+      "loss": 1.2905,
+      "step": 9792
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.986726633220767e-05,
+      "loss": 1.3272,
+      "step": 9793
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.986720310100104e-05,
+      "loss": 1.2697,
+      "step": 9794
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.986713985475711e-05,
+      "loss": 1.3147,
+      "step": 9795
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.986707659347588e-05,
+      "loss": 1.3375,
+      "step": 9796
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.986701331715735e-05,
+      "loss": 1.3258,
+      "step": 9797
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.986695002580158e-05,
+      "loss": 1.313,
+      "step": 9798
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.986688671940856e-05,
+      "loss": 1.2938,
+      "step": 9799
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.986682339797832e-05,
+      "loss": 1.3123,
+      "step": 9800
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.986676006151088e-05,
+      "loss": 1.3128,
+      "step": 9801
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.986669671000624e-05,
+      "loss": 1.2792,
+      "step": 9802
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.986663334346445e-05,
+      "loss": 1.3718,
+      "step": 9803
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.986656996188549e-05,
+      "loss": 1.3181,
+      "step": 9804
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.986650656526941e-05,
+      "loss": 1.3538,
+      "step": 9805
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.986644315361624e-05,
+      "loss": 1.3303,
+      "step": 9806
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.986637972692596e-05,
+      "loss": 1.2827,
+      "step": 9807
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.986631628519863e-05,
+      "loss": 1.3548,
+      "step": 9808
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.986625282843422e-05,
+      "loss": 1.3186,
+      "step": 9809
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.986618935663281e-05,
+      "loss": 1.3293,
+      "step": 9810
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.986612586979437e-05,
+      "loss": 1.2972,
+      "step": 9811
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.986606236791893e-05,
+      "loss": 1.3136,
+      "step": 9812
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.986599885100654e-05,
+      "loss": 1.3079,
+      "step": 9813
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.986593531905717e-05,
+      "loss": 1.3312,
+      "step": 9814
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.986587177207088e-05,
+      "loss": 1.3153,
+      "step": 9815
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.986580821004765e-05,
+      "loss": 1.3295,
+      "step": 9816
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.986574463298755e-05,
+      "loss": 1.2935,
+      "step": 9817
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.986568104089055e-05,
+      "loss": 1.2984,
+      "step": 9818
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.986561743375669e-05,
+      "loss": 1.3049,
+      "step": 9819
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.9865553811586e-05,
+      "loss": 1.2904,
+      "step": 9820
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.986549017437848e-05,
+      "loss": 1.3225,
+      "step": 9821
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.986542652213417e-05,
+      "loss": 1.3163,
+      "step": 9822
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.986536285485306e-05,
+      "loss": 1.2985,
+      "step": 9823
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.98652991725352e-05,
+      "loss": 1.305,
+      "step": 9824
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.98652354751806e-05,
+      "loss": 1.3222,
+      "step": 9825
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.986517176278926e-05,
+      "loss": 1.2992,
+      "step": 9826
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.986510803536122e-05,
+      "loss": 1.3094,
+      "step": 9827
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.986504429289649e-05,
+      "loss": 1.2671,
+      "step": 9828
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.98649805353951e-05,
+      "loss": 1.3102,
+      "step": 9829
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.986491676285707e-05,
+      "loss": 1.3413,
+      "step": 9830
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.986485297528238e-05,
+      "loss": 1.3256,
+      "step": 9831
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.98647891726711e-05,
+      "loss": 1.3055,
+      "step": 9832
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.986472535502324e-05,
+      "loss": 1.3187,
+      "step": 9833
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.98646615223388e-05,
+      "loss": 1.3282,
+      "step": 9834
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.98645976746178e-05,
+      "loss": 1.3175,
+      "step": 9835
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.986453381186027e-05,
+      "loss": 1.2886,
+      "step": 9836
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.986446993406622e-05,
+      "loss": 1.3335,
+      "step": 9837
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.986440604123569e-05,
+      "loss": 1.3117,
+      "step": 9838
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.986434213336866e-05,
+      "loss": 1.3517,
+      "step": 9839
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.98642782104652e-05,
+      "loss": 1.3222,
+      "step": 9840
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.986421427252529e-05,
+      "loss": 1.3167,
+      "step": 9841
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.986415031954897e-05,
+      "loss": 1.3245,
+      "step": 9842
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.986408635153625e-05,
+      "loss": 1.2951,
+      "step": 9843
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.986402236848715e-05,
+      "loss": 1.3253,
+      "step": 9844
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.986395837040169e-05,
+      "loss": 1.3038,
+      "step": 9845
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.98638943572799e-05,
+      "loss": 1.3139,
+      "step": 9846
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.986383032912176e-05,
+      "loss": 1.3308,
+      "step": 9847
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.986376628592735e-05,
+      "loss": 1.3173,
+      "step": 9848
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.986370222769665e-05,
+      "loss": 1.3026,
+      "step": 9849
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.986363815442968e-05,
+      "loss": 1.3441,
+      "step": 9850
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.986357406612645e-05,
+      "loss": 1.3037,
+      "step": 9851
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.986350996278702e-05,
+      "loss": 1.2992,
+      "step": 9852
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.986344584441139e-05,
+      "loss": 1.2791,
+      "step": 9853
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.986338171099956e-05,
+      "loss": 1.302,
+      "step": 9854
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.986331756255156e-05,
+      "loss": 1.2647,
+      "step": 9855
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.986325339906742e-05,
+      "loss": 1.2965,
+      "step": 9856
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.986318922054717e-05,
+      "loss": 1.2886,
+      "step": 9857
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.986312502699078e-05,
+      "loss": 1.3086,
+      "step": 9858
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.986306081839831e-05,
+      "loss": 1.302,
+      "step": 9859
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.986299659476979e-05,
+      "loss": 1.3196,
+      "step": 9860
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.98629323561052e-05,
+      "loss": 1.2747,
+      "step": 9861
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.986286810240458e-05,
+      "loss": 1.3084,
+      "step": 9862
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.986280383366795e-05,
+      "loss": 1.2983,
+      "step": 9863
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.986273954989534e-05,
+      "loss": 1.299,
+      "step": 9864
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.986267525108675e-05,
+      "loss": 1.291,
+      "step": 9865
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.98626109372422e-05,
+      "loss": 1.3139,
+      "step": 9866
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.986254660836172e-05,
+      "loss": 1.3324,
+      "step": 9867
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.986248226444532e-05,
+      "loss": 1.3087,
+      "step": 9868
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.986241790549303e-05,
+      "loss": 1.3216,
+      "step": 9869
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.986235353150487e-05,
+      "loss": 1.3068,
+      "step": 9870
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.986228914248085e-05,
+      "loss": 1.3569,
+      "step": 9871
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.986222473842099e-05,
+      "loss": 1.3292,
+      "step": 9872
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.986216031932532e-05,
+      "loss": 1.3109,
+      "step": 9873
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.986209588519384e-05,
+      "loss": 1.3007,
+      "step": 9874
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.986203143602659e-05,
+      "loss": 1.3367,
+      "step": 9875
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.986196697182357e-05,
+      "loss": 1.2672,
+      "step": 9876
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.986190249258483e-05,
+      "loss": 1.2811,
+      "step": 9877
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.986183799831035e-05,
+      "loss": 1.3128,
+      "step": 9878
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.986177348900017e-05,
+      "loss": 1.2823,
+      "step": 9879
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.986170896465432e-05,
+      "loss": 1.3066,
+      "step": 9880
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.986164442527281e-05,
+      "loss": 1.3108,
+      "step": 9881
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.986157987085565e-05,
+      "loss": 1.302,
+      "step": 9882
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.986151530140286e-05,
+      "loss": 1.2748,
+      "step": 9883
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.986145071691449e-05,
+      "loss": 1.3237,
+      "step": 9884
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.986138611739051e-05,
+      "loss": 1.2992,
+      "step": 9885
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.986132150283099e-05,
+      "loss": 1.3284,
+      "step": 9886
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.986125687323591e-05,
+      "loss": 1.2776,
+      "step": 9887
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.98611922286053e-05,
+      "loss": 1.325,
+      "step": 9888
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.986112756893919e-05,
+      "loss": 1.3384,
+      "step": 9889
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.986106289423759e-05,
+      "loss": 1.3062,
+      "step": 9890
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.986099820450053e-05,
+      "loss": 1.2939,
+      "step": 9891
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.986093349972802e-05,
+      "loss": 1.3033,
+      "step": 9892
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.986086877992007e-05,
+      "loss": 1.3065,
+      "step": 9893
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.986080404507673e-05,
+      "loss": 1.3076,
+      "step": 9894
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.986073929519799e-05,
+      "loss": 1.3121,
+      "step": 9895
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.986067453028388e-05,
+      "loss": 1.3262,
+      "step": 9896
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.986060975033442e-05,
+      "loss": 1.3136,
+      "step": 9897
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.986054495534964e-05,
+      "loss": 1.3247,
+      "step": 9898
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.986048014532954e-05,
+      "loss": 1.3125,
+      "step": 9899
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.986041532027415e-05,
+      "loss": 1.2918,
+      "step": 9900
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.986035048018349e-05,
+      "loss": 1.3081,
+      "step": 9901
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.986028562505756e-05,
+      "loss": 1.3417,
+      "step": 9902
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.986022075489643e-05,
+      "loss": 1.3434,
+      "step": 9903
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.986015586970006e-05,
+      "loss": 1.302,
+      "step": 9904
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.986009096946851e-05,
+      "loss": 1.308,
+      "step": 9905
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.986002605420179e-05,
+      "loss": 1.299,
+      "step": 9906
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.98599611238999e-05,
+      "loss": 1.3209,
+      "step": 9907
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.985989617856289e-05,
+      "loss": 1.3445,
+      "step": 9908
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.985983121819076e-05,
+      "loss": 1.2919,
+      "step": 9909
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.985976624278353e-05,
+      "loss": 1.3087,
+      "step": 9910
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.985970125234122e-05,
+      "loss": 1.281,
+      "step": 9911
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.985963624686385e-05,
+      "loss": 1.3151,
+      "step": 9912
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.985957122635145e-05,
+      "loss": 1.3388,
+      "step": 9913
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.985950619080404e-05,
+      "loss": 1.3171,
+      "step": 9914
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.985944114022164e-05,
+      "loss": 1.2866,
+      "step": 9915
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.985937607460425e-05,
+      "loss": 1.3156,
+      "step": 9916
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.98593109939519e-05,
+      "loss": 1.2663,
+      "step": 9917
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.985924589826461e-05,
+      "loss": 1.3102,
+      "step": 9918
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.98591807875424e-05,
+      "loss": 1.3349,
+      "step": 9919
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.985911566178529e-05,
+      "loss": 1.307,
+      "step": 9920
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.985905052099331e-05,
+      "loss": 1.2951,
+      "step": 9921
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.985898536516646e-05,
+      "loss": 1.3016,
+      "step": 9922
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.985892019430477e-05,
+      "loss": 1.3544,
+      "step": 9923
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.985885500840826e-05,
+      "loss": 1.2871,
+      "step": 9924
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.985878980747695e-05,
+      "loss": 1.2537,
+      "step": 9925
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.985872459151086e-05,
+      "loss": 1.3048,
+      "step": 9926
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.985865936051e-05,
+      "loss": 1.3014,
+      "step": 9927
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.98585941144744e-05,
+      "loss": 1.2752,
+      "step": 9928
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.985852885340408e-05,
+      "loss": 1.2878,
+      "step": 9929
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.985846357729907e-05,
+      "loss": 1.3067,
+      "step": 9930
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.985839828615936e-05,
+      "loss": 1.3034,
+      "step": 9931
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.9858332979985e-05,
+      "loss": 1.2947,
+      "step": 9932
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.985826765877597e-05,
+      "loss": 1.295,
+      "step": 9933
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.985820232253234e-05,
+      "loss": 1.2886,
+      "step": 9934
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.98581369712541e-05,
+      "loss": 1.2627,
+      "step": 9935
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.985807160494127e-05,
+      "loss": 1.3009,
+      "step": 9936
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.985800622359388e-05,
+      "loss": 1.2792,
+      "step": 9937
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.985794082721194e-05,
+      "loss": 1.2596,
+      "step": 9938
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.985787541579548e-05,
+      "loss": 1.3217,
+      "step": 9939
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.985780998934451e-05,
+      "loss": 1.301,
+      "step": 9940
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.985774454785905e-05,
+      "loss": 1.281,
+      "step": 9941
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.985767909133914e-05,
+      "loss": 1.3173,
+      "step": 9942
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.985761361978476e-05,
+      "loss": 1.2786,
+      "step": 9943
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.985754813319597e-05,
+      "loss": 1.3296,
+      "step": 9944
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.985748263157277e-05,
+      "loss": 1.2947,
+      "step": 9945
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.985741711491519e-05,
+      "loss": 1.3056,
+      "step": 9946
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.985735158322322e-05,
+      "loss": 1.3108,
+      "step": 9947
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.985728603649691e-05,
+      "loss": 1.2902,
+      "step": 9948
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.985722047473628e-05,
+      "loss": 1.3294,
+      "step": 9949
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.985715489794135e-05,
+      "loss": 1.326,
+      "step": 9950
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.985708930611211e-05,
+      "loss": 1.2825,
+      "step": 9951
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.985702369924861e-05,
+      "loss": 1.3121,
+      "step": 9952
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.985695807735085e-05,
+      "loss": 1.3369,
+      "step": 9953
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.985689244041887e-05,
+      "loss": 1.269,
+      "step": 9954
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.98568267884527e-05,
+      "loss": 1.2898,
+      "step": 9955
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.985676112145231e-05,
+      "loss": 1.3033,
+      "step": 9956
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.985669543941776e-05,
+      "loss": 1.3177,
+      "step": 9957
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.985662974234906e-05,
+      "loss": 1.305,
+      "step": 9958
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.985656403024624e-05,
+      "loss": 1.3225,
+      "step": 9959
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.98564983031093e-05,
+      "loss": 1.2982,
+      "step": 9960
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.985643256093827e-05,
+      "loss": 1.3038,
+      "step": 9961
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.985636680373317e-05,
+      "loss": 1.2596,
+      "step": 9962
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.985630103149402e-05,
+      "loss": 1.2809,
+      "step": 9963
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.985623524422083e-05,
+      "loss": 1.3167,
+      "step": 9964
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.985616944191365e-05,
+      "loss": 1.3083,
+      "step": 9965
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.985610362457247e-05,
+      "loss": 1.2779,
+      "step": 9966
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.98560377921973e-05,
+      "loss": 1.3144,
+      "step": 9967
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.98559719447882e-05,
+      "loss": 1.3148,
+      "step": 9968
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.985590608234515e-05,
+      "loss": 1.2945,
+      "step": 9969
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.98558402048682e-05,
+      "loss": 1.328,
+      "step": 9970
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.985577431235735e-05,
+      "loss": 1.3069,
+      "step": 9971
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.985570840481265e-05,
+      "loss": 1.3194,
+      "step": 9972
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.985564248223407e-05,
+      "loss": 1.2743,
+      "step": 9973
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.985557654462167e-05,
+      "loss": 1.3171,
+      "step": 9974
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.985551059197544e-05,
+      "loss": 1.2662,
+      "step": 9975
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.985544462429544e-05,
+      "loss": 1.3108,
+      "step": 9976
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.985537864158166e-05,
+      "loss": 1.2812,
+      "step": 9977
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.985531264383412e-05,
+      "loss": 1.3208,
+      "step": 9978
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.985524663105286e-05,
+      "loss": 1.3028,
+      "step": 9979
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.985518060323788e-05,
+      "loss": 1.3493,
+      "step": 9980
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.985511456038921e-05,
+      "loss": 1.2823,
+      "step": 9981
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.985504850250686e-05,
+      "loss": 1.2897,
+      "step": 9982
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.985498242959085e-05,
+      "loss": 1.3262,
+      "step": 9983
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.985491634164121e-05,
+      "loss": 1.3069,
+      "step": 9984
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.985485023865797e-05,
+      "loss": 1.2843,
+      "step": 9985
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.985478412064114e-05,
+      "loss": 1.3013,
+      "step": 9986
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.985471798759072e-05,
+      "loss": 1.2985,
+      "step": 9987
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.985465183950674e-05,
+      "loss": 1.3245,
+      "step": 9988
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.985458567638923e-05,
+      "loss": 1.3324,
+      "step": 9989
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.985451949823822e-05,
+      "loss": 1.2941,
+      "step": 9990
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.985445330505371e-05,
+      "loss": 1.2913,
+      "step": 9991
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.985438709683572e-05,
+      "loss": 1.2856,
+      "step": 9992
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.985432087358428e-05,
+      "loss": 1.3112,
+      "step": 9993
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.98542546352994e-05,
+      "loss": 1.2886,
+      "step": 9994
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.985418838198113e-05,
+      "loss": 1.3324,
+      "step": 9995
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.985412211362944e-05,
+      "loss": 1.3085,
+      "step": 9996
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.985405583024438e-05,
+      "loss": 1.2941,
+      "step": 9997
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.985398953182598e-05,
+      "loss": 1.2896,
+      "step": 9998
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.985392321837423e-05,
+      "loss": 1.2852,
+      "step": 9999
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.985385688988917e-05,
+      "loss": 1.3443,
+      "step": 10000
+    },
+    {
+      "epoch": 1.61,
+      "eval_loss": 1.3242932558059692,
+      "eval_runtime": 61.3692,
+      "eval_samples_per_second": 162.948,
+      "eval_steps_per_second": 0.228,
+      "step": 10000
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.985379054637082e-05,
+      "loss": 1.3338,
+      "step": 10001
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.98537241878192e-05,
+      "loss": 1.3046,
+      "step": 10002
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.985365781423431e-05,
+      "loss": 1.2935,
+      "step": 10003
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.985359142561618e-05,
+      "loss": 1.3046,
+      "step": 10004
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.985352502196486e-05,
+      "loss": 1.3139,
+      "step": 10005
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.985345860328035e-05,
+      "loss": 1.3347,
+      "step": 10006
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.985339216956265e-05,
+      "loss": 1.2982,
+      "step": 10007
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.98533257208118e-05,
+      "loss": 1.3079,
+      "step": 10008
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.985325925702781e-05,
+      "loss": 1.2991,
+      "step": 10009
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.98531927782107e-05,
+      "loss": 1.3467,
+      "step": 10010
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.985312628436053e-05,
+      "loss": 1.3226,
+      "step": 10011
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.985305977547726e-05,
+      "loss": 1.3037,
+      "step": 10012
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.985299325156092e-05,
+      "loss": 1.3448,
+      "step": 10013
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.985292671261157e-05,
+      "loss": 1.3022,
+      "step": 10014
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.98528601586292e-05,
+      "loss": 1.2907,
+      "step": 10015
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.985279358961384e-05,
+      "loss": 1.3062,
+      "step": 10016
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.985272700556551e-05,
+      "loss": 1.2966,
+      "step": 10017
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.985266040648422e-05,
+      "loss": 1.2876,
+      "step": 10018
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.985259379237e-05,
+      "loss": 1.2866,
+      "step": 10019
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.985252716322287e-05,
+      "loss": 1.3327,
+      "step": 10020
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.985246051904283e-05,
+      "loss": 1.3389,
+      "step": 10021
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.985239385982991e-05,
+      "loss": 1.2979,
+      "step": 10022
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.985232718558417e-05,
+      "loss": 1.3206,
+      "step": 10023
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.985226049630559e-05,
+      "loss": 1.2858,
+      "step": 10024
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.985219379199418e-05,
+      "loss": 1.32,
+      "step": 10025
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.985212707264999e-05,
+      "loss": 1.2982,
+      "step": 10026
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.985206033827303e-05,
+      "loss": 1.3228,
+      "step": 10027
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.98519935888633e-05,
+      "loss": 1.3358,
+      "step": 10028
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.985192682442084e-05,
+      "loss": 1.3129,
+      "step": 10029
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.985186004494569e-05,
+      "loss": 1.3159,
+      "step": 10030
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.985179325043783e-05,
+      "loss": 1.3092,
+      "step": 10031
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.985172644089731e-05,
+      "loss": 1.355,
+      "step": 10032
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.985165961632412e-05,
+      "loss": 1.3176,
+      "step": 10033
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.985159277671831e-05,
+      "loss": 1.3014,
+      "step": 10034
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.98515259220799e-05,
+      "loss": 1.2916,
+      "step": 10035
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.985145905240888e-05,
+      "loss": 1.3183,
+      "step": 10036
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.98513921677053e-05,
+      "loss": 1.2751,
+      "step": 10037
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.985132526796916e-05,
+      "loss": 1.333,
+      "step": 10038
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.985125835320049e-05,
+      "loss": 1.3142,
+      "step": 10039
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.985119142339931e-05,
+      "loss": 1.3034,
+      "step": 10040
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.985112447856566e-05,
+      "loss": 1.3026,
+      "step": 10041
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.985105751869952e-05,
+      "loss": 1.3315,
+      "step": 10042
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.985099054380093e-05,
+      "loss": 1.3166,
+      "step": 10043
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.985092355386991e-05,
+      "loss": 1.2946,
+      "step": 10044
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.98508565489065e-05,
+      "loss": 1.3319,
+      "step": 10045
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.985078952891068e-05,
+      "loss": 1.2943,
+      "step": 10046
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.985072249388247e-05,
+      "loss": 1.3346,
+      "step": 10047
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.985065544382195e-05,
+      "loss": 1.2848,
+      "step": 10048
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.985058837872908e-05,
+      "loss": 1.3001,
+      "step": 10049
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.985052129860391e-05,
+      "loss": 1.3114,
+      "step": 10050
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.985045420344645e-05,
+      "loss": 1.2986,
+      "step": 10051
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.985038709325673e-05,
+      "loss": 1.3314,
+      "step": 10052
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.985031996803475e-05,
+      "loss": 1.3243,
+      "step": 10053
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.985025282778055e-05,
+      "loss": 1.3012,
+      "step": 10054
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.985018567249413e-05,
+      "loss": 1.3375,
+      "step": 10055
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.985011850217553e-05,
+      "loss": 1.296,
+      "step": 10056
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.985005131682476e-05,
+      "loss": 1.3258,
+      "step": 10057
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.984998411644186e-05,
+      "loss": 1.308,
+      "step": 10058
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.984991690102682e-05,
+      "loss": 1.3095,
+      "step": 10059
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.984984967057968e-05,
+      "loss": 1.3093,
+      "step": 10060
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.984978242510044e-05,
+      "loss": 1.299,
+      "step": 10061
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.984971516458914e-05,
+      "loss": 1.325,
+      "step": 10062
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.984964788904581e-05,
+      "loss": 1.2641,
+      "step": 10063
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.984958059847045e-05,
+      "loss": 1.328,
+      "step": 10064
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.984951329286307e-05,
+      "loss": 1.3163,
+      "step": 10065
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.984944597222372e-05,
+      "loss": 1.3101,
+      "step": 10066
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.98493786365524e-05,
+      "loss": 1.306,
+      "step": 10067
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.984931128584913e-05,
+      "loss": 1.3117,
+      "step": 10068
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.984924392011394e-05,
+      "loss": 1.3193,
+      "step": 10069
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.984917653934686e-05,
+      "loss": 1.2881,
+      "step": 10070
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.984910914354788e-05,
+      "loss": 1.314,
+      "step": 10071
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.984904173271705e-05,
+      "loss": 1.2878,
+      "step": 10072
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.984897430685438e-05,
+      "loss": 1.2579,
+      "step": 10073
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.984890686595987e-05,
+      "loss": 1.2858,
+      "step": 10074
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.984883941003358e-05,
+      "loss": 1.2584,
+      "step": 10075
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.98487719390755e-05,
+      "loss": 1.3403,
+      "step": 10076
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.984870445308565e-05,
+      "loss": 1.3352,
+      "step": 10077
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.984863695206407e-05,
+      "loss": 1.3322,
+      "step": 10078
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.984856943601077e-05,
+      "loss": 1.3032,
+      "step": 10079
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.984850190492577e-05,
+      "loss": 1.2911,
+      "step": 10080
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.984843435880909e-05,
+      "loss": 1.3111,
+      "step": 10081
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.984836679766074e-05,
+      "loss": 1.2748,
+      "step": 10082
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.984829922148076e-05,
+      "loss": 1.3199,
+      "step": 10083
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.984823163026916e-05,
+      "loss": 1.2892,
+      "step": 10084
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.984816402402597e-05,
+      "loss": 1.338,
+      "step": 10085
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.984809640275119e-05,
+      "loss": 1.3206,
+      "step": 10086
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.984802876644486e-05,
+      "loss": 1.2752,
+      "step": 10087
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.9847961115107e-05,
+      "loss": 1.3403,
+      "step": 10088
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.984789344873762e-05,
+      "loss": 1.3056,
+      "step": 10089
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.984782576733673e-05,
+      "loss": 1.3011,
+      "step": 10090
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.984775807090438e-05,
+      "loss": 1.306,
+      "step": 10091
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.984769035944057e-05,
+      "loss": 1.3088,
+      "step": 10092
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.984762263294532e-05,
+      "loss": 1.3058,
+      "step": 10093
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.984755489141866e-05,
+      "loss": 1.3209,
+      "step": 10094
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.98474871348606e-05,
+      "loss": 1.3337,
+      "step": 10095
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.984741936327118e-05,
+      "loss": 1.2836,
+      "step": 10096
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.984735157665041e-05,
+      "loss": 1.3111,
+      "step": 10097
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.984728377499829e-05,
+      "loss": 1.3226,
+      "step": 10098
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.984721595831486e-05,
+      "loss": 1.3263,
+      "step": 10099
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.984714812660016e-05,
+      "loss": 1.3553,
+      "step": 10100
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.984708027985417e-05,
+      "loss": 1.3186,
+      "step": 10101
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.984701241807692e-05,
+      "loss": 1.3095,
+      "step": 10102
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.984694454126846e-05,
+      "loss": 1.2847,
+      "step": 10103
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.984687664942878e-05,
+      "loss": 1.3366,
+      "step": 10104
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.984680874255791e-05,
+      "loss": 1.3195,
+      "step": 10105
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.984674082065588e-05,
+      "loss": 1.3352,
+      "step": 10106
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.984667288372269e-05,
+      "loss": 1.3438,
+      "step": 10107
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.984660493175839e-05,
+      "loss": 1.3135,
+      "step": 10108
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.984653696476296e-05,
+      "loss": 1.3073,
+      "step": 10109
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.984646898273647e-05,
+      "loss": 1.3039,
+      "step": 10110
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.984640098567888e-05,
+      "loss": 1.3118,
+      "step": 10111
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.984633297359028e-05,
+      "loss": 1.3124,
+      "step": 10112
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.984626494647063e-05,
+      "loss": 1.3362,
+      "step": 10113
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.984619690431998e-05,
+      "loss": 1.3187,
+      "step": 10114
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.984612884713834e-05,
+      "loss": 1.2983,
+      "step": 10115
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.984606077492575e-05,
+      "loss": 1.3325,
+      "step": 10116
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.98459926876822e-05,
+      "loss": 1.2821,
+      "step": 10117
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.984592458540775e-05,
+      "loss": 1.298,
+      "step": 10118
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.984585646810239e-05,
+      "loss": 1.3396,
+      "step": 10119
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.984578833576614e-05,
+      "loss": 1.3231,
+      "step": 10120
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.984572018839903e-05,
+      "loss": 1.3055,
+      "step": 10121
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.984565202600108e-05,
+      "loss": 1.3,
+      "step": 10122
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.984558384857232e-05,
+      "loss": 1.3248,
+      "step": 10123
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.984551565611274e-05,
+      "loss": 1.3131,
+      "step": 10124
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.984544744862239e-05,
+      "loss": 1.3039,
+      "step": 10125
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.984537922610129e-05,
+      "loss": 1.315,
+      "step": 10126
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.984531098854945e-05,
+      "loss": 1.3339,
+      "step": 10127
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.98452427359669e-05,
+      "loss": 1.317,
+      "step": 10128
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.984517446835363e-05,
+      "loss": 1.3293,
+      "step": 10129
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.98451061857097e-05,
+      "loss": 1.298,
+      "step": 10130
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.98450378880351e-05,
+      "loss": 1.3181,
+      "step": 10131
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.98449695753299e-05,
+      "loss": 1.3359,
+      "step": 10132
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.984490124759405e-05,
+      "loss": 1.3071,
+      "step": 10133
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.984483290482763e-05,
+      "loss": 1.2961,
+      "step": 10134
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.984476454703061e-05,
+      "loss": 1.3118,
+      "step": 10135
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.984469617420306e-05,
+      "loss": 1.312,
+      "step": 10136
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.984462778634496e-05,
+      "loss": 1.3059,
+      "step": 10137
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.984455938345637e-05,
+      "loss": 1.3306,
+      "step": 10138
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.984449096553727e-05,
+      "loss": 1.2671,
+      "step": 10139
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.984442253258772e-05,
+      "loss": 1.3178,
+      "step": 10140
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.98443540846077e-05,
+      "loss": 1.2879,
+      "step": 10141
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.984428562159727e-05,
+      "loss": 1.3252,
+      "step": 10142
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.984421714355643e-05,
+      "loss": 1.3099,
+      "step": 10143
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.984414865048519e-05,
+      "loss": 1.2961,
+      "step": 10144
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.98440801423836e-05,
+      "loss": 1.3341,
+      "step": 10145
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.984401161925165e-05,
+      "loss": 1.326,
+      "step": 10146
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.984394308108938e-05,
+      "loss": 1.3025,
+      "step": 10147
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.984387452789681e-05,
+      "loss": 1.3267,
+      "step": 10148
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.984380595967396e-05,
+      "loss": 1.285,
+      "step": 10149
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.984373737642083e-05,
+      "loss": 1.2912,
+      "step": 10150
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.984366877813746e-05,
+      "loss": 1.3093,
+      "step": 10151
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.984360016482387e-05,
+      "loss": 1.277,
+      "step": 10152
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.984353153648007e-05,
+      "loss": 1.3096,
+      "step": 10153
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.984346289310612e-05,
+      "loss": 1.3119,
+      "step": 10154
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.984339423470198e-05,
+      "loss": 1.2759,
+      "step": 10155
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.984332556126772e-05,
+      "loss": 1.3035,
+      "step": 10156
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.984325687280333e-05,
+      "loss": 1.3301,
+      "step": 10157
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.984318816930884e-05,
+      "loss": 1.2966,
+      "step": 10158
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.984311945078428e-05,
+      "loss": 1.2828,
+      "step": 10159
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.984305071722966e-05,
+      "loss": 1.3093,
+      "step": 10160
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.984298196864501e-05,
+      "loss": 1.2623,
+      "step": 10161
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.984291320503034e-05,
+      "loss": 1.3067,
+      "step": 10162
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.984284442638568e-05,
+      "loss": 1.3118,
+      "step": 10163
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.984277563271104e-05,
+      "loss": 1.3122,
+      "step": 10164
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.984270682400645e-05,
+      "loss": 1.3486,
+      "step": 10165
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.984263800027193e-05,
+      "loss": 1.3089,
+      "step": 10166
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.98425691615075e-05,
+      "loss": 1.325,
+      "step": 10167
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.984250030771317e-05,
+      "loss": 1.2923,
+      "step": 10168
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.984243143888898e-05,
+      "loss": 1.3526,
+      "step": 10169
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.984236255503494e-05,
+      "loss": 1.3027,
+      "step": 10170
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.984229365615107e-05,
+      "loss": 1.2754,
+      "step": 10171
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.984222474223739e-05,
+      "loss": 1.3009,
+      "step": 10172
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.984215581329392e-05,
+      "loss": 1.3666,
+      "step": 10173
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.984208686932071e-05,
+      "loss": 1.2843,
+      "step": 10174
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.984201791031773e-05,
+      "loss": 1.3071,
+      "step": 10175
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.984194893628503e-05,
+      "loss": 1.348,
+      "step": 10176
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.984187994722263e-05,
+      "loss": 1.3399,
+      "step": 10177
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.984181094313054e-05,
+      "loss": 1.2614,
+      "step": 10178
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.98417419240088e-05,
+      "loss": 1.3171,
+      "step": 10179
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.984167288985742e-05,
+      "loss": 1.3241,
+      "step": 10180
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.98416038406764e-05,
+      "loss": 1.2937,
+      "step": 10181
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.984153477646581e-05,
+      "loss": 1.2962,
+      "step": 10182
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.984146569722561e-05,
+      "loss": 1.3221,
+      "step": 10183
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.984139660295588e-05,
+      "loss": 1.3107,
+      "step": 10184
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.98413274936566e-05,
+      "loss": 1.3226,
+      "step": 10185
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.98412583693278e-05,
+      "loss": 1.2559,
+      "step": 10186
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.984118922996951e-05,
+      "loss": 1.3011,
+      "step": 10187
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.984112007558176e-05,
+      "loss": 1.3233,
+      "step": 10188
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.984105090616453e-05,
+      "loss": 1.2776,
+      "step": 10189
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.984098172171787e-05,
+      "loss": 1.2983,
+      "step": 10190
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.984091252224182e-05,
+      "loss": 1.2896,
+      "step": 10191
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.984084330773637e-05,
+      "loss": 1.2975,
+      "step": 10192
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.984077407820154e-05,
+      "loss": 1.3022,
+      "step": 10193
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.984070483363737e-05,
+      "loss": 1.2724,
+      "step": 10194
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.984063557404387e-05,
+      "loss": 1.3046,
+      "step": 10195
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.984056629942107e-05,
+      "loss": 1.327,
+      "step": 10196
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.984049700976896e-05,
+      "loss": 1.3504,
+      "step": 10197
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.984042770508761e-05,
+      "loss": 1.2997,
+      "step": 10198
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.984035838537699e-05,
+      "loss": 1.2703,
+      "step": 10199
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.984028905063718e-05,
+      "loss": 1.2604,
+      "step": 10200
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.984021970086814e-05,
+      "loss": 1.3504,
+      "step": 10201
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.984015033606992e-05,
+      "loss": 1.333,
+      "step": 10202
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.984008095624256e-05,
+      "loss": 1.293,
+      "step": 10203
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.984001156138605e-05,
+      "loss": 1.3253,
+      "step": 10204
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.983994215150042e-05,
+      "loss": 1.2987,
+      "step": 10205
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.983987272658568e-05,
+      "loss": 1.3016,
+      "step": 10206
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.983980328664188e-05,
+      "loss": 1.3184,
+      "step": 10207
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.9839733831669e-05,
+      "loss": 1.2773,
+      "step": 10208
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.983966436166711e-05,
+      "loss": 1.307,
+      "step": 10209
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.98395948766362e-05,
+      "loss": 1.3434,
+      "step": 10210
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.983952537657629e-05,
+      "loss": 1.286,
+      "step": 10211
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.98394558614874e-05,
+      "loss": 1.3235,
+      "step": 10212
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.983938633136958e-05,
+      "loss": 1.3245,
+      "step": 10213
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.98393167862228e-05,
+      "loss": 1.3115,
+      "step": 10214
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.983924722604713e-05,
+      "loss": 1.3121,
+      "step": 10215
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.983917765084258e-05,
+      "loss": 1.2894,
+      "step": 10216
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.983910806060914e-05,
+      "loss": 1.2973,
+      "step": 10217
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.983903845534688e-05,
+      "loss": 1.3338,
+      "step": 10218
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.983896883505576e-05,
+      "loss": 1.3253,
+      "step": 10219
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.983889919973586e-05,
+      "loss": 1.3413,
+      "step": 10220
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.983882954938717e-05,
+      "loss": 1.3247,
+      "step": 10221
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.983875988400972e-05,
+      "loss": 1.3187,
+      "step": 10222
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.983869020360353e-05,
+      "loss": 1.2907,
+      "step": 10223
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.983862050816861e-05,
+      "loss": 1.313,
+      "step": 10224
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.9838550797705e-05,
+      "loss": 1.2912,
+      "step": 10225
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.98384810722127e-05,
+      "loss": 1.2911,
+      "step": 10226
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.983841133169175e-05,
+      "loss": 1.2931,
+      "step": 10227
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.983834157614216e-05,
+      "loss": 1.2882,
+      "step": 10228
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.983827180556395e-05,
+      "loss": 1.2932,
+      "step": 10229
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.983820201995717e-05,
+      "loss": 1.3235,
+      "step": 10230
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.983813221932178e-05,
+      "loss": 1.317,
+      "step": 10231
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.983806240365787e-05,
+      "loss": 1.337,
+      "step": 10232
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.983799257296543e-05,
+      "loss": 1.3169,
+      "step": 10233
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.983792272724445e-05,
+      "loss": 1.3145,
+      "step": 10234
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.9837852866495e-05,
+      "loss": 1.3265,
+      "step": 10235
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.983778299071706e-05,
+      "loss": 1.2941,
+      "step": 10236
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.98377130999107e-05,
+      "loss": 1.316,
+      "step": 10237
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.983764319407591e-05,
+      "loss": 1.3172,
+      "step": 10238
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.98375732732127e-05,
+      "loss": 1.3095,
+      "step": 10239
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.983750333732111e-05,
+      "loss": 1.3061,
+      "step": 10240
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.983743338640117e-05,
+      "loss": 1.3054,
+      "step": 10241
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.983736342045288e-05,
+      "loss": 1.3078,
+      "step": 10242
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.983729343947628e-05,
+      "loss": 1.3175,
+      "step": 10243
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.983722344347137e-05,
+      "loss": 1.3049,
+      "step": 10244
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.983715343243818e-05,
+      "loss": 1.2939,
+      "step": 10245
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.983708340637673e-05,
+      "loss": 1.3098,
+      "step": 10246
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.983701336528704e-05,
+      "loss": 1.2926,
+      "step": 10247
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.983694330916915e-05,
+      "loss": 1.3072,
+      "step": 10248
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.983687323802307e-05,
+      "loss": 1.2922,
+      "step": 10249
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.98368031518488e-05,
+      "loss": 1.3078,
+      "step": 10250
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.983673305064639e-05,
+      "loss": 1.3137,
+      "step": 10251
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.983666293441584e-05,
+      "loss": 1.2953,
+      "step": 10252
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.983659280315719e-05,
+      "loss": 1.3039,
+      "step": 10253
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.983652265687043e-05,
+      "loss": 1.3171,
+      "step": 10254
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.983645249555562e-05,
+      "loss": 1.3242,
+      "step": 10255
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.983638231921277e-05,
+      "loss": 1.3402,
+      "step": 10256
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.98363121278419e-05,
+      "loss": 1.2874,
+      "step": 10257
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.983624192144301e-05,
+      "loss": 1.31,
+      "step": 10258
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.983617170001614e-05,
+      "loss": 1.2921,
+      "step": 10259
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.98361014635613e-05,
+      "loss": 1.3316,
+      "step": 10260
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.983603121207855e-05,
+      "loss": 1.3433,
+      "step": 10261
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.983596094556785e-05,
+      "loss": 1.2853,
+      "step": 10262
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.983589066402927e-05,
+      "loss": 1.2983,
+      "step": 10263
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.983582036746281e-05,
+      "loss": 1.2623,
+      "step": 10264
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.98357500558685e-05,
+      "loss": 1.2969,
+      "step": 10265
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.983567972924634e-05,
+      "loss": 1.3365,
+      "step": 10266
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.983560938759637e-05,
+      "loss": 1.2933,
+      "step": 10267
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.983553903091862e-05,
+      "loss": 1.3193,
+      "step": 10268
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.983546865921309e-05,
+      "loss": 1.3042,
+      "step": 10269
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.983539827247982e-05,
+      "loss": 1.3011,
+      "step": 10270
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.983532787071881e-05,
+      "loss": 1.2917,
+      "step": 10271
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.983525745393009e-05,
+      "loss": 1.3129,
+      "step": 10272
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.98351870221137e-05,
+      "loss": 1.3002,
+      "step": 10273
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.983511657526966e-05,
+      "loss": 1.3052,
+      "step": 10274
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.983504611339794e-05,
+      "loss": 1.3354,
+      "step": 10275
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.983497563649861e-05,
+      "loss": 1.2886,
+      "step": 10276
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.98349051445717e-05,
+      "loss": 1.3018,
+      "step": 10277
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.983483463761719e-05,
+      "loss": 1.3101,
+      "step": 10278
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.983476411563513e-05,
+      "loss": 1.2994,
+      "step": 10279
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.983469357862553e-05,
+      "loss": 1.295,
+      "step": 10280
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.983462302658842e-05,
+      "loss": 1.2989,
+      "step": 10281
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.98345524595238e-05,
+      "loss": 1.2893,
+      "step": 10282
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.983448187743172e-05,
+      "loss": 1.3387,
+      "step": 10283
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.98344112803122e-05,
+      "loss": 1.3081,
+      "step": 10284
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.983434066816522e-05,
+      "loss": 1.2979,
+      "step": 10285
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.983427004099084e-05,
+      "loss": 1.2958,
+      "step": 10286
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.983419939878909e-05,
+      "loss": 1.3023,
+      "step": 10287
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.983412874155995e-05,
+      "loss": 1.3347,
+      "step": 10288
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.983405806930348e-05,
+      "loss": 1.3229,
+      "step": 10289
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.983398738201968e-05,
+      "loss": 1.3001,
+      "step": 10290
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.983391667970859e-05,
+      "loss": 1.3348,
+      "step": 10291
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.983384596237021e-05,
+      "loss": 1.3473,
+      "step": 10292
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.983377523000457e-05,
+      "loss": 1.3191,
+      "step": 10293
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.98337044826117e-05,
+      "loss": 1.308,
+      "step": 10294
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.983363372019159e-05,
+      "loss": 1.3032,
+      "step": 10295
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.983356294274429e-05,
+      "loss": 1.308,
+      "step": 10296
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.983349215026983e-05,
+      "loss": 1.33,
+      "step": 10297
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.98334213427682e-05,
+      "loss": 1.3296,
+      "step": 10298
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.983335052023945e-05,
+      "loss": 1.2711,
+      "step": 10299
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.983327968268359e-05,
+      "loss": 1.2838,
+      "step": 10300
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.983320883010064e-05,
+      "loss": 1.2829,
+      "step": 10301
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.983313796249063e-05,
+      "loss": 1.2958,
+      "step": 10302
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.983306707985356e-05,
+      "loss": 1.2881,
+      "step": 10303
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.983299618218946e-05,
+      "loss": 1.2341,
+      "step": 10304
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.983292526949836e-05,
+      "loss": 1.2934,
+      "step": 10305
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.983285434178029e-05,
+      "loss": 1.2809,
+      "step": 10306
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.983278339903526e-05,
+      "loss": 1.3107,
+      "step": 10307
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.983271244126327e-05,
+      "loss": 1.3222,
+      "step": 10308
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.983264146846438e-05,
+      "loss": 1.2898,
+      "step": 10309
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.983257048063859e-05,
+      "loss": 1.2777,
+      "step": 10310
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.983249947778592e-05,
+      "loss": 1.2708,
+      "step": 10311
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.983242845990639e-05,
+      "loss": 1.2893,
+      "step": 10312
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.983235742700003e-05,
+      "loss": 1.3026,
+      "step": 10313
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.983228637906688e-05,
+      "loss": 1.3066,
+      "step": 10314
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.983221531610692e-05,
+      "loss": 1.2959,
+      "step": 10315
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.98321442381202e-05,
+      "loss": 1.2768,
+      "step": 10316
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.983207314510671e-05,
+      "loss": 1.3084,
+      "step": 10317
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.983200203706654e-05,
+      "loss": 1.3048,
+      "step": 10318
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.983193091399963e-05,
+      "loss": 1.2999,
+      "step": 10319
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.983185977590604e-05,
+      "loss": 1.3302,
+      "step": 10320
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.98317886227858e-05,
+      "loss": 1.303,
+      "step": 10321
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.983171745463891e-05,
+      "loss": 1.2958,
+      "step": 10322
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.983164627146542e-05,
+      "loss": 1.2632,
+      "step": 10323
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.983157507326532e-05,
+      "loss": 1.3296,
+      "step": 10324
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.983150386003864e-05,
+      "loss": 1.3172,
+      "step": 10325
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.983143263178543e-05,
+      "loss": 1.3307,
+      "step": 10326
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.983136138850566e-05,
+      "loss": 1.2976,
+      "step": 10327
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.983129013019939e-05,
+      "loss": 1.2958,
+      "step": 10328
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.983121885686664e-05,
+      "loss": 1.3315,
+      "step": 10329
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.983114756850741e-05,
+      "loss": 1.2833,
+      "step": 10330
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.983107626512174e-05,
+      "loss": 1.3005,
+      "step": 10331
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.983100494670964e-05,
+      "loss": 1.3025,
+      "step": 10332
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.983093361327114e-05,
+      "loss": 1.3195,
+      "step": 10333
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.983086226480625e-05,
+      "loss": 1.3227,
+      "step": 10334
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.983079090131502e-05,
+      "loss": 1.2913,
+      "step": 10335
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.983071952279743e-05,
+      "loss": 1.3249,
+      "step": 10336
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.983064812925354e-05,
+      "loss": 1.3513,
+      "step": 10337
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.983057672068335e-05,
+      "loss": 1.2969,
+      "step": 10338
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.983050529708687e-05,
+      "loss": 1.3129,
+      "step": 10339
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.983043385846415e-05,
+      "loss": 1.3037,
+      "step": 10340
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.983036240481521e-05,
+      "loss": 1.3084,
+      "step": 10341
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.983029093614003e-05,
+      "loss": 1.2904,
+      "step": 10342
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.98302194524387e-05,
+      "loss": 1.319,
+      "step": 10343
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.983014795371117e-05,
+      "loss": 1.3152,
+      "step": 10344
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.983007643995752e-05,
+      "loss": 1.2976,
+      "step": 10345
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.983000491117773e-05,
+      "loss": 1.3101,
+      "step": 10346
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.982993336737184e-05,
+      "loss": 1.2784,
+      "step": 10347
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.982986180853988e-05,
+      "loss": 1.3219,
+      "step": 10348
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.982979023468186e-05,
+      "loss": 1.2732,
+      "step": 10349
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.98297186457978e-05,
+      "loss": 1.3005,
+      "step": 10350
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.982964704188772e-05,
+      "loss": 1.285,
+      "step": 10351
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.982957542295165e-05,
+      "loss": 1.3238,
+      "step": 10352
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.98295037889896e-05,
+      "loss": 1.2951,
+      "step": 10353
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.982943214000162e-05,
+      "loss": 1.314,
+      "step": 10354
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.98293604759877e-05,
+      "loss": 1.2958,
+      "step": 10355
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.982928879694786e-05,
+      "loss": 1.3137,
+      "step": 10356
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.982921710288215e-05,
+      "loss": 1.3189,
+      "step": 10357
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.982914539379056e-05,
+      "loss": 1.3264,
+      "step": 10358
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.982907366967314e-05,
+      "loss": 1.2712,
+      "step": 10359
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.982900193052991e-05,
+      "loss": 1.3321,
+      "step": 10360
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.982893017636086e-05,
+      "loss": 1.3122,
+      "step": 10361
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.982885840716605e-05,
+      "loss": 1.2879,
+      "step": 10362
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.982878662294547e-05,
+      "loss": 1.3225,
+      "step": 10363
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.982871482369915e-05,
+      "loss": 1.3265,
+      "step": 10364
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.982864300942714e-05,
+      "loss": 1.309,
+      "step": 10365
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.982857118012943e-05,
+      "loss": 1.2749,
+      "step": 10366
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.982849933580604e-05,
+      "loss": 1.3047,
+      "step": 10367
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.982842747645702e-05,
+      "loss": 1.2993,
+      "step": 10368
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.982835560208236e-05,
+      "loss": 1.3207,
+      "step": 10369
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.982828371268208e-05,
+      "loss": 1.2954,
+      "step": 10370
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.982821180825626e-05,
+      "loss": 1.3344,
+      "step": 10371
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.982813988880485e-05,
+      "loss": 1.3271,
+      "step": 10372
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.982806795432791e-05,
+      "loss": 1.3092,
+      "step": 10373
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.982799600482544e-05,
+      "loss": 1.2818,
+      "step": 10374
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.982792404029748e-05,
+      "loss": 1.3173,
+      "step": 10375
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.982785206074406e-05,
+      "loss": 1.312,
+      "step": 10376
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.982778006616518e-05,
+      "loss": 1.2869,
+      "step": 10377
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.982770805656086e-05,
+      "loss": 1.3274,
+      "step": 10378
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.982763603193113e-05,
+      "loss": 1.3337,
+      "step": 10379
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.982756399227602e-05,
+      "loss": 1.3103,
+      "step": 10380
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.982749193759555e-05,
+      "loss": 1.3236,
+      "step": 10381
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.982741986788973e-05,
+      "loss": 1.2857,
+      "step": 10382
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.982734778315859e-05,
+      "loss": 1.317,
+      "step": 10383
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.982727568340215e-05,
+      "loss": 1.3125,
+      "step": 10384
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.982720356862043e-05,
+      "loss": 1.2911,
+      "step": 10385
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.982713143881344e-05,
+      "loss": 1.2981,
+      "step": 10386
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.982705929398123e-05,
+      "loss": 1.2906,
+      "step": 10387
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.982698713412381e-05,
+      "loss": 1.2775,
+      "step": 10388
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.982691495924119e-05,
+      "loss": 1.2636,
+      "step": 10389
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.98268427693334e-05,
+      "loss": 1.285,
+      "step": 10390
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.982677056440046e-05,
+      "loss": 1.306,
+      "step": 10391
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.98266983444424e-05,
+      "loss": 1.2998,
+      "step": 10392
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.982662610945923e-05,
+      "loss": 1.2995,
+      "step": 10393
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.982655385945098e-05,
+      "loss": 1.2913,
+      "step": 10394
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.982648159441767e-05,
+      "loss": 1.2991,
+      "step": 10395
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.982640931435932e-05,
+      "loss": 1.3148,
+      "step": 10396
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.982633701927595e-05,
+      "loss": 1.3145,
+      "step": 10397
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.982626470916758e-05,
+      "loss": 1.3299,
+      "step": 10398
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.982619238403424e-05,
+      "loss": 1.3131,
+      "step": 10399
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.982612004387595e-05,
+      "loss": 1.3124,
+      "step": 10400
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.982604768869273e-05,
+      "loss": 1.3055,
+      "step": 10401
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.982597531848461e-05,
+      "loss": 1.3209,
+      "step": 10402
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.982590293325159e-05,
+      "loss": 1.2895,
+      "step": 10403
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.982583053299371e-05,
+      "loss": 1.3092,
+      "step": 10404
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.982575811771098e-05,
+      "loss": 1.3015,
+      "step": 10405
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.982568568740345e-05,
+      "loss": 1.3167,
+      "step": 10406
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.98256132420711e-05,
+      "loss": 1.3022,
+      "step": 10407
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.982554078171398e-05,
+      "loss": 1.3287,
+      "step": 10408
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.982546830633211e-05,
+      "loss": 1.2929,
+      "step": 10409
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.982539581592549e-05,
+      "loss": 1.2812,
+      "step": 10410
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.982532331049417e-05,
+      "loss": 1.2933,
+      "step": 10411
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.982525079003818e-05,
+      "loss": 1.3512,
+      "step": 10412
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.982517825455748e-05,
+      "loss": 1.2834,
+      "step": 10413
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.982510570405216e-05,
+      "loss": 1.3138,
+      "step": 10414
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.98250331385222e-05,
+      "loss": 1.2813,
+      "step": 10415
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.982496055796766e-05,
+      "loss": 1.2668,
+      "step": 10416
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.982488796238853e-05,
+      "loss": 1.309,
+      "step": 10417
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.982481535178483e-05,
+      "loss": 1.2999,
+      "step": 10418
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.98247427261566e-05,
+      "loss": 1.2867,
+      "step": 10419
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.982467008550386e-05,
+      "loss": 1.2897,
+      "step": 10420
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.982459742982661e-05,
+      "loss": 1.289,
+      "step": 10421
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.982452475912492e-05,
+      "loss": 1.3325,
+      "step": 10422
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.982445207339876e-05,
+      "loss": 1.3032,
+      "step": 10423
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.982437937264816e-05,
+      "loss": 1.3158,
+      "step": 10424
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.982430665687319e-05,
+      "loss": 1.3202,
+      "step": 10425
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.982423392607381e-05,
+      "loss": 1.3128,
+      "step": 10426
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.982416118025008e-05,
+      "loss": 1.3232,
+      "step": 10427
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.982408841940201e-05,
+      "loss": 1.2799,
+      "step": 10428
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.982401564352962e-05,
+      "loss": 1.3239,
+      "step": 10429
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.982394285263292e-05,
+      "loss": 1.3165,
+      "step": 10430
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.982387004671196e-05,
+      "loss": 1.2979,
+      "step": 10431
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.982379722576676e-05,
+      "loss": 1.306,
+      "step": 10432
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.982372438979732e-05,
+      "loss": 1.3015,
+      "step": 10433
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.982365153880366e-05,
+      "loss": 1.2966,
+      "step": 10434
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.982357867278582e-05,
+      "loss": 1.3421,
+      "step": 10435
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.982350579174382e-05,
+      "loss": 1.2904,
+      "step": 10436
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.982343289567769e-05,
+      "loss": 1.3195,
+      "step": 10437
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.982335998458742e-05,
+      "loss": 1.3155,
+      "step": 10438
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.982328705847306e-05,
+      "loss": 1.2634,
+      "step": 10439
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.982321411733463e-05,
+      "loss": 1.2881,
+      "step": 10440
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.982314116117214e-05,
+      "loss": 1.3218,
+      "step": 10441
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.982306818998561e-05,
+      "loss": 1.3043,
+      "step": 10442
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.98229952037751e-05,
+      "loss": 1.3424,
+      "step": 10443
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.982292220254058e-05,
+      "loss": 1.2998,
+      "step": 10444
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.982284918628209e-05,
+      "loss": 1.3018,
+      "step": 10445
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.982277615499967e-05,
+      "loss": 1.2905,
+      "step": 10446
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.98227031086933e-05,
+      "loss": 1.2924,
+      "step": 10447
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.982263004736306e-05,
+      "loss": 1.3359,
+      "step": 10448
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.982255697100893e-05,
+      "loss": 1.3362,
+      "step": 10449
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.982248387963095e-05,
+      "loss": 1.2814,
+      "step": 10450
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.982241077322913e-05,
+      "loss": 1.3324,
+      "step": 10451
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.98223376518035e-05,
+      "loss": 1.3493,
+      "step": 10452
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.982226451535409e-05,
+      "loss": 1.335,
+      "step": 10453
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.982219136388089e-05,
+      "loss": 1.3341,
+      "step": 10454
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.982211819738396e-05,
+      "loss": 1.3299,
+      "step": 10455
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.982204501586329e-05,
+      "loss": 1.31,
+      "step": 10456
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.982197181931895e-05,
+      "loss": 1.3018,
+      "step": 10457
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.98218986077509e-05,
+      "loss": 1.3236,
+      "step": 10458
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.982182538115922e-05,
+      "loss": 1.2776,
+      "step": 10459
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.982175213954389e-05,
+      "loss": 1.2803,
+      "step": 10460
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.982167888290495e-05,
+      "loss": 1.2995,
+      "step": 10461
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.982160561124241e-05,
+      "loss": 1.295,
+      "step": 10462
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.982153232455631e-05,
+      "loss": 1.3513,
+      "step": 10463
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.982145902284666e-05,
+      "loss": 1.3154,
+      "step": 10464
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.98213857061135e-05,
+      "loss": 1.2888,
+      "step": 10465
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.982131237435681e-05,
+      "loss": 1.294,
+      "step": 10466
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.982123902757665e-05,
+      "loss": 1.2879,
+      "step": 10467
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.982116566577304e-05,
+      "loss": 1.3093,
+      "step": 10468
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.9821092288946e-05,
+      "loss": 1.2927,
+      "step": 10469
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.982101889709554e-05,
+      "loss": 1.3221,
+      "step": 10470
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.982094549022167e-05,
+      "loss": 1.3196,
+      "step": 10471
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.982087206832444e-05,
+      "loss": 1.3171,
+      "step": 10472
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.982079863140388e-05,
+      "loss": 1.2799,
+      "step": 10473
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.982072517945997e-05,
+      "loss": 1.3336,
+      "step": 10474
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.982065171249277e-05,
+      "loss": 1.3071,
+      "step": 10475
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.982057823050229e-05,
+      "loss": 1.2853,
+      "step": 10476
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.982050473348854e-05,
+      "loss": 1.3129,
+      "step": 10477
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.982043122145156e-05,
+      "loss": 1.3078,
+      "step": 10478
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.982035769439136e-05,
+      "loss": 1.3293,
+      "step": 10479
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.982028415230797e-05,
+      "loss": 1.2981,
+      "step": 10480
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.982021059520141e-05,
+      "loss": 1.3634,
+      "step": 10481
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.982013702307171e-05,
+      "loss": 1.2881,
+      "step": 10482
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.982006343591887e-05,
+      "loss": 1.2883,
+      "step": 10483
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.981998983374294e-05,
+      "loss": 1.3505,
+      "step": 10484
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.981991621654392e-05,
+      "loss": 1.3174,
+      "step": 10485
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.981984258432184e-05,
+      "loss": 1.3431,
+      "step": 10486
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.981976893707673e-05,
+      "loss": 1.2951,
+      "step": 10487
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.981969527480861e-05,
+      "loss": 1.2936,
+      "step": 10488
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.981962159751748e-05,
+      "loss": 1.2937,
+      "step": 10489
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.981954790520339e-05,
+      "loss": 1.3059,
+      "step": 10490
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.981947419786635e-05,
+      "loss": 1.2969,
+      "step": 10491
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.981940047550637e-05,
+      "loss": 1.2889,
+      "step": 10492
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.98193267381235e-05,
+      "loss": 1.2988,
+      "step": 10493
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.981925298571774e-05,
+      "loss": 1.2958,
+      "step": 10494
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.981917921828914e-05,
+      "loss": 1.2993,
+      "step": 10495
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.981910543583767e-05,
+      "loss": 1.2957,
+      "step": 10496
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.981903163836341e-05,
+      "loss": 1.3378,
+      "step": 10497
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.981895782586636e-05,
+      "loss": 1.3031,
+      "step": 10498
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.981888399834654e-05,
+      "loss": 1.3074,
+      "step": 10499
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.981881015580396e-05,
+      "loss": 1.2845,
+      "step": 10500
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.981873629823866e-05,
+      "loss": 1.2971,
+      "step": 10501
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.981866242565064e-05,
+      "loss": 1.299,
+      "step": 10502
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.981858853803996e-05,
+      "loss": 1.3314,
+      "step": 10503
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.981851463540661e-05,
+      "loss": 1.279,
+      "step": 10504
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.981844071775062e-05,
+      "loss": 1.3017,
+      "step": 10505
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.981836678507203e-05,
+      "loss": 1.331,
+      "step": 10506
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.981829283737083e-05,
+      "loss": 1.3239,
+      "step": 10507
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.981821887464708e-05,
+      "loss": 1.2955,
+      "step": 10508
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.981814489690076e-05,
+      "loss": 1.2785,
+      "step": 10509
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.981807090413193e-05,
+      "loss": 1.3285,
+      "step": 10510
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.981799689634059e-05,
+      "loss": 1.335,
+      "step": 10511
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.981792287352678e-05,
+      "loss": 1.293,
+      "step": 10512
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.981784883569051e-05,
+      "loss": 1.2737,
+      "step": 10513
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.98177747828318e-05,
+      "loss": 1.2862,
+      "step": 10514
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.981770071495067e-05,
+      "loss": 1.3204,
+      "step": 10515
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.981762663204715e-05,
+      "loss": 1.2876,
+      "step": 10516
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.981755253412126e-05,
+      "loss": 1.3083,
+      "step": 10517
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.981747842117304e-05,
+      "loss": 1.3331,
+      "step": 10518
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.981740429320248e-05,
+      "loss": 1.2827,
+      "step": 10519
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.981733015020962e-05,
+      "loss": 1.2832,
+      "step": 10520
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.981725599219448e-05,
+      "loss": 1.3099,
+      "step": 10521
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.981718181915708e-05,
+      "loss": 1.2908,
+      "step": 10522
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.981710763109746e-05,
+      "loss": 1.2908,
+      "step": 10523
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.98170334280156e-05,
+      "loss": 1.2708,
+      "step": 10524
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.981695920991158e-05,
+      "loss": 1.284,
+      "step": 10525
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.981688497678537e-05,
+      "loss": 1.3244,
+      "step": 10526
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.981681072863702e-05,
+      "loss": 1.3117,
+      "step": 10527
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.981673646546657e-05,
+      "loss": 1.3143,
+      "step": 10528
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.981666218727399e-05,
+      "loss": 1.3014,
+      "step": 10529
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.981658789405934e-05,
+      "loss": 1.3176,
+      "step": 10530
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.981651358582264e-05,
+      "loss": 1.2868,
+      "step": 10531
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.98164392625639e-05,
+      "loss": 1.2894,
+      "step": 10532
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.981636492428314e-05,
+      "loss": 1.3183,
+      "step": 10533
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.981629057098041e-05,
+      "loss": 1.2654,
+      "step": 10534
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.981621620265571e-05,
+      "loss": 1.3043,
+      "step": 10535
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.981614181930907e-05,
+      "loss": 1.3161,
+      "step": 10536
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.98160674209405e-05,
+      "loss": 1.3131,
+      "step": 10537
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.981599300755003e-05,
+      "loss": 1.3095,
+      "step": 10538
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.981591857913767e-05,
+      "loss": 1.3001,
+      "step": 10539
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.981584413570348e-05,
+      "loss": 1.2955,
+      "step": 10540
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.981576967724747e-05,
+      "loss": 1.3073,
+      "step": 10541
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.981569520376963e-05,
+      "loss": 1.3351,
+      "step": 10542
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.981562071527e-05,
+      "loss": 1.3027,
+      "step": 10543
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.981554621174861e-05,
+      "loss": 1.3312,
+      "step": 10544
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.981547169320547e-05,
+      "loss": 1.3237,
+      "step": 10545
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.981539715964063e-05,
+      "loss": 1.2954,
+      "step": 10546
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.981532261105409e-05,
+      "loss": 1.2898,
+      "step": 10547
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.981524804744587e-05,
+      "loss": 1.288,
+      "step": 10548
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.981517346881599e-05,
+      "loss": 1.3169,
+      "step": 10549
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.981509887516448e-05,
+      "loss": 1.3188,
+      "step": 10550
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.981502426649137e-05,
+      "loss": 1.2676,
+      "step": 10551
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.981494964279667e-05,
+      "loss": 1.3384,
+      "step": 10552
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.981487500408043e-05,
+      "loss": 1.3341,
+      "step": 10553
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.981480035034263e-05,
+      "loss": 1.272,
+      "step": 10554
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.981472568158332e-05,
+      "loss": 1.3107,
+      "step": 10555
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.981465099780252e-05,
+      "loss": 1.2429,
+      "step": 10556
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.981457629900024e-05,
+      "loss": 1.3227,
+      "step": 10557
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.981450158517651e-05,
+      "loss": 1.3236,
+      "step": 10558
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.981442685633136e-05,
+      "loss": 1.2783,
+      "step": 10559
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.98143521124648e-05,
+      "loss": 1.2851,
+      "step": 10560
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.981427735357686e-05,
+      "loss": 1.2993,
+      "step": 10561
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.981420257966756e-05,
+      "loss": 1.2854,
+      "step": 10562
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.981412779073694e-05,
+      "loss": 1.3002,
+      "step": 10563
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.981405298678498e-05,
+      "loss": 1.3029,
+      "step": 10564
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.981397816781174e-05,
+      "loss": 1.2941,
+      "step": 10565
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.981390333381724e-05,
+      "loss": 1.3037,
+      "step": 10566
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.981382848480148e-05,
+      "loss": 1.3393,
+      "step": 10567
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.981375362076451e-05,
+      "loss": 1.2892,
+      "step": 10568
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.981367874170631e-05,
+      "loss": 1.3131,
+      "step": 10569
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.981360384762696e-05,
+      "loss": 1.3108,
+      "step": 10570
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.981352893852644e-05,
+      "loss": 1.3054,
+      "step": 10571
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.98134540144048e-05,
+      "loss": 1.3137,
+      "step": 10572
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.981337907526204e-05,
+      "loss": 1.2997,
+      "step": 10573
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.981330412109819e-05,
+      "loss": 1.3019,
+      "step": 10574
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.981322915191328e-05,
+      "loss": 1.2785,
+      "step": 10575
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.981315416770733e-05,
+      "loss": 1.2981,
+      "step": 10576
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.981307916848035e-05,
+      "loss": 1.2633,
+      "step": 10577
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.981300415423238e-05,
+      "loss": 1.3246,
+      "step": 10578
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.981292912496342e-05,
+      "loss": 1.328,
+      "step": 10579
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.981285408067352e-05,
+      "loss": 1.2918,
+      "step": 10580
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.981277902136269e-05,
+      "loss": 1.2741,
+      "step": 10581
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.981270394703096e-05,
+      "loss": 1.3109,
+      "step": 10582
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.981262885767832e-05,
+      "loss": 1.3341,
+      "step": 10583
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.981255375330485e-05,
+      "loss": 1.2822,
+      "step": 10584
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.981247863391053e-05,
+      "loss": 1.2683,
+      "step": 10585
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.981240349949538e-05,
+      "loss": 1.2896,
+      "step": 10586
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.981232835005945e-05,
+      "loss": 1.2987,
+      "step": 10587
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.981225318560274e-05,
+      "loss": 1.3009,
+      "step": 10588
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.981217800612527e-05,
+      "loss": 1.2757,
+      "step": 10589
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.981210281162709e-05,
+      "loss": 1.2924,
+      "step": 10590
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.981202760210822e-05,
+      "loss": 1.2553,
+      "step": 10591
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.981195237756865e-05,
+      "loss": 1.3177,
+      "step": 10592
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.981187713800844e-05,
+      "loss": 1.3127,
+      "step": 10593
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.981180188342756e-05,
+      "loss": 1.3038,
+      "step": 10594
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.98117266138261e-05,
+      "loss": 1.316,
+      "step": 10595
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.981165132920404e-05,
+      "loss": 1.3016,
+      "step": 10596
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.981157602956142e-05,
+      "loss": 1.2989,
+      "step": 10597
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.981150071489824e-05,
+      "loss": 1.2864,
+      "step": 10598
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.981142538521454e-05,
+      "loss": 1.3112,
+      "step": 10599
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.981135004051034e-05,
+      "loss": 1.2931,
+      "step": 10600
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.981127468078568e-05,
+      "loss": 1.2981,
+      "step": 10601
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.981119930604055e-05,
+      "loss": 1.3128,
+      "step": 10602
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.981112391627501e-05,
+      "loss": 1.3004,
+      "step": 10603
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.981104851148904e-05,
+      "loss": 1.3045,
+      "step": 10604
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.98109730916827e-05,
+      "loss": 1.3096,
+      "step": 10605
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.981089765685599e-05,
+      "loss": 1.3047,
+      "step": 10606
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.981082220700892e-05,
+      "loss": 1.3131,
+      "step": 10607
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.981074674214156e-05,
+      "loss": 1.35,
+      "step": 10608
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.981067126225389e-05,
+      "loss": 1.3009,
+      "step": 10609
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.981059576734597e-05,
+      "loss": 1.279,
+      "step": 10610
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.981052025741779e-05,
+      "loss": 1.3169,
+      "step": 10611
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.981044473246937e-05,
+      "loss": 1.2964,
+      "step": 10612
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.981036919250076e-05,
+      "loss": 1.3262,
+      "step": 10613
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.981029363751196e-05,
+      "loss": 1.2989,
+      "step": 10614
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.981021806750301e-05,
+      "loss": 1.3082,
+      "step": 10615
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.981014248247393e-05,
+      "loss": 1.2902,
+      "step": 10616
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.981006688242474e-05,
+      "loss": 1.3346,
+      "step": 10617
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.980999126735544e-05,
+      "loss": 1.2984,
+      "step": 10618
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.980991563726609e-05,
+      "loss": 1.2949,
+      "step": 10619
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.980983999215669e-05,
+      "loss": 1.2885,
+      "step": 10620
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.980976433202727e-05,
+      "loss": 1.2972,
+      "step": 10621
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.980968865687785e-05,
+      "loss": 1.3241,
+      "step": 10622
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.980961296670846e-05,
+      "loss": 1.3017,
+      "step": 10623
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.980953726151912e-05,
+      "loss": 1.2787,
+      "step": 10624
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.980946154130983e-05,
+      "loss": 1.3049,
+      "step": 10625
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.980938580608065e-05,
+      "loss": 1.2756,
+      "step": 10626
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.98093100558316e-05,
+      "loss": 1.2624,
+      "step": 10627
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.980923429056266e-05,
+      "loss": 1.3006,
+      "step": 10628
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.980915851027391e-05,
+      "loss": 1.3129,
+      "step": 10629
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.980908271496532e-05,
+      "loss": 1.336,
+      "step": 10630
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.980900690463694e-05,
+      "loss": 1.3392,
+      "step": 10631
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.98089310792888e-05,
+      "loss": 1.3087,
+      "step": 10632
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.980885523892092e-05,
+      "loss": 1.3262,
+      "step": 10633
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.98087793835333e-05,
+      "loss": 1.3135,
+      "step": 10634
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.980870351312599e-05,
+      "loss": 1.3129,
+      "step": 10635
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.980862762769899e-05,
+      "loss": 1.3418,
+      "step": 10636
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.980855172725234e-05,
+      "loss": 1.3207,
+      "step": 10637
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.980847581178607e-05,
+      "loss": 1.3338,
+      "step": 10638
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.980839988130017e-05,
+      "loss": 1.3407,
+      "step": 10639
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.98083239357947e-05,
+      "loss": 1.3336,
+      "step": 10640
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.980824797526964e-05,
+      "loss": 1.3249,
+      "step": 10641
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.980817199972507e-05,
+      "loss": 1.2669,
+      "step": 10642
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.980809600916097e-05,
+      "loss": 1.3183,
+      "step": 10643
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.980802000357736e-05,
+      "loss": 1.2811,
+      "step": 10644
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.980794398297431e-05,
+      "loss": 1.3245,
+      "step": 10645
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.980786794735178e-05,
+      "loss": 1.3184,
+      "step": 10646
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.980779189670983e-05,
+      "loss": 1.3213,
+      "step": 10647
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.980771583104849e-05,
+      "loss": 1.3205,
+      "step": 10648
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.980763975036777e-05,
+      "loss": 1.3072,
+      "step": 10649
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.980756365466767e-05,
+      "loss": 1.3297,
+      "step": 10650
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.980748754394826e-05,
+      "loss": 1.3164,
+      "step": 10651
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.980741141820953e-05,
+      "loss": 1.3299,
+      "step": 10652
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.98073352774515e-05,
+      "loss": 1.2655,
+      "step": 10653
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.980725912167422e-05,
+      "loss": 1.3184,
+      "step": 10654
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.98071829508777e-05,
+      "loss": 1.2842,
+      "step": 10655
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.980710676506195e-05,
+      "loss": 1.3241,
+      "step": 10656
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.9807030564227e-05,
+      "loss": 1.3308,
+      "step": 10657
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.980695434837288e-05,
+      "loss": 1.2949,
+      "step": 10658
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.980687811749961e-05,
+      "loss": 1.3174,
+      "step": 10659
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.980680187160721e-05,
+      "loss": 1.3358,
+      "step": 10660
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.98067256106957e-05,
+      "loss": 1.2993,
+      "step": 10661
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.980664933476513e-05,
+      "loss": 1.2927,
+      "step": 10662
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.980657304381548e-05,
+      "loss": 1.3146,
+      "step": 10663
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.98064967378468e-05,
+      "loss": 1.3001,
+      "step": 10664
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.980642041685911e-05,
+      "loss": 1.2657,
+      "step": 10665
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.980634408085243e-05,
+      "loss": 1.3042,
+      "step": 10666
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.980626772982678e-05,
+      "loss": 1.3246,
+      "step": 10667
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.980619136378219e-05,
+      "loss": 1.3164,
+      "step": 10668
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.980611498271869e-05,
+      "loss": 1.2904,
+      "step": 10669
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.980603858663626e-05,
+      "loss": 1.3473,
+      "step": 10670
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.980596217553498e-05,
+      "loss": 1.3314,
+      "step": 10671
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.980588574941484e-05,
+      "loss": 1.3382,
+      "step": 10672
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.980580930827588e-05,
+      "loss": 1.3151,
+      "step": 10673
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.98057328521181e-05,
+      "loss": 1.3488,
+      "step": 10674
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.980565638094155e-05,
+      "loss": 1.2961,
+      "step": 10675
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.980557989474623e-05,
+      "loss": 1.2659,
+      "step": 10676
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.980550339353218e-05,
+      "loss": 1.2686,
+      "step": 10677
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.980542687729943e-05,
+      "loss": 1.285,
+      "step": 10678
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.980535034604796e-05,
+      "loss": 1.3246,
+      "step": 10679
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.980527379977784e-05,
+      "loss": 1.3384,
+      "step": 10680
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.980519723848909e-05,
+      "loss": 1.2855,
+      "step": 10681
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.98051206621817e-05,
+      "loss": 1.3021,
+      "step": 10682
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.980504407085571e-05,
+      "loss": 1.3017,
+      "step": 10683
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.980496746451115e-05,
+      "loss": 1.3271,
+      "step": 10684
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.980489084314805e-05,
+      "loss": 1.2856,
+      "step": 10685
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.98048142067664e-05,
+      "loss": 1.3072,
+      "step": 10686
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.980473755536625e-05,
+      "loss": 1.3093,
+      "step": 10687
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.980466088894765e-05,
+      "loss": 1.3133,
+      "step": 10688
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.980458420751056e-05,
+      "loss": 1.2863,
+      "step": 10689
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.980450751105504e-05,
+      "loss": 1.3069,
+      "step": 10690
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.980443079958109e-05,
+      "loss": 1.2875,
+      "step": 10691
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.980435407308877e-05,
+      "loss": 1.3215,
+      "step": 10692
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.980427733157808e-05,
+      "loss": 1.2648,
+      "step": 10693
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.980420057504905e-05,
+      "loss": 1.3044,
+      "step": 10694
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.980412380350168e-05,
+      "loss": 1.3095,
+      "step": 10695
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.980404701693603e-05,
+      "loss": 1.2961,
+      "step": 10696
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.98039702153521e-05,
+      "loss": 1.3375,
+      "step": 10697
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.980389339874992e-05,
+      "loss": 1.2573,
+      "step": 10698
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.980381656712951e-05,
+      "loss": 1.2688,
+      "step": 10699
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.980373972049091e-05,
+      "loss": 1.3072,
+      "step": 10700
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.980366285883411e-05,
+      "loss": 1.3277,
+      "step": 10701
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.980358598215918e-05,
+      "loss": 1.3104,
+      "step": 10702
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.980350909046608e-05,
+      "loss": 1.3284,
+      "step": 10703
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.980343218375488e-05,
+      "loss": 1.2896,
+      "step": 10704
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.98033552620256e-05,
+      "loss": 1.3377,
+      "step": 10705
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.980327832527825e-05,
+      "loss": 1.295,
+      "step": 10706
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.980320137351284e-05,
+      "loss": 1.3269,
+      "step": 10707
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.980312440672943e-05,
+      "loss": 1.3014,
+      "step": 10708
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.980304742492803e-05,
+      "loss": 1.3272,
+      "step": 10709
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.980297042810864e-05,
+      "loss": 1.3178,
+      "step": 10710
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.980289341627132e-05,
+      "loss": 1.3169,
+      "step": 10711
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.980281638941606e-05,
+      "loss": 1.2911,
+      "step": 10712
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.98027393475429e-05,
+      "loss": 1.2894,
+      "step": 10713
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.980266229065185e-05,
+      "loss": 1.285,
+      "step": 10714
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.980258521874296e-05,
+      "loss": 1.2927,
+      "step": 10715
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.980250813181624e-05,
+      "loss": 1.2905,
+      "step": 10716
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.980243102987168e-05,
+      "loss": 1.3376,
+      "step": 10717
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.980235391290935e-05,
+      "loss": 1.2927,
+      "step": 10718
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.980227678092928e-05,
+      "loss": 1.3133,
+      "step": 10719
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.980219963393143e-05,
+      "loss": 1.3006,
+      "step": 10720
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.980212247191589e-05,
+      "loss": 1.2889,
+      "step": 10721
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.980204529488264e-05,
+      "loss": 1.2892,
+      "step": 10722
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.980196810283173e-05,
+      "loss": 1.3056,
+      "step": 10723
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.980189089576317e-05,
+      "loss": 1.2839,
+      "step": 10724
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.9801813673677e-05,
+      "loss": 1.2926,
+      "step": 10725
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.98017364365732e-05,
+      "loss": 1.3189,
+      "step": 10726
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.980165918445183e-05,
+      "loss": 1.3247,
+      "step": 10727
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.980158191731292e-05,
+      "loss": 1.2876,
+      "step": 10728
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.980150463515647e-05,
+      "loss": 1.2941,
+      "step": 10729
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.980142733798252e-05,
+      "loss": 1.3174,
+      "step": 10730
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.980135002579108e-05,
+      "loss": 1.3308,
+      "step": 10731
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.980127269858218e-05,
+      "loss": 1.3209,
+      "step": 10732
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.980119535635585e-05,
+      "loss": 1.2888,
+      "step": 10733
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.98011179991121e-05,
+      "loss": 1.2993,
+      "step": 10734
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.980104062685097e-05,
+      "loss": 1.3238,
+      "step": 10735
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.980096323957244e-05,
+      "loss": 1.3222,
+      "step": 10736
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.980088583727658e-05,
+      "loss": 1.3199,
+      "step": 10737
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.980080841996342e-05,
+      "loss": 1.3266,
+      "step": 10738
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.980073098763295e-05,
+      "loss": 1.3087,
+      "step": 10739
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.980065354028519e-05,
+      "loss": 1.3631,
+      "step": 10740
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.980057607792019e-05,
+      "loss": 1.3291,
+      "step": 10741
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.980049860053797e-05,
+      "loss": 1.3315,
+      "step": 10742
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.980042110813854e-05,
+      "loss": 1.295,
+      "step": 10743
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.980034360072192e-05,
+      "loss": 1.3398,
+      "step": 10744
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.980026607828816e-05,
+      "loss": 1.2874,
+      "step": 10745
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.980018854083725e-05,
+      "loss": 1.2819,
+      "step": 10746
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.980011098836925e-05,
+      "loss": 1.3709,
+      "step": 10747
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.980003342088416e-05,
+      "loss": 1.2548,
+      "step": 10748
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.9799955838382e-05,
+      "loss": 1.3215,
+      "step": 10749
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.979987824086279e-05,
+      "loss": 1.3284,
+      "step": 10750
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.979980062832657e-05,
+      "loss": 1.3376,
+      "step": 10751
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.979972300077334e-05,
+      "loss": 1.3199,
+      "step": 10752
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.979964535820317e-05,
+      "loss": 1.3034,
+      "step": 10753
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.979956770061602e-05,
+      "loss": 1.3117,
+      "step": 10754
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.979949002801198e-05,
+      "loss": 1.3029,
+      "step": 10755
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.979941234039102e-05,
+      "loss": 1.3213,
+      "step": 10756
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.979933463775319e-05,
+      "loss": 1.2914,
+      "step": 10757
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.97992569200985e-05,
+      "loss": 1.3338,
+      "step": 10758
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.979917918742698e-05,
+      "loss": 1.2928,
+      "step": 10759
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.979910143973867e-05,
+      "loss": 1.3247,
+      "step": 10760
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.979902367703357e-05,
+      "loss": 1.3037,
+      "step": 10761
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.97989458993117e-05,
+      "loss": 1.3221,
+      "step": 10762
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.979886810657311e-05,
+      "loss": 1.3425,
+      "step": 10763
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.979879029881778e-05,
+      "loss": 1.3397,
+      "step": 10764
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.97987124760458e-05,
+      "loss": 1.309,
+      "step": 10765
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.979863463825713e-05,
+      "loss": 1.3326,
+      "step": 10766
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.979855678545182e-05,
+      "loss": 1.3252,
+      "step": 10767
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.979847891762989e-05,
+      "loss": 1.3087,
+      "step": 10768
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.979840103479136e-05,
+      "loss": 1.315,
+      "step": 10769
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.979832313693627e-05,
+      "loss": 1.3297,
+      "step": 10770
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.979824522406463e-05,
+      "loss": 1.2962,
+      "step": 10771
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.979816729617646e-05,
+      "loss": 1.2978,
+      "step": 10772
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.979808935327179e-05,
+      "loss": 1.3109,
+      "step": 10773
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.979801139535066e-05,
+      "loss": 1.281,
+      "step": 10774
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.979793342241306e-05,
+      "loss": 1.3016,
+      "step": 10775
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.979785543445903e-05,
+      "loss": 1.3483,
+      "step": 10776
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.979777743148859e-05,
+      "loss": 1.2978,
+      "step": 10777
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.979769941350177e-05,
+      "loss": 1.2536,
+      "step": 10778
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.979762138049859e-05,
+      "loss": 1.3006,
+      "step": 10779
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.979754333247907e-05,
+      "loss": 1.3164,
+      "step": 10780
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.979746526944325e-05,
+      "loss": 1.2873,
+      "step": 10781
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.979738719139113e-05,
+      "loss": 1.3152,
+      "step": 10782
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.979730909832275e-05,
+      "loss": 1.3237,
+      "step": 10783
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.97972309902381e-05,
+      "loss": 1.3151,
+      "step": 10784
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.979715286713726e-05,
+      "loss": 1.3036,
+      "step": 10785
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.979707472902022e-05,
+      "loss": 1.269,
+      "step": 10786
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.979699657588702e-05,
+      "loss": 1.3203,
+      "step": 10787
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.979691840773766e-05,
+      "loss": 1.3144,
+      "step": 10788
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.979684022457216e-05,
+      "loss": 1.3516,
+      "step": 10789
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.979676202639056e-05,
+      "loss": 1.2675,
+      "step": 10790
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.979668381319291e-05,
+      "loss": 1.2801,
+      "step": 10791
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.979660558497918e-05,
+      "loss": 1.3321,
+      "step": 10792
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.979652734174944e-05,
+      "loss": 1.2974,
+      "step": 10793
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.979644908350369e-05,
+      "loss": 1.2877,
+      "step": 10794
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.979637081024196e-05,
+      "loss": 1.2921,
+      "step": 10795
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.979629252196426e-05,
+      "loss": 1.2805,
+      "step": 10796
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.979621421867063e-05,
+      "loss": 1.2932,
+      "step": 10797
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.979613590036108e-05,
+      "loss": 1.3063,
+      "step": 10798
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.979605756703564e-05,
+      "loss": 1.2778,
+      "step": 10799
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.979597921869435e-05,
+      "loss": 1.2716,
+      "step": 10800
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.979590085533722e-05,
+      "loss": 1.2926,
+      "step": 10801
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.979582247696424e-05,
+      "loss": 1.2617,
+      "step": 10802
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.979574408357549e-05,
+      "loss": 1.3032,
+      "step": 10803
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.979566567517097e-05,
+      "loss": 1.306,
+      "step": 10804
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.97955872517507e-05,
+      "loss": 1.2724,
+      "step": 10805
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.979550881331471e-05,
+      "loss": 1.2963,
+      "step": 10806
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.979543035986302e-05,
+      "loss": 1.2838,
+      "step": 10807
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.979535189139565e-05,
+      "loss": 1.2942,
+      "step": 10808
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.979527340791263e-05,
+      "loss": 1.2999,
+      "step": 10809
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.979519490941398e-05,
+      "loss": 1.2881,
+      "step": 10810
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.979511639589973e-05,
+      "loss": 1.3215,
+      "step": 10811
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.97950378673699e-05,
+      "loss": 1.297,
+      "step": 10812
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.97949593238245e-05,
+      "loss": 1.2624,
+      "step": 10813
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.979488076526358e-05,
+      "loss": 1.2805,
+      "step": 10814
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.979480219168714e-05,
+      "loss": 1.307,
+      "step": 10815
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.979472360309522e-05,
+      "loss": 1.2984,
+      "step": 10816
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.979464499948783e-05,
+      "loss": 1.303,
+      "step": 10817
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.979456638086502e-05,
+      "loss": 1.2945,
+      "step": 10818
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.979448774722678e-05,
+      "loss": 1.3124,
+      "step": 10819
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.979440909857313e-05,
+      "loss": 1.2993,
+      "step": 10820
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.979433043490414e-05,
+      "loss": 1.2937,
+      "step": 10821
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.979425175621981e-05,
+      "loss": 1.2916,
+      "step": 10822
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.979417306252015e-05,
+      "loss": 1.3206,
+      "step": 10823
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.979409435380518e-05,
+      "loss": 1.3073,
+      "step": 10824
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.979401563007495e-05,
+      "loss": 1.3078,
+      "step": 10825
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.979393689132946e-05,
+      "loss": 1.2541,
+      "step": 10826
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.979385813756876e-05,
+      "loss": 1.2987,
+      "step": 10827
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.979377936879285e-05,
+      "loss": 1.3,
+      "step": 10828
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.979370058500177e-05,
+      "loss": 1.3193,
+      "step": 10829
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.979362178619553e-05,
+      "loss": 1.2939,
+      "step": 10830
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.979354297237416e-05,
+      "loss": 1.3099,
+      "step": 10831
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.979346414353768e-05,
+      "loss": 1.3098,
+      "step": 10832
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.979338529968612e-05,
+      "loss": 1.2819,
+      "step": 10833
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.97933064408195e-05,
+      "loss": 1.2881,
+      "step": 10834
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.979322756693783e-05,
+      "loss": 1.2556,
+      "step": 10835
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.979314867804118e-05,
+      "loss": 1.2872,
+      "step": 10836
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.97930697741295e-05,
+      "loss": 1.3036,
+      "step": 10837
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.97929908552029e-05,
+      "loss": 1.2647,
+      "step": 10838
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.979291192126133e-05,
+      "loss": 1.2996,
+      "step": 10839
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.979283297230486e-05,
+      "loss": 1.3409,
+      "step": 10840
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.97927540083335e-05,
+      "loss": 1.3298,
+      "step": 10841
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.979267502934726e-05,
+      "loss": 1.2865,
+      "step": 10842
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.979259603534617e-05,
+      "loss": 1.2409,
+      "step": 10843
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.979251702633027e-05,
+      "loss": 1.3265,
+      "step": 10844
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.979243800229956e-05,
+      "loss": 1.296,
+      "step": 10845
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.97923589632541e-05,
+      "loss": 1.2996,
+      "step": 10846
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.979227990919387e-05,
+      "loss": 1.3041,
+      "step": 10847
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.979220084011892e-05,
+      "loss": 1.2743,
+      "step": 10848
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.979212175602927e-05,
+      "loss": 1.3143,
+      "step": 10849
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.979204265692494e-05,
+      "loss": 1.309,
+      "step": 10850
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.979196354280596e-05,
+      "loss": 1.282,
+      "step": 10851
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.979188441367234e-05,
+      "loss": 1.2863,
+      "step": 10852
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.979180526952412e-05,
+      "loss": 1.3003,
+      "step": 10853
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.979172611036131e-05,
+      "loss": 1.2939,
+      "step": 10854
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.979164693618395e-05,
+      "loss": 1.2683,
+      "step": 10855
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.979156774699206e-05,
+      "loss": 1.2928,
+      "step": 10856
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.979148854278565e-05,
+      "loss": 1.2822,
+      "step": 10857
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.979140932356476e-05,
+      "loss": 1.3404,
+      "step": 10858
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.97913300893294e-05,
+      "loss": 1.2911,
+      "step": 10859
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.97912508400796e-05,
+      "loss": 1.3164,
+      "step": 10860
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.979117157581538e-05,
+      "loss": 1.2851,
+      "step": 10861
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.979109229653678e-05,
+      "loss": 1.3248,
+      "step": 10862
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.97910130022438e-05,
+      "loss": 1.3095,
+      "step": 10863
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.979093369293649e-05,
+      "loss": 1.315,
+      "step": 10864
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.979085436861486e-05,
+      "loss": 1.3135,
+      "step": 10865
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.979077502927892e-05,
+      "loss": 1.2985,
+      "step": 10866
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.979069567492872e-05,
+      "loss": 1.3102,
+      "step": 10867
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.979061630556427e-05,
+      "loss": 1.2916,
+      "step": 10868
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.979053692118559e-05,
+      "loss": 1.3277,
+      "step": 10869
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.979045752179271e-05,
+      "loss": 1.333,
+      "step": 10870
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.979037810738566e-05,
+      "loss": 1.33,
+      "step": 10871
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.979029867796446e-05,
+      "loss": 1.3214,
+      "step": 10872
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.979021923352911e-05,
+      "loss": 1.3229,
+      "step": 10873
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.979013977407969e-05,
+      "loss": 1.3215,
+      "step": 10874
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.979006029961616e-05,
+      "loss": 1.3018,
+      "step": 10875
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.97899808101386e-05,
+      "loss": 1.2801,
+      "step": 10876
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.978990130564698e-05,
+      "loss": 1.3188,
+      "step": 10877
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.978982178614136e-05,
+      "loss": 1.3099,
+      "step": 10878
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.978974225162177e-05,
+      "loss": 1.2465,
+      "step": 10879
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.978966270208821e-05,
+      "loss": 1.3291,
+      "step": 10880
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.97895831375407e-05,
+      "loss": 1.2712,
+      "step": 10881
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.97895035579793e-05,
+      "loss": 1.2965,
+      "step": 10882
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.978942396340399e-05,
+      "loss": 1.313,
+      "step": 10883
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.978934435381483e-05,
+      "loss": 1.2909,
+      "step": 10884
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.978926472921182e-05,
+      "loss": 1.308,
+      "step": 10885
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.978918508959501e-05,
+      "loss": 1.319,
+      "step": 10886
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.978910543496439e-05,
+      "loss": 1.3137,
+      "step": 10887
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.978902576532002e-05,
+      "loss": 1.3012,
+      "step": 10888
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.978894608066188e-05,
+      "loss": 1.3147,
+      "step": 10889
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.978886638099004e-05,
+      "loss": 1.2805,
+      "step": 10890
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.978878666630449e-05,
+      "loss": 1.2852,
+      "step": 10891
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.978870693660527e-05,
+      "loss": 1.3324,
+      "step": 10892
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.978862719189242e-05,
+      "loss": 1.2721,
+      "step": 10893
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.978854743216592e-05,
+      "loss": 1.2959,
+      "step": 10894
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.978846765742583e-05,
+      "loss": 1.3062,
+      "step": 10895
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.978838786767217e-05,
+      "loss": 1.3108,
+      "step": 10896
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.978830806290494e-05,
+      "loss": 1.2866,
+      "step": 10897
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.978822824312419e-05,
+      "loss": 1.3195,
+      "step": 10898
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.978814840832995e-05,
+      "loss": 1.3299,
+      "step": 10899
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.978806855852224e-05,
+      "loss": 1.2972,
+      "step": 10900
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.978798869370103e-05,
+      "loss": 1.3141,
+      "step": 10901
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.978790881386642e-05,
+      "loss": 1.3367,
+      "step": 10902
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.978782891901838e-05,
+      "loss": 1.2602,
+      "step": 10903
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.978774900915698e-05,
+      "loss": 1.3066,
+      "step": 10904
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.978766908428221e-05,
+      "loss": 1.3121,
+      "step": 10905
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.978758914439412e-05,
+      "loss": 1.2959,
+      "step": 10906
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.97875091894927e-05,
+      "loss": 1.3079,
+      "step": 10907
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.9787429219578e-05,
+      "loss": 1.3121,
+      "step": 10908
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.978734923465003e-05,
+      "loss": 1.3448,
+      "step": 10909
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.978726923470883e-05,
+      "loss": 1.2877,
+      "step": 10910
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 9.97871892197544e-05,
+      "loss": 1.2817,
+      "step": 10911
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 9.97871091897868e-05,
+      "loss": 1.2873,
+      "step": 10912
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 9.978702914480602e-05,
+      "loss": 1.3372,
+      "step": 10913
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 9.97869490848121e-05,
+      "loss": 1.3162,
+      "step": 10914
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 9.978686900980506e-05,
+      "loss": 1.278,
+      "step": 10915
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 9.978678891978494e-05,
+      "loss": 1.3425,
+      "step": 10916
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 9.978670881475172e-05,
+      "loss": 1.271,
+      "step": 10917
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 9.978662869470549e-05,
+      "loss": 1.3349,
+      "step": 10918
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 9.978654855964622e-05,
+      "loss": 1.3106,
+      "step": 10919
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 9.978646840957396e-05,
+      "loss": 1.282,
+      "step": 10920
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 9.978638824448872e-05,
+      "loss": 1.2717,
+      "step": 10921
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 9.978630806439052e-05,
+      "loss": 1.2746,
+      "step": 10922
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 9.97862278692794e-05,
+      "loss": 1.2877,
+      "step": 10923
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 9.97861476591554e-05,
+      "loss": 1.3158,
+      "step": 10924
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 9.97860674340185e-05,
+      "loss": 1.3262,
+      "step": 10925
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 9.978598719386876e-05,
+      "loss": 1.2949,
+      "step": 10926
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 9.978590693870618e-05,
+      "loss": 1.2962,
+      "step": 10927
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 9.97858266685308e-05,
+      "loss": 1.3202,
+      "step": 10928
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 9.978574638334264e-05,
+      "loss": 1.313,
+      "step": 10929
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 9.978566608314173e-05,
+      "loss": 1.2968,
+      "step": 10930
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 9.97855857679281e-05,
+      "loss": 1.2902,
+      "step": 10931
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 9.978550543770174e-05,
+      "loss": 1.3217,
+      "step": 10932
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 9.978542509246272e-05,
+      "loss": 1.3088,
+      "step": 10933
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 9.978534473221102e-05,
+      "loss": 1.2772,
+      "step": 10934
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 9.978526435694669e-05,
+      "loss": 1.3146,
+      "step": 10935
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 9.978518396666976e-05,
+      "loss": 1.3538,
+      "step": 10936
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 9.978510356138022e-05,
+      "loss": 1.3166,
+      "step": 10937
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 9.978502314107813e-05,
+      "loss": 1.3059,
+      "step": 10938
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 9.978494270576353e-05,
+      "loss": 1.2912,
+      "step": 10939
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 9.978486225543638e-05,
+      "loss": 1.3098,
+      "step": 10940
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 9.978478179009676e-05,
+      "loss": 1.3074,
+      "step": 10941
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 9.978470130974467e-05,
+      "loss": 1.2737,
+      "step": 10942
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 9.978462081438013e-05,
+      "loss": 1.3018,
+      "step": 10943
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 9.978454030400319e-05,
+      "loss": 1.3041,
+      "step": 10944
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 9.978445977861386e-05,
+      "loss": 1.2673,
+      "step": 10945
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 9.978437923821216e-05,
+      "loss": 1.2941,
+      "step": 10946
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 9.97842986827981e-05,
+      "loss": 1.2968,
+      "step": 10947
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 9.978421811237173e-05,
+      "loss": 1.2819,
+      "step": 10948
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 9.978413752693307e-05,
+      "loss": 1.2948,
+      "step": 10949
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 9.978405692648214e-05,
+      "loss": 1.3559,
+      "step": 10950
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 9.978397631101896e-05,
+      "loss": 1.285,
+      "step": 10951
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 9.978389568054356e-05,
+      "loss": 1.3461,
+      "step": 10952
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 9.978381503505596e-05,
+      "loss": 1.2896,
+      "step": 10953
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 9.978373437455618e-05,
+      "loss": 1.3019,
+      "step": 10954
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 9.978365369904425e-05,
+      "loss": 1.2608,
+      "step": 10955
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 9.978357300852022e-05,
+      "loss": 1.2992,
+      "step": 10956
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 9.978349230298407e-05,
+      "loss": 1.3285,
+      "step": 10957
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 9.978341158243585e-05,
+      "loss": 1.2974,
+      "step": 10958
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 9.978333084687556e-05,
+      "loss": 1.3373,
+      "step": 10959
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 9.978325009630328e-05,
+      "loss": 1.328,
+      "step": 10960
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 9.978316933071896e-05,
+      "loss": 1.327,
+      "step": 10961
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 9.978308855012268e-05,
+      "loss": 1.3406,
+      "step": 10962
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 9.978300775451444e-05,
+      "loss": 1.3061,
+      "step": 10963
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 9.978292694389428e-05,
+      "loss": 1.3116,
+      "step": 10964
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 9.978284611826221e-05,
+      "loss": 1.3048,
+      "step": 10965
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 9.978276527761825e-05,
+      "loss": 1.2748,
+      "step": 10966
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 9.978268442196244e-05,
+      "loss": 1.3217,
+      "step": 10967
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 9.978260355129481e-05,
+      "loss": 1.3396,
+      "step": 10968
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 9.978252266561535e-05,
+      "loss": 1.2697,
+      "step": 10969
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 9.978244176492412e-05,
+      "loss": 1.3425,
+      "step": 10970
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 9.978236084922112e-05,
+      "loss": 1.3092,
+      "step": 10971
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 9.97822799185064e-05,
+      "loss": 1.3011,
+      "step": 10972
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 9.978219897277996e-05,
+      "loss": 1.2731,
+      "step": 10973
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.978211801204183e-05,
+      "loss": 1.2982,
+      "step": 10974
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.978203703629206e-05,
+      "loss": 1.3078,
+      "step": 10975
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.978195604553062e-05,
+      "loss": 1.2828,
+      "step": 10976
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.978187503975758e-05,
+      "loss": 1.3024,
+      "step": 10977
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.978179401897297e-05,
+      "loss": 1.3033,
+      "step": 10978
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.978171298317678e-05,
+      "loss": 1.2778,
+      "step": 10979
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.978163193236905e-05,
+      "loss": 1.2716,
+      "step": 10980
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.978155086654982e-05,
+      "loss": 1.2929,
+      "step": 10981
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.978146978571907e-05,
+      "loss": 1.3047,
+      "step": 10982
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.978138868987687e-05,
+      "loss": 1.3131,
+      "step": 10983
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.978130757902324e-05,
+      "loss": 1.2892,
+      "step": 10984
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.978122645315819e-05,
+      "loss": 1.2811,
+      "step": 10985
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.978114531228174e-05,
+      "loss": 1.2694,
+      "step": 10986
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.978106415639393e-05,
+      "loss": 1.3011,
+      "step": 10987
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.978098298549477e-05,
+      "loss": 1.2921,
+      "step": 10988
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.978090179958428e-05,
+      "loss": 1.2535,
+      "step": 10989
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.978082059866249e-05,
+      "loss": 1.2803,
+      "step": 10990
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.978073938272945e-05,
+      "loss": 1.2927,
+      "step": 10991
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.978065815178517e-05,
+      "loss": 1.2678,
+      "step": 10992
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.978057690582965e-05,
+      "loss": 1.3011,
+      "step": 10993
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.978049564486293e-05,
+      "loss": 1.2741,
+      "step": 10994
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.978041436888505e-05,
+      "loss": 1.3166,
+      "step": 10995
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.978033307789603e-05,
+      "loss": 1.3036,
+      "step": 10996
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.978025177189588e-05,
+      "loss": 1.3256,
+      "step": 10997
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.978017045088462e-05,
+      "loss": 1.2552,
+      "step": 10998
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.97800891148623e-05,
+      "loss": 1.2789,
+      "step": 10999
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.978000776382892e-05,
+      "loss": 1.301,
+      "step": 11000
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.977992639778451e-05,
+      "loss": 1.3012,
+      "step": 11001
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.977984501672912e-05,
+      "loss": 1.2835,
+      "step": 11002
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.977976362066275e-05,
+      "loss": 1.2955,
+      "step": 11003
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.977968220958541e-05,
+      "loss": 1.3092,
+      "step": 11004
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.977960078349715e-05,
+      "loss": 1.3151,
+      "step": 11005
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.977951934239799e-05,
+      "loss": 1.3103,
+      "step": 11006
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.977943788628796e-05,
+      "loss": 1.2602,
+      "step": 11007
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.977935641516707e-05,
+      "loss": 1.2825,
+      "step": 11008
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.977927492903534e-05,
+      "loss": 1.2704,
+      "step": 11009
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.977919342789284e-05,
+      "loss": 1.3125,
+      "step": 11010
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.977911191173951e-05,
+      "loss": 1.3325,
+      "step": 11011
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.977903038057546e-05,
+      "loss": 1.2947,
+      "step": 11012
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.977894883440069e-05,
+      "loss": 1.2974,
+      "step": 11013
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.977886727321519e-05,
+      "loss": 1.3203,
+      "step": 11014
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.977878569701902e-05,
+      "loss": 1.2861,
+      "step": 11015
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.977870410581218e-05,
+      "loss": 1.3111,
+      "step": 11016
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.977862249959471e-05,
+      "loss": 1.2888,
+      "step": 11017
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.977854087836665e-05,
+      "loss": 1.3078,
+      "step": 11018
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.977845924212799e-05,
+      "loss": 1.2944,
+      "step": 11019
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.977837759087878e-05,
+      "loss": 1.2944,
+      "step": 11020
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.977829592461903e-05,
+      "loss": 1.3025,
+      "step": 11021
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.977821424334878e-05,
+      "loss": 1.2982,
+      "step": 11022
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.977813254706804e-05,
+      "loss": 1.2524,
+      "step": 11023
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.977805083577684e-05,
+      "loss": 1.3112,
+      "step": 11024
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.977796910947521e-05,
+      "loss": 1.291,
+      "step": 11025
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.977788736816317e-05,
+      "loss": 1.3025,
+      "step": 11026
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.977780561184074e-05,
+      "loss": 1.2998,
+      "step": 11027
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.977772384050794e-05,
+      "loss": 1.2802,
+      "step": 11028
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.977764205416482e-05,
+      "loss": 1.273,
+      "step": 11029
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.977756025281138e-05,
+      "loss": 1.317,
+      "step": 11030
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.977747843644765e-05,
+      "loss": 1.2908,
+      "step": 11031
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.977739660507366e-05,
+      "loss": 1.3036,
+      "step": 11032
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.977731475868944e-05,
+      "loss": 1.2988,
+      "step": 11033
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.977723289729499e-05,
+      "loss": 1.315,
+      "step": 11034
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.977715102089036e-05,
+      "loss": 1.3221,
+      "step": 11035
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.977706912947556e-05,
+      "loss": 1.3239,
+      "step": 11036
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.977698722305064e-05,
+      "loss": 1.303,
+      "step": 11037
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.977690530161559e-05,
+      "loss": 1.287,
+      "step": 11038
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.977682336517045e-05,
+      "loss": 1.3013,
+      "step": 11039
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.977674141371524e-05,
+      "loss": 1.2952,
+      "step": 11040
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.977665944725001e-05,
+      "loss": 1.3009,
+      "step": 11041
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.977657746577476e-05,
+      "loss": 1.2851,
+      "step": 11042
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.977649546928951e-05,
+      "loss": 1.2844,
+      "step": 11043
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.977641345779429e-05,
+      "loss": 1.258,
+      "step": 11044
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.977633143128914e-05,
+      "loss": 1.3294,
+      "step": 11045
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.977624938977407e-05,
+      "loss": 1.2638,
+      "step": 11046
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.97761673332491e-05,
+      "loss": 1.3426,
+      "step": 11047
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.977608526171428e-05,
+      "loss": 1.31,
+      "step": 11048
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.97760031751696e-05,
+      "loss": 1.3164,
+      "step": 11049
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.977592107361511e-05,
+      "loss": 1.2919,
+      "step": 11050
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.977583895705083e-05,
+      "loss": 1.295,
+      "step": 11051
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.977575682547677e-05,
+      "loss": 1.3417,
+      "step": 11052
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.9775674678893e-05,
+      "loss": 1.288,
+      "step": 11053
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.977559251729948e-05,
+      "loss": 1.3257,
+      "step": 11054
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.977551034069627e-05,
+      "loss": 1.3073,
+      "step": 11055
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.977542814908339e-05,
+      "loss": 1.2904,
+      "step": 11056
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.977534594246087e-05,
+      "loss": 1.3629,
+      "step": 11057
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.977526372082873e-05,
+      "loss": 1.2959,
+      "step": 11058
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.977518148418701e-05,
+      "loss": 1.3107,
+      "step": 11059
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.97750992325357e-05,
+      "loss": 1.3014,
+      "step": 11060
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.977501696587484e-05,
+      "loss": 1.2487,
+      "step": 11061
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.977493468420448e-05,
+      "loss": 1.3392,
+      "step": 11062
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.977485238752462e-05,
+      "loss": 1.295,
+      "step": 11063
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.977477007583528e-05,
+      "loss": 1.291,
+      "step": 11064
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.97746877491365e-05,
+      "loss": 1.3168,
+      "step": 11065
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.97746054074283e-05,
+      "loss": 1.2905,
+      "step": 11066
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.97745230507107e-05,
+      "loss": 1.3036,
+      "step": 11067
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.977444067898374e-05,
+      "loss": 1.3308,
+      "step": 11068
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.977435829224742e-05,
+      "loss": 1.3421,
+      "step": 11069
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.977427589050178e-05,
+      "loss": 1.3112,
+      "step": 11070
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.977419347374684e-05,
+      "loss": 1.3223,
+      "step": 11071
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.977411104198264e-05,
+      "loss": 1.3064,
+      "step": 11072
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.977402859520919e-05,
+      "loss": 1.3247,
+      "step": 11073
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.977394613342652e-05,
+      "loss": 1.3029,
+      "step": 11074
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.977386365663463e-05,
+      "loss": 1.2865,
+      "step": 11075
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.97737811648336e-05,
+      "loss": 1.2832,
+      "step": 11076
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.977369865802341e-05,
+      "loss": 1.3109,
+      "step": 11077
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.977361613620409e-05,
+      "loss": 1.3117,
+      "step": 11078
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.977353359937569e-05,
+      "loss": 1.3215,
+      "step": 11079
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.97734510475382e-05,
+      "loss": 1.299,
+      "step": 11080
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.977336848069166e-05,
+      "loss": 1.3251,
+      "step": 11081
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.97732858988361e-05,
+      "loss": 1.2723,
+      "step": 11082
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.977320330197155e-05,
+      "loss": 1.348,
+      "step": 11083
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.977312069009802e-05,
+      "loss": 1.3221,
+      "step": 11084
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.977303806321554e-05,
+      "loss": 1.3121,
+      "step": 11085
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.977295542132415e-05,
+      "loss": 1.3037,
+      "step": 11086
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.977287276442385e-05,
+      "loss": 1.3234,
+      "step": 11087
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.977279009251468e-05,
+      "loss": 1.2852,
+      "step": 11088
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.977270740559666e-05,
+      "loss": 1.3317,
+      "step": 11089
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.977262470366981e-05,
+      "loss": 1.3166,
+      "step": 11090
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.977254198673416e-05,
+      "loss": 1.3291,
+      "step": 11091
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.977245925478975e-05,
+      "loss": 1.3048,
+      "step": 11092
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.977237650783658e-05,
+      "loss": 1.3156,
+      "step": 11093
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.977229374587469e-05,
+      "loss": 1.2995,
+      "step": 11094
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.97722109689041e-05,
+      "loss": 1.3168,
+      "step": 11095
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.977212817692482e-05,
+      "loss": 1.3064,
+      "step": 11096
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.977204536993691e-05,
+      "loss": 1.3269,
+      "step": 11097
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 9.977196254794039e-05,
+      "loss": 1.3259,
+      "step": 11098
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 9.977187971093525e-05,
+      "loss": 1.3106,
+      "step": 11099
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 9.977179685892154e-05,
+      "loss": 1.298,
+      "step": 11100
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 9.977171399189929e-05,
+      "loss": 1.2776,
+      "step": 11101
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 9.977163110986848e-05,
+      "loss": 1.314,
+      "step": 11102
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 9.977154821282921e-05,
+      "loss": 1.3013,
+      "step": 11103
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 9.977146530078146e-05,
+      "loss": 1.3014,
+      "step": 11104
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 9.977138237372526e-05,
+      "loss": 1.2804,
+      "step": 11105
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 9.977129943166061e-05,
+      "loss": 1.3125,
+      "step": 11106
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 9.97712164745876e-05,
+      "loss": 1.288,
+      "step": 11107
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 9.977113350250619e-05,
+      "loss": 1.2742,
+      "step": 11108
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 9.977105051541644e-05,
+      "loss": 1.2587,
+      "step": 11109
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 9.977096751331837e-05,
+      "loss": 1.3198,
+      "step": 11110
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 9.977088449621198e-05,
+      "loss": 1.3594,
+      "step": 11111
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 9.977080146409733e-05,
+      "loss": 1.307,
+      "step": 11112
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 9.977071841697443e-05,
+      "loss": 1.3094,
+      "step": 11113
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 9.97706353548433e-05,
+      "loss": 1.3085,
+      "step": 11114
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 9.977055227770399e-05,
+      "loss": 1.2646,
+      "step": 11115
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 9.97704691855565e-05,
+      "loss": 1.2826,
+      "step": 11116
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 9.977038607840085e-05,
+      "loss": 1.3251,
+      "step": 11117
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 9.977030295623708e-05,
+      "loss": 1.2726,
+      "step": 11118
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 9.97702198190652e-05,
+      "loss": 1.3288,
+      "step": 11119
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 9.977013666688526e-05,
+      "loss": 1.3026,
+      "step": 11120
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 9.977005349969728e-05,
+      "loss": 1.3155,
+      "step": 11121
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 9.976997031750126e-05,
+      "loss": 1.3028,
+      "step": 11122
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 9.976988712029725e-05,
+      "loss": 1.2857,
+      "step": 11123
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 9.976980390808527e-05,
+      "loss": 1.3301,
+      "step": 11124
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 9.976972068086533e-05,
+      "loss": 1.3043,
+      "step": 11125
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 9.976963743863748e-05,
+      "loss": 1.3171,
+      "step": 11126
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 9.976955418140172e-05,
+      "loss": 1.2956,
+      "step": 11127
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 9.976947090915809e-05,
+      "loss": 1.2913,
+      "step": 11128
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 9.976938762190662e-05,
+      "loss": 1.3024,
+      "step": 11129
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 9.976930431964732e-05,
+      "loss": 1.2928,
+      "step": 11130
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 9.976922100238021e-05,
+      "loss": 1.2962,
+      "step": 11131
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 9.976913767010536e-05,
+      "loss": 1.3174,
+      "step": 11132
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 9.976905432282274e-05,
+      "loss": 1.292,
+      "step": 11133
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 9.97689709605324e-05,
+      "loss": 1.3052,
+      "step": 11134
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 9.976888758323437e-05,
+      "loss": 1.3237,
+      "step": 11135
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 9.976880419092866e-05,
+      "loss": 1.2947,
+      "step": 11136
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 9.976872078361531e-05,
+      "loss": 1.3069,
+      "step": 11137
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 9.976863736129434e-05,
+      "loss": 1.2786,
+      "step": 11138
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 9.976855392396576e-05,
+      "loss": 1.3152,
+      "step": 11139
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 9.976847047162963e-05,
+      "loss": 1.337,
+      "step": 11140
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 9.976838700428593e-05,
+      "loss": 1.3065,
+      "step": 11141
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 9.976830352193473e-05,
+      "loss": 1.3078,
+      "step": 11142
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 9.976822002457602e-05,
+      "loss": 1.2857,
+      "step": 11143
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 9.976813651220985e-05,
+      "loss": 1.2931,
+      "step": 11144
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 9.976805298483623e-05,
+      "loss": 1.3182,
+      "step": 11145
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 9.976796944245518e-05,
+      "loss": 1.2822,
+      "step": 11146
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 9.976788588506674e-05,
+      "loss": 1.3306,
+      "step": 11147
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 9.976780231267094e-05,
+      "loss": 1.3217,
+      "step": 11148
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 9.976771872526777e-05,
+      "loss": 1.3333,
+      "step": 11149
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 9.976763512285731e-05,
+      "loss": 1.2993,
+      "step": 11150
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 9.976755150543955e-05,
+      "loss": 1.3159,
+      "step": 11151
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 9.976746787301452e-05,
+      "loss": 1.3211,
+      "step": 11152
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 9.976738422558223e-05,
+      "loss": 1.3167,
+      "step": 11153
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 9.976730056314274e-05,
+      "loss": 1.2995,
+      "step": 11154
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 9.976721688569604e-05,
+      "loss": 1.2808,
+      "step": 11155
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 9.976713319324218e-05,
+      "loss": 1.2976,
+      "step": 11156
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 9.976704948578119e-05,
+      "loss": 1.2846,
+      "step": 11157
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 9.976696576331306e-05,
+      "loss": 1.3391,
+      "step": 11158
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 9.976688202583786e-05,
+      "loss": 1.3396,
+      "step": 11159
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 9.976679827335557e-05,
+      "loss": 1.3014,
+      "step": 11160
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 9.976671450586625e-05,
+      "loss": 1.2964,
+      "step": 11161
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 9.976663072336993e-05,
+      "loss": 1.3099,
+      "step": 11162
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 9.97665469258666e-05,
+      "loss": 1.2798,
+      "step": 11163
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 9.97664631133563e-05,
+      "loss": 1.315,
+      "step": 11164
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 9.976637928583907e-05,
+      "loss": 1.3303,
+      "step": 11165
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 9.976629544331492e-05,
+      "loss": 1.3108,
+      "step": 11166
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 9.976621158578388e-05,
+      "loss": 1.2928,
+      "step": 11167
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 9.976612771324598e-05,
+      "loss": 1.3089,
+      "step": 11168
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 9.976604382570123e-05,
+      "loss": 1.3131,
+      "step": 11169
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 9.976595992314967e-05,
+      "loss": 1.3413,
+      "step": 11170
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 9.976587600559131e-05,
+      "loss": 1.2879,
+      "step": 11171
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 9.97657920730262e-05,
+      "loss": 1.2751,
+      "step": 11172
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 9.976570812545436e-05,
+      "loss": 1.2813,
+      "step": 11173
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 9.976562416287578e-05,
+      "loss": 1.3184,
+      "step": 11174
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 9.976554018529055e-05,
+      "loss": 1.3456,
+      "step": 11175
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 9.976545619269861e-05,
+      "loss": 1.3122,
+      "step": 11176
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 9.976537218510007e-05,
+      "loss": 1.3232,
+      "step": 11177
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 9.97652881624949e-05,
+      "loss": 1.3111,
+      "step": 11178
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 9.976520412488315e-05,
+      "loss": 1.3004,
+      "step": 11179
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 9.976512007226484e-05,
+      "loss": 1.3024,
+      "step": 11180
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 9.976503600464e-05,
+      "loss": 1.291,
+      "step": 11181
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 9.976495192200863e-05,
+      "loss": 1.3142,
+      "step": 11182
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 9.97648678243708e-05,
+      "loss": 1.2817,
+      "step": 11183
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 9.97647837117265e-05,
+      "loss": 1.2995,
+      "step": 11184
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 9.976469958407574e-05,
+      "loss": 1.3204,
+      "step": 11185
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 9.97646154414186e-05,
+      "loss": 1.3351,
+      "step": 11186
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 9.976453128375508e-05,
+      "loss": 1.3071,
+      "step": 11187
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 9.976444711108519e-05,
+      "loss": 1.3057,
+      "step": 11188
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 9.976436292340897e-05,
+      "loss": 1.3107,
+      "step": 11189
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 9.976427872072645e-05,
+      "loss": 1.3352,
+      "step": 11190
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 9.976419450303763e-05,
+      "loss": 1.3075,
+      "step": 11191
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 9.976411027034256e-05,
+      "loss": 1.3069,
+      "step": 11192
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 9.976402602264127e-05,
+      "loss": 1.3213,
+      "step": 11193
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 9.976394175993375e-05,
+      "loss": 1.321,
+      "step": 11194
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 9.976385748222008e-05,
+      "loss": 1.3173,
+      "step": 11195
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 9.976377318950023e-05,
+      "loss": 1.3236,
+      "step": 11196
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 9.976368888177427e-05,
+      "loss": 1.3035,
+      "step": 11197
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 9.976360455904219e-05,
+      "loss": 1.2902,
+      "step": 11198
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 9.976352022130404e-05,
+      "loss": 1.3192,
+      "step": 11199
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 9.976343586855983e-05,
+      "loss": 1.3162,
+      "step": 11200
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 9.976335150080959e-05,
+      "loss": 1.2931,
+      "step": 11201
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 9.976326711805336e-05,
+      "loss": 1.3376,
+      "step": 11202
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 9.976318272029115e-05,
+      "loss": 1.3109,
+      "step": 11203
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 9.976309830752297e-05,
+      "loss": 1.3253,
+      "step": 11204
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 9.976301387974889e-05,
+      "loss": 1.317,
+      "step": 11205
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 9.97629294369689e-05,
+      "loss": 1.3311,
+      "step": 11206
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 9.976284497918303e-05,
+      "loss": 1.3031,
+      "step": 11207
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 9.976276050639131e-05,
+      "loss": 1.3153,
+      "step": 11208
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 9.976267601859378e-05,
+      "loss": 1.3173,
+      "step": 11209
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 9.976259151579042e-05,
+      "loss": 1.2998,
+      "step": 11210
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 9.976250699798132e-05,
+      "loss": 1.339,
+      "step": 11211
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 9.976242246516644e-05,
+      "loss": 1.3281,
+      "step": 11212
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 9.976233791734586e-05,
+      "loss": 1.2935,
+      "step": 11213
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 9.976225335451959e-05,
+      "loss": 1.3181,
+      "step": 11214
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 9.976216877668762e-05,
+      "loss": 1.3361,
+      "step": 11215
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 9.976208418385003e-05,
+      "loss": 1.3307,
+      "step": 11216
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 9.976199957600681e-05,
+      "loss": 1.2829,
+      "step": 11217
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 9.9761914953158e-05,
+      "loss": 1.3037,
+      "step": 11218
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 9.976183031530362e-05,
+      "loss": 1.3151,
+      "step": 11219
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 9.976174566244369e-05,
+      "loss": 1.2741,
+      "step": 11220
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 9.976166099457824e-05,
+      "loss": 1.2986,
+      "step": 11221
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.976157631170731e-05,
+      "loss": 1.3004,
+      "step": 11222
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.97614916138309e-05,
+      "loss": 1.3135,
+      "step": 11223
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.976140690094906e-05,
+      "loss": 1.3119,
+      "step": 11224
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.97613221730618e-05,
+      "loss": 1.3054,
+      "step": 11225
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.976123743016914e-05,
+      "loss": 1.2867,
+      "step": 11226
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.976115267227112e-05,
+      "loss": 1.3221,
+      "step": 11227
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.976106789936776e-05,
+      "loss": 1.2995,
+      "step": 11228
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.97609831114591e-05,
+      "loss": 1.3237,
+      "step": 11229
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.976089830854514e-05,
+      "loss": 1.2844,
+      "step": 11230
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.97608134906259e-05,
+      "loss": 1.285,
+      "step": 11231
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.976072865770146e-05,
+      "loss": 1.2893,
+      "step": 11232
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.976064380977176e-05,
+      "loss": 1.3019,
+      "step": 11233
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.976055894683692e-05,
+      "loss": 1.2972,
+      "step": 11234
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.97604740688969e-05,
+      "loss": 1.3219,
+      "step": 11235
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.976038917595174e-05,
+      "loss": 1.3282,
+      "step": 11236
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.976030426800148e-05,
+      "loss": 1.2875,
+      "step": 11237
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.976021934504613e-05,
+      "loss": 1.2873,
+      "step": 11238
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.976013440708574e-05,
+      "loss": 1.3084,
+      "step": 11239
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.976004945412029e-05,
+      "loss": 1.2832,
+      "step": 11240
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.975996448614985e-05,
+      "loss": 1.3001,
+      "step": 11241
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.975987950317442e-05,
+      "loss": 1.2668,
+      "step": 11242
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.975979450519403e-05,
+      "loss": 1.308,
+      "step": 11243
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.975970949220874e-05,
+      "loss": 1.313,
+      "step": 11244
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.975962446421851e-05,
+      "loss": 1.3259,
+      "step": 11245
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.975953942122342e-05,
+      "loss": 1.3273,
+      "step": 11246
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.975945436322348e-05,
+      "loss": 1.2956,
+      "step": 11247
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.97593692902187e-05,
+      "loss": 1.291,
+      "step": 11248
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.975928420220912e-05,
+      "loss": 1.3049,
+      "step": 11249
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.975919909919478e-05,
+      "loss": 1.2829,
+      "step": 11250
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.975911398117567e-05,
+      "loss": 1.3226,
+      "step": 11251
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.975902884815185e-05,
+      "loss": 1.3279,
+      "step": 11252
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.975894370012332e-05,
+      "loss": 1.3346,
+      "step": 11253
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.975885853709012e-05,
+      "loss": 1.3158,
+      "step": 11254
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.975877335905228e-05,
+      "loss": 1.3267,
+      "step": 11255
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.975868816600982e-05,
+      "loss": 1.3075,
+      "step": 11256
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.975860295796275e-05,
+      "loss": 1.2747,
+      "step": 11257
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.975851773491111e-05,
+      "loss": 1.3103,
+      "step": 11258
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.975843249685493e-05,
+      "loss": 1.2943,
+      "step": 11259
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.975834724379424e-05,
+      "loss": 1.288,
+      "step": 11260
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.975826197572904e-05,
+      "loss": 1.3086,
+      "step": 11261
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.975817669265938e-05,
+      "loss": 1.3164,
+      "step": 11262
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.975809139458528e-05,
+      "loss": 1.3157,
+      "step": 11263
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.975800608150676e-05,
+      "loss": 1.3179,
+      "step": 11264
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.975792075342385e-05,
+      "loss": 1.3474,
+      "step": 11265
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.975783541033657e-05,
+      "loss": 1.2934,
+      "step": 11266
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.975775005224496e-05,
+      "loss": 1.3358,
+      "step": 11267
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.975766467914903e-05,
+      "loss": 1.2748,
+      "step": 11268
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.975757929104882e-05,
+      "loss": 1.3295,
+      "step": 11269
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.975749388794434e-05,
+      "loss": 1.2861,
+      "step": 11270
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.975740846983563e-05,
+      "loss": 1.278,
+      "step": 11271
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.975732303672271e-05,
+      "loss": 1.3375,
+      "step": 11272
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.97572375886056e-05,
+      "loss": 1.3375,
+      "step": 11273
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.975715212548433e-05,
+      "loss": 1.3122,
+      "step": 11274
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.975706664735894e-05,
+      "loss": 1.3212,
+      "step": 11275
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.975698115422942e-05,
+      "loss": 1.336,
+      "step": 11276
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.975689564609582e-05,
+      "loss": 1.317,
+      "step": 11277
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.975681012295819e-05,
+      "loss": 1.3152,
+      "step": 11278
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.97567245848165e-05,
+      "loss": 1.2721,
+      "step": 11279
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.975663903167083e-05,
+      "loss": 1.3103,
+      "step": 11280
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.975655346352117e-05,
+      "loss": 1.3063,
+      "step": 11281
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.975646788036757e-05,
+      "loss": 1.3285,
+      "step": 11282
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.975638228221002e-05,
+      "loss": 1.2763,
+      "step": 11283
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 9.975629666904859e-05,
+      "loss": 1.2704,
+      "step": 11284
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 9.975621104088328e-05,
+      "loss": 1.2982,
+      "step": 11285
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 9.97561253977141e-05,
+      "loss": 1.3307,
+      "step": 11286
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 9.975603973954112e-05,
+      "loss": 1.2703,
+      "step": 11287
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 9.975595406636435e-05,
+      "loss": 1.3154,
+      "step": 11288
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 9.975586837818378e-05,
+      "loss": 1.3156,
+      "step": 11289
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 9.975578267499949e-05,
+      "loss": 1.3288,
+      "step": 11290
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 9.975569695681146e-05,
+      "loss": 1.2832,
+      "step": 11291
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 9.975561122361976e-05,
+      "loss": 1.3183,
+      "step": 11292
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 9.975552547542436e-05,
+      "loss": 1.298,
+      "step": 11293
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 9.975543971222534e-05,
+      "loss": 1.3269,
+      "step": 11294
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 9.975535393402268e-05,
+      "loss": 1.3112,
+      "step": 11295
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 9.975526814081646e-05,
+      "loss": 1.3405,
+      "step": 11296
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 9.975518233260665e-05,
+      "loss": 1.2904,
+      "step": 11297
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 9.975509650939331e-05,
+      "loss": 1.3451,
+      "step": 11298
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 9.975501067117648e-05,
+      "loss": 1.3279,
+      "step": 11299
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 9.975492481795612e-05,
+      "loss": 1.3245,
+      "step": 11300
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 9.975483894973232e-05,
+      "loss": 1.2953,
+      "step": 11301
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 9.975475306650509e-05,
+      "loss": 1.3025,
+      "step": 11302
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 9.975466716827444e-05,
+      "loss": 1.3319,
+      "step": 11303
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 9.975458125504042e-05,
+      "loss": 1.3199,
+      "step": 11304
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 9.975449532680303e-05,
+      "loss": 1.2759,
+      "step": 11305
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 9.975440938356231e-05,
+      "loss": 1.2828,
+      "step": 11306
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 9.975432342531827e-05,
+      "loss": 1.3317,
+      "step": 11307
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 9.975423745207098e-05,
+      "loss": 1.3158,
+      "step": 11308
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 9.975415146382042e-05,
+      "loss": 1.2924,
+      "step": 11309
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 9.975406546056662e-05,
+      "loss": 1.2758,
+      "step": 11310
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 9.975397944230962e-05,
+      "loss": 1.3291,
+      "step": 11311
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 9.975389340904946e-05,
+      "loss": 1.3113,
+      "step": 11312
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 9.975380736078614e-05,
+      "loss": 1.2906,
+      "step": 11313
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 9.97537212975197e-05,
+      "loss": 1.3156,
+      "step": 11314
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 9.975363521925016e-05,
+      "loss": 1.3375,
+      "step": 11315
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 9.975354912597755e-05,
+      "loss": 1.2906,
+      "step": 11316
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 9.97534630177019e-05,
+      "loss": 1.3263,
+      "step": 11317
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 9.975337689442321e-05,
+      "loss": 1.3253,
+      "step": 11318
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 9.975329075614153e-05,
+      "loss": 1.2831,
+      "step": 11319
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 9.975320460285689e-05,
+      "loss": 1.3193,
+      "step": 11320
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 9.97531184345693e-05,
+      "loss": 1.3062,
+      "step": 11321
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 9.97530322512788e-05,
+      "loss": 1.3017,
+      "step": 11322
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 9.97529460529854e-05,
+      "loss": 1.3167,
+      "step": 11323
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 9.975285983968914e-05,
+      "loss": 1.295,
+      "step": 11324
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 9.975277361139004e-05,
+      "loss": 1.3169,
+      "step": 11325
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 9.975268736808814e-05,
+      "loss": 1.2931,
+      "step": 11326
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 9.975260110978344e-05,
+      "loss": 1.3348,
+      "step": 11327
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 9.975251483647599e-05,
+      "loss": 1.3195,
+      "step": 11328
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 9.975242854816577e-05,
+      "loss": 1.3112,
+      "step": 11329
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 9.975234224485288e-05,
+      "loss": 1.3287,
+      "step": 11330
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 9.97522559265373e-05,
+      "loss": 1.2936,
+      "step": 11331
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 9.975216959321906e-05,
+      "loss": 1.3366,
+      "step": 11332
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 9.975208324489817e-05,
+      "loss": 1.295,
+      "step": 11333
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 9.97519968815747e-05,
+      "loss": 1.288,
+      "step": 11334
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 9.975191050324866e-05,
+      "loss": 1.3222,
+      "step": 11335
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 9.975182410992005e-05,
+      "loss": 1.3252,
+      "step": 11336
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 9.97517377015889e-05,
+      "loss": 1.3239,
+      "step": 11337
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 9.975165127825527e-05,
+      "loss": 1.3793,
+      "step": 11338
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 9.975156483991917e-05,
+      "loss": 1.2766,
+      "step": 11339
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 9.975147838658061e-05,
+      "loss": 1.3384,
+      "step": 11340
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 9.975139191823963e-05,
+      "loss": 1.3205,
+      "step": 11341
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 9.975130543489626e-05,
+      "loss": 1.3297,
+      "step": 11342
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 9.97512189365505e-05,
+      "loss": 1.2972,
+      "step": 11343
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 9.975113242320242e-05,
+      "loss": 1.3291,
+      "step": 11344
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 9.9751045894852e-05,
+      "loss": 1.2776,
+      "step": 11345
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 9.975095935149931e-05,
+      "loss": 1.3007,
+      "step": 11346
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 9.975087279314434e-05,
+      "loss": 1.325,
+      "step": 11347
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 9.975078621978714e-05,
+      "loss": 1.288,
+      "step": 11348
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 9.975069963142771e-05,
+      "loss": 1.3306,
+      "step": 11349
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 9.975061302806612e-05,
+      "loss": 1.2973,
+      "step": 11350
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 9.975052640970235e-05,
+      "loss": 1.283,
+      "step": 11351
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 9.975043977633644e-05,
+      "loss": 1.321,
+      "step": 11352
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 9.975035312796843e-05,
+      "loss": 1.3002,
+      "step": 11353
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 9.975026646459834e-05,
+      "loss": 1.2712,
+      "step": 11354
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 9.975017978622619e-05,
+      "loss": 1.3126,
+      "step": 11355
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 9.9750093092852e-05,
+      "loss": 1.2937,
+      "step": 11356
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 9.975000638447582e-05,
+      "loss": 1.3207,
+      "step": 11357
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 9.974991966109765e-05,
+      "loss": 1.3327,
+      "step": 11358
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 9.974983292271753e-05,
+      "loss": 1.2709,
+      "step": 11359
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 9.97497461693355e-05,
+      "loss": 1.3018,
+      "step": 11360
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 9.974965940095156e-05,
+      "loss": 1.3057,
+      "step": 11361
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 9.974957261756573e-05,
+      "loss": 1.3293,
+      "step": 11362
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 9.974948581917807e-05,
+      "loss": 1.3214,
+      "step": 11363
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 9.97493990057886e-05,
+      "loss": 1.3101,
+      "step": 11364
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 9.974931217739732e-05,
+      "loss": 1.3049,
+      "step": 11365
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 9.974922533400426e-05,
+      "loss": 1.3084,
+      "step": 11366
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 9.974913847560947e-05,
+      "loss": 1.2962,
+      "step": 11367
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 9.974905160221295e-05,
+      "loss": 1.2915,
+      "step": 11368
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 9.974896471381477e-05,
+      "loss": 1.3415,
+      "step": 11369
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 9.974887781041489e-05,
+      "loss": 1.2542,
+      "step": 11370
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 9.974879089201339e-05,
+      "loss": 1.3158,
+      "step": 11371
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 9.974870395861028e-05,
+      "loss": 1.2945,
+      "step": 11372
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 9.974861701020558e-05,
+      "loss": 1.3144,
+      "step": 11373
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 9.974853004679932e-05,
+      "loss": 1.321,
+      "step": 11374
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 9.974844306839153e-05,
+      "loss": 1.3089,
+      "step": 11375
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 9.974835607498222e-05,
+      "loss": 1.3057,
+      "step": 11376
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 9.974826906657145e-05,
+      "loss": 1.302,
+      "step": 11377
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 9.974818204315921e-05,
+      "loss": 1.2846,
+      "step": 11378
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 9.974809500474556e-05,
+      "loss": 1.2706,
+      "step": 11379
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 9.974800795133049e-05,
+      "loss": 1.3027,
+      "step": 11380
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 9.974792088291405e-05,
+      "loss": 1.283,
+      "step": 11381
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 9.974783379949626e-05,
+      "loss": 1.3095,
+      "step": 11382
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 9.974774670107716e-05,
+      "loss": 1.3316,
+      "step": 11383
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 9.974765958765676e-05,
+      "loss": 1.3068,
+      "step": 11384
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 9.974757245923506e-05,
+      "loss": 1.2738,
+      "step": 11385
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 9.974748531581213e-05,
+      "loss": 1.3147,
+      "step": 11386
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 9.974739815738799e-05,
+      "loss": 1.3115,
+      "step": 11387
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 9.974731098396267e-05,
+      "loss": 1.303,
+      "step": 11388
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 9.974722379553616e-05,
+      "loss": 1.3246,
+      "step": 11389
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 9.974713659210852e-05,
+      "loss": 1.276,
+      "step": 11390
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 9.974704937367977e-05,
+      "loss": 1.3177,
+      "step": 11391
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 9.974696214024993e-05,
+      "loss": 1.3254,
+      "step": 11392
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 9.974687489181903e-05,
+      "loss": 1.3109,
+      "step": 11393
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 9.974678762838709e-05,
+      "loss": 1.2969,
+      "step": 11394
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 9.974670034995415e-05,
+      "loss": 1.3262,
+      "step": 11395
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 9.974661305652024e-05,
+      "loss": 1.3168,
+      "step": 11396
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 9.974652574808536e-05,
+      "loss": 1.3222,
+      "step": 11397
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 9.974643842464955e-05,
+      "loss": 1.3204,
+      "step": 11398
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 9.974635108621284e-05,
+      "loss": 1.3074,
+      "step": 11399
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 9.974626373277526e-05,
+      "loss": 1.2922,
+      "step": 11400
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 9.974617636433683e-05,
+      "loss": 1.3307,
+      "step": 11401
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 9.974608898089756e-05,
+      "loss": 1.3274,
+      "step": 11402
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 9.974600158245752e-05,
+      "loss": 1.3402,
+      "step": 11403
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 9.974591416901669e-05,
+      "loss": 1.308,
+      "step": 11404
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 9.974582674057512e-05,
+      "loss": 1.3269,
+      "step": 11405
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 9.974573929713282e-05,
+      "loss": 1.3497,
+      "step": 11406
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 9.974565183868985e-05,
+      "loss": 1.2809,
+      "step": 11407
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 9.97455643652462e-05,
+      "loss": 1.2837,
+      "step": 11408
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.974547687680192e-05,
+      "loss": 1.3148,
+      "step": 11409
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.974538937335702e-05,
+      "loss": 1.3297,
+      "step": 11410
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.974530185491154e-05,
+      "loss": 1.2695,
+      "step": 11411
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.974521432146551e-05,
+      "loss": 1.3203,
+      "step": 11412
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.974512677301892e-05,
+      "loss": 1.3244,
+      "step": 11413
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.974503920957184e-05,
+      "loss": 1.2712,
+      "step": 11414
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.974495163112428e-05,
+      "loss": 1.2968,
+      "step": 11415
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.974486403767627e-05,
+      "loss": 1.3044,
+      "step": 11416
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.974477642922781e-05,
+      "loss": 1.3189,
+      "step": 11417
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.974468880577897e-05,
+      "loss": 1.2708,
+      "step": 11418
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.974460116732976e-05,
+      "loss": 1.2983,
+      "step": 11419
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.97445135138802e-05,
+      "loss": 1.2925,
+      "step": 11420
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.974442584543029e-05,
+      "loss": 1.2714,
+      "step": 11421
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.97443381619801e-05,
+      "loss": 1.2777,
+      "step": 11422
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.974425046352965e-05,
+      "loss": 1.3401,
+      "step": 11423
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.974416275007896e-05,
+      "loss": 1.3379,
+      "step": 11424
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.974407502162804e-05,
+      "loss": 1.2944,
+      "step": 11425
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.974398727817693e-05,
+      "loss": 1.3063,
+      "step": 11426
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.974389951972567e-05,
+      "loss": 1.3207,
+      "step": 11427
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.974381174627427e-05,
+      "loss": 1.3118,
+      "step": 11428
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.974372395782274e-05,
+      "loss": 1.3002,
+      "step": 11429
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.974363615437115e-05,
+      "loss": 1.3153,
+      "step": 11430
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.97435483359195e-05,
+      "loss": 1.3478,
+      "step": 11431
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.97434605024678e-05,
+      "loss": 1.3034,
+      "step": 11432
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.974337265401612e-05,
+      "loss": 1.3006,
+      "step": 11433
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.974328479056445e-05,
+      "loss": 1.2857,
+      "step": 11434
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.974319691211282e-05,
+      "loss": 1.2885,
+      "step": 11435
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.974310901866128e-05,
+      "loss": 1.3189,
+      "step": 11436
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.974302111020984e-05,
+      "loss": 1.3239,
+      "step": 11437
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.974293318675851e-05,
+      "loss": 1.326,
+      "step": 11438
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.974284524830735e-05,
+      "loss": 1.3042,
+      "step": 11439
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.974275729485638e-05,
+      "loss": 1.2905,
+      "step": 11440
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.97426693264056e-05,
+      "loss": 1.306,
+      "step": 11441
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.974258134295507e-05,
+      "loss": 1.2788,
+      "step": 11442
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.974249334450478e-05,
+      "loss": 1.3161,
+      "step": 11443
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.97424053310548e-05,
+      "loss": 1.2797,
+      "step": 11444
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.974231730260511e-05,
+      "loss": 1.3056,
+      "step": 11445
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.974222925915578e-05,
+      "loss": 1.2718,
+      "step": 11446
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.97421412007068e-05,
+      "loss": 1.3122,
+      "step": 11447
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.974205312725824e-05,
+      "loss": 1.2924,
+      "step": 11448
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.974196503881009e-05,
+      "loss": 1.3185,
+      "step": 11449
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.974187693536237e-05,
+      "loss": 1.289,
+      "step": 11450
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.974178881691513e-05,
+      "loss": 1.3302,
+      "step": 11451
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.974170068346839e-05,
+      "loss": 1.3341,
+      "step": 11452
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.974161253502218e-05,
+      "loss": 1.299,
+      "step": 11453
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.974152437157651e-05,
+      "loss": 1.2951,
+      "step": 11454
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.974143619313144e-05,
+      "loss": 1.3239,
+      "step": 11455
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.974134799968698e-05,
+      "loss": 1.3066,
+      "step": 11456
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.974125979124313e-05,
+      "loss": 1.325,
+      "step": 11457
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.974117156779994e-05,
+      "loss": 1.3096,
+      "step": 11458
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.974108332935746e-05,
+      "loss": 1.3061,
+      "step": 11459
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.974099507591567e-05,
+      "loss": 1.327,
+      "step": 11460
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.974090680747463e-05,
+      "loss": 1.2887,
+      "step": 11461
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.974081852403434e-05,
+      "loss": 1.3523,
+      "step": 11462
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.974073022559485e-05,
+      "loss": 1.275,
+      "step": 11463
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.974064191215619e-05,
+      "loss": 1.3004,
+      "step": 11464
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.974055358371837e-05,
+      "loss": 1.3349,
+      "step": 11465
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.974046524028142e-05,
+      "loss": 1.3029,
+      "step": 11466
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.974037688184537e-05,
+      "loss": 1.3,
+      "step": 11467
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.974028850841024e-05,
+      "loss": 1.3314,
+      "step": 11468
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.974020011997607e-05,
+      "loss": 1.3162,
+      "step": 11469
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.974011171654287e-05,
+      "loss": 1.2867,
+      "step": 11470
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 9.974002329811068e-05,
+      "loss": 1.3076,
+      "step": 11471
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 9.973993486467951e-05,
+      "loss": 1.3445,
+      "step": 11472
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 9.973984641624943e-05,
+      "loss": 1.2932,
+      "step": 11473
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 9.97397579528204e-05,
+      "loss": 1.2969,
+      "step": 11474
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 9.973966947439251e-05,
+      "loss": 1.2933,
+      "step": 11475
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 9.973958098096574e-05,
+      "loss": 1.2806,
+      "step": 11476
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 9.973949247254014e-05,
+      "loss": 1.274,
+      "step": 11477
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 9.973940394911574e-05,
+      "loss": 1.3319,
+      "step": 11478
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 9.973931541069255e-05,
+      "loss": 1.3376,
+      "step": 11479
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 9.97392268572706e-05,
+      "loss": 1.3001,
+      "step": 11480
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 9.973913828884994e-05,
+      "loss": 1.3196,
+      "step": 11481
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 9.973904970543058e-05,
+      "loss": 1.3273,
+      "step": 11482
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 9.973896110701252e-05,
+      "loss": 1.2729,
+      "step": 11483
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 9.973887249359583e-05,
+      "loss": 1.2942,
+      "step": 11484
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 9.973878386518051e-05,
+      "loss": 1.2515,
+      "step": 11485
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 9.973869522176662e-05,
+      "loss": 1.3154,
+      "step": 11486
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 9.973860656335414e-05,
+      "loss": 1.325,
+      "step": 11487
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 9.973851788994313e-05,
+      "loss": 1.2933,
+      "step": 11488
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 9.97384292015336e-05,
+      "loss": 1.2996,
+      "step": 11489
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 9.973834049812558e-05,
+      "loss": 1.2635,
+      "step": 11490
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 9.973825177971909e-05,
+      "loss": 1.2962,
+      "step": 11491
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 9.973816304631417e-05,
+      "loss": 1.2863,
+      "step": 11492
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 9.973807429791087e-05,
+      "loss": 1.2998,
+      "step": 11493
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 9.973798553450917e-05,
+      "loss": 1.2725,
+      "step": 11494
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 9.973789675610912e-05,
+      "loss": 1.3146,
+      "step": 11495
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 9.973780796271074e-05,
+      "loss": 1.3029,
+      "step": 11496
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 9.973771915431406e-05,
+      "loss": 1.2823,
+      "step": 11497
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 9.97376303309191e-05,
+      "loss": 1.27,
+      "step": 11498
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 9.97375414925259e-05,
+      "loss": 1.3005,
+      "step": 11499
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 9.97374526391345e-05,
+      "loss": 1.2978,
+      "step": 11500
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 9.973736377074487e-05,
+      "loss": 1.3084,
+      "step": 11501
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 9.97372748873571e-05,
+      "loss": 1.3025,
+      "step": 11502
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 9.97371859889712e-05,
+      "loss": 1.2996,
+      "step": 11503
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 9.973709707558716e-05,
+      "loss": 1.3337,
+      "step": 11504
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 9.973700814720505e-05,
+      "loss": 1.3173,
+      "step": 11505
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 9.973691920382488e-05,
+      "loss": 1.3203,
+      "step": 11506
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 9.973683024544666e-05,
+      "loss": 1.2863,
+      "step": 11507
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 9.973674127207047e-05,
+      "loss": 1.3117,
+      "step": 11508
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 9.973665228369626e-05,
+      "loss": 1.2902,
+      "step": 11509
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 9.973656328032413e-05,
+      "loss": 1.3367,
+      "step": 11510
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 9.973647426195407e-05,
+      "loss": 1.3084,
+      "step": 11511
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 9.97363852285861e-05,
+      "loss": 1.301,
+      "step": 11512
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 9.973629618022026e-05,
+      "loss": 1.3197,
+      "step": 11513
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 9.973620711685659e-05,
+      "loss": 1.3242,
+      "step": 11514
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 9.97361180384951e-05,
+      "loss": 1.2986,
+      "step": 11515
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 9.973602894513582e-05,
+      "loss": 1.3142,
+      "step": 11516
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 9.973593983677877e-05,
+      "loss": 1.2731,
+      "step": 11517
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 9.973585071342399e-05,
+      "loss": 1.2766,
+      "step": 11518
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 9.97357615750715e-05,
+      "loss": 1.3374,
+      "step": 11519
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 9.973567242172131e-05,
+      "loss": 1.2968,
+      "step": 11520
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 9.973558325337349e-05,
+      "loss": 1.3266,
+      "step": 11521
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 9.973549407002802e-05,
+      "loss": 1.2596,
+      "step": 11522
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 9.973540487168496e-05,
+      "loss": 1.338,
+      "step": 11523
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 9.97353156583443e-05,
+      "loss": 1.2914,
+      "step": 11524
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 9.973522643000613e-05,
+      "loss": 1.2869,
+      "step": 11525
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 9.973513718667042e-05,
+      "loss": 1.2955,
+      "step": 11526
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 9.973504792833721e-05,
+      "loss": 1.316,
+      "step": 11527
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 9.973495865500654e-05,
+      "loss": 1.303,
+      "step": 11528
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 9.973486936667842e-05,
+      "loss": 1.2919,
+      "step": 11529
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 9.97347800633529e-05,
+      "loss": 1.3345,
+      "step": 11530
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 9.973469074502999e-05,
+      "loss": 1.2846,
+      "step": 11531
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 9.973460141170972e-05,
+      "loss": 1.3147,
+      "step": 11532
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 9.97345120633921e-05,
+      "loss": 1.3132,
+      "step": 11533
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 9.97344227000772e-05,
+      "loss": 1.3019,
+      "step": 11534
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 9.973433332176501e-05,
+      "loss": 1.2685,
+      "step": 11535
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 9.973424392845556e-05,
+      "loss": 1.3089,
+      "step": 11536
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 9.973415452014889e-05,
+      "loss": 1.319,
+      "step": 11537
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 9.973406509684502e-05,
+      "loss": 1.306,
+      "step": 11538
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 9.973397565854399e-05,
+      "loss": 1.2843,
+      "step": 11539
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 9.97338862052458e-05,
+      "loss": 1.2869,
+      "step": 11540
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 9.97337967369505e-05,
+      "loss": 1.3205,
+      "step": 11541
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 9.973370725365812e-05,
+      "loss": 1.3012,
+      "step": 11542
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 9.973361775536866e-05,
+      "loss": 1.2915,
+      "step": 11543
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 9.973352824208217e-05,
+      "loss": 1.2823,
+      "step": 11544
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 9.973343871379868e-05,
+      "loss": 1.2998,
+      "step": 11545
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 9.973334917051818e-05,
+      "loss": 1.312,
+      "step": 11546
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 9.973325961224076e-05,
+      "loss": 1.3381,
+      "step": 11547
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 9.973317003896639e-05,
+      "loss": 1.3089,
+      "step": 11548
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 9.973308045069511e-05,
+      "loss": 1.2946,
+      "step": 11549
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 9.973299084742697e-05,
+      "loss": 1.3076,
+      "step": 11550
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 9.973290122916199e-05,
+      "loss": 1.305,
+      "step": 11551
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 9.973281159590017e-05,
+      "loss": 1.3094,
+      "step": 11552
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 9.973272194764159e-05,
+      "loss": 1.3274,
+      "step": 11553
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 9.973263228438621e-05,
+      "loss": 1.3012,
+      "step": 11554
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 9.97325426061341e-05,
+      "loss": 1.3167,
+      "step": 11555
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 9.973245291288528e-05,
+      "loss": 1.3113,
+      "step": 11556
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 9.973236320463977e-05,
+      "loss": 1.3014,
+      "step": 11557
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 9.97322734813976e-05,
+      "loss": 1.321,
+      "step": 11558
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 9.973218374315882e-05,
+      "loss": 1.3272,
+      "step": 11559
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 9.973209398992342e-05,
+      "loss": 1.2838,
+      "step": 11560
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 9.973200422169145e-05,
+      "loss": 1.3175,
+      "step": 11561
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 9.973191443846291e-05,
+      "loss": 1.2948,
+      "step": 11562
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 9.973182464023787e-05,
+      "loss": 1.242,
+      "step": 11563
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 9.973173482701632e-05,
+      "loss": 1.3211,
+      "step": 11564
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 9.97316449987983e-05,
+      "loss": 1.322,
+      "step": 11565
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 9.973155515558385e-05,
+      "loss": 1.2843,
+      "step": 11566
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 9.973146529737299e-05,
+      "loss": 1.3625,
+      "step": 11567
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 9.973137542416572e-05,
+      "loss": 1.3307,
+      "step": 11568
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 9.973128553596211e-05,
+      "loss": 1.2994,
+      "step": 11569
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 9.973119563276216e-05,
+      "loss": 1.3117,
+      "step": 11570
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 9.973110571456592e-05,
+      "loss": 1.3224,
+      "step": 11571
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 9.973101578137338e-05,
+      "loss": 1.2992,
+      "step": 11572
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 9.97309258331846e-05,
+      "loss": 1.2754,
+      "step": 11573
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 9.973083586999958e-05,
+      "loss": 1.3009,
+      "step": 11574
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 9.973074589181837e-05,
+      "loss": 1.3009,
+      "step": 11575
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 9.973065589864099e-05,
+      "loss": 1.2582,
+      "step": 11576
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 9.973056589046747e-05,
+      "loss": 1.2869,
+      "step": 11577
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 9.973047586729785e-05,
+      "loss": 1.269,
+      "step": 11578
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 9.973038582913211e-05,
+      "loss": 1.2946,
+      "step": 11579
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 9.973029577597033e-05,
+      "loss": 1.2989,
+      "step": 11580
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 9.97302057078125e-05,
+      "loss": 1.3337,
+      "step": 11581
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 9.973011562465867e-05,
+      "loss": 1.2925,
+      "step": 11582
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 9.973002552650885e-05,
+      "loss": 1.3227,
+      "step": 11583
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 9.97299354133631e-05,
+      "loss": 1.3435,
+      "step": 11584
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 9.97298452852214e-05,
+      "loss": 1.3063,
+      "step": 11585
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 9.972975514208381e-05,
+      "loss": 1.3106,
+      "step": 11586
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 9.972966498395036e-05,
+      "loss": 1.3216,
+      "step": 11587
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 9.972957481082105e-05,
+      "loss": 1.2775,
+      "step": 11588
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 9.972948462269593e-05,
+      "loss": 1.3248,
+      "step": 11589
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 9.972939441957501e-05,
+      "loss": 1.3349,
+      "step": 11590
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 9.972930420145834e-05,
+      "loss": 1.272,
+      "step": 11591
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 9.972921396834594e-05,
+      "loss": 1.2892,
+      "step": 11592
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 9.972912372023781e-05,
+      "loss": 1.3111,
+      "step": 11593
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 9.9729033457134e-05,
+      "loss": 1.2953,
+      "step": 11594
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 9.972894317903455e-05,
+      "loss": 1.2978,
+      "step": 11595
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 9.972885288593948e-05,
+      "loss": 1.3062,
+      "step": 11596
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 9.972876257784879e-05,
+      "loss": 1.2799,
+      "step": 11597
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 9.972867225476255e-05,
+      "loss": 1.282,
+      "step": 11598
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 9.972858191668074e-05,
+      "loss": 1.3427,
+      "step": 11599
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 9.972849156360343e-05,
+      "loss": 1.2879,
+      "step": 11600
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 9.972840119553062e-05,
+      "loss": 1.2772,
+      "step": 11601
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 9.972831081246235e-05,
+      "loss": 1.3225,
+      "step": 11602
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 9.972822041439864e-05,
+      "loss": 1.2846,
+      "step": 11603
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 9.972813000133953e-05,
+      "loss": 1.3147,
+      "step": 11604
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 9.972803957328503e-05,
+      "loss": 1.3495,
+      "step": 11605
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 9.972794913023517e-05,
+      "loss": 1.3038,
+      "step": 11606
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 9.972785867219e-05,
+      "loss": 1.3188,
+      "step": 11607
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 9.972776819914952e-05,
+      "loss": 1.2976,
+      "step": 11608
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 9.972767771111378e-05,
+      "loss": 1.3193,
+      "step": 11609
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 9.972758720808278e-05,
+      "loss": 1.3097,
+      "step": 11610
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 9.972749669005656e-05,
+      "loss": 1.2736,
+      "step": 11611
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 9.972740615703516e-05,
+      "loss": 1.3333,
+      "step": 11612
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 9.972731560901859e-05,
+      "loss": 1.2973,
+      "step": 11613
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 9.972722504600688e-05,
+      "loss": 1.3003,
+      "step": 11614
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 9.972713446800008e-05,
+      "loss": 1.3217,
+      "step": 11615
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 9.972704387499819e-05,
+      "loss": 1.3107,
+      "step": 11616
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 9.972695326700124e-05,
+      "loss": 1.3035,
+      "step": 11617
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 9.972686264400926e-05,
+      "loss": 1.3083,
+      "step": 11618
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 9.972677200602229e-05,
+      "loss": 1.3015,
+      "step": 11619
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 9.972668135304034e-05,
+      "loss": 1.3107,
+      "step": 11620
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 9.972659068506344e-05,
+      "loss": 1.3278,
+      "step": 11621
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 9.972650000209164e-05,
+      "loss": 1.3201,
+      "step": 11622
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 9.972640930412495e-05,
+      "loss": 1.3479,
+      "step": 11623
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 9.972631859116338e-05,
+      "loss": 1.3205,
+      "step": 11624
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 9.9726227863207e-05,
+      "loss": 1.3196,
+      "step": 11625
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 9.972613712025578e-05,
+      "loss": 1.3177,
+      "step": 11626
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 9.97260463623098e-05,
+      "loss": 1.3128,
+      "step": 11627
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 9.972595558936905e-05,
+      "loss": 1.3407,
+      "step": 11628
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 9.97258648014336e-05,
+      "loss": 1.3049,
+      "step": 11629
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 9.972577399850343e-05,
+      "loss": 1.3031,
+      "step": 11630
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 9.97256831805786e-05,
+      "loss": 1.3233,
+      "step": 11631
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 9.972559234765912e-05,
+      "loss": 1.321,
+      "step": 11632
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 9.972550149974502e-05,
+      "loss": 1.3113,
+      "step": 11633
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 9.972541063683634e-05,
+      "loss": 1.3164,
+      "step": 11634
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 9.97253197589331e-05,
+      "loss": 1.3139,
+      "step": 11635
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 9.972522886603532e-05,
+      "loss": 1.3144,
+      "step": 11636
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 9.972513795814305e-05,
+      "loss": 1.3062,
+      "step": 11637
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 9.972504703525628e-05,
+      "loss": 1.319,
+      "step": 11638
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 9.972495609737507e-05,
+      "loss": 1.304,
+      "step": 11639
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 9.972486514449941e-05,
+      "loss": 1.2781,
+      "step": 11640
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 9.972477417662937e-05,
+      "loss": 1.2716,
+      "step": 11641
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 9.972468319376498e-05,
+      "loss": 1.268,
+      "step": 11642
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 9.972459219590621e-05,
+      "loss": 1.2833,
+      "step": 11643
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 9.972450118305317e-05,
+      "loss": 1.2655,
+      "step": 11644
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 9.97244101552058e-05,
+      "loss": 1.2886,
+      "step": 11645
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 9.97243191123642e-05,
+      "loss": 1.274,
+      "step": 11646
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 9.972422805452834e-05,
+      "loss": 1.2938,
+      "step": 11647
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 9.97241369816983e-05,
+      "loss": 1.2917,
+      "step": 11648
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 9.972404589387408e-05,
+      "loss": 1.2854,
+      "step": 11649
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 9.97239547910557e-05,
+      "loss": 1.2911,
+      "step": 11650
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 9.97238636732432e-05,
+      "loss": 1.293,
+      "step": 11651
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 9.97237725404366e-05,
+      "loss": 1.3329,
+      "step": 11652
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 9.972368139263595e-05,
+      "loss": 1.3086,
+      "step": 11653
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 9.972359022984123e-05,
+      "loss": 1.3194,
+      "step": 11654
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 9.972349905205252e-05,
+      "loss": 1.2786,
+      "step": 11655
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 9.972340785926983e-05,
+      "loss": 1.3249,
+      "step": 11656
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.972331665149317e-05,
+      "loss": 1.2938,
+      "step": 11657
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.972322542872258e-05,
+      "loss": 1.3316,
+      "step": 11658
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.972313419095809e-05,
+      "loss": 1.308,
+      "step": 11659
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.972304293819973e-05,
+      "loss": 1.3176,
+      "step": 11660
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.972295167044752e-05,
+      "loss": 1.2944,
+      "step": 11661
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.972286038770148e-05,
+      "loss": 1.3099,
+      "step": 11662
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.972276908996166e-05,
+      "loss": 1.3097,
+      "step": 11663
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.972267777722805e-05,
+      "loss": 1.2903,
+      "step": 11664
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.972258644950074e-05,
+      "loss": 1.3378,
+      "step": 11665
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.97224951067797e-05,
+      "loss": 1.3241,
+      "step": 11666
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.972240374906499e-05,
+      "loss": 1.278,
+      "step": 11667
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.972231237635661e-05,
+      "loss": 1.3133,
+      "step": 11668
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.97222209886546e-05,
+      "loss": 1.325,
+      "step": 11669
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.9722129585959e-05,
+      "loss": 1.3037,
+      "step": 11670
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.972203816826983e-05,
+      "loss": 1.2822,
+      "step": 11671
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.97219467355871e-05,
+      "loss": 1.3145,
+      "step": 11672
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.972185528791086e-05,
+      "loss": 1.2965,
+      "step": 11673
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.972176382524114e-05,
+      "loss": 1.3005,
+      "step": 11674
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.972167234757795e-05,
+      "loss": 1.3265,
+      "step": 11675
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.972158085492133e-05,
+      "loss": 1.2764,
+      "step": 11676
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.97214893472713e-05,
+      "loss": 1.2954,
+      "step": 11677
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.97213978246279e-05,
+      "loss": 1.2457,
+      "step": 11678
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.972130628699115e-05,
+      "loss": 1.3022,
+      "step": 11679
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.972121473436106e-05,
+      "loss": 1.2939,
+      "step": 11680
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.972112316673768e-05,
+      "loss": 1.3084,
+      "step": 11681
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.972103158412102e-05,
+      "loss": 1.3174,
+      "step": 11682
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.972093998651114e-05,
+      "loss": 1.2961,
+      "step": 11683
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.972084837390804e-05,
+      "loss": 1.3062,
+      "step": 11684
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.972075674631175e-05,
+      "loss": 1.3322,
+      "step": 11685
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.972066510372231e-05,
+      "loss": 1.2936,
+      "step": 11686
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.972057344613972e-05,
+      "loss": 1.261,
+      "step": 11687
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.972048177356405e-05,
+      "loss": 1.2881,
+      "step": 11688
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.97203900859953e-05,
+      "loss": 1.2888,
+      "step": 11689
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.972029838343348e-05,
+      "loss": 1.334,
+      "step": 11690
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.972020666587867e-05,
+      "loss": 1.2833,
+      "step": 11691
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.972011493333085e-05,
+      "loss": 1.3182,
+      "step": 11692
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.972002318579008e-05,
+      "loss": 1.3295,
+      "step": 11693
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.971993142325635e-05,
+      "loss": 1.2764,
+      "step": 11694
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.971983964572973e-05,
+      "loss": 1.2704,
+      "step": 11695
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.971974785321022e-05,
+      "loss": 1.3114,
+      "step": 11696
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.971965604569787e-05,
+      "loss": 1.2681,
+      "step": 11697
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.971956422319267e-05,
+      "loss": 1.3104,
+      "step": 11698
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.971947238569467e-05,
+      "loss": 1.2834,
+      "step": 11699
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.971938053320393e-05,
+      "loss": 1.313,
+      "step": 11700
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.971928866572042e-05,
+      "loss": 1.2711,
+      "step": 11701
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.971919678324421e-05,
+      "loss": 1.3236,
+      "step": 11702
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.971910488577532e-05,
+      "loss": 1.3059,
+      "step": 11703
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.971901297331375e-05,
+      "loss": 1.3173,
+      "step": 11704
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.971892104585955e-05,
+      "loss": 1.3085,
+      "step": 11705
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.971882910341275e-05,
+      "loss": 1.3145,
+      "step": 11706
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.971873714597337e-05,
+      "loss": 1.2973,
+      "step": 11707
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.971864517354143e-05,
+      "loss": 1.3017,
+      "step": 11708
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.971855318611699e-05,
+      "loss": 1.2747,
+      "step": 11709
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.971846118370006e-05,
+      "loss": 1.2944,
+      "step": 11710
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.971836916629065e-05,
+      "loss": 1.2925,
+      "step": 11711
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.971827713388879e-05,
+      "loss": 1.3385,
+      "step": 11712
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.971818508649453e-05,
+      "loss": 1.28,
+      "step": 11713
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.97180930241079e-05,
+      "loss": 1.313,
+      "step": 11714
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.971800094672891e-05,
+      "loss": 1.2967,
+      "step": 11715
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.971790885435758e-05,
+      "loss": 1.2994,
+      "step": 11716
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.971781674699396e-05,
+      "loss": 1.2717,
+      "step": 11717
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.971772462463807e-05,
+      "loss": 1.3041,
+      "step": 11718
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.971763248728994e-05,
+      "loss": 1.338,
+      "step": 11719
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 9.971754033494958e-05,
+      "loss": 1.283,
+      "step": 11720
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 9.971744816761704e-05,
+      "loss": 1.2869,
+      "step": 11721
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 9.971735598529233e-05,
+      "loss": 1.277,
+      "step": 11722
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 9.97172637879755e-05,
+      "loss": 1.344,
+      "step": 11723
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 9.971717157566655e-05,
+      "loss": 1.3295,
+      "step": 11724
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 9.971707934836553e-05,
+      "loss": 1.3172,
+      "step": 11725
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 9.971698710607248e-05,
+      "loss": 1.3314,
+      "step": 11726
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 9.971689484878739e-05,
+      "loss": 1.2717,
+      "step": 11727
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 9.97168025765103e-05,
+      "loss": 1.3162,
+      "step": 11728
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 9.971671028924126e-05,
+      "loss": 1.3328,
+      "step": 11729
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 9.971661798698029e-05,
+      "loss": 1.3313,
+      "step": 11730
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 9.971652566972738e-05,
+      "loss": 1.2922,
+      "step": 11731
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 9.97164333374826e-05,
+      "loss": 1.3359,
+      "step": 11732
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 9.971634099024599e-05,
+      "loss": 1.2832,
+      "step": 11733
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 9.971624862801751e-05,
+      "loss": 1.297,
+      "step": 11734
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 9.971615625079726e-05,
+      "loss": 1.2917,
+      "step": 11735
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 9.971606385858523e-05,
+      "loss": 1.2853,
+      "step": 11736
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 9.971597145138146e-05,
+      "loss": 1.3358,
+      "step": 11737
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 9.971587902918598e-05,
+      "loss": 1.3311,
+      "step": 11738
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 9.97157865919988e-05,
+      "loss": 1.3382,
+      "step": 11739
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 9.971569413981997e-05,
+      "loss": 1.2999,
+      "step": 11740
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 9.97156016726495e-05,
+      "loss": 1.276,
+      "step": 11741
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 9.971550919048744e-05,
+      "loss": 1.2895,
+      "step": 11742
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 9.97154166933338e-05,
+      "loss": 1.2984,
+      "step": 11743
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 9.971532418118861e-05,
+      "loss": 1.3441,
+      "step": 11744
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 9.971523165405189e-05,
+      "loss": 1.3392,
+      "step": 11745
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 9.971513911192368e-05,
+      "loss": 1.3239,
+      "step": 11746
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 9.971504655480401e-05,
+      "loss": 1.3169,
+      "step": 11747
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 9.97149539826929e-05,
+      "loss": 1.2555,
+      "step": 11748
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 9.97148613955904e-05,
+      "loss": 1.295,
+      "step": 11749
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 9.97147687934965e-05,
+      "loss": 1.3217,
+      "step": 11750
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 9.971467617641126e-05,
+      "loss": 1.2892,
+      "step": 11751
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 9.97145835443347e-05,
+      "loss": 1.2822,
+      "step": 11752
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 9.971449089726682e-05,
+      "loss": 1.3048,
+      "step": 11753
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 9.971439823520768e-05,
+      "loss": 1.2795,
+      "step": 11754
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 9.971430555815731e-05,
+      "loss": 1.2892,
+      "step": 11755
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 9.971421286611571e-05,
+      "loss": 1.3005,
+      "step": 11756
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 9.971412015908295e-05,
+      "loss": 1.2832,
+      "step": 11757
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 9.971402743705901e-05,
+      "loss": 1.2868,
+      "step": 11758
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 9.971393470004395e-05,
+      "loss": 1.2837,
+      "step": 11759
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 9.971384194803779e-05,
+      "loss": 1.2743,
+      "step": 11760
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 9.971374918104055e-05,
+      "loss": 1.3025,
+      "step": 11761
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 9.971365639905229e-05,
+      "loss": 1.3062,
+      "step": 11762
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 9.971356360207298e-05,
+      "loss": 1.3257,
+      "step": 11763
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 9.971347079010269e-05,
+      "loss": 1.2769,
+      "step": 11764
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 9.971337796314144e-05,
+      "loss": 1.2834,
+      "step": 11765
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 9.971328512118926e-05,
+      "loss": 1.2923,
+      "step": 11766
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 9.971319226424618e-05,
+      "loss": 1.2835,
+      "step": 11767
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 9.971309939231221e-05,
+      "loss": 1.265,
+      "step": 11768
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 9.97130065053874e-05,
+      "loss": 1.2858,
+      "step": 11769
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 9.971291360347178e-05,
+      "loss": 1.3219,
+      "step": 11770
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 9.971282068656533e-05,
+      "loss": 1.2975,
+      "step": 11771
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 9.971272775466816e-05,
+      "loss": 1.3078,
+      "step": 11772
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 9.971263480778022e-05,
+      "loss": 1.2818,
+      "step": 11773
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 9.971254184590159e-05,
+      "loss": 1.2939,
+      "step": 11774
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 9.971244886903226e-05,
+      "loss": 1.3114,
+      "step": 11775
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 9.971235587717229e-05,
+      "loss": 1.2738,
+      "step": 11776
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 9.97122628703217e-05,
+      "loss": 1.3037,
+      "step": 11777
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 9.97121698484805e-05,
+      "loss": 1.2975,
+      "step": 11778
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 9.971207681164874e-05,
+      "loss": 1.3138,
+      "step": 11779
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 9.971198375982644e-05,
+      "loss": 1.3101,
+      "step": 11780
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 9.971189069301363e-05,
+      "loss": 1.3445,
+      "step": 11781
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.971179761121032e-05,
+      "loss": 1.2856,
+      "step": 11782
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.971170451441657e-05,
+      "loss": 1.3221,
+      "step": 11783
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.971161140263239e-05,
+      "loss": 1.3264,
+      "step": 11784
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.97115182758578e-05,
+      "loss": 1.3063,
+      "step": 11785
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.971142513409284e-05,
+      "loss": 1.287,
+      "step": 11786
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.971133197733753e-05,
+      "loss": 1.2913,
+      "step": 11787
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.971123880559193e-05,
+      "loss": 1.3298,
+      "step": 11788
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.971114561885601e-05,
+      "loss": 1.2726,
+      "step": 11789
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.971105241712985e-05,
+      "loss": 1.326,
+      "step": 11790
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.971095920041344e-05,
+      "loss": 1.3222,
+      "step": 11791
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.971086596870685e-05,
+      "loss": 1.3605,
+      "step": 11792
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.971077272201006e-05,
+      "loss": 1.3257,
+      "step": 11793
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.971067946032315e-05,
+      "loss": 1.3002,
+      "step": 11794
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.971058618364609e-05,
+      "loss": 1.3022,
+      "step": 11795
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.971049289197896e-05,
+      "loss": 1.3194,
+      "step": 11796
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.971039958532176e-05,
+      "loss": 1.2941,
+      "step": 11797
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.971030626367453e-05,
+      "loss": 1.2888,
+      "step": 11798
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.971021292703728e-05,
+      "loss": 1.2782,
+      "step": 11799
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.971011957541007e-05,
+      "loss": 1.2639,
+      "step": 11800
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.97100262087929e-05,
+      "loss": 1.3151,
+      "step": 11801
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.970993282718579e-05,
+      "loss": 1.32,
+      "step": 11802
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.970983943058882e-05,
+      "loss": 1.2886,
+      "step": 11803
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.970974601900196e-05,
+      "loss": 1.2936,
+      "step": 11804
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.970965259242527e-05,
+      "loss": 1.2941,
+      "step": 11805
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.970955915085877e-05,
+      "loss": 1.3122,
+      "step": 11806
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.970946569430249e-05,
+      "loss": 1.2999,
+      "step": 11807
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.970937222275645e-05,
+      "loss": 1.299,
+      "step": 11808
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.970927873622069e-05,
+      "loss": 1.2861,
+      "step": 11809
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.970918523469523e-05,
+      "loss": 1.3204,
+      "step": 11810
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.97090917181801e-05,
+      "loss": 1.3122,
+      "step": 11811
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.970899818667534e-05,
+      "loss": 1.3293,
+      "step": 11812
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.970890464018095e-05,
+      "loss": 1.3513,
+      "step": 11813
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.970881107869699e-05,
+      "loss": 1.3227,
+      "step": 11814
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.970871750222348e-05,
+      "loss": 1.3323,
+      "step": 11815
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.970862391076042e-05,
+      "loss": 1.296,
+      "step": 11816
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.970853030430787e-05,
+      "loss": 1.2933,
+      "step": 11817
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.970843668286584e-05,
+      "loss": 1.3142,
+      "step": 11818
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.97083430464344e-05,
+      "loss": 1.305,
+      "step": 11819
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.970824939501351e-05,
+      "loss": 1.3209,
+      "step": 11820
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.970815572860325e-05,
+      "loss": 1.3179,
+      "step": 11821
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.970806204720363e-05,
+      "loss": 1.3339,
+      "step": 11822
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.970796835081467e-05,
+      "loss": 1.2835,
+      "step": 11823
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.970787463943642e-05,
+      "loss": 1.2924,
+      "step": 11824
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.97077809130689e-05,
+      "loss": 1.2739,
+      "step": 11825
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.970768717171211e-05,
+      "loss": 1.3048,
+      "step": 11826
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.970759341536613e-05,
+      "loss": 1.3079,
+      "step": 11827
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.970749964403094e-05,
+      "loss": 1.3188,
+      "step": 11828
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.970740585770661e-05,
+      "loss": 1.2679,
+      "step": 11829
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.970731205639315e-05,
+      "loss": 1.2886,
+      "step": 11830
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.970721824009056e-05,
+      "loss": 1.2847,
+      "step": 11831
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.970712440879891e-05,
+      "loss": 1.3089,
+      "step": 11832
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.970703056251821e-05,
+      "loss": 1.2938,
+      "step": 11833
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.970693670124849e-05,
+      "loss": 1.2687,
+      "step": 11834
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.970684282498978e-05,
+      "loss": 1.276,
+      "step": 11835
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.970674893374211e-05,
+      "loss": 1.3347,
+      "step": 11836
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.97066550275055e-05,
+      "loss": 1.3469,
+      "step": 11837
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.970656110628e-05,
+      "loss": 1.3299,
+      "step": 11838
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.97064671700656e-05,
+      "loss": 1.2742,
+      "step": 11839
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.970637321886236e-05,
+      "loss": 1.327,
+      "step": 11840
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.970627925267031e-05,
+      "loss": 1.3167,
+      "step": 11841
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.970618527148944e-05,
+      "loss": 1.3135,
+      "step": 11842
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.970609127531983e-05,
+      "loss": 1.3236,
+      "step": 11843
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.970599726416149e-05,
+      "loss": 1.2951,
+      "step": 11844
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.970590323801443e-05,
+      "loss": 1.3464,
+      "step": 11845
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.970580919687869e-05,
+      "loss": 1.2717,
+      "step": 11846
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.97057151407543e-05,
+      "loss": 1.2975,
+      "step": 11847
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.970562106964128e-05,
+      "loss": 1.3177,
+      "step": 11848
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.970552698353967e-05,
+      "loss": 1.2863,
+      "step": 11849
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.970543288244951e-05,
+      "loss": 1.2842,
+      "step": 11850
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.97053387663708e-05,
+      "loss": 1.3464,
+      "step": 11851
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.970524463530357e-05,
+      "loss": 1.2929,
+      "step": 11852
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.970515048924789e-05,
+      "loss": 1.3399,
+      "step": 11853
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.970505632820373e-05,
+      "loss": 1.3038,
+      "step": 11854
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.970496215217116e-05,
+      "loss": 1.3496,
+      "step": 11855
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.97048679611502e-05,
+      "loss": 1.3113,
+      "step": 11856
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.970477375514084e-05,
+      "loss": 1.2985,
+      "step": 11857
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.970467953414318e-05,
+      "loss": 1.354,
+      "step": 11858
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.97045852981572e-05,
+      "loss": 1.3006,
+      "step": 11859
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.970449104718293e-05,
+      "loss": 1.3046,
+      "step": 11860
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.970439678122041e-05,
+      "loss": 1.3095,
+      "step": 11861
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.970430250026966e-05,
+      "loss": 1.3089,
+      "step": 11862
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.970420820433072e-05,
+      "loss": 1.3151,
+      "step": 11863
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.970411389340361e-05,
+      "loss": 1.2971,
+      "step": 11864
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.970401956748836e-05,
+      "loss": 1.3189,
+      "step": 11865
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.9703925226585e-05,
+      "loss": 1.2816,
+      "step": 11866
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.970383087069355e-05,
+      "loss": 1.2914,
+      "step": 11867
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.970373649981406e-05,
+      "loss": 1.3012,
+      "step": 11868
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.970364211394653e-05,
+      "loss": 1.2958,
+      "step": 11869
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.9703547713091e-05,
+      "loss": 1.299,
+      "step": 11870
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.970345329724753e-05,
+      "loss": 1.2901,
+      "step": 11871
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.970335886641608e-05,
+      "loss": 1.299,
+      "step": 11872
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.970326442059674e-05,
+      "loss": 1.2859,
+      "step": 11873
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.970316995978952e-05,
+      "loss": 1.3219,
+      "step": 11874
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.970307548399445e-05,
+      "loss": 1.321,
+      "step": 11875
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.970298099321154e-05,
+      "loss": 1.3151,
+      "step": 11876
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.970288648744084e-05,
+      "loss": 1.3008,
+      "step": 11877
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.970279196668235e-05,
+      "loss": 1.2621,
+      "step": 11878
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.970269743093614e-05,
+      "loss": 1.3041,
+      "step": 11879
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.970260288020221e-05,
+      "loss": 1.2799,
+      "step": 11880
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.97025083144806e-05,
+      "loss": 1.3065,
+      "step": 11881
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.970241373377134e-05,
+      "loss": 1.3324,
+      "step": 11882
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.970231913807444e-05,
+      "loss": 1.2826,
+      "step": 11883
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.970222452738995e-05,
+      "loss": 1.2888,
+      "step": 11884
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.970212990171788e-05,
+      "loss": 1.3145,
+      "step": 11885
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.970203526105828e-05,
+      "loss": 1.2867,
+      "step": 11886
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.970194060541117e-05,
+      "loss": 1.3128,
+      "step": 11887
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.970184593477656e-05,
+      "loss": 1.3457,
+      "step": 11888
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.97017512491545e-05,
+      "loss": 1.2696,
+      "step": 11889
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.970165654854504e-05,
+      "loss": 1.3086,
+      "step": 11890
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.970156183294814e-05,
+      "loss": 1.3184,
+      "step": 11891
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.97014671023639e-05,
+      "loss": 1.2948,
+      "step": 11892
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.97013723567923e-05,
+      "loss": 1.3019,
+      "step": 11893
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.97012775962334e-05,
+      "loss": 1.3045,
+      "step": 11894
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.97011828206872e-05,
+      "loss": 1.3058,
+      "step": 11895
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.970108803015375e-05,
+      "loss": 1.2942,
+      "step": 11896
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.970099322463308e-05,
+      "loss": 1.3258,
+      "step": 11897
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.97008984041252e-05,
+      "loss": 1.3106,
+      "step": 11898
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.970080356863016e-05,
+      "loss": 1.3129,
+      "step": 11899
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.970070871814798e-05,
+      "loss": 1.2599,
+      "step": 11900
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.970061385267869e-05,
+      "loss": 1.2671,
+      "step": 11901
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.970051897222232e-05,
+      "loss": 1.2576,
+      "step": 11902
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.970042407677887e-05,
+      "loss": 1.3094,
+      "step": 11903
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.970032916634842e-05,
+      "loss": 1.3168,
+      "step": 11904
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.970023424093096e-05,
+      "loss": 1.3039,
+      "step": 11905
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.970013930052651e-05,
+      "loss": 1.301,
+      "step": 11906
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.970004434513515e-05,
+      "loss": 1.324,
+      "step": 11907
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.969994937475686e-05,
+      "loss": 1.3164,
+      "step": 11908
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.969985438939168e-05,
+      "loss": 1.302,
+      "step": 11909
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.969975938903967e-05,
+      "loss": 1.3189,
+      "step": 11910
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.969966437370082e-05,
+      "loss": 1.3201,
+      "step": 11911
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.969956934337517e-05,
+      "loss": 1.3233,
+      "step": 11912
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.969947429806276e-05,
+      "loss": 1.3185,
+      "step": 11913
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.96993792377636e-05,
+      "loss": 1.3129,
+      "step": 11914
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.969928416247774e-05,
+      "loss": 1.2614,
+      "step": 11915
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.969918907220518e-05,
+      "loss": 1.2982,
+      "step": 11916
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.969909396694597e-05,
+      "loss": 1.3065,
+      "step": 11917
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.969899884670013e-05,
+      "loss": 1.2928,
+      "step": 11918
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.969890371146771e-05,
+      "loss": 1.2931,
+      "step": 11919
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.969880856124871e-05,
+      "loss": 1.3191,
+      "step": 11920
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.969871339604317e-05,
+      "loss": 1.3315,
+      "step": 11921
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.969861821585113e-05,
+      "loss": 1.3213,
+      "step": 11922
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.96985230206726e-05,
+      "loss": 1.3248,
+      "step": 11923
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.96984278105076e-05,
+      "loss": 1.2933,
+      "step": 11924
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.96983325853562e-05,
+      "loss": 1.3064,
+      "step": 11925
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.96982373452184e-05,
+      "loss": 1.2833,
+      "step": 11926
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.969814209009421e-05,
+      "loss": 1.2687,
+      "step": 11927
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.96980468199837e-05,
+      "loss": 1.3121,
+      "step": 11928
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.969795153488689e-05,
+      "loss": 1.3187,
+      "step": 11929
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.969785623480378e-05,
+      "loss": 1.3057,
+      "step": 11930
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.969776091973441e-05,
+      "loss": 1.2865,
+      "step": 11931
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.969766558967884e-05,
+      "loss": 1.2917,
+      "step": 11932
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.969757024463706e-05,
+      "loss": 1.3167,
+      "step": 11933
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.969747488460913e-05,
+      "loss": 1.2689,
+      "step": 11934
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.969737950959503e-05,
+      "loss": 1.3227,
+      "step": 11935
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.969728411959485e-05,
+      "loss": 1.3098,
+      "step": 11936
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.969718871460857e-05,
+      "loss": 1.2767,
+      "step": 11937
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.969709329463626e-05,
+      "loss": 1.2864,
+      "step": 11938
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.969699785967792e-05,
+      "loss": 1.2856,
+      "step": 11939
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.969690240973358e-05,
+      "loss": 1.2702,
+      "step": 11940
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.969680694480328e-05,
+      "loss": 1.289,
+      "step": 11941
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.969671146488703e-05,
+      "loss": 1.3029,
+      "step": 11942
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.96966159699849e-05,
+      "loss": 1.2857,
+      "step": 11943
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.969652046009687e-05,
+      "loss": 1.3018,
+      "step": 11944
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.9696424935223e-05,
+      "loss": 1.2905,
+      "step": 11945
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.969632939536331e-05,
+      "loss": 1.2731,
+      "step": 11946
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.969623384051782e-05,
+      "loss": 1.2656,
+      "step": 11947
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.969613827068657e-05,
+      "loss": 1.296,
+      "step": 11948
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.969604268586958e-05,
+      "loss": 1.2974,
+      "step": 11949
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.969594708606689e-05,
+      "loss": 1.2584,
+      "step": 11950
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.969585147127854e-05,
+      "loss": 1.3283,
+      "step": 11951
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.969575584150451e-05,
+      "loss": 1.3057,
+      "step": 11952
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.969566019674487e-05,
+      "loss": 1.3067,
+      "step": 11953
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.969556453699965e-05,
+      "loss": 1.3115,
+      "step": 11954
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.969546886226886e-05,
+      "loss": 1.3043,
+      "step": 11955
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.969537317255254e-05,
+      "loss": 1.3192,
+      "step": 11956
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.969527746785072e-05,
+      "loss": 1.3271,
+      "step": 11957
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.969518174816342e-05,
+      "loss": 1.2886,
+      "step": 11958
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.969508601349068e-05,
+      "loss": 1.2929,
+      "step": 11959
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.969499026383251e-05,
+      "loss": 1.3174,
+      "step": 11960
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.969489449918896e-05,
+      "loss": 1.315,
+      "step": 11961
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.969479871956005e-05,
+      "loss": 1.2966,
+      "step": 11962
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.969470292494581e-05,
+      "loss": 1.2722,
+      "step": 11963
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.969460711534629e-05,
+      "loss": 1.3131,
+      "step": 11964
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.969451129076146e-05,
+      "loss": 1.311,
+      "step": 11965
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.96944154511914e-05,
+      "loss": 1.3122,
+      "step": 11966
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.969431959663613e-05,
+      "loss": 1.3339,
+      "step": 11967
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.969422372709567e-05,
+      "loss": 1.3176,
+      "step": 11968
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.969412784257005e-05,
+      "loss": 1.3165,
+      "step": 11969
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.96940319430593e-05,
+      "loss": 1.2917,
+      "step": 11970
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.969393602856346e-05,
+      "loss": 1.2899,
+      "step": 11971
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.969384009908255e-05,
+      "loss": 1.283,
+      "step": 11972
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.969374415461659e-05,
+      "loss": 1.3168,
+      "step": 11973
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.969364819516563e-05,
+      "loss": 1.3392,
+      "step": 11974
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.969355222072967e-05,
+      "loss": 1.2928,
+      "step": 11975
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.969345623130876e-05,
+      "loss": 1.2812,
+      "step": 11976
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.969336022690292e-05,
+      "loss": 1.2683,
+      "step": 11977
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.969326420751219e-05,
+      "loss": 1.3353,
+      "step": 11978
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.96931681731366e-05,
+      "loss": 1.2829,
+      "step": 11979
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.969307212377617e-05,
+      "loss": 1.3193,
+      "step": 11980
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.96929760594309e-05,
+      "loss": 1.3097,
+      "step": 11981
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.969287998010088e-05,
+      "loss": 1.3029,
+      "step": 11982
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.96927838857861e-05,
+      "loss": 1.2955,
+      "step": 11983
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.96926877764866e-05,
+      "loss": 1.2968,
+      "step": 11984
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.96925916522024e-05,
+      "loss": 1.3107,
+      "step": 11985
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.969249551293353e-05,
+      "loss": 1.2731,
+      "step": 11986
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.969239935868004e-05,
+      "loss": 1.3361,
+      "step": 11987
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.969230318944194e-05,
+      "loss": 1.3197,
+      "step": 11988
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.969220700521925e-05,
+      "loss": 1.2768,
+      "step": 11989
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.969211080601202e-05,
+      "loss": 1.262,
+      "step": 11990
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.969201459182027e-05,
+      "loss": 1.3206,
+      "step": 11991
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.969191836264402e-05,
+      "loss": 1.3015,
+      "step": 11992
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.969182211848331e-05,
+      "loss": 1.2576,
+      "step": 11993
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.969172585933818e-05,
+      "loss": 1.332,
+      "step": 11994
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.969162958520864e-05,
+      "loss": 1.33,
+      "step": 11995
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.969153329609471e-05,
+      "loss": 1.308,
+      "step": 11996
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.969143699199645e-05,
+      "loss": 1.2936,
+      "step": 11997
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.969134067291387e-05,
+      "loss": 1.2462,
+      "step": 11998
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.9691244338847e-05,
+      "loss": 1.3173,
+      "step": 11999
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.969114798979586e-05,
+      "loss": 1.2987,
+      "step": 12000
+    },
+    {
+      "epoch": 1.93,
+      "eval_loss": 1.328537106513977,
+      "eval_runtime": 61.5278,
+      "eval_samples_per_second": 162.528,
+      "eval_steps_per_second": 0.228,
+      "step": 12000
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.969105162576052e-05,
+      "loss": 1.3202,
+      "step": 12001
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.969095524674096e-05,
+      "loss": 1.3044,
+      "step": 12002
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.969085885273722e-05,
+      "loss": 1.3043,
+      "step": 12003
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.969076244374935e-05,
+      "loss": 1.3167,
+      "step": 12004
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.969066601977736e-05,
+      "loss": 1.3054,
+      "step": 12005
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.96905695808213e-05,
+      "loss": 1.3335,
+      "step": 12006
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.969047312688118e-05,
+      "loss": 1.2814,
+      "step": 12007
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.969037665795702e-05,
+      "loss": 1.297,
+      "step": 12008
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.969028017404887e-05,
+      "loss": 1.3075,
+      "step": 12009
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.969018367515675e-05,
+      "loss": 1.2842,
+      "step": 12010
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.96900871612807e-05,
+      "loss": 1.3104,
+      "step": 12011
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.968999063242073e-05,
+      "loss": 1.3398,
+      "step": 12012
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.968989408857688e-05,
+      "loss": 1.281,
+      "step": 12013
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.968979752974918e-05,
+      "loss": 1.2755,
+      "step": 12014
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.968970095593766e-05,
+      "loss": 1.3213,
+      "step": 12015
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.968960436714234e-05,
+      "loss": 1.3086,
+      "step": 12016
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.968950776336327e-05,
+      "loss": 1.2907,
+      "step": 12017
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.968941114460046e-05,
+      "loss": 1.2789,
+      "step": 12018
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.968931451085394e-05,
+      "loss": 1.3013,
+      "step": 12019
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.968921786212374e-05,
+      "loss": 1.2899,
+      "step": 12020
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.96891211984099e-05,
+      "loss": 1.3114,
+      "step": 12021
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.968902451971243e-05,
+      "loss": 1.2846,
+      "step": 12022
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.968892782603139e-05,
+      "loss": 1.3094,
+      "step": 12023
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.968883111736677e-05,
+      "loss": 1.3131,
+      "step": 12024
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.968873439371862e-05,
+      "loss": 1.3153,
+      "step": 12025
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.968863765508697e-05,
+      "loss": 1.2981,
+      "step": 12026
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.968854090147186e-05,
+      "loss": 1.2623,
+      "step": 12027
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.96884441328733e-05,
+      "loss": 1.3064,
+      "step": 12028
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.968834734929132e-05,
+      "loss": 1.2748,
+      "step": 12029
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.968825055072596e-05,
+      "loss": 1.2874,
+      "step": 12030
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.968815373717724e-05,
+      "loss": 1.294,
+      "step": 12031
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.96880569086452e-05,
+      "loss": 1.3404,
+      "step": 12032
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.968796006512986e-05,
+      "loss": 1.3091,
+      "step": 12033
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.968786320663124e-05,
+      "loss": 1.3314,
+      "step": 12034
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.968776633314939e-05,
+      "loss": 1.2939,
+      "step": 12035
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.968766944468433e-05,
+      "loss": 1.316,
+      "step": 12036
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.968757254123608e-05,
+      "loss": 1.3153,
+      "step": 12037
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.968747562280469e-05,
+      "loss": 1.3152,
+      "step": 12038
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.968737868939019e-05,
+      "loss": 1.277,
+      "step": 12039
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.968728174099258e-05,
+      "loss": 1.3105,
+      "step": 12040
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.96871847776119e-05,
+      "loss": 1.3171,
+      "step": 12041
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.96870877992482e-05,
+      "loss": 1.3078,
+      "step": 12042
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.96869908059015e-05,
+      "loss": 1.2959,
+      "step": 12043
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.968689379757178e-05,
+      "loss": 1.2716,
+      "step": 12044
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.968679677425917e-05,
+      "loss": 1.3259,
+      "step": 12045
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.968669973596361e-05,
+      "loss": 1.3281,
+      "step": 12046
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.968660268268516e-05,
+      "loss": 1.3262,
+      "step": 12047
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.968650561442387e-05,
+      "loss": 1.2683,
+      "step": 12048
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.968640853117975e-05,
+      "loss": 1.3682,
+      "step": 12049
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.968631143295281e-05,
+      "loss": 1.2976,
+      "step": 12050
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.96862143197431e-05,
+      "loss": 1.2669,
+      "step": 12051
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.968611719155066e-05,
+      "loss": 1.3194,
+      "step": 12052
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.968602004837551e-05,
+      "loss": 1.2673,
+      "step": 12053
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.968592289021766e-05,
+      "loss": 1.3088,
+      "step": 12054
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.968582571707717e-05,
+      "loss": 1.3007,
+      "step": 12055
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.968572852895406e-05,
+      "loss": 1.333,
+      "step": 12056
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.968563132584833e-05,
+      "loss": 1.3168,
+      "step": 12057
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.968553410776005e-05,
+      "loss": 1.2986,
+      "step": 12058
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.968543687468923e-05,
+      "loss": 1.2995,
+      "step": 12059
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.968533962663592e-05,
+      "loss": 1.3173,
+      "step": 12060
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.96852423636001e-05,
+      "loss": 1.3282,
+      "step": 12061
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.968514508558186e-05,
+      "loss": 1.2956,
+      "step": 12062
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.968504779258119e-05,
+      "loss": 1.316,
+      "step": 12063
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.968495048459812e-05,
+      "loss": 1.3027,
+      "step": 12064
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.96848531616327e-05,
+      "loss": 1.3086,
+      "step": 12065
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.968475582368495e-05,
+      "loss": 1.3609,
+      "step": 12066
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.968465847075488e-05,
+      "loss": 1.3457,
+      "step": 12067
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.968456110284255e-05,
+      "loss": 1.3284,
+      "step": 12068
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.968446371994797e-05,
+      "loss": 1.3017,
+      "step": 12069
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.968436632207119e-05,
+      "loss": 1.2755,
+      "step": 12070
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.968426890921222e-05,
+      "loss": 1.3318,
+      "step": 12071
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.96841714813711e-05,
+      "loss": 1.3044,
+      "step": 12072
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.968407403854785e-05,
+      "loss": 1.3345,
+      "step": 12073
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.968397658074249e-05,
+      "loss": 1.2764,
+      "step": 12074
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.968387910795507e-05,
+      "loss": 1.2861,
+      "step": 12075
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.968378162018561e-05,
+      "loss": 1.2687,
+      "step": 12076
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.968368411743414e-05,
+      "loss": 1.276,
+      "step": 12077
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.968358659970071e-05,
+      "loss": 1.2975,
+      "step": 12078
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.968348906698532e-05,
+      "loss": 1.3359,
+      "step": 12079
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.9683391519288e-05,
+      "loss": 1.3009,
+      "step": 12080
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.96832939566088e-05,
+      "loss": 1.3241,
+      "step": 12081
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.968319637894774e-05,
+      "loss": 1.3255,
+      "step": 12082
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.968309878630483e-05,
+      "loss": 1.3046,
+      "step": 12083
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.968300117868013e-05,
+      "loss": 1.3331,
+      "step": 12084
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.968290355607366e-05,
+      "loss": 1.2777,
+      "step": 12085
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.968280591848544e-05,
+      "loss": 1.3453,
+      "step": 12086
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.968270826591552e-05,
+      "loss": 1.2804,
+      "step": 12087
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.96826105983639e-05,
+      "loss": 1.2906,
+      "step": 12088
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.968251291583063e-05,
+      "loss": 1.3182,
+      "step": 12089
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.968241521831574e-05,
+      "loss": 1.2792,
+      "step": 12090
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.968231750581924e-05,
+      "loss": 1.2676,
+      "step": 12091
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.968221977834118e-05,
+      "loss": 1.2698,
+      "step": 12092
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.968212203588159e-05,
+      "loss": 1.3144,
+      "step": 12093
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.968202427844048e-05,
+      "loss": 1.2933,
+      "step": 12094
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.968192650601789e-05,
+      "loss": 1.2999,
+      "step": 12095
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.968182871861387e-05,
+      "loss": 1.2838,
+      "step": 12096
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.968173091622842e-05,
+      "loss": 1.2882,
+      "step": 12097
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.968163309886155e-05,
+      "loss": 1.2952,
+      "step": 12098
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.968153526651336e-05,
+      "loss": 1.3209,
+      "step": 12099
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.968143741918384e-05,
+      "loss": 1.2714,
+      "step": 12100
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.9681339556873e-05,
+      "loss": 1.321,
+      "step": 12101
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.968124167958089e-05,
+      "loss": 1.3021,
+      "step": 12102
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.968114378730752e-05,
+      "loss": 1.2882,
+      "step": 12103
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.968104588005297e-05,
+      "loss": 1.2855,
+      "step": 12104
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.968094795781722e-05,
+      "loss": 1.2703,
+      "step": 12105
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.968085002060033e-05,
+      "loss": 1.3096,
+      "step": 12106
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.96807520684023e-05,
+      "loss": 1.2807,
+      "step": 12107
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.968065410122319e-05,
+      "loss": 1.3462,
+      "step": 12108
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.9680556119063e-05,
+      "loss": 1.3201,
+      "step": 12109
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.968045812192177e-05,
+      "loss": 1.2712,
+      "step": 12110
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.968036010979955e-05,
+      "loss": 1.2842,
+      "step": 12111
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.968026208269635e-05,
+      "loss": 1.2855,
+      "step": 12112
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.968016404061218e-05,
+      "loss": 1.3428,
+      "step": 12113
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.968006598354711e-05,
+      "loss": 1.3041,
+      "step": 12114
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.967996791150116e-05,
+      "loss": 1.3008,
+      "step": 12115
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.967986982447435e-05,
+      "loss": 1.309,
+      "step": 12116
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.96797717224667e-05,
+      "loss": 1.2642,
+      "step": 12117
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.967967360547825e-05,
+      "loss": 1.3611,
+      "step": 12118
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.967957547350903e-05,
+      "loss": 1.3005,
+      "step": 12119
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.967947732655909e-05,
+      "loss": 1.2956,
+      "step": 12120
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.967937916462841e-05,
+      "loss": 1.3275,
+      "step": 12121
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.967928098771708e-05,
+      "loss": 1.3172,
+      "step": 12122
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.967918279582509e-05,
+      "loss": 1.3092,
+      "step": 12123
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.967908458895247e-05,
+      "loss": 1.3134,
+      "step": 12124
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.967898636709925e-05,
+      "loss": 1.2974,
+      "step": 12125
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.967888813026547e-05,
+      "loss": 1.3233,
+      "step": 12126
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.967878987845118e-05,
+      "loss": 1.2927,
+      "step": 12127
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.967869161165638e-05,
+      "loss": 1.2845,
+      "step": 12128
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.96785933298811e-05,
+      "loss": 1.2727,
+      "step": 12129
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.967849503312536e-05,
+      "loss": 1.2598,
+      "step": 12130
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.967839672138922e-05,
+      "loss": 1.3208,
+      "step": 12131
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.967829839467269e-05,
+      "loss": 1.2904,
+      "step": 12132
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.967820005297582e-05,
+      "loss": 1.2808,
+      "step": 12133
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.96781016962986e-05,
+      "loss": 1.3165,
+      "step": 12134
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.96780033246411e-05,
+      "loss": 1.2853,
+      "step": 12135
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.967790493800334e-05,
+      "loss": 1.2928,
+      "step": 12136
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.967780653638534e-05,
+      "loss": 1.2717,
+      "step": 12137
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.967770811978713e-05,
+      "loss": 1.3328,
+      "step": 12138
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.967760968820873e-05,
+      "loss": 1.2517,
+      "step": 12139
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.967751124165021e-05,
+      "loss": 1.2803,
+      "step": 12140
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.967741278011155e-05,
+      "loss": 1.337,
+      "step": 12141
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.967731430359281e-05,
+      "loss": 1.2743,
+      "step": 12142
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.967721581209401e-05,
+      "loss": 1.3006,
+      "step": 12143
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.96771173056152e-05,
+      "loss": 1.3164,
+      "step": 12144
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.967701878415636e-05,
+      "loss": 1.2975,
+      "step": 12145
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.967692024771757e-05,
+      "loss": 1.3104,
+      "step": 12146
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.967682169629884e-05,
+      "loss": 1.2535,
+      "step": 12147
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.96767231299002e-05,
+      "loss": 1.3071,
+      "step": 12148
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.967662454852166e-05,
+      "loss": 1.2936,
+      "step": 12149
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.96765259521633e-05,
+      "loss": 1.3106,
+      "step": 12150
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.967642734082509e-05,
+      "loss": 1.2841,
+      "step": 12151
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.967632871450711e-05,
+      "loss": 1.3327,
+      "step": 12152
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.967623007320937e-05,
+      "loss": 1.2958,
+      "step": 12153
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.967613141693189e-05,
+      "loss": 1.294,
+      "step": 12154
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.967603274567471e-05,
+      "loss": 1.2899,
+      "step": 12155
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.967593405943785e-05,
+      "loss": 1.299,
+      "step": 12156
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.967583535822135e-05,
+      "loss": 1.3118,
+      "step": 12157
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.967573664202526e-05,
+      "loss": 1.3557,
+      "step": 12158
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.967563791084957e-05,
+      "loss": 1.2661,
+      "step": 12159
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.967553916469433e-05,
+      "loss": 1.3217,
+      "step": 12160
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.967544040355956e-05,
+      "loss": 1.289,
+      "step": 12161
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.967534162744532e-05,
+      "loss": 1.2888,
+      "step": 12162
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.96752428363516e-05,
+      "loss": 1.3303,
+      "step": 12163
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.967514403027845e-05,
+      "loss": 1.2612,
+      "step": 12164
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.967504520922589e-05,
+      "loss": 1.2972,
+      "step": 12165
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.967494637319394e-05,
+      "loss": 1.3161,
+      "step": 12166
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.967484752218268e-05,
+      "loss": 1.2675,
+      "step": 12167
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.967474865619209e-05,
+      "loss": 1.3065,
+      "step": 12168
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.96746497752222e-05,
+      "loss": 1.3043,
+      "step": 12169
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.967455087927309e-05,
+      "loss": 1.2842,
+      "step": 12170
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.967445196834473e-05,
+      "loss": 1.3089,
+      "step": 12171
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.967435304243717e-05,
+      "loss": 1.2604,
+      "step": 12172
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.967425410155046e-05,
+      "loss": 1.3079,
+      "step": 12173
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.967415514568461e-05,
+      "loss": 1.2867,
+      "step": 12174
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.967405617483965e-05,
+      "loss": 1.2939,
+      "step": 12175
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.967395718901562e-05,
+      "loss": 1.3004,
+      "step": 12176
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.967385818821254e-05,
+      "loss": 1.2776,
+      "step": 12177
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.967375917243045e-05,
+      "loss": 1.2715,
+      "step": 12178
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.967366014166936e-05,
+      "loss": 1.2845,
+      "step": 12179
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.967356109592934e-05,
+      "loss": 1.2923,
+      "step": 12180
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.967346203521038e-05,
+      "loss": 1.3436,
+      "step": 12181
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.967336295951251e-05,
+      "loss": 1.3349,
+      "step": 12182
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.967326386883579e-05,
+      "loss": 1.2827,
+      "step": 12183
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.967316476318022e-05,
+      "loss": 1.3442,
+      "step": 12184
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.967306564254586e-05,
+      "loss": 1.3158,
+      "step": 12185
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.96729665069327e-05,
+      "loss": 1.2527,
+      "step": 12186
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.967286735634082e-05,
+      "loss": 1.2787,
+      "step": 12187
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.967276819077021e-05,
+      "loss": 1.3304,
+      "step": 12188
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.96726690102209e-05,
+      "loss": 1.2985,
+      "step": 12189
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.967256981469295e-05,
+      "loss": 1.3111,
+      "step": 12190
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.967247060418636e-05,
+      "loss": 1.2868,
+      "step": 12191
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.967237137870118e-05,
+      "loss": 1.313,
+      "step": 12192
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.967227213823743e-05,
+      "loss": 1.2816,
+      "step": 12193
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.967217288279515e-05,
+      "loss": 1.3136,
+      "step": 12194
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.967207361237435e-05,
+      "loss": 1.2992,
+      "step": 12195
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.967197432697509e-05,
+      "loss": 1.3229,
+      "step": 12196
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.967187502659736e-05,
+      "loss": 1.3478,
+      "step": 12197
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.967177571124123e-05,
+      "loss": 1.311,
+      "step": 12198
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.967167638090669e-05,
+      "loss": 1.2883,
+      "step": 12199
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.96715770355938e-05,
+      "loss": 1.3063,
+      "step": 12200
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.96714776753026e-05,
+      "loss": 1.2922,
+      "step": 12201
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.967137830003308e-05,
+      "loss": 1.3448,
+      "step": 12202
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.96712789097853e-05,
+      "loss": 1.3255,
+      "step": 12203
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.967117950455929e-05,
+      "loss": 1.3077,
+      "step": 12204
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.967108008435506e-05,
+      "loss": 1.302,
+      "step": 12205
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.967098064917265e-05,
+      "loss": 1.292,
+      "step": 12206
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.96708811990121e-05,
+      "loss": 1.3253,
+      "step": 12207
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.967078173387343e-05,
+      "loss": 1.3151,
+      "step": 12208
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.967068225375666e-05,
+      "loss": 1.3181,
+      "step": 12209
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.967058275866186e-05,
+      "loss": 1.2786,
+      "step": 12210
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.967048324858901e-05,
+      "loss": 1.3357,
+      "step": 12211
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.967038372353816e-05,
+      "loss": 1.2861,
+      "step": 12212
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.967028418350934e-05,
+      "loss": 1.3083,
+      "step": 12213
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.967018462850259e-05,
+      "loss": 1.3203,
+      "step": 12214
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.967008505851794e-05,
+      "loss": 1.3088,
+      "step": 12215
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.966998547355539e-05,
+      "loss": 1.307,
+      "step": 12216
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.966988587361501e-05,
+      "loss": 1.2898,
+      "step": 12217
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.96697862586968e-05,
+      "loss": 1.2858,
+      "step": 12218
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.966968662880081e-05,
+      "loss": 1.322,
+      "step": 12219
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.966958698392705e-05,
+      "loss": 1.2924,
+      "step": 12220
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.966948732407558e-05,
+      "loss": 1.2983,
+      "step": 12221
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.96693876492464e-05,
+      "loss": 1.3257,
+      "step": 12222
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.966928795943956e-05,
+      "loss": 1.3031,
+      "step": 12223
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.966918825465508e-05,
+      "loss": 1.265,
+      "step": 12224
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.966908853489298e-05,
+      "loss": 1.3139,
+      "step": 12225
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.966898880015332e-05,
+      "loss": 1.3232,
+      "step": 12226
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.96688890504361e-05,
+      "loss": 1.3351,
+      "step": 12227
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.966878928574137e-05,
+      "loss": 1.2728,
+      "step": 12228
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.966868950606915e-05,
+      "loss": 1.3392,
+      "step": 12229
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.966858971141948e-05,
+      "loss": 1.3416,
+      "step": 12230
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.966848990179236e-05,
+      "loss": 1.3114,
+      "step": 12231
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.966839007718786e-05,
+      "loss": 1.2838,
+      "step": 12232
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.966829023760598e-05,
+      "loss": 1.3091,
+      "step": 12233
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.966819038304678e-05,
+      "loss": 1.3466,
+      "step": 12234
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.966809051351027e-05,
+      "loss": 1.3216,
+      "step": 12235
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.966799062899648e-05,
+      "loss": 1.2842,
+      "step": 12236
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.966789072950543e-05,
+      "loss": 1.3149,
+      "step": 12237
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.966779081503717e-05,
+      "loss": 1.2947,
+      "step": 12238
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.966769088559173e-05,
+      "loss": 1.3417,
+      "step": 12239
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.966759094116913e-05,
+      "loss": 1.2724,
+      "step": 12240
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.966749098176942e-05,
+      "loss": 1.304,
+      "step": 12241
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.96673910073926e-05,
+      "loss": 1.3081,
+      "step": 12242
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.966729101803871e-05,
+      "loss": 1.2713,
+      "step": 12243
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.966719101370779e-05,
+      "loss": 1.2695,
+      "step": 12244
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.966709099439986e-05,
+      "loss": 1.3089,
+      "step": 12245
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.966699096011495e-05,
+      "loss": 1.2983,
+      "step": 12246
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.966689091085311e-05,
+      "loss": 1.291,
+      "step": 12247
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.966679084661434e-05,
+      "loss": 1.3207,
+      "step": 12248
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.96666907673987e-05,
+      "loss": 1.3264,
+      "step": 12249
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.966659067320619e-05,
+      "loss": 1.3153,
+      "step": 12250
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.966649056403688e-05,
+      "loss": 1.2997,
+      "step": 12251
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.966639043989075e-05,
+      "loss": 1.3,
+      "step": 12252
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.966629030076786e-05,
+      "loss": 1.3223,
+      "step": 12253
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.966619014666823e-05,
+      "loss": 1.3107,
+      "step": 12254
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.966608997759191e-05,
+      "loss": 1.3252,
+      "step": 12255
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.966598979353892e-05,
+      "loss": 1.3253,
+      "step": 12256
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.966588959450928e-05,
+      "loss": 1.2846,
+      "step": 12257
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.966578938050302e-05,
+      "loss": 1.2894,
+      "step": 12258
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.96656891515202e-05,
+      "loss": 1.3133,
+      "step": 12259
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.96655889075608e-05,
+      "loss": 1.3395,
+      "step": 12260
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.966548864862489e-05,
+      "loss": 1.3077,
+      "step": 12261
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.966538837471249e-05,
+      "loss": 1.3457,
+      "step": 12262
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.966528808582362e-05,
+      "loss": 1.2831,
+      "step": 12263
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.966518778195832e-05,
+      "loss": 1.2902,
+      "step": 12264
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.966508746311663e-05,
+      "loss": 1.3202,
+      "step": 12265
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.966498712929855e-05,
+      "loss": 1.3358,
+      "step": 12266
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.966488678050414e-05,
+      "loss": 1.3201,
+      "step": 12267
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.966478641673341e-05,
+      "loss": 1.3302,
+      "step": 12268
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.966468603798642e-05,
+      "loss": 1.3236,
+      "step": 12269
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.966458564426315e-05,
+      "loss": 1.3001,
+      "step": 12270
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.966448523556368e-05,
+      "loss": 1.2806,
+      "step": 12271
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.966438481188803e-05,
+      "loss": 1.3162,
+      "step": 12272
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.966428437323621e-05,
+      "loss": 1.3239,
+      "step": 12273
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.966418391960825e-05,
+      "loss": 1.3008,
+      "step": 12274
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.96640834510042e-05,
+      "loss": 1.2899,
+      "step": 12275
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.966398296742408e-05,
+      "loss": 1.3279,
+      "step": 12276
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.966388246886792e-05,
+      "loss": 1.2973,
+      "step": 12277
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.966378195533575e-05,
+      "loss": 1.2922,
+      "step": 12278
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.96636814268276e-05,
+      "loss": 1.2938,
+      "step": 12279
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.966358088334353e-05,
+      "loss": 1.3097,
+      "step": 12280
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.96634803248835e-05,
+      "loss": 1.2951,
+      "step": 12281
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.966337975144761e-05,
+      "loss": 1.292,
+      "step": 12282
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.966327916303587e-05,
+      "loss": 1.2758,
+      "step": 12283
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.96631785596483e-05,
+      "loss": 1.277,
+      "step": 12284
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.966307794128493e-05,
+      "loss": 1.2721,
+      "step": 12285
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.966297730794578e-05,
+      "loss": 1.2881,
+      "step": 12286
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.966287665963092e-05,
+      "loss": 1.308,
+      "step": 12287
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.966277599634034e-05,
+      "loss": 1.2989,
+      "step": 12288
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.96626753180741e-05,
+      "loss": 1.3087,
+      "step": 12289
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.96625746248322e-05,
+      "loss": 1.3101,
+      "step": 12290
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.966247391661469e-05,
+      "loss": 1.2865,
+      "step": 12291
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.96623731934216e-05,
+      "loss": 1.2773,
+      "step": 12292
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.966227245525295e-05,
+      "loss": 1.2745,
+      "step": 12293
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.966217170210878e-05,
+      "loss": 1.2885,
+      "step": 12294
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.966207093398913e-05,
+      "loss": 1.2913,
+      "step": 12295
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.966197015089401e-05,
+      "loss": 1.2973,
+      "step": 12296
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.966186935282346e-05,
+      "loss": 1.332,
+      "step": 12297
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.96617685397775e-05,
+      "loss": 1.309,
+      "step": 12298
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.966166771175617e-05,
+      "loss": 1.2978,
+      "step": 12299
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.966156686875951e-05,
+      "loss": 1.2662,
+      "step": 12300
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.966146601078754e-05,
+      "loss": 1.2832,
+      "step": 12301
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.96613651378403e-05,
+      "loss": 1.3301,
+      "step": 12302
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.96612642499178e-05,
+      "loss": 1.2869,
+      "step": 12303
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.966116334702007e-05,
+      "loss": 1.2942,
+      "step": 12304
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.966106242914717e-05,
+      "loss": 1.252,
+      "step": 12305
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.966096149629911e-05,
+      "loss": 1.3299,
+      "step": 12306
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.966086054847591e-05,
+      "loss": 1.2786,
+      "step": 12307
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.966075958567762e-05,
+      "loss": 1.3072,
+      "step": 12308
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.966065860790426e-05,
+      "loss": 1.3118,
+      "step": 12309
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.966055761515588e-05,
+      "loss": 1.3091,
+      "step": 12310
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.966045660743248e-05,
+      "loss": 1.2882,
+      "step": 12311
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.966035558473411e-05,
+      "loss": 1.2435,
+      "step": 12312
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.96602545470608e-05,
+      "loss": 1.2802,
+      "step": 12313
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.966015349441258e-05,
+      "loss": 1.248,
+      "step": 12314
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.966005242678946e-05,
+      "loss": 1.2821,
+      "step": 12315
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.96599513441915e-05,
+      "loss": 1.2894,
+      "step": 12316
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.96598502466187e-05,
+      "loss": 1.296,
+      "step": 12317
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.965974913407112e-05,
+      "loss": 1.322,
+      "step": 12318
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.965964800654879e-05,
+      "loss": 1.3126,
+      "step": 12319
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.965954686405171e-05,
+      "loss": 1.3322,
+      "step": 12320
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.965944570657995e-05,
+      "loss": 1.3,
+      "step": 12321
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.96593445341335e-05,
+      "loss": 1.2974,
+      "step": 12322
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.965924334671241e-05,
+      "loss": 1.3084,
+      "step": 12323
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.965914214431673e-05,
+      "loss": 1.34,
+      "step": 12324
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.965904092694646e-05,
+      "loss": 1.3039,
+      "step": 12325
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.965893969460163e-05,
+      "loss": 1.2685,
+      "step": 12326
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.965883844728229e-05,
+      "loss": 1.2916,
+      "step": 12327
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.965873718498848e-05,
+      "loss": 1.2615,
+      "step": 12328
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.96586359077202e-05,
+      "loss": 1.3001,
+      "step": 12329
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.96585346154775e-05,
+      "loss": 1.2912,
+      "step": 12330
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.96584333082604e-05,
+      "loss": 1.3205,
+      "step": 12331
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.965833198606893e-05,
+      "loss": 1.2892,
+      "step": 12332
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.965823064890313e-05,
+      "loss": 1.3159,
+      "step": 12333
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.965812929676303e-05,
+      "loss": 1.2716,
+      "step": 12334
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.965802792964865e-05,
+      "loss": 1.3199,
+      "step": 12335
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.965792654756004e-05,
+      "loss": 1.2831,
+      "step": 12336
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.965782515049722e-05,
+      "loss": 1.2981,
+      "step": 12337
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.96577237384602e-05,
+      "loss": 1.2661,
+      "step": 12338
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.965762231144903e-05,
+      "loss": 1.3039,
+      "step": 12339
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.965752086946376e-05,
+      "loss": 1.2734,
+      "step": 12340
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.96574194125044e-05,
+      "loss": 1.3147,
+      "step": 12341
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.965731794057096e-05,
+      "loss": 1.3071,
+      "step": 12342
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.96572164536635e-05,
+      "loss": 1.2503,
+      "step": 12343
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.965711495178205e-05,
+      "loss": 1.3083,
+      "step": 12344
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.965701343492662e-05,
+      "loss": 1.2954,
+      "step": 12345
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.965691190309725e-05,
+      "loss": 1.3205,
+      "step": 12346
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.9656810356294e-05,
+      "loss": 1.2998,
+      "step": 12347
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.965670879451685e-05,
+      "loss": 1.293,
+      "step": 12348
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.965660721776587e-05,
+      "loss": 1.2757,
+      "step": 12349
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.965650562604106e-05,
+      "loss": 1.3033,
+      "step": 12350
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.965640401934249e-05,
+      "loss": 1.3103,
+      "step": 12351
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.965630239767015e-05,
+      "loss": 1.3358,
+      "step": 12352
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.96562007610241e-05,
+      "loss": 1.3083,
+      "step": 12353
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.965609910940433e-05,
+      "loss": 1.3519,
+      "step": 12354
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.965599744281092e-05,
+      "loss": 1.3056,
+      "step": 12355
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.965589576124387e-05,
+      "loss": 1.3006,
+      "step": 12356
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.965579406470324e-05,
+      "loss": 1.288,
+      "step": 12357
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.965569235318903e-05,
+      "loss": 1.2926,
+      "step": 12358
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.96555906267013e-05,
+      "loss": 1.3108,
+      "step": 12359
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.965548888524004e-05,
+      "loss": 1.2908,
+      "step": 12360
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.965538712880528e-05,
+      "loss": 1.3114,
+      "step": 12361
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.965528535739712e-05,
+      "loss": 1.2947,
+      "step": 12362
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.965518357101552e-05,
+      "loss": 1.291,
+      "step": 12363
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.965508176966054e-05,
+      "loss": 1.2566,
+      "step": 12364
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.965497995333222e-05,
+      "loss": 1.2828,
+      "step": 12365
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.965487812203055e-05,
+      "loss": 1.3067,
+      "step": 12366
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.965477627575561e-05,
+      "loss": 1.3197,
+      "step": 12367
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.96546744145074e-05,
+      "loss": 1.3222,
+      "step": 12368
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.965457253828595e-05,
+      "loss": 1.2504,
+      "step": 12369
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.96544706470913e-05,
+      "loss": 1.2995,
+      "step": 12370
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.96543687409235e-05,
+      "loss": 1.3143,
+      "step": 12371
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.965426681978254e-05,
+      "loss": 1.33,
+      "step": 12372
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.965416488366848e-05,
+      "loss": 1.2886,
+      "step": 12373
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.965406293258135e-05,
+      "loss": 1.2907,
+      "step": 12374
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.965396096652116e-05,
+      "loss": 1.282,
+      "step": 12375
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.965385898548795e-05,
+      "loss": 1.2882,
+      "step": 12376
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.965375698948179e-05,
+      "loss": 1.2975,
+      "step": 12377
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.965365497850263e-05,
+      "loss": 1.2867,
+      "step": 12378
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.965355295255057e-05,
+      "loss": 1.3333,
+      "step": 12379
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.965345091162561e-05,
+      "loss": 1.3132,
+      "step": 12380
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.96533488557278e-05,
+      "loss": 1.3075,
+      "step": 12381
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.965324678485715e-05,
+      "loss": 1.2941,
+      "step": 12382
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.965314469901369e-05,
+      "loss": 1.2983,
+      "step": 12383
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.965304259819747e-05,
+      "loss": 1.3216,
+      "step": 12384
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.96529404824085e-05,
+      "loss": 1.3099,
+      "step": 12385
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.965283835164684e-05,
+      "loss": 1.267,
+      "step": 12386
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.965273620591249e-05,
+      "loss": 1.2716,
+      "step": 12387
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.96526340452055e-05,
+      "loss": 1.3023,
+      "step": 12388
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.96525318695259e-05,
+      "loss": 1.345,
+      "step": 12389
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.96524296788737e-05,
+      "loss": 1.3023,
+      "step": 12390
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.965232747324894e-05,
+      "loss": 1.292,
+      "step": 12391
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.965222525265168e-05,
+      "loss": 1.2802,
+      "step": 12392
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.965212301708191e-05,
+      "loss": 1.2948,
+      "step": 12393
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.965202076653968e-05,
+      "loss": 1.294,
+      "step": 12394
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.965191850102502e-05,
+      "loss": 1.3429,
+      "step": 12395
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.965181622053796e-05,
+      "loss": 1.2736,
+      "step": 12396
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.965171392507853e-05,
+      "loss": 1.313,
+      "step": 12397
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.965161161464677e-05,
+      "loss": 1.2721,
+      "step": 12398
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.96515092892427e-05,
+      "loss": 1.2996,
+      "step": 12399
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.965140694886634e-05,
+      "loss": 1.2917,
+      "step": 12400
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.965130459351775e-05,
+      "loss": 1.3152,
+      "step": 12401
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.965120222319693e-05,
+      "loss": 1.3107,
+      "step": 12402
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.965109983790393e-05,
+      "loss": 1.3403,
+      "step": 12403
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.965099743763879e-05,
+      "loss": 1.2603,
+      "step": 12404
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.96508950224015e-05,
+      "loss": 1.2917,
+      "step": 12405
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.965079259219214e-05,
+      "loss": 1.3248,
+      "step": 12406
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.965069014701072e-05,
+      "loss": 1.2762,
+      "step": 12407
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.965058768685726e-05,
+      "loss": 1.3257,
+      "step": 12408
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.965048521173182e-05,
+      "loss": 1.295,
+      "step": 12409
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.96503827216344e-05,
+      "loss": 1.2955,
+      "step": 12410
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.965028021656504e-05,
+      "loss": 1.3044,
+      "step": 12411
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.965017769652378e-05,
+      "loss": 1.3179,
+      "step": 12412
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.965007516151064e-05,
+      "loss": 1.3098,
+      "step": 12413
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.964997261152564e-05,
+      "loss": 1.3276,
+      "step": 12414
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.964987004656886e-05,
+      "loss": 1.2762,
+      "step": 12415
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.964976746664028e-05,
+      "loss": 1.2725,
+      "step": 12416
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.964966487173993e-05,
+      "loss": 1.2919,
+      "step": 12417
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.964956226186788e-05,
+      "loss": 1.2866,
+      "step": 12418
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.964945963702414e-05,
+      "loss": 1.3065,
+      "step": 12419
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.964935699720874e-05,
+      "loss": 1.3367,
+      "step": 12420
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.964925434242171e-05,
+      "loss": 1.3205,
+      "step": 12421
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.964915167266308e-05,
+      "loss": 1.2725,
+      "step": 12422
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.96490489879329e-05,
+      "loss": 1.3232,
+      "step": 12423
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.964894628823117e-05,
+      "loss": 1.3067,
+      "step": 12424
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.964884357355794e-05,
+      "loss": 1.3035,
+      "step": 12425
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.964874084391324e-05,
+      "loss": 1.2869,
+      "step": 12426
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.96486380992971e-05,
+      "loss": 1.2825,
+      "step": 12427
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.964853533970954e-05,
+      "loss": 1.2815,
+      "step": 12428
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.96484325651506e-05,
+      "loss": 1.2872,
+      "step": 12429
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.964832977562031e-05,
+      "loss": 1.2963,
+      "step": 12430
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.96482269711187e-05,
+      "loss": 1.3208,
+      "step": 12431
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.964812415164581e-05,
+      "loss": 1.3084,
+      "step": 12432
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.964802131720168e-05,
+      "loss": 1.3032,
+      "step": 12433
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.96479184677863e-05,
+      "loss": 1.2723,
+      "step": 12434
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.964781560339974e-05,
+      "loss": 1.2937,
+      "step": 12435
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.9647712724042e-05,
+      "loss": 1.2832,
+      "step": 12436
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.964760982971315e-05,
+      "loss": 1.3016,
+      "step": 12437
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.964750692041317e-05,
+      "loss": 1.3054,
+      "step": 12438
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.964740399614215e-05,
+      "loss": 1.3397,
+      "step": 12439
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.964730105690008e-05,
+      "loss": 1.2949,
+      "step": 12440
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.964719810268701e-05,
+      "loss": 1.2735,
+      "step": 12441
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.964709513350294e-05,
+      "loss": 1.2637,
+      "step": 12442
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.964699214934795e-05,
+      "loss": 1.2867,
+      "step": 12443
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.964688915022204e-05,
+      "loss": 1.337,
+      "step": 12444
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.964678613612523e-05,
+      "loss": 1.2682,
+      "step": 12445
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.964668310705759e-05,
+      "loss": 1.2507,
+      "step": 12446
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.964658006301912e-05,
+      "loss": 1.3001,
+      "step": 12447
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.964647700400985e-05,
+      "loss": 1.3102,
+      "step": 12448
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.964637393002982e-05,
+      "loss": 1.2447,
+      "step": 12449
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.964627084107907e-05,
+      "loss": 1.2735,
+      "step": 12450
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.964616773715762e-05,
+      "loss": 1.2739,
+      "step": 12451
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.96460646182655e-05,
+      "loss": 1.2976,
+      "step": 12452
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.964596148440277e-05,
+      "loss": 1.3157,
+      "step": 12453
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.964585833556941e-05,
+      "loss": 1.3255,
+      "step": 12454
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.964575517176549e-05,
+      "loss": 1.2758,
+      "step": 12455
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.964565199299102e-05,
+      "loss": 1.3085,
+      "step": 12456
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.964554879924604e-05,
+      "loss": 1.2611,
+      "step": 12457
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.964544559053058e-05,
+      "loss": 1.2531,
+      "step": 12458
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.964534236684467e-05,
+      "loss": 1.3074,
+      "step": 12459
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.964523912818835e-05,
+      "loss": 1.2782,
+      "step": 12460
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.964513587456164e-05,
+      "loss": 1.2719,
+      "step": 12461
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.964503260596459e-05,
+      "loss": 1.2681,
+      "step": 12462
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.964492932239719e-05,
+      "loss": 1.2908,
+      "step": 12463
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.964482602385951e-05,
+      "loss": 1.2867,
+      "step": 12464
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.964472271035156e-05,
+      "loss": 1.25,
+      "step": 12465
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.96446193818734e-05,
+      "loss": 1.2902,
+      "step": 12466
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.964451603842503e-05,
+      "loss": 1.2528,
+      "step": 12467
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.964441268000648e-05,
+      "loss": 1.2815,
+      "step": 12468
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.964430930661782e-05,
+      "loss": 1.2799,
+      "step": 12469
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.964420591825903e-05,
+      "loss": 1.2683,
+      "step": 12470
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.964410251493018e-05,
+      "loss": 1.2899,
+      "step": 12471
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.964399909663127e-05,
+      "loss": 1.2904,
+      "step": 12472
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.964389566336236e-05,
+      "loss": 1.2573,
+      "step": 12473
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.964379221512347e-05,
+      "loss": 1.2621,
+      "step": 12474
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.964368875191463e-05,
+      "loss": 1.2867,
+      "step": 12475
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.964358527373586e-05,
+      "loss": 1.3143,
+      "step": 12476
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.964348178058722e-05,
+      "loss": 1.311,
+      "step": 12477
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.96433782724687e-05,
+      "loss": 1.2696,
+      "step": 12478
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.964327474938037e-05,
+      "loss": 1.2883,
+      "step": 12479
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.964317121132225e-05,
+      "loss": 1.3081,
+      "step": 12480
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.964306765829437e-05,
+      "loss": 1.2611,
+      "step": 12481
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.964296409029675e-05,
+      "loss": 1.2763,
+      "step": 12482
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.964286050732943e-05,
+      "loss": 1.2791,
+      "step": 12483
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.964275690939244e-05,
+      "loss": 1.2728,
+      "step": 12484
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.964265329648583e-05,
+      "loss": 1.2841,
+      "step": 12485
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.96425496686096e-05,
+      "loss": 1.28,
+      "step": 12486
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.964244602576379e-05,
+      "loss": 1.2742,
+      "step": 12487
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.964234236794845e-05,
+      "loss": 1.2774,
+      "step": 12488
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.964223869516358e-05,
+      "loss": 1.3269,
+      "step": 12489
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.964213500740924e-05,
+      "loss": 1.2881,
+      "step": 12490
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.964203130468544e-05,
+      "loss": 1.2791,
+      "step": 12491
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.964192758699223e-05,
+      "loss": 1.2767,
+      "step": 12492
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.964182385432964e-05,
+      "loss": 1.3191,
+      "step": 12493
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.964172010669768e-05,
+      "loss": 1.3094,
+      "step": 12494
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.96416163440964e-05,
+      "loss": 1.3166,
+      "step": 12495
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.964151256652583e-05,
+      "loss": 1.296,
+      "step": 12496
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.964140877398599e-05,
+      "loss": 1.3074,
+      "step": 12497
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.964130496647692e-05,
+      "loss": 1.3004,
+      "step": 12498
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.964120114399866e-05,
+      "loss": 1.3024,
+      "step": 12499
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.964109730655122e-05,
+      "loss": 1.2653,
+      "step": 12500
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.964099345413465e-05,
+      "loss": 1.2873,
+      "step": 12501
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.964088958674898e-05,
+      "loss": 1.2997,
+      "step": 12502
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.964078570439421e-05,
+      "loss": 1.2902,
+      "step": 12503
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.964068180707043e-05,
+      "loss": 1.2917,
+      "step": 12504
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.96405778947776e-05,
+      "loss": 1.3009,
+      "step": 12505
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.964047396751583e-05,
+      "loss": 1.2666,
+      "step": 12506
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.964037002528509e-05,
+      "loss": 1.2875,
+      "step": 12507
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.964026606808543e-05,
+      "loss": 1.3103,
+      "step": 12508
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.964016209591689e-05,
+      "loss": 1.3031,
+      "step": 12509
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.96400581087795e-05,
+      "loss": 1.3243,
+      "step": 12510
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.963995410667328e-05,
+      "loss": 1.287,
+      "step": 12511
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.963985008959827e-05,
+      "loss": 1.3019,
+      "step": 12512
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.963974605755448e-05,
+      "loss": 1.2634,
+      "step": 12513
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.963964201054199e-05,
+      "loss": 1.2825,
+      "step": 12514
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.963953794856078e-05,
+      "loss": 1.2489,
+      "step": 12515
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.963943387161092e-05,
+      "loss": 1.2664,
+      "step": 12516
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.963932977969242e-05,
+      "loss": 1.2864,
+      "step": 12517
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.96392256728053e-05,
+      "loss": 1.2754,
+      "step": 12518
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.963912155094963e-05,
+      "loss": 1.2698,
+      "step": 12519
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.96390174141254e-05,
+      "loss": 1.2878,
+      "step": 12520
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.963891326233268e-05,
+      "loss": 1.2897,
+      "step": 12521
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.963880909557146e-05,
+      "loss": 1.2663,
+      "step": 12522
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.96387049138418e-05,
+      "loss": 1.2554,
+      "step": 12523
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.963860071714372e-05,
+      "loss": 1.2699,
+      "step": 12524
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.963849650547727e-05,
+      "loss": 1.3033,
+      "step": 12525
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.963839227884246e-05,
+      "loss": 1.2892,
+      "step": 12526
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.963828803723933e-05,
+      "loss": 1.324,
+      "step": 12527
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.963818378066791e-05,
+      "loss": 1.2834,
+      "step": 12528
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.963807950912823e-05,
+      "loss": 1.2652,
+      "step": 12529
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.963797522262033e-05,
+      "loss": 1.3271,
+      "step": 12530
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.963787092114421e-05,
+      "loss": 1.3061,
+      "step": 12531
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.963776660469995e-05,
+      "loss": 1.3066,
+      "step": 12532
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.963766227328754e-05,
+      "loss": 1.291,
+      "step": 12533
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.963755792690705e-05,
+      "loss": 1.2911,
+      "step": 12534
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.963745356555847e-05,
+      "loss": 1.255,
+      "step": 12535
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.963734918924187e-05,
+      "loss": 1.2861,
+      "step": 12536
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.963724479795725e-05,
+      "loss": 1.2868,
+      "step": 12537
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.963714039170466e-05,
+      "loss": 1.3163,
+      "step": 12538
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.963703597048413e-05,
+      "loss": 1.2953,
+      "step": 12539
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.963693153429567e-05,
+      "loss": 1.27,
+      "step": 12540
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.963682708313936e-05,
+      "loss": 1.2699,
+      "step": 12541
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.963672261701517e-05,
+      "loss": 1.2716,
+      "step": 12542
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.963661813592318e-05,
+      "loss": 1.3029,
+      "step": 12543
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.96365136398634e-05,
+      "loss": 1.3313,
+      "step": 12544
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.963640912883585e-05,
+      "loss": 1.3077,
+      "step": 12545
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.96363046028406e-05,
+      "loss": 1.279,
+      "step": 12546
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.963620006187763e-05,
+      "loss": 1.3055,
+      "step": 12547
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.963609550594702e-05,
+      "loss": 1.2623,
+      "step": 12548
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.963599093504878e-05,
+      "loss": 1.2826,
+      "step": 12549
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.963588634918294e-05,
+      "loss": 1.2977,
+      "step": 12550
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.963578174834953e-05,
+      "loss": 1.2721,
+      "step": 12551
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.96356771325486e-05,
+      "loss": 1.3055,
+      "step": 12552
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.963557250178014e-05,
+      "loss": 1.2752,
+      "step": 12553
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.963546785604423e-05,
+      "loss": 1.2884,
+      "step": 12554
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.963536319534089e-05,
+      "loss": 1.2769,
+      "step": 12555
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.963525851967013e-05,
+      "loss": 1.28,
+      "step": 12556
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.9635153829032e-05,
+      "loss": 1.2586,
+      "step": 12557
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.963504912342651e-05,
+      "loss": 1.2254,
+      "step": 12558
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.96349444028537e-05,
+      "loss": 1.2704,
+      "step": 12559
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.963483966731364e-05,
+      "loss": 1.3064,
+      "step": 12560
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.96347349168063e-05,
+      "loss": 1.2988,
+      "step": 12561
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.963463015133177e-05,
+      "loss": 1.2669,
+      "step": 12562
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.963452537089002e-05,
+      "loss": 1.2682,
+      "step": 12563
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.963442057548114e-05,
+      "loss": 1.2889,
+      "step": 12564
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.963431576510512e-05,
+      "loss": 1.2708,
+      "step": 12565
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.963421093976202e-05,
+      "loss": 1.2966,
+      "step": 12566
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.963410609945186e-05,
+      "loss": 1.2894,
+      "step": 12567
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.963400124417466e-05,
+      "loss": 1.3116,
+      "step": 12568
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.963389637393048e-05,
+      "loss": 1.2812,
+      "step": 12569
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.963379148871932e-05,
+      "loss": 1.2439,
+      "step": 12570
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.963368658854122e-05,
+      "loss": 1.273,
+      "step": 12571
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.963358167339623e-05,
+      "loss": 1.2708,
+      "step": 12572
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.963347674328436e-05,
+      "loss": 1.3045,
+      "step": 12573
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.963337179820567e-05,
+      "loss": 1.3032,
+      "step": 12574
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.963326683816016e-05,
+      "loss": 1.3017,
+      "step": 12575
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.963316186314786e-05,
+      "loss": 1.2595,
+      "step": 12576
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.963305687316884e-05,
+      "loss": 1.286,
+      "step": 12577
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.96329518682231e-05,
+      "loss": 1.2934,
+      "step": 12578
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.963284684831067e-05,
+      "loss": 1.3033,
+      "step": 12579
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.96327418134316e-05,
+      "loss": 1.3109,
+      "step": 12580
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.963263676358591e-05,
+      "loss": 1.2779,
+      "step": 12581
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.963253169877363e-05,
+      "loss": 1.285,
+      "step": 12582
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.96324266189948e-05,
+      "loss": 1.2972,
+      "step": 12583
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.963232152424944e-05,
+      "loss": 1.2708,
+      "step": 12584
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.963221641453759e-05,
+      "loss": 1.2874,
+      "step": 12585
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.96321112898593e-05,
+      "loss": 1.2845,
+      "step": 12586
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.963200615021456e-05,
+      "loss": 1.3004,
+      "step": 12587
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.963190099560344e-05,
+      "loss": 1.262,
+      "step": 12588
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.963179582602595e-05,
+      "loss": 1.3238,
+      "step": 12589
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.963169064148212e-05,
+      "loss": 1.3101,
+      "step": 12590
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.9631585441972e-05,
+      "loss": 1.2579,
+      "step": 12591
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.963148022749561e-05,
+      "loss": 1.2967,
+      "step": 12592
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.963137499805298e-05,
+      "loss": 1.2914,
+      "step": 12593
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.963126975364414e-05,
+      "loss": 1.2749,
+      "step": 12594
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.963116449426914e-05,
+      "loss": 1.3557,
+      "step": 12595
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.963105921992799e-05,
+      "loss": 1.2701,
+      "step": 12596
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.963095393062073e-05,
+      "loss": 1.3023,
+      "step": 12597
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.96308486263474e-05,
+      "loss": 1.2587,
+      "step": 12598
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.9630743307108e-05,
+      "loss": 1.2926,
+      "step": 12599
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.96306379729026e-05,
+      "loss": 1.3292,
+      "step": 12600
+    }
+  ],
+  "max_steps": 186510,
+  "num_train_epochs": 30,
+  "total_flos": 2.406271807515525e+19,
+  "trial_name": null,
+  "trial_params": null
+}