diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,49273 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 4.0,
+  "global_step": 16416,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.113590263691684e-08,
+      "loss": 1.2614,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6227180527383367e-07,
+      "loss": 0.5289,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.434077079107505e-07,
+      "loss": 0.4145,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.2454361054766735e-07,
+      "loss": 0.4136,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.056795131845842e-07,
+      "loss": 0.414,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.86815415821501e-07,
+      "loss": 0.4016,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.679513184584178e-07,
+      "loss": 0.4036,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.490872210953347e-07,
+      "loss": 0.3919,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.302231237322515e-07,
+      "loss": 0.3831,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.113590263691684e-07,
+      "loss": 0.3727,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.924949290060852e-07,
+      "loss": 0.3496,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.73630831643002e-07,
+      "loss": 0.3871,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0547667342799188e-06,
+      "loss": 0.3724,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1359026369168357e-06,
+      "loss": 0.3791,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2170385395537525e-06,
+      "loss": 0.3617,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2981744421906694e-06,
+      "loss": 0.3574,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3793103448275862e-06,
+      "loss": 0.3616,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.460446247464503e-06,
+      "loss": 0.3585,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.54158215010142e-06,
+      "loss": 0.3509,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6227180527383368e-06,
+      "loss": 0.3897,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7038539553752536e-06,
+      "loss": 0.3582,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7849898580121705e-06,
+      "loss": 0.3566,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8661257606490873e-06,
+      "loss": 0.3368,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.947261663286004e-06,
+      "loss": 0.3712,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.028397565922921e-06,
+      "loss": 0.3633,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1095334685598377e-06,
+      "loss": 0.3518,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1906693711967548e-06,
+      "loss": 0.3629,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2718052738336714e-06,
+      "loss": 0.3377,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3529411764705885e-06,
+      "loss": 0.3539,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.434077079107505e-06,
+      "loss": 0.3736,
+      "step": 60
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.515212981744422e-06,
+      "loss": 0.3617,
+      "step": 62
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.596348884381339e-06,
+      "loss": 0.3517,
+      "step": 64
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.677484787018256e-06,
+      "loss": 0.3694,
+      "step": 66
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.7586206896551725e-06,
+      "loss": 0.3483,
+      "step": 68
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.8397565922920896e-06,
+      "loss": 0.3322,
+      "step": 70
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.920892494929006e-06,
+      "loss": 0.3295,
+      "step": 72
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.0020283975659233e-06,
+      "loss": 0.3514,
+      "step": 74
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.08316430020284e-06,
+      "loss": 0.3366,
+      "step": 76
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.164300202839757e-06,
+      "loss": 0.3086,
+      "step": 78
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.2454361054766736e-06,
+      "loss": 0.3452,
+      "step": 80
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.3265720081135907e-06,
+      "loss": 0.2971,
+      "step": 82
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.4077079107505073e-06,
+      "loss": 0.3384,
+      "step": 84
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.4888438133874244e-06,
+      "loss": 0.3428,
+      "step": 86
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.569979716024341e-06,
+      "loss": 0.3635,
+      "step": 88
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.651115618661258e-06,
+      "loss": 0.3625,
+      "step": 90
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7322515212981747e-06,
+      "loss": 0.3502,
+      "step": 92
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.8133874239350913e-06,
+      "loss": 0.3678,
+      "step": 94
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.894523326572008e-06,
+      "loss": 0.3365,
+      "step": 96
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.975659229208925e-06,
+      "loss": 0.3364,
+      "step": 98
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.056795131845842e-06,
+      "loss": 0.3526,
+      "step": 100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.137931034482759e-06,
+      "loss": 0.3411,
+      "step": 102
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.219066937119675e-06,
+      "loss": 0.3537,
+      "step": 104
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.3002028397565924e-06,
+      "loss": 0.3394,
+      "step": 106
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.3813387423935095e-06,
+      "loss": 0.3474,
+      "step": 108
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.4624746450304266e-06,
+      "loss": 0.3248,
+      "step": 110
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.543610547667343e-06,
+      "loss": 0.3571,
+      "step": 112
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.62474645030426e-06,
+      "loss": 0.3544,
+      "step": 114
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.705882352941177e-06,
+      "loss": 0.3519,
+      "step": 116
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.787018255578094e-06,
+      "loss": 0.3528,
+      "step": 118
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.86815415821501e-06,
+      "loss": 0.3595,
+      "step": 120
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.949290060851927e-06,
+      "loss": 0.3573,
+      "step": 122
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.030425963488844e-06,
+      "loss": 0.3196,
+      "step": 124
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.111561866125761e-06,
+      "loss": 0.3401,
+      "step": 126
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.192697768762678e-06,
+      "loss": 0.3253,
+      "step": 128
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.2738336713995955e-06,
+      "loss": 0.3246,
+      "step": 130
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.354969574036512e-06,
+      "loss": 0.3478,
+      "step": 132
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.436105476673429e-06,
+      "loss": 0.3498,
+      "step": 134
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.517241379310345e-06,
+      "loss": 0.3404,
+      "step": 136
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.598377281947263e-06,
+      "loss": 0.3578,
+      "step": 138
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.679513184584179e-06,
+      "loss": 0.3445,
+      "step": 140
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.760649087221096e-06,
+      "loss": 0.3386,
+      "step": 142
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.841784989858012e-06,
+      "loss": 0.3562,
+      "step": 144
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.92292089249493e-06,
+      "loss": 0.3339,
+      "step": 146
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.0040567951318465e-06,
+      "loss": 0.3415,
+      "step": 148
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.0851926977687636e-06,
+      "loss": 0.3352,
+      "step": 150
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.16632860040568e-06,
+      "loss": 0.3339,
+      "step": 152
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.247464503042598e-06,
+      "loss": 0.3511,
+      "step": 154
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.328600405679514e-06,
+      "loss": 0.3339,
+      "step": 156
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.409736308316431e-06,
+      "loss": 0.312,
+      "step": 158
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.490872210953347e-06,
+      "loss": 0.3313,
+      "step": 160
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.572008113590265e-06,
+      "loss": 0.3529,
+      "step": 162
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.653144016227181e-06,
+      "loss": 0.3632,
+      "step": 164
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.734279918864098e-06,
+      "loss": 0.3218,
+      "step": 166
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.815415821501015e-06,
+      "loss": 0.3444,
+      "step": 168
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.896551724137932e-06,
+      "loss": 0.3473,
+      "step": 170
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 6.977687626774849e-06,
+      "loss": 0.3323,
+      "step": 172
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.058823529411766e-06,
+      "loss": 0.3382,
+      "step": 174
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.139959432048682e-06,
+      "loss": 0.324,
+      "step": 176
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.221095334685599e-06,
+      "loss": 0.3314,
+      "step": 178
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.302231237322516e-06,
+      "loss": 0.3148,
+      "step": 180
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.383367139959433e-06,
+      "loss": 0.3384,
+      "step": 182
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.464503042596349e-06,
+      "loss": 0.3375,
+      "step": 184
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.5456389452332665e-06,
+      "loss": 0.3501,
+      "step": 186
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.626774847870183e-06,
+      "loss": 0.3491,
+      "step": 188
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.7079107505071e-06,
+      "loss": 0.353,
+      "step": 190
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.789046653144016e-06,
+      "loss": 0.3278,
+      "step": 192
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.870182555780935e-06,
+      "loss": 0.3406,
+      "step": 194
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.95131845841785e-06,
+      "loss": 0.3584,
+      "step": 196
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.032454361054767e-06,
+      "loss": 0.3401,
+      "step": 198
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.113590263691684e-06,
+      "loss": 0.3356,
+      "step": 200
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.194726166328601e-06,
+      "loss": 0.3439,
+      "step": 202
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.275862068965518e-06,
+      "loss": 0.3372,
+      "step": 204
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.356997971602435e-06,
+      "loss": 0.3373,
+      "step": 206
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.43813387423935e-06,
+      "loss": 0.3266,
+      "step": 208
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.51926977687627e-06,
+      "loss": 0.3466,
+      "step": 210
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.600405679513185e-06,
+      "loss": 0.3455,
+      "step": 212
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.681541582150102e-06,
+      "loss": 0.3269,
+      "step": 214
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.762677484787019e-06,
+      "loss": 0.3052,
+      "step": 216
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.843813387423936e-06,
+      "loss": 0.3405,
+      "step": 218
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 8.924949290060853e-06,
+      "loss": 0.3277,
+      "step": 220
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.00608519269777e-06,
+      "loss": 0.3151,
+      "step": 222
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.087221095334686e-06,
+      "loss": 0.3159,
+      "step": 224
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.168356997971604e-06,
+      "loss": 0.3493,
+      "step": 226
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.24949290060852e-06,
+      "loss": 0.3403,
+      "step": 228
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.330628803245437e-06,
+      "loss": 0.3386,
+      "step": 230
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.411764705882354e-06,
+      "loss": 0.3393,
+      "step": 232
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.492900608519271e-06,
+      "loss": 0.3426,
+      "step": 234
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.574036511156188e-06,
+      "loss": 0.3531,
+      "step": 236
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.655172413793105e-06,
+      "loss": 0.351,
+      "step": 238
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.73630831643002e-06,
+      "loss": 0.3543,
+      "step": 240
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.817444219066939e-06,
+      "loss": 0.3378,
+      "step": 242
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.898580121703854e-06,
+      "loss": 0.3288,
+      "step": 244
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.979716024340772e-06,
+      "loss": 0.3512,
+      "step": 246
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.0060851926977689e-05,
+      "loss": 0.3391,
+      "step": 248
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.0141987829614606e-05,
+      "loss": 0.3461,
+      "step": 250
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.0223123732251523e-05,
+      "loss": 0.3246,
+      "step": 252
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.0304259634888438e-05,
+      "loss": 0.325,
+      "step": 254
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.0385395537525355e-05,
+      "loss": 0.3457,
+      "step": 256
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.0466531440162272e-05,
+      "loss": 0.3072,
+      "step": 258
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.0547667342799191e-05,
+      "loss": 0.3352,
+      "step": 260
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.0628803245436106e-05,
+      "loss": 0.3282,
+      "step": 262
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.0709939148073023e-05,
+      "loss": 0.3274,
+      "step": 264
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.079107505070994e-05,
+      "loss": 0.3467,
+      "step": 266
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.0872210953346858e-05,
+      "loss": 0.3319,
+      "step": 268
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.0953346855983773e-05,
+      "loss": 0.3206,
+      "step": 270
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.103448275862069e-05,
+      "loss": 0.3468,
+      "step": 272
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1115618661257607e-05,
+      "loss": 0.3423,
+      "step": 274
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1196754563894526e-05,
+      "loss": 0.3467,
+      "step": 276
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1277890466531441e-05,
+      "loss": 0.3367,
+      "step": 278
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1359026369168358e-05,
+      "loss": 0.3399,
+      "step": 280
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1440162271805275e-05,
+      "loss": 0.33,
+      "step": 282
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1521298174442192e-05,
+      "loss": 0.3267,
+      "step": 284
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1602434077079108e-05,
+      "loss": 0.3496,
+      "step": 286
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1683569979716025e-05,
+      "loss": 0.3487,
+      "step": 288
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1764705882352942e-05,
+      "loss": 0.3477,
+      "step": 290
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.184584178498986e-05,
+      "loss": 0.3298,
+      "step": 292
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.1926977687626774e-05,
+      "loss": 0.3296,
+      "step": 294
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.2008113590263693e-05,
+      "loss": 0.3371,
+      "step": 296
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.208924949290061e-05,
+      "loss": 0.3444,
+      "step": 298
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.2170385395537527e-05,
+      "loss": 0.3382,
+      "step": 300
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.2251521298174443e-05,
+      "loss": 0.3259,
+      "step": 302
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.233265720081136e-05,
+      "loss": 0.3469,
+      "step": 304
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.2413793103448277e-05,
+      "loss": 0.317,
+      "step": 306
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.2494929006085195e-05,
+      "loss": 0.3263,
+      "step": 308
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.2576064908722109e-05,
+      "loss": 0.3204,
+      "step": 310
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.2657200811359028e-05,
+      "loss": 0.3197,
+      "step": 312
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.2738336713995945e-05,
+      "loss": 0.3275,
+      "step": 314
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.2819472616632862e-05,
+      "loss": 0.3183,
+      "step": 316
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.2900608519269777e-05,
+      "loss": 0.3363,
+      "step": 318
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.2981744421906694e-05,
+      "loss": 0.3328,
+      "step": 320
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.3062880324543611e-05,
+      "loss": 0.3409,
+      "step": 322
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.314401622718053e-05,
+      "loss": 0.3373,
+      "step": 324
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.3225152129817444e-05,
+      "loss": 0.3331,
+      "step": 326
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.3306288032454363e-05,
+      "loss": 0.3247,
+      "step": 328
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.338742393509128e-05,
+      "loss": 0.3305,
+      "step": 330
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.3468559837728197e-05,
+      "loss": 0.3329,
+      "step": 332
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.3549695740365112e-05,
+      "loss": 0.332,
+      "step": 334
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.363083164300203e-05,
+      "loss": 0.3168,
+      "step": 336
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.3711967545638946e-05,
+      "loss": 0.3176,
+      "step": 338
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.3793103448275863e-05,
+      "loss": 0.312,
+      "step": 340
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.3874239350912779e-05,
+      "loss": 0.3378,
+      "step": 342
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.3955375253549697e-05,
+      "loss": 0.338,
+      "step": 344
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.4036511156186615e-05,
+      "loss": 0.3286,
+      "step": 346
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.4117647058823532e-05,
+      "loss": 0.3103,
+      "step": 348
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4198782961460447e-05,
+      "loss": 0.3186,
+      "step": 350
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4279918864097364e-05,
+      "loss": 0.3559,
+      "step": 352
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4361054766734281e-05,
+      "loss": 0.313,
+      "step": 354
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4442190669371198e-05,
+      "loss": 0.3274,
+      "step": 356
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4523326572008113e-05,
+      "loss": 0.3427,
+      "step": 358
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4604462474645032e-05,
+      "loss": 0.3379,
+      "step": 360
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.468559837728195e-05,
+      "loss": 0.3384,
+      "step": 362
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4766734279918866e-05,
+      "loss": 0.3395,
+      "step": 364
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4847870182555782e-05,
+      "loss": 0.3361,
+      "step": 366
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.4929006085192699e-05,
+      "loss": 0.3459,
+      "step": 368
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.5010141987829616e-05,
+      "loss": 0.3415,
+      "step": 370
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.5091277890466533e-05,
+      "loss": 0.3399,
+      "step": 372
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.5172413793103448e-05,
+      "loss": 0.3289,
+      "step": 374
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.5253549695740365e-05,
+      "loss": 0.3346,
+      "step": 376
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.5334685598377284e-05,
+      "loss": 0.3278,
+      "step": 378
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.54158215010142e-05,
+      "loss": 0.3178,
+      "step": 380
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.5496957403651115e-05,
+      "loss": 0.3179,
+      "step": 382
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.5578093306288032e-05,
+      "loss": 0.3195,
+      "step": 384
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.5659229208924952e-05,
+      "loss": 0.3376,
+      "step": 386
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.574036511156187e-05,
+      "loss": 0.3371,
+      "step": 388
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.5821501014198783e-05,
+      "loss": 0.3258,
+      "step": 390
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.59026369168357e-05,
+      "loss": 0.3207,
+      "step": 392
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.5983772819472617e-05,
+      "loss": 0.3065,
+      "step": 394
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.6064908722109534e-05,
+      "loss": 0.3562,
+      "step": 396
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.614604462474645e-05,
+      "loss": 0.3228,
+      "step": 398
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.622718052738337e-05,
+      "loss": 0.3368,
+      "step": 400
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.6308316430020285e-05,
+      "loss": 0.3478,
+      "step": 402
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.6389452332657203e-05,
+      "loss": 0.3398,
+      "step": 404
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.647058823529412e-05,
+      "loss": 0.3244,
+      "step": 406
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.6551724137931037e-05,
+      "loss": 0.3294,
+      "step": 408
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.6632860040567954e-05,
+      "loss": 0.3381,
+      "step": 410
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.671399594320487e-05,
+      "loss": 0.3363,
+      "step": 412
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.6795131845841784e-05,
+      "loss": 0.3429,
+      "step": 414
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.68762677484787e-05,
+      "loss": 0.337,
+      "step": 416
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.6957403651115622e-05,
+      "loss": 0.327,
+      "step": 418
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.703853955375254e-05,
+      "loss": 0.3235,
+      "step": 420
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.7119675456389453e-05,
+      "loss": 0.3387,
+      "step": 422
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.720081135902637e-05,
+      "loss": 0.336,
+      "step": 424
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.7281947261663287e-05,
+      "loss": 0.3452,
+      "step": 426
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.7363083164300204e-05,
+      "loss": 0.3181,
+      "step": 428
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.744421906693712e-05,
+      "loss": 0.3369,
+      "step": 430
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.7525354969574038e-05,
+      "loss": 0.3506,
+      "step": 432
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.7606490872210955e-05,
+      "loss": 0.3135,
+      "step": 434
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.7687626774847872e-05,
+      "loss": 0.3127,
+      "step": 436
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.776876267748479e-05,
+      "loss": 0.35,
+      "step": 438
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.7849898580121706e-05,
+      "loss": 0.3423,
+      "step": 440
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.7931034482758623e-05,
+      "loss": 0.3362,
+      "step": 442
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.801217038539554e-05,
+      "loss": 0.3,
+      "step": 444
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.8093306288032454e-05,
+      "loss": 0.3476,
+      "step": 446
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.817444219066937e-05,
+      "loss": 0.3289,
+      "step": 448
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.8255578093306288e-05,
+      "loss": 0.3418,
+      "step": 450
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.833671399594321e-05,
+      "loss": 0.3152,
+      "step": 452
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.8417849898580122e-05,
+      "loss": 0.3405,
+      "step": 454
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.849898580121704e-05,
+      "loss": 0.3378,
+      "step": 456
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.8580121703853956e-05,
+      "loss": 0.3447,
+      "step": 458
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.8661257606490873e-05,
+      "loss": 0.3415,
+      "step": 460
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.874239350912779e-05,
+      "loss": 0.2988,
+      "step": 462
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.8823529411764708e-05,
+      "loss": 0.3443,
+      "step": 464
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.8904665314401625e-05,
+      "loss": 0.3468,
+      "step": 466
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.8985801217038542e-05,
+      "loss": 0.3389,
+      "step": 468
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.906693711967546e-05,
+      "loss": 0.3202,
+      "step": 470
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9148073022312376e-05,
+      "loss": 0.3352,
+      "step": 472
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9229208924949293e-05,
+      "loss": 0.3316,
+      "step": 474
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.931034482758621e-05,
+      "loss": 0.3248,
+      "step": 476
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9391480730223124e-05,
+      "loss": 0.3419,
+      "step": 478
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.947261663286004e-05,
+      "loss": 0.3124,
+      "step": 480
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9553752535496958e-05,
+      "loss": 0.4068,
+      "step": 482
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9634888438133878e-05,
+      "loss": 0.3444,
+      "step": 484
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9716024340770792e-05,
+      "loss": 0.3553,
+      "step": 486
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.979716024340771e-05,
+      "loss": 0.3286,
+      "step": 488
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9878296146044626e-05,
+      "loss": 0.3579,
+      "step": 490
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9959432048681543e-05,
+      "loss": 0.3307,
+      "step": 492
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999999980536544e-05,
+      "loss": 0.3232,
+      "step": 494
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9999998248289005e-05,
+      "loss": 0.3437,
+      "step": 496
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9999995134136373e-05,
+      "loss": 0.328,
+      "step": 498
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9999990462908037e-05,
+      "loss": 0.3468,
+      "step": 500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9999984234604716e-05,
+      "loss": 0.3406,
+      "step": 502
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9999976449227386e-05,
+      "loss": 0.3182,
+      "step": 504
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9999967106777253e-05,
+      "loss": 0.3249,
+      "step": 506
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9999956207255773e-05,
+      "loss": 0.3209,
+      "step": 508
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9999943750664653e-05,
+      "loss": 0.3341,
+      "step": 510
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9999929737005818e-05,
+      "loss": 0.3335,
+      "step": 512
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999991416628146e-05,
+      "loss": 0.3344,
+      "step": 514
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9999897038494e-05,
+      "loss": 0.3374,
+      "step": 516
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999987835364611e-05,
+      "loss": 0.3262,
+      "step": 518
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9999858111740688e-05,
+      "loss": 0.3144,
+      "step": 520
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9999836312780895e-05,
+      "loss": 0.3221,
+      "step": 522
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9999812956770125e-05,
+      "loss": 0.3367,
+      "step": 524
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999978804371201e-05,
+      "loss": 0.3143,
+      "step": 526
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9999761573610432e-05,
+      "loss": 0.3397,
+      "step": 528
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9999733546469514e-05,
+      "loss": 0.3183,
+      "step": 530
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9999703962293612e-05,
+      "loss": 0.3474,
+      "step": 532
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9999672821087347e-05,
+      "loss": 0.3183,
+      "step": 534
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9999640122855556e-05,
+      "loss": 0.3377,
+      "step": 536
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9999605867603333e-05,
+      "loss": 0.3219,
+      "step": 538
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9999570055336014e-05,
+      "loss": 0.3197,
+      "step": 540
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9999532686059175e-05,
+      "loss": 0.3282,
+      "step": 542
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9999493759778635e-05,
+      "loss": 0.3137,
+      "step": 544
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999945327650045e-05,
+      "loss": 0.3431,
+      "step": 546
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999941123623093e-05,
+      "loss": 0.3354,
+      "step": 548
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999936763897662e-05,
+      "loss": 0.3173,
+      "step": 550
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9999322484744305e-05,
+      "loss": 0.3322,
+      "step": 552
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9999275773541016e-05,
+      "loss": 0.3082,
+      "step": 554
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9999227505374033e-05,
+      "loss": 0.3208,
+      "step": 556
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9999177680250863e-05,
+      "loss": 0.3308,
+      "step": 558
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.999912629817927e-05,
+      "loss": 0.3151,
+      "step": 560
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.999907335916725e-05,
+      "loss": 0.295,
+      "step": 562
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.999901886322305e-05,
+      "loss": 0.3134,
+      "step": 564
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9998962810355152e-05,
+      "loss": 0.3051,
+      "step": 566
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.999890520057229e-05,
+      "loss": 0.327,
+      "step": 568
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9998846033883427e-05,
+      "loss": 0.3553,
+      "step": 570
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.999878531029778e-05,
+      "loss": 0.3285,
+      "step": 572
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.99987230298248e-05,
+      "loss": 0.2928,
+      "step": 574
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9998659192474193e-05,
+      "loss": 0.3042,
+      "step": 576
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.999859379825589e-05,
+      "loss": 0.3498,
+      "step": 578
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.999852684718008e-05,
+      "loss": 0.3228,
+      "step": 580
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.999845833925718e-05,
+      "loss": 0.3335,
+      "step": 582
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9998388274497864e-05,
+      "loss": 0.3374,
+      "step": 584
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9998316652913038e-05,
+      "loss": 0.3328,
+      "step": 586
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.999824347451386e-05,
+      "loss": 0.3158,
+      "step": 588
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9998168739311715e-05,
+      "loss": 0.2963,
+      "step": 590
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9998092447318247e-05,
+      "loss": 0.3467,
+      "step": 592
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9998014598545335e-05,
+      "loss": 0.3319,
+      "step": 594
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9997935193005093e-05,
+      "loss": 0.3027,
+      "step": 596
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9997854230709896e-05,
+      "loss": 0.32,
+      "step": 598
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9997771711672343e-05,
+      "loss": 0.315,
+      "step": 600
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9997687635905284e-05,
+      "loss": 0.331,
+      "step": 602
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9997602003421815e-05,
+      "loss": 0.3158,
+      "step": 604
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9997514814235264e-05,
+      "loss": 0.2994,
+      "step": 606
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.999742606835921e-05,
+      "loss": 0.3224,
+      "step": 608
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9997335765807463e-05,
+      "loss": 0.3146,
+      "step": 610
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9997243906594098e-05,
+      "loss": 0.3013,
+      "step": 612
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9997150490733412e-05,
+      "loss": 0.3083,
+      "step": 614
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9997055518239947e-05,
+      "loss": 0.3039,
+      "step": 616
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9996958989128498e-05,
+      "loss": 0.3265,
+      "step": 618
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9996860903414085e-05,
+      "loss": 0.3186,
+      "step": 620
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.999676126111199e-05,
+      "loss": 0.3373,
+      "step": 622
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9996660062237723e-05,
+      "loss": 0.2947,
+      "step": 624
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9996557306807046e-05,
+      "loss": 0.3157,
+      "step": 626
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9996452994835954e-05,
+      "loss": 0.3326,
+      "step": 628
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9996347126340692e-05,
+      "loss": 0.3033,
+      "step": 630
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9996239701337744e-05,
+      "loss": 0.3376,
+      "step": 632
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.999613071984384e-05,
+      "loss": 0.3279,
+      "step": 634
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9996020181875942e-05,
+      "loss": 0.3207,
+      "step": 636
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9995908087451264e-05,
+      "loss": 0.3149,
+      "step": 638
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9995794436587266e-05,
+      "loss": 0.3375,
+      "step": 640
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999567922930164e-05,
+      "loss": 0.3287,
+      "step": 642
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9995562465612322e-05,
+      "loss": 0.3353,
+      "step": 644
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9995444145537494e-05,
+      "loss": 0.3311,
+      "step": 646
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9995324269095585e-05,
+      "loss": 0.3122,
+      "step": 648
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9995202836305255e-05,
+      "loss": 0.2925,
+      "step": 650
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999507984718541e-05,
+      "loss": 0.3346,
+      "step": 652
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999495530175521e-05,
+      "loss": 0.3232,
+      "step": 654
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9994829200034037e-05,
+      "loss": 0.3189,
+      "step": 656
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9994701542041533e-05,
+      "loss": 0.3218,
+      "step": 658
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999457232779757e-05,
+      "loss": 0.3283,
+      "step": 660
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9994441557322273e-05,
+      "loss": 0.3224,
+      "step": 662
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9994309230636003e-05,
+      "loss": 0.3298,
+      "step": 664
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999417534775936e-05,
+      "loss": 0.3048,
+      "step": 666
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9994039908713194e-05,
+      "loss": 0.3109,
+      "step": 668
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9993902913518593e-05,
+      "loss": 0.3329,
+      "step": 670
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999376436219689e-05,
+      "loss": 0.3356,
+      "step": 672
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9993624254769655e-05,
+      "loss": 0.314,
+      "step": 674
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9993482591258706e-05,
+      "loss": 0.3093,
+      "step": 676
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.99933393716861e-05,
+      "loss": 0.3187,
+      "step": 678
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.999319459607414e-05,
+      "loss": 0.3038,
+      "step": 680
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9993048264445368e-05,
+      "loss": 0.3096,
+      "step": 682
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.999290037682257e-05,
+      "loss": 0.3198,
+      "step": 684
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9992750933228768e-05,
+      "loss": 0.3216,
+      "step": 686
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9992599933687233e-05,
+      "loss": 0.3128,
+      "step": 688
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.999244737822148e-05,
+      "loss": 0.3169,
+      "step": 690
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.999229326685526e-05,
+      "loss": 0.3119,
+      "step": 692
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9992137599612573e-05,
+      "loss": 0.3272,
+      "step": 694
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.999198037651765e-05,
+      "loss": 0.3413,
+      "step": 696
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9991821597594983e-05,
+      "loss": 0.3305,
+      "step": 698
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.999166126286929e-05,
+      "loss": 0.3104,
+      "step": 700
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.999149937236553e-05,
+      "loss": 0.3277,
+      "step": 702
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9991335926108923e-05,
+      "loss": 0.3228,
+      "step": 704
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9991170924124906e-05,
+      "loss": 0.3249,
+      "step": 706
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.999100436643918e-05,
+      "loss": 0.3304,
+      "step": 708
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9990836253077677e-05,
+      "loss": 0.3435,
+      "step": 710
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.999066658406657e-05,
+      "loss": 0.3404,
+      "step": 712
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9990495359432286e-05,
+      "loss": 0.3116,
+      "step": 714
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9990322579201476e-05,
+      "loss": 0.3237,
+      "step": 716
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9990148243401048e-05,
+      "loss": 0.3107,
+      "step": 718
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.998997235205815e-05,
+      "loss": 0.3326,
+      "step": 720
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9989794905200167e-05,
+      "loss": 0.3296,
+      "step": 722
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9989615902854726e-05,
+      "loss": 0.3097,
+      "step": 724
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9989435345049704e-05,
+      "loss": 0.3424,
+      "step": 726
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.998925323181321e-05,
+      "loss": 0.3295,
+      "step": 728
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9989069563173606e-05,
+      "loss": 0.3117,
+      "step": 730
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9988884339159487e-05,
+      "loss": 0.3138,
+      "step": 732
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9988697559799696e-05,
+      "loss": 0.3237,
+      "step": 734
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9988509225123317e-05,
+      "loss": 0.2955,
+      "step": 736
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.998831933515967e-05,
+      "loss": 0.3375,
+      "step": 738
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9988127889938324e-05,
+      "loss": 0.3289,
+      "step": 740
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.998793488948909e-05,
+      "loss": 0.3463,
+      "step": 742
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.998774033384202e-05,
+      "loss": 0.3198,
+      "step": 744
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9987544223027412e-05,
+      "loss": 0.3082,
+      "step": 746
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9987346557075792e-05,
+      "loss": 0.3383,
+      "step": 748
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9987147336017945e-05,
+      "loss": 0.325,
+      "step": 750
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.998694655988489e-05,
+      "loss": 0.3152,
+      "step": 752
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9986744228707888e-05,
+      "loss": 0.3224,
+      "step": 754
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9986540342518445e-05,
+      "loss": 0.3282,
+      "step": 756
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9986334901348307e-05,
+      "loss": 0.3152,
+      "step": 758
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9986127905229463e-05,
+      "loss": 0.3278,
+      "step": 760
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9985919354194142e-05,
+      "loss": 0.3435,
+      "step": 762
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9985709248274822e-05,
+      "loss": 0.3143,
+      "step": 764
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9985497587504213e-05,
+      "loss": 0.326,
+      "step": 766
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9985284371915273e-05,
+      "loss": 0.3432,
+      "step": 768
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.99850696015412e-05,
+      "loss": 0.3417,
+      "step": 770
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9984853276415444e-05,
+      "loss": 0.3378,
+      "step": 772
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.998463539657168e-05,
+      "loss": 0.324,
+      "step": 774
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9984415962043835e-05,
+      "loss": 0.3303,
+      "step": 776
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9984194972866076e-05,
+      "loss": 0.2976,
+      "step": 778
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9983972429072814e-05,
+      "loss": 0.3077,
+      "step": 780
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9983748330698703e-05,
+      "loss": 0.3179,
+      "step": 782
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9983522677778634e-05,
+      "loss": 0.3089,
+      "step": 784
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9983295470347745e-05,
+      "loss": 0.3315,
+      "step": 786
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.998306670844141e-05,
+      "loss": 0.3128,
+      "step": 788
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.998283639209525e-05,
+      "loss": 0.3377,
+      "step": 790
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9982604521345132e-05,
+      "loss": 0.3234,
+      "step": 792
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9982371096227155e-05,
+      "loss": 0.3373,
+      "step": 794
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9982136116777666e-05,
+      "loss": 0.3241,
+      "step": 796
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9981899583033255e-05,
+      "loss": 0.3161,
+      "step": 798
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.998166149503075e-05,
+      "loss": 0.3227,
+      "step": 800
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9981421852807227e-05,
+      "loss": 0.3276,
+      "step": 802
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9981180656399993e-05,
+      "loss": 0.3242,
+      "step": 804
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.998093790584661e-05,
+      "loss": 0.3239,
+      "step": 806
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9980693601184874e-05,
+      "loss": 0.3241,
+      "step": 808
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9980447742452823e-05,
+      "loss": 0.3365,
+      "step": 810
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9980200329688742e-05,
+      "loss": 0.3305,
+      "step": 812
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.997995136293116e-05,
+      "loss": 0.3351,
+      "step": 814
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9979700842218833e-05,
+      "loss": 0.3169,
+      "step": 816
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9979448767590776e-05,
+      "loss": 0.3424,
+      "step": 818
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9979195139086235e-05,
+      "loss": 0.3177,
+      "step": 820
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.99789399567447e-05,
+      "loss": 0.3276,
+      "step": 822
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9978683220605912e-05,
+      "loss": 0.335,
+      "step": 824
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9978424930709842e-05,
+      "loss": 0.3291,
+      "step": 826
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.997816508709671e-05,
+      "loss": 0.3103,
+      "step": 828
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9977903689806975e-05,
+      "loss": 0.3105,
+      "step": 830
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9977640738881337e-05,
+      "loss": 0.3115,
+      "step": 832
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.997737623436074e-05,
+      "loss": 0.3296,
+      "step": 834
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.997711017628637e-05,
+      "loss": 0.3187,
+      "step": 836
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9976842564699654e-05,
+      "loss": 0.299,
+      "step": 838
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.997657339964226e-05,
+      "loss": 0.3225,
+      "step": 840
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9976302681156103e-05,
+      "loss": 0.3118,
+      "step": 842
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.997603040928333e-05,
+      "loss": 0.3034,
+      "step": 844
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.997575658406634e-05,
+      "loss": 0.2916,
+      "step": 846
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.997548120554777e-05,
+      "loss": 0.3229,
+      "step": 848
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9975204273770497e-05,
+      "loss": 0.3357,
+      "step": 850
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9974925788777642e-05,
+      "loss": 0.3207,
+      "step": 852
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9974645750612568e-05,
+      "loss": 0.3423,
+      "step": 854
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9974364159318876e-05,
+      "loss": 0.3072,
+      "step": 856
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9974081014940415e-05,
+      "loss": 0.3288,
+      "step": 858
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.997379631752127e-05,
+      "loss": 0.3285,
+      "step": 860
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9973510067105774e-05,
+      "loss": 0.307,
+      "step": 862
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9973222263738495e-05,
+      "loss": 0.3245,
+      "step": 864
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.997293290746425e-05,
+      "loss": 0.3332,
+      "step": 866
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9972641998328086e-05,
+      "loss": 0.3261,
+      "step": 868
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9972349536375312e-05,
+      "loss": 0.3223,
+      "step": 870
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.997205552165146e-05,
+      "loss": 0.3134,
+      "step": 872
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9971759954202306e-05,
+      "loss": 0.3199,
+      "step": 874
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9971462834073878e-05,
+      "loss": 0.3102,
+      "step": 876
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.997116416131244e-05,
+      "loss": 0.3278,
+      "step": 878
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9970863935964496e-05,
+      "loss": 0.3101,
+      "step": 880
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9970562158076793e-05,
+      "loss": 0.3196,
+      "step": 882
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.997025882769632e-05,
+      "loss": 0.2994,
+      "step": 884
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.996995394487031e-05,
+      "loss": 0.3033,
+      "step": 886
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9969647509646234e-05,
+      "loss": 0.3227,
+      "step": 888
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.99693395220718e-05,
+      "loss": 0.3207,
+      "step": 890
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9969029982194978e-05,
+      "loss": 0.3291,
+      "step": 892
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9968718890063952e-05,
+      "loss": 0.3326,
+      "step": 894
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9968406245727175e-05,
+      "loss": 0.309,
+      "step": 896
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9968092049233317e-05,
+      "loss": 0.3268,
+      "step": 898
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9967776300631302e-05,
+      "loss": 0.3132,
+      "step": 900
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9967458999970302e-05,
+      "loss": 0.3,
+      "step": 902
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9967140147299714e-05,
+      "loss": 0.3339,
+      "step": 904
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.996681974266919e-05,
+      "loss": 0.3136,
+      "step": 906
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.996649778612862e-05,
+      "loss": 0.3094,
+      "step": 908
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9966174277728135e-05,
+      "loss": 0.3212,
+      "step": 910
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9965849217518107e-05,
+      "loss": 0.2945,
+      "step": 912
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.996552260554915e-05,
+      "loss": 0.3019,
+      "step": 914
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.996519444187212e-05,
+      "loss": 0.3295,
+      "step": 916
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9964864726538117e-05,
+      "loss": 0.3355,
+      "step": 918
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9964533459598473e-05,
+      "loss": 0.3212,
+      "step": 920
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.996420064110478e-05,
+      "loss": 0.3067,
+      "step": 922
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9963866271108854e-05,
+      "loss": 0.3124,
+      "step": 924
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9963530349662754e-05,
+      "loss": 0.3017,
+      "step": 926
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9963192876818797e-05,
+      "loss": 0.3132,
+      "step": 928
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.996285385262952e-05,
+      "loss": 0.31,
+      "step": 930
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.996251327714772e-05,
+      "loss": 0.3187,
+      "step": 932
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9962171150426418e-05,
+      "loss": 0.323,
+      "step": 934
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.996182747251889e-05,
+      "loss": 0.3116,
+      "step": 936
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9961482243478654e-05,
+      "loss": 0.3026,
+      "step": 938
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.996113546335946e-05,
+      "loss": 0.3032,
+      "step": 940
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9960787132215305e-05,
+      "loss": 0.2994,
+      "step": 942
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9960437250100427e-05,
+      "loss": 0.325,
+      "step": 944
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.99600858170693e-05,
+      "loss": 0.3098,
+      "step": 946
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9959732833176656e-05,
+      "loss": 0.3191,
+      "step": 948
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.995937829847745e-05,
+      "loss": 0.3162,
+      "step": 950
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.995902221302689e-05,
+      "loss": 0.2874,
+      "step": 952
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9958664576880412e-05,
+      "loss": 0.3069,
+      "step": 954
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9958305390093714e-05,
+      "loss": 0.3371,
+      "step": 956
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9957944652722716e-05,
+      "loss": 0.3269,
+      "step": 958
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.995758236482359e-05,
+      "loss": 0.3234,
+      "step": 960
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.995721852645275e-05,
+      "loss": 0.2994,
+      "step": 962
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9956853137666842e-05,
+      "loss": 0.2938,
+      "step": 964
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9956486198522767e-05,
+      "loss": 0.3126,
+      "step": 966
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.995611770907766e-05,
+      "loss": 0.32,
+      "step": 968
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9955747669388893e-05,
+      "loss": 0.3068,
+      "step": 970
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9955376079514083e-05,
+      "loss": 0.3029,
+      "step": 972
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9955002939511093e-05,
+      "loss": 0.3227,
+      "step": 974
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9954628249438023e-05,
+      "loss": 0.2945,
+      "step": 976
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9954252009353217e-05,
+      "loss": 0.325,
+      "step": 978
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9953874219315256e-05,
+      "loss": 0.3219,
+      "step": 980
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9953494879382963e-05,
+      "loss": 0.315,
+      "step": 982
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.995311398961541e-05,
+      "loss": 0.3171,
+      "step": 984
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9952731550071894e-05,
+      "loss": 0.3189,
+      "step": 986
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9952347560811977e-05,
+      "loss": 0.3194,
+      "step": 988
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.995196202189544e-05,
+      "loss": 0.3283,
+      "step": 990
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.995157493338232e-05,
+      "loss": 0.3157,
+      "step": 992
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9951186295332882e-05,
+      "loss": 0.3113,
+      "step": 994
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9950796107807648e-05,
+      "loss": 0.309,
+      "step": 996
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9950404370867368e-05,
+      "loss": 0.2968,
+      "step": 998
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9950011084573042e-05,
+      "loss": 0.3049,
+      "step": 1000
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9949616248985904e-05,
+      "loss": 0.3273,
+      "step": 1002
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.994921986416744e-05,
+      "loss": 0.3085,
+      "step": 1004
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9948821930179357e-05,
+      "loss": 0.3198,
+      "step": 1006
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9948422447083628e-05,
+      "loss": 0.292,
+      "step": 1008
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.994802141494245e-05,
+      "loss": 0.3178,
+      "step": 1010
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.994761883381827e-05,
+      "loss": 0.3165,
+      "step": 1012
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9947214703773773e-05,
+      "loss": 0.2986,
+      "step": 1014
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9946809024871884e-05,
+      "loss": 0.3241,
+      "step": 1016
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9946401797175767e-05,
+      "loss": 0.3061,
+      "step": 1018
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9945993020748834e-05,
+      "loss": 0.3012,
+      "step": 1020
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9945582695654738e-05,
+      "loss": 0.3203,
+      "step": 1022
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.994517082195736e-05,
+      "loss": 0.3154,
+      "step": 1024
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9944757399720843e-05,
+      "loss": 0.323,
+      "step": 1026
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.994434242900955e-05,
+      "loss": 0.3123,
+      "step": 1028
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9943925909888103e-05,
+      "loss": 0.3145,
+      "step": 1030
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9943507842421357e-05,
+      "loss": 0.3033,
+      "step": 1032
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.99430882266744e-05,
+      "loss": 0.3297,
+      "step": 1034
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.994266706271258e-05,
+      "loss": 0.3268,
+      "step": 1036
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9942244350601462e-05,
+      "loss": 0.3109,
+      "step": 1038
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.994182009040688e-05,
+      "loss": 0.2908,
+      "step": 1040
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9941394282194887e-05,
+      "loss": 0.3152,
+      "step": 1042
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9940966926031788e-05,
+      "loss": 0.29,
+      "step": 1044
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.994053802198412e-05,
+      "loss": 0.3145,
+      "step": 1046
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.994010757011867e-05,
+      "loss": 0.3104,
+      "step": 1048
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9939675570502467e-05,
+      "loss": 0.2953,
+      "step": 1050
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.993924202320277e-05,
+      "loss": 0.3167,
+      "step": 1052
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9938806928287086e-05,
+      "loss": 0.3192,
+      "step": 1054
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9938370285823167e-05,
+      "loss": 0.326,
+      "step": 1056
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9937932095879e-05,
+      "loss": 0.3085,
+      "step": 1058
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.993749235852281e-05,
+      "loss": 0.3399,
+      "step": 1060
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9937051073823074e-05,
+      "loss": 0.3125,
+      "step": 1062
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.99366082418485e-05,
+      "loss": 0.3179,
+      "step": 1064
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9936163862668043e-05,
+      "loss": 0.3049,
+      "step": 1066
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9935717936350894e-05,
+      "loss": 0.3066,
+      "step": 1068
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9935270462966484e-05,
+      "loss": 0.2742,
+      "step": 1070
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9934821442584495e-05,
+      "loss": 0.3001,
+      "step": 1072
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9934370875274836e-05,
+      "loss": 0.3153,
+      "step": 1074
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.993391876110767e-05,
+      "loss": 0.3057,
+      "step": 1076
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9933465100153388e-05,
+      "loss": 0.3336,
+      "step": 1078
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9933009892482636e-05,
+      "loss": 0.3095,
+      "step": 1080
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9932553138166287e-05,
+      "loss": 0.3327,
+      "step": 1082
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9932094837275465e-05,
+      "loss": 0.3228,
+      "step": 1084
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.993163498988153e-05,
+      "loss": 0.309,
+      "step": 1086
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9931173596056085e-05,
+      "loss": 0.3207,
+      "step": 1088
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9930710655870967e-05,
+      "loss": 0.32,
+      "step": 1090
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.993024616939826e-05,
+      "loss": 0.325,
+      "step": 1092
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.99297801367103e-05,
+      "loss": 0.3103,
+      "step": 1094
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9929312557879638e-05,
+      "loss": 0.316,
+      "step": 1096
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9928843432979084e-05,
+      "loss": 0.3174,
+      "step": 1098
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9928372762081686e-05,
+      "loss": 0.3038,
+      "step": 1100
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9927900545260735e-05,
+      "loss": 0.287,
+      "step": 1102
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.992742678258975e-05,
+      "loss": 0.3089,
+      "step": 1104
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.99269514741425e-05,
+      "loss": 0.2913,
+      "step": 1106
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9926474619993e-05,
+      "loss": 0.2994,
+      "step": 1108
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.99259962202155e-05,
+      "loss": 0.317,
+      "step": 1110
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9925516274884487e-05,
+      "loss": 0.3194,
+      "step": 1112
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9925034784074692e-05,
+      "loss": 0.3157,
+      "step": 1114
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9924551747861088e-05,
+      "loss": 0.3334,
+      "step": 1116
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9924067166318884e-05,
+      "loss": 0.3092,
+      "step": 1118
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.992358103952354e-05,
+      "loss": 0.3108,
+      "step": 1120
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9923093367550747e-05,
+      "loss": 0.277,
+      "step": 1122
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9922604150476435e-05,
+      "loss": 0.3012,
+      "step": 1124
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9922113388376786e-05,
+      "loss": 0.276,
+      "step": 1126
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9921621081328207e-05,
+      "loss": 0.3107,
+      "step": 1128
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9921127229407363e-05,
+      "loss": 0.3169,
+      "step": 1130
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9920631832691143e-05,
+      "loss": 0.3211,
+      "step": 1132
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9920134891256685e-05,
+      "loss": 0.3012,
+      "step": 1134
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9919636405181375e-05,
+      "loss": 0.317,
+      "step": 1136
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.991913637454282e-05,
+      "loss": 0.3059,
+      "step": 1138
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9918634799418886e-05,
+      "loss": 0.308,
+      "step": 1140
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9918131679887668e-05,
+      "loss": 0.3348,
+      "step": 1142
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9917627016027504e-05,
+      "loss": 0.3083,
+      "step": 1144
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.991712080791698e-05,
+      "loss": 0.2949,
+      "step": 1146
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9916613055634914e-05,
+      "loss": 0.3144,
+      "step": 1148
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9916103759260363e-05,
+      "loss": 0.3472,
+      "step": 1150
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9915592918872635e-05,
+      "loss": 0.3034,
+      "step": 1152
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.991508053455127e-05,
+      "loss": 0.3044,
+      "step": 1154
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9914566606376045e-05,
+      "loss": 0.3265,
+      "step": 1156
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9914051134426987e-05,
+      "loss": 0.3078,
+      "step": 1158
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9913534118784358e-05,
+      "loss": 0.2931,
+      "step": 1160
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9913015559528662e-05,
+      "loss": 0.351,
+      "step": 1162
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9912495456740642e-05,
+      "loss": 0.3184,
+      "step": 1164
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.991197381050128e-05,
+      "loss": 0.3217,
+      "step": 1166
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9911450620891807e-05,
+      "loss": 0.2977,
+      "step": 1168
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9910925887993676e-05,
+      "loss": 0.3109,
+      "step": 1170
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9910399611888604e-05,
+      "loss": 0.3246,
+      "step": 1172
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.990987179265853e-05,
+      "loss": 0.2781,
+      "step": 1174
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.990934243038564e-05,
+      "loss": 0.2934,
+      "step": 1176
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9908811525152362e-05,
+      "loss": 0.3316,
+      "step": 1178
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.990827907704136e-05,
+      "loss": 0.346,
+      "step": 1180
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9907745086135538e-05,
+      "loss": 0.3196,
+      "step": 1182
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9907209552518046e-05,
+      "loss": 0.3159,
+      "step": 1184
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.990667247627227e-05,
+      "loss": 0.3162,
+      "step": 1186
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9906133857481837e-05,
+      "loss": 0.3112,
+      "step": 1188
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9905593696230615e-05,
+      "loss": 0.3185,
+      "step": 1190
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9905051992602708e-05,
+      "loss": 0.3425,
+      "step": 1192
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9904508746682465e-05,
+      "loss": 0.307,
+      "step": 1194
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9903963958554474e-05,
+      "loss": 0.3286,
+      "step": 1196
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9903417628303565e-05,
+      "loss": 0.3038,
+      "step": 1198
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.99028697560148e-05,
+      "loss": 0.3197,
+      "step": 1200
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.990232034177349e-05,
+      "loss": 0.3308,
+      "step": 1202
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9901769385665185e-05,
+      "loss": 0.2893,
+      "step": 1204
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.990121688777567e-05,
+      "loss": 0.2965,
+      "step": 1206
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9900662848190977e-05,
+      "loss": 0.3098,
+      "step": 1208
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9900107266997367e-05,
+      "loss": 0.3128,
+      "step": 1210
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9899550144281358e-05,
+      "loss": 0.3294,
+      "step": 1212
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9898991480129692e-05,
+      "loss": 0.3076,
+      "step": 1214
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9898431274629356e-05,
+      "loss": 0.3294,
+      "step": 1216
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9897869527867582e-05,
+      "loss": 0.3314,
+      "step": 1218
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9897306239931837e-05,
+      "loss": 0.3265,
+      "step": 1220
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.989674141090983e-05,
+      "loss": 0.3109,
+      "step": 1222
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.989617504088951e-05,
+      "loss": 0.2841,
+      "step": 1224
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9895607129959058e-05,
+      "loss": 0.2836,
+      "step": 1226
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.989503767820691e-05,
+      "loss": 0.2899,
+      "step": 1228
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9894466685721734e-05,
+      "loss": 0.3002,
+      "step": 1230
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9893894152592433e-05,
+      "loss": 0.3189,
+      "step": 1232
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.989332007890816e-05,
+      "loss": 0.3337,
+      "step": 1234
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9892744464758295e-05,
+      "loss": 0.3012,
+      "step": 1236
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9892167310232473e-05,
+      "loss": 0.3007,
+      "step": 1238
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9891588615420558e-05,
+      "loss": 0.3122,
+      "step": 1240
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.989100838041266e-05,
+      "loss": 0.3058,
+      "step": 1242
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9890426605299125e-05,
+      "loss": 0.3151,
+      "step": 1244
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9889843290170535e-05,
+      "loss": 0.2995,
+      "step": 1246
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9889258435117723e-05,
+      "loss": 0.3267,
+      "step": 1248
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9888672040231753e-05,
+      "loss": 0.3216,
+      "step": 1250
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.988808410560393e-05,
+      "loss": 0.3212,
+      "step": 1252
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9887494631325805e-05,
+      "loss": 0.303,
+      "step": 1254
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9886903617489156e-05,
+      "loss": 0.3135,
+      "step": 1256
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9886311064186012e-05,
+      "loss": 0.3279,
+      "step": 1258
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.988571697150864e-05,
+      "loss": 0.3013,
+      "step": 1260
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.988512133954954e-05,
+      "loss": 0.3068,
+      "step": 1262
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.988452416840146e-05,
+      "loss": 0.3294,
+      "step": 1264
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9883925458157386e-05,
+      "loss": 0.3274,
+      "step": 1266
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9883325208910537e-05,
+      "loss": 0.3023,
+      "step": 1268
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.988272342075438e-05,
+      "loss": 0.327,
+      "step": 1270
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9882120093782616e-05,
+      "loss": 0.2948,
+      "step": 1272
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9881515228089188e-05,
+      "loss": 0.3254,
+      "step": 1274
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.988090882376828e-05,
+      "loss": 0.2967,
+      "step": 1276
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.988030088091431e-05,
+      "loss": 0.2976,
+      "step": 1278
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.987969139962194e-05,
+      "loss": 0.2674,
+      "step": 1280
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9879080379986074e-05,
+      "loss": 0.3203,
+      "step": 1282
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9878467822101853e-05,
+      "loss": 0.2864,
+      "step": 1284
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9877853726064655e-05,
+      "loss": 0.327,
+      "step": 1286
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.98772380919701e-05,
+      "loss": 0.3233,
+      "step": 1288
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9876620919914044e-05,
+      "loss": 0.3,
+      "step": 1290
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9876002209992586e-05,
+      "loss": 0.3088,
+      "step": 1292
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.987538196230207e-05,
+      "loss": 0.2999,
+      "step": 1294
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9874760176939066e-05,
+      "loss": 0.3079,
+      "step": 1296
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9874136854000398e-05,
+      "loss": 0.3162,
+      "step": 1298
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9873511993583114e-05,
+      "loss": 0.3089,
+      "step": 1300
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.987288559578451e-05,
+      "loss": 0.3147,
+      "step": 1302
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9872257660702126e-05,
+      "loss": 0.3282,
+      "step": 1304
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.987162818843374e-05,
+      "loss": 0.3074,
+      "step": 1306
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9870997179077353e-05,
+      "loss": 0.3278,
+      "step": 1308
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.987036463273123e-05,
+      "loss": 0.3033,
+      "step": 1310
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9869730549493857e-05,
+      "loss": 0.3363,
+      "step": 1312
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9869094929463967e-05,
+      "loss": 0.3086,
+      "step": 1314
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.986845777274053e-05,
+      "loss": 0.3084,
+      "step": 1316
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9867819079422758e-05,
+      "loss": 0.3092,
+      "step": 1318
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.98671788496101e-05,
+      "loss": 0.2931,
+      "step": 1320
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9866537083402243e-05,
+      "loss": 0.3154,
+      "step": 1322
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.986589378089912e-05,
+      "loss": 0.2928,
+      "step": 1324
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.986524894220089e-05,
+      "loss": 0.3224,
+      "step": 1326
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9864602567407962e-05,
+      "loss": 0.2886,
+      "step": 1328
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9863954656620987e-05,
+      "loss": 0.3036,
+      "step": 1330
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9863305209940843e-05,
+      "loss": 0.3011,
+      "step": 1332
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.986265422746866e-05,
+      "loss": 0.34,
+      "step": 1334
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9862001709305793e-05,
+      "loss": 0.3024,
+      "step": 1336
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9861347655553852e-05,
+      "loss": 0.3307,
+      "step": 1338
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9860692066314676e-05,
+      "loss": 0.2974,
+      "step": 1340
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9860034941690342e-05,
+      "loss": 0.3023,
+      "step": 1342
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9859376281783168e-05,
+      "loss": 0.3131,
+      "step": 1344
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9858716086695723e-05,
+      "loss": 0.3038,
+      "step": 1346
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.985805435653079e-05,
+      "loss": 0.3196,
+      "step": 1348
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9857391091391415e-05,
+      "loss": 0.3102,
+      "step": 1350
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9856726291380872e-05,
+      "loss": 0.307,
+      "step": 1352
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9856059956602675e-05,
+      "loss": 0.2894,
+      "step": 1354
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9855392087160576e-05,
+      "loss": 0.3243,
+      "step": 1356
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.985472268315857e-05,
+      "loss": 0.3037,
+      "step": 1358
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9854051744700884e-05,
+      "loss": 0.3132,
+      "step": 1360
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9853379271891994e-05,
+      "loss": 0.2907,
+      "step": 1362
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9852705264836602e-05,
+      "loss": 0.328,
+      "step": 1364
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9852029723639663e-05,
+      "loss": 0.2926,
+      "step": 1366
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.985135264840636e-05,
+      "loss": 0.2987,
+      "step": 1368
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9850674039242117e-05,
+      "loss": 0.3277,
+      "step": 1370
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9849993896252604e-05,
+      "loss": 0.2985,
+      "step": 1372
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9849312219543723e-05,
+      "loss": 0.3122,
+      "step": 1374
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9848629009221615e-05,
+      "loss": 0.3129,
+      "step": 1376
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.984794426539266e-05,
+      "loss": 0.3185,
+      "step": 1378
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.984725798816348e-05,
+      "loss": 0.2946,
+      "step": 1380
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.984657017764093e-05,
+      "loss": 0.3036,
+      "step": 1382
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9845880833932113e-05,
+      "loss": 0.2939,
+      "step": 1384
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9845189957144358e-05,
+      "loss": 0.3104,
+      "step": 1386
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.984449754738525e-05,
+      "loss": 0.3117,
+      "step": 1388
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9843803604762594e-05,
+      "loss": 0.3189,
+      "step": 1390
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9843108129384444e-05,
+      "loss": 0.316,
+      "step": 1392
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.984241112135909e-05,
+      "loss": 0.3162,
+      "step": 1394
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.984171258079507e-05,
+      "loss": 0.3082,
+      "step": 1396
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9841012507801136e-05,
+      "loss": 0.27,
+      "step": 1398
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9840310902486308e-05,
+      "loss": 0.292,
+      "step": 1400
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9839607764959827e-05,
+      "loss": 0.2984,
+      "step": 1402
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9838903095331175e-05,
+      "loss": 0.3186,
+      "step": 1404
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.983819689371008e-05,
+      "loss": 0.3414,
+      "step": 1406
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9837489160206495e-05,
+      "loss": 0.3212,
+      "step": 1408
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9836779894930623e-05,
+      "loss": 0.28,
+      "step": 1410
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9836069097992906e-05,
+      "loss": 0.2762,
+      "step": 1412
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.983535676950402e-05,
+      "loss": 0.3076,
+      "step": 1414
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9834642909574875e-05,
+      "loss": 0.324,
+      "step": 1416
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9833927518316625e-05,
+      "loss": 0.305,
+      "step": 1418
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9833210595840667e-05,
+      "loss": 0.2914,
+      "step": 1420
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.983249214225863e-05,
+      "loss": 0.277,
+      "step": 1422
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9831772157682375e-05,
+      "loss": 0.302,
+      "step": 1424
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9831050642224017e-05,
+      "loss": 0.3169,
+      "step": 1426
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9830327595995898e-05,
+      "loss": 0.3141,
+      "step": 1428
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9829603019110607e-05,
+      "loss": 0.3157,
+      "step": 1430
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9828876911680962e-05,
+      "loss": 0.2737,
+      "step": 1432
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9828149273820017e-05,
+      "loss": 0.339,
+      "step": 1434
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9827420105641086e-05,
+      "loss": 0.3179,
+      "step": 1436
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9826689407257694e-05,
+      "loss": 0.3278,
+      "step": 1438
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9825957178783622e-05,
+      "loss": 0.3079,
+      "step": 1440
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.982522342033288e-05,
+      "loss": 0.2928,
+      "step": 1442
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9824488132019717e-05,
+      "loss": 0.3072,
+      "step": 1444
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9823751313958634e-05,
+      "loss": 0.3153,
+      "step": 1446
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9823012966264353e-05,
+      "loss": 0.3207,
+      "step": 1448
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9822273089051834e-05,
+      "loss": 0.3214,
+      "step": 1450
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9821531682436293e-05,
+      "loss": 0.3116,
+      "step": 1452
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9820788746533165e-05,
+      "loss": 0.3103,
+      "step": 1454
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9820044281458136e-05,
+      "loss": 0.3026,
+      "step": 1456
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9819298287327114e-05,
+      "loss": 0.3033,
+      "step": 1458
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9818550764256273e-05,
+      "loss": 0.3103,
+      "step": 1460
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.981780171236199e-05,
+      "loss": 0.2929,
+      "step": 1462
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9817051131760915e-05,
+      "loss": 0.3138,
+      "step": 1464
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.981629902256991e-05,
+      "loss": 0.3056,
+      "step": 1466
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9815545384906083e-05,
+      "loss": 0.3235,
+      "step": 1468
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9814790218886783e-05,
+      "loss": 0.3198,
+      "step": 1470
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.98140335246296e-05,
+      "loss": 0.3209,
+      "step": 1472
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9813275302252347e-05,
+      "loss": 0.3038,
+      "step": 1474
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9812515551873093e-05,
+      "loss": 0.3263,
+      "step": 1476
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9811754273610138e-05,
+      "loss": 0.3314,
+      "step": 1478
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9810991467582013e-05,
+      "loss": 0.2907,
+      "step": 1480
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9810227133907492e-05,
+      "loss": 0.3088,
+      "step": 1482
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9809461272705595e-05,
+      "loss": 0.3254,
+      "step": 1484
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9808693884095568e-05,
+      "loss": 0.3263,
+      "step": 1486
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9807924968196897e-05,
+      "loss": 0.3091,
+      "step": 1488
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9807154525129314e-05,
+      "loss": 0.3114,
+      "step": 1490
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9806382555012777e-05,
+      "loss": 0.2993,
+      "step": 1492
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.980560905796749e-05,
+      "loss": 0.3116,
+      "step": 1494
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9804834034113893e-05,
+      "loss": 0.2834,
+      "step": 1496
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9804057483572663e-05,
+      "loss": 0.2976,
+      "step": 1498
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9803279406464714e-05,
+      "loss": 0.3019,
+      "step": 1500
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.98024998029112e-05,
+      "loss": 0.3148,
+      "step": 1502
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.980171867303351e-05,
+      "loss": 0.2935,
+      "step": 1504
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9800936016953277e-05,
+      "loss": 0.3229,
+      "step": 1506
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9800151834792355e-05,
+      "loss": 0.3158,
+      "step": 1508
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9799366126672858e-05,
+      "loss": 0.2974,
+      "step": 1510
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.979857889271712e-05,
+      "loss": 0.2913,
+      "step": 1512
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9797790133047722e-05,
+      "loss": 0.2826,
+      "step": 1514
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9796999847787485e-05,
+      "loss": 0.2959,
+      "step": 1516
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9796208037059454e-05,
+      "loss": 0.327,
+      "step": 1518
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9795414700986922e-05,
+      "loss": 0.3133,
+      "step": 1520
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9794619839693417e-05,
+      "loss": 0.2951,
+      "step": 1522
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9793823453302712e-05,
+      "loss": 0.3277,
+      "step": 1524
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9793025541938805e-05,
+      "loss": 0.3191,
+      "step": 1526
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9792226105725935e-05,
+      "loss": 0.3309,
+      "step": 1528
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9791425144788583e-05,
+      "loss": 0.2971,
+      "step": 1530
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9790622659251465e-05,
+      "loss": 0.316,
+      "step": 1532
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9789818649239533e-05,
+      "loss": 0.3063,
+      "step": 1534
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9789013114877978e-05,
+      "loss": 0.3034,
+      "step": 1536
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.978820605629223e-05,
+      "loss": 0.3228,
+      "step": 1538
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9787397473607947e-05,
+      "loss": 0.3186,
+      "step": 1540
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.978658736695104e-05,
+      "loss": 0.3237,
+      "step": 1542
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9785775736447644e-05,
+      "loss": 0.2878,
+      "step": 1544
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.978496258222414e-05,
+      "loss": 0.288,
+      "step": 1546
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.978414790440714e-05,
+      "loss": 0.3085,
+      "step": 1548
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.978333170312349e-05,
+      "loss": 0.3123,
+      "step": 1550
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.978251397850029e-05,
+      "loss": 0.337,
+      "step": 1552
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9781694730664855e-05,
+      "loss": 0.2834,
+      "step": 1554
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9780873959744754e-05,
+      "loss": 0.3375,
+      "step": 1556
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9780051665867792e-05,
+      "loss": 0.3127,
+      "step": 1558
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9779227849161998e-05,
+      "loss": 0.3209,
+      "step": 1560
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9778402509755645e-05,
+      "loss": 0.3221,
+      "step": 1562
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9777575647777256e-05,
+      "loss": 0.3113,
+      "step": 1564
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9776747263355566e-05,
+      "loss": 0.3399,
+      "step": 1566
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9775917356619575e-05,
+      "loss": 0.3173,
+      "step": 1568
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9775085927698496e-05,
+      "loss": 0.3113,
+      "step": 1570
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.977425297672179e-05,
+      "loss": 0.3062,
+      "step": 1572
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9773418503819153e-05,
+      "loss": 0.3285,
+      "step": 1574
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9772582509120525e-05,
+      "loss": 0.3211,
+      "step": 1576
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.977174499275607e-05,
+      "loss": 0.3244,
+      "step": 1578
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.97709059548562e-05,
+      "loss": 0.3001,
+      "step": 1580
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9770065395551562e-05,
+      "loss": 0.3094,
+      "step": 1582
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9769223314973027e-05,
+      "loss": 0.3091,
+      "step": 1584
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9768379713251725e-05,
+      "loss": 0.3053,
+      "step": 1586
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9767534590519e-05,
+      "loss": 0.3255,
+      "step": 1588
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.976668794690646e-05,
+      "loss": 0.3164,
+      "step": 1590
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9765839782545916e-05,
+      "loss": 0.3025,
+      "step": 1592
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9764990097569445e-05,
+      "loss": 0.2943,
+      "step": 1594
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9764138892109344e-05,
+      "loss": 0.3292,
+      "step": 1596
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9763286166298157e-05,
+      "loss": 0.3143,
+      "step": 1598
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.976243192026866e-05,
+      "loss": 0.2977,
+      "step": 1600
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9761576154153855e-05,
+      "loss": 0.3237,
+      "step": 1602
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9760718868087006e-05,
+      "loss": 0.3173,
+      "step": 1604
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.975986006220159e-05,
+      "loss": 0.324,
+      "step": 1606
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9758999736631336e-05,
+      "loss": 0.3094,
+      "step": 1608
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9758137891510194e-05,
+      "loss": 0.3136,
+      "step": 1610
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9757274526972367e-05,
+      "loss": 0.2919,
+      "step": 1612
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9756409643152287e-05,
+      "loss": 0.2911,
+      "step": 1614
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.975554324018462e-05,
+      "loss": 0.3079,
+      "step": 1616
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9754675318204275e-05,
+      "loss": 0.3119,
+      "step": 1618
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.975380587734639e-05,
+      "loss": 0.294,
+      "step": 1620
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9752934917746346e-05,
+      "loss": 0.3157,
+      "step": 1622
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.975206243953976e-05,
+      "loss": 0.3027,
+      "step": 1624
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9751188442862485e-05,
+      "loss": 0.2889,
+      "step": 1626
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.97503129278506e-05,
+      "loss": 0.2995,
+      "step": 1628
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9749435894640438e-05,
+      "loss": 0.2985,
+      "step": 1630
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9748557343368554e-05,
+      "loss": 0.2941,
+      "step": 1632
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9747677274171745e-05,
+      "loss": 0.2984,
+      "step": 1634
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9746795687187054e-05,
+      "loss": 0.3193,
+      "step": 1636
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.974591258255174e-05,
+      "loss": 0.3129,
+      "step": 1638
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9745027960403312e-05,
+      "loss": 0.3223,
+      "step": 1640
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.974414182087952e-05,
+      "loss": 0.3171,
+      "step": 1642
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.974325416411833e-05,
+      "loss": 0.3139,
+      "step": 1644
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9742364990257966e-05,
+      "loss": 0.3186,
+      "step": 1646
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9741474299436877e-05,
+      "loss": 0.3171,
+      "step": 1648
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.974058209179375e-05,
+      "loss": 0.275,
+      "step": 1650
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9739688367467508e-05,
+      "loss": 0.2959,
+      "step": 1652
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.973879312659731e-05,
+      "loss": 0.3034,
+      "step": 1654
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9737896369322555e-05,
+      "loss": 0.3057,
+      "step": 1656
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.973699809578287e-05,
+      "loss": 0.3162,
+      "step": 1658
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.973609830611813e-05,
+      "loss": 0.3393,
+      "step": 1660
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9735197000468435e-05,
+      "loss": 0.3355,
+      "step": 1662
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9734294178974123e-05,
+      "loss": 0.3039,
+      "step": 1664
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9733389841775773e-05,
+      "loss": 0.2998,
+      "step": 1666
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9732483989014197e-05,
+      "loss": 0.3127,
+      "step": 1668
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9731576620830444e-05,
+      "loss": 0.2932,
+      "step": 1670
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9730667737365796e-05,
+      "loss": 0.3171,
+      "step": 1672
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9729757338761775e-05,
+      "loss": 0.3,
+      "step": 1674
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9728845425160137e-05,
+      "loss": 0.3099,
+      "step": 1676
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.972793199670287e-05,
+      "loss": 0.2798,
+      "step": 1678
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9727017053532207e-05,
+      "loss": 0.3084,
+      "step": 1680
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.972610059579061e-05,
+      "loss": 0.3135,
+      "step": 1682
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9725182623620777e-05,
+      "loss": 0.3098,
+      "step": 1684
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9724263137165648e-05,
+      "loss": 0.2916,
+      "step": 1686
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9723342136568385e-05,
+      "loss": 0.319,
+      "step": 1688
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9722419621972405e-05,
+      "loss": 0.2902,
+      "step": 1690
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9721495593521343e-05,
+      "loss": 0.3347,
+      "step": 1692
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9720570051359084e-05,
+      "loss": 0.3115,
+      "step": 1694
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9719642995629735e-05,
+      "loss": 0.3027,
+      "step": 1696
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9718714426477646e-05,
+      "loss": 0.3016,
+      "step": 1698
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9717784344047408e-05,
+      "loss": 0.3083,
+      "step": 1700
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.971685274848384e-05,
+      "loss": 0.3114,
+      "step": 1702
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9715919639931996e-05,
+      "loss": 0.3163,
+      "step": 1704
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9714985018537173e-05,
+      "loss": 0.3084,
+      "step": 1706
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9714048884444894e-05,
+      "loss": 0.3248,
+      "step": 1708
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.971311123780092e-05,
+      "loss": 0.3078,
+      "step": 1710
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.971217207875126e-05,
+      "loss": 0.3056,
+      "step": 1712
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9711231407442136e-05,
+      "loss": 0.3068,
+      "step": 1714
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9710289224020028e-05,
+      "loss": 0.3245,
+      "step": 1716
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9709345528631637e-05,
+      "loss": 0.2868,
+      "step": 1718
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9708400321423897e-05,
+      "loss": 0.3025,
+      "step": 1720
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9707453602543994e-05,
+      "loss": 0.2995,
+      "step": 1722
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9706505372139337e-05,
+      "loss": 0.3032,
+      "step": 1724
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.970555563035757e-05,
+      "loss": 0.3164,
+      "step": 1726
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9704604377346577e-05,
+      "loss": 0.2949,
+      "step": 1728
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9703651613254475e-05,
+      "loss": 0.318,
+      "step": 1730
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9702697338229613e-05,
+      "loss": 0.3144,
+      "step": 1732
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9701741552420587e-05,
+      "loss": 0.2944,
+      "step": 1734
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9700784255976212e-05,
+      "loss": 0.3185,
+      "step": 1736
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9699825449045548e-05,
+      "loss": 0.3026,
+      "step": 1738
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9698865131777896e-05,
+      "loss": 0.3052,
+      "step": 1740
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9697903304322775e-05,
+      "loss": 0.314,
+      "step": 1742
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9696939966829957e-05,
+      "loss": 0.3105,
+      "step": 1744
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9695975119449434e-05,
+      "loss": 0.3298,
+      "step": 1746
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9695008762331444e-05,
+      "loss": 0.3048,
+      "step": 1748
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9694040895626452e-05,
+      "loss": 0.2939,
+      "step": 1750
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.969307151948517e-05,
+      "loss": 0.3049,
+      "step": 1752
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.969210063405853e-05,
+      "loss": 0.2966,
+      "step": 1754
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.969112823949771e-05,
+      "loss": 0.3224,
+      "step": 1756
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.969015433595412e-05,
+      "loss": 0.3018,
+      "step": 1758
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9689178923579404e-05,
+      "loss": 0.2984,
+      "step": 1760
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9688202002525437e-05,
+      "loss": 0.2696,
+      "step": 1762
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9687223572944337e-05,
+      "loss": 0.2899,
+      "step": 1764
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.968624363498845e-05,
+      "loss": 0.3124,
+      "step": 1766
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9685262188810365e-05,
+      "loss": 0.3092,
+      "step": 1768
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9684279234562894e-05,
+      "loss": 0.3199,
+      "step": 1770
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9683294772399094e-05,
+      "loss": 0.3017,
+      "step": 1772
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9682308802472256e-05,
+      "loss": 0.3229,
+      "step": 1774
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9681321324935897e-05,
+      "loss": 0.3124,
+      "step": 1776
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.968033233994378e-05,
+      "loss": 0.3186,
+      "step": 1778
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9679341847649894e-05,
+      "loss": 0.3194,
+      "step": 1780
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.967834984820847e-05,
+      "loss": 0.3029,
+      "step": 1782
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9677356341773968e-05,
+      "loss": 0.314,
+      "step": 1784
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.967636132850108e-05,
+      "loss": 0.2938,
+      "step": 1786
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9675364808544745e-05,
+      "loss": 0.3223,
+      "step": 1788
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9674366782060126e-05,
+      "loss": 0.3082,
+      "step": 1790
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9673367249202623e-05,
+      "loss": 0.309,
+      "step": 1792
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9672366210127863e-05,
+      "loss": 0.3034,
+      "step": 1794
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9671363664991728e-05,
+      "loss": 0.2849,
+      "step": 1796
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.967035961395032e-05,
+      "loss": 0.2915,
+      "step": 1798
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.966935405715997e-05,
+      "loss": 0.3152,
+      "step": 1800
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9668346994777257e-05,
+      "loss": 0.3072,
+      "step": 1802
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9667338426958986e-05,
+      "loss": 0.3013,
+      "step": 1804
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.96663283538622e-05,
+      "loss": 0.303,
+      "step": 1806
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9665316775644174e-05,
+      "loss": 0.2886,
+      "step": 1808
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9664303692462417e-05,
+      "loss": 0.3094,
+      "step": 1810
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9663289104474675e-05,
+      "loss": 0.3039,
+      "step": 1812
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.966227301183893e-05,
+      "loss": 0.3065,
+      "step": 1814
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9661255414713394e-05,
+      "loss": 0.3259,
+      "step": 1816
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9660236313256508e-05,
+      "loss": 0.3194,
+      "step": 1818
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9659215707626968e-05,
+      "loss": 0.3159,
+      "step": 1820
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9658193597983673e-05,
+      "loss": 0.3046,
+      "step": 1822
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9657169984485785e-05,
+      "loss": 0.3013,
+      "step": 1824
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.965614486729269e-05,
+      "loss": 0.3137,
+      "step": 1826
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9655118246563996e-05,
+      "loss": 0.2932,
+      "step": 1828
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9654090122459566e-05,
+      "loss": 0.3086,
+      "step": 1830
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.965306049513948e-05,
+      "loss": 0.2965,
+      "step": 1832
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.965202936476406e-05,
+      "loss": 0.297,
+      "step": 1834
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9650996731493866e-05,
+      "loss": 0.308,
+      "step": 1836
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.964996259548968e-05,
+      "loss": 0.3188,
+      "step": 1838
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.964892695691253e-05,
+      "loss": 0.309,
+      "step": 1840
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9647889815923672e-05,
+      "loss": 0.3366,
+      "step": 1842
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9646851172684593e-05,
+      "loss": 0.2949,
+      "step": 1844
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9645811027357017e-05,
+      "loss": 0.3055,
+      "step": 1846
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9644769380102912e-05,
+      "loss": 0.2972,
+      "step": 1848
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.964372623108446e-05,
+      "loss": 0.3134,
+      "step": 1850
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9642681580464095e-05,
+      "loss": 0.2692,
+      "step": 1852
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9641635428404475e-05,
+      "loss": 0.3146,
+      "step": 1854
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.964058777506849e-05,
+      "loss": 0.3048,
+      "step": 1856
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9639538620619266e-05,
+      "loss": 0.3155,
+      "step": 1858
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9638487965220176e-05,
+      "loss": 0.3291,
+      "step": 1860
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.96374358090348e-05,
+      "loss": 0.2986,
+      "step": 1862
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.963638215222698e-05,
+      "loss": 0.2878,
+      "step": 1864
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9635326994960773e-05,
+      "loss": 0.3019,
+      "step": 1866
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9634270337400474e-05,
+      "loss": 0.2835,
+      "step": 1868
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9633212179710613e-05,
+      "loss": 0.2899,
+      "step": 1870
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9632152522055953e-05,
+      "loss": 0.2793,
+      "step": 1872
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9631091364601495e-05,
+      "loss": 0.29,
+      "step": 1874
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9630028707512462e-05,
+      "loss": 0.3209,
+      "step": 1876
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9628964550954325e-05,
+      "loss": 0.3049,
+      "step": 1878
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9627898895092773e-05,
+      "loss": 0.3086,
+      "step": 1880
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9626831740093745e-05,
+      "loss": 0.3065,
+      "step": 1882
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9625763086123403e-05,
+      "loss": 0.3178,
+      "step": 1884
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9624692933348142e-05,
+      "loss": 0.3102,
+      "step": 1886
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.96236212819346e-05,
+      "loss": 0.2425,
+      "step": 1888
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.962254813204963e-05,
+      "loss": 0.2794,
+      "step": 1890
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9621473483860335e-05,
+      "loss": 0.3106,
+      "step": 1892
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.962039733753405e-05,
+      "loss": 0.3126,
+      "step": 1894
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9619319693238333e-05,
+      "loss": 0.3055,
+      "step": 1896
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9618240551140985e-05,
+      "loss": 0.2972,
+      "step": 1898
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9617159911410037e-05,
+      "loss": 0.2988,
+      "step": 1900
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.961607777421375e-05,
+      "loss": 0.2913,
+      "step": 1902
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9614994139720624e-05,
+      "loss": 0.3034,
+      "step": 1904
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9613909008099387e-05,
+      "loss": 0.3074,
+      "step": 1906
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9612822379519004e-05,
+      "loss": 0.3059,
+      "step": 1908
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.961173425414867e-05,
+      "loss": 0.3056,
+      "step": 1910
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9610644632157814e-05,
+      "loss": 0.2983,
+      "step": 1912
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.96095535137161e-05,
+      "loss": 0.2777,
+      "step": 1914
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9608460898993425e-05,
+      "loss": 0.2874,
+      "step": 1916
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9607366788159914e-05,
+      "loss": 0.2735,
+      "step": 1918
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.960627118138593e-05,
+      "loss": 0.2974,
+      "step": 1920
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.960517407884207e-05,
+      "loss": 0.3067,
+      "step": 1922
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9604075480699155e-05,
+      "loss": 0.2966,
+      "step": 1924
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.960297538712825e-05,
+      "loss": 0.2996,
+      "step": 1926
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9601873798300647e-05,
+      "loss": 0.2881,
+      "step": 1928
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.960077071438787e-05,
+      "loss": 0.2806,
+      "step": 1930
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.959966613556168e-05,
+      "loss": 0.2916,
+      "step": 1932
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9598560061994067e-05,
+      "loss": 0.3156,
+      "step": 1934
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9597452493857253e-05,
+      "loss": 0.3111,
+      "step": 1936
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.95963434313237e-05,
+      "loss": 0.3016,
+      "step": 1938
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9595232874566098e-05,
+      "loss": 0.3144,
+      "step": 1940
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.959412082375736e-05,
+      "loss": 0.3074,
+      "step": 1942
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9593007279070654e-05,
+      "loss": 0.3026,
+      "step": 1944
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9591892240679353e-05,
+      "loss": 0.2951,
+      "step": 1946
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.959077570875709e-05,
+      "loss": 0.3124,
+      "step": 1948
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9589657683477708e-05,
+      "loss": 0.2941,
+      "step": 1950
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9588538165015297e-05,
+      "loss": 0.298,
+      "step": 1952
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.958741715354417e-05,
+      "loss": 0.3097,
+      "step": 1954
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9586294649238885e-05,
+      "loss": 0.2838,
+      "step": 1956
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.958517065227422e-05,
+      "loss": 0.3099,
+      "step": 1958
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9584045162825188e-05,
+      "loss": 0.3099,
+      "step": 1960
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9582918181067043e-05,
+      "loss": 0.2977,
+      "step": 1962
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9581789707175254e-05,
+      "loss": 0.3225,
+      "step": 1964
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9580659741325544e-05,
+      "loss": 0.3154,
+      "step": 1966
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9579528283693846e-05,
+      "loss": 0.303,
+      "step": 1968
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9578395334456348e-05,
+      "loss": 0.2989,
+      "step": 1970
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.957726089378945e-05,
+      "loss": 0.3123,
+      "step": 1972
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.95761249618698e-05,
+      "loss": 0.3198,
+      "step": 1974
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.957498753887427e-05,
+      "loss": 0.2754,
+      "step": 1976
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.957384862497996e-05,
+      "loss": 0.2932,
+      "step": 1978
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9572708220364212e-05,
+      "loss": 0.3162,
+      "step": 1980
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9571566325204593e-05,
+      "loss": 0.31,
+      "step": 1982
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.957042293967891e-05,
+      "loss": 0.3051,
+      "step": 1984
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9569278063965193e-05,
+      "loss": 0.2821,
+      "step": 1986
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9568131698241708e-05,
+      "loss": 0.3124,
+      "step": 1988
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9566983842686954e-05,
+      "loss": 0.2982,
+      "step": 1990
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.956583449747966e-05,
+      "loss": 0.3054,
+      "step": 1992
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9564683662798788e-05,
+      "loss": 0.3064,
+      "step": 1994
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9563531338823528e-05,
+      "loss": 0.2784,
+      "step": 1996
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9562377525733314e-05,
+      "loss": 0.297,
+      "step": 1998
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.95612222237078e-05,
+      "loss": 0.279,
+      "step": 2000
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.956006543292687e-05,
+      "loss": 0.2912,
+      "step": 2002
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9558907153570654e-05,
+      "loss": 0.324,
+      "step": 2004
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.95577473858195e-05,
+      "loss": 0.2992,
+      "step": 2006
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9556586129853993e-05,
+      "loss": 0.2856,
+      "step": 2008
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9555423385854948e-05,
+      "loss": 0.2596,
+      "step": 2010
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9554259154003415e-05,
+      "loss": 0.2967,
+      "step": 2012
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9553093434480673e-05,
+      "loss": 0.3143,
+      "step": 2014
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9551926227468238e-05,
+      "loss": 0.3187,
+      "step": 2016
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9550757533147845e-05,
+      "loss": 0.3061,
+      "step": 2018
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9549587351701474e-05,
+      "loss": 0.3104,
+      "step": 2020
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.954841568331133e-05,
+      "loss": 0.3155,
+      "step": 2022
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9547242528159852e-05,
+      "loss": 0.2866,
+      "step": 2024
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9546067886429705e-05,
+      "loss": 0.3206,
+      "step": 2026
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9544891758303795e-05,
+      "loss": 0.3331,
+      "step": 2028
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9543714143965252e-05,
+      "loss": 0.31,
+      "step": 2030
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.954253504359744e-05,
+      "loss": 0.3137,
+      "step": 2032
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.954135445738395e-05,
+      "loss": 0.3016,
+      "step": 2034
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9540172385508614e-05,
+      "loss": 0.2827,
+      "step": 2036
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9538988828155484e-05,
+      "loss": 0.32,
+      "step": 2038
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9537803785508854e-05,
+      "loss": 0.303,
+      "step": 2040
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9536617257753246e-05,
+      "loss": 0.3173,
+      "step": 2042
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9535429245073407e-05,
+      "loss": 0.3248,
+      "step": 2044
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9534239747654316e-05,
+      "loss": 0.3261,
+      "step": 2046
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.953304876568119e-05,
+      "loss": 0.3067,
+      "step": 2048
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9531856299339483e-05,
+      "loss": 0.2753,
+      "step": 2050
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9530662348814858e-05,
+      "loss": 0.3137,
+      "step": 2052
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.952946691429323e-05,
+      "loss": 0.3059,
+      "step": 2054
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9528269995960732e-05,
+      "loss": 0.2983,
+      "step": 2056
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.952707159400374e-05,
+      "loss": 0.2921,
+      "step": 2058
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9525871708608847e-05,
+      "loss": 0.3104,
+      "step": 2060
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.952467033996289e-05,
+      "loss": 0.2975,
+      "step": 2062
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.952346748825293e-05,
+      "loss": 0.2984,
+      "step": 2064
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9522263153666257e-05,
+      "loss": 0.2981,
+      "step": 2066
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9521057336390398e-05,
+      "loss": 0.3075,
+      "step": 2068
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.951985003661311e-05,
+      "loss": 0.308,
+      "step": 2070
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9518641254522374e-05,
+      "loss": 0.3135,
+      "step": 2072
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.951743099030641e-05,
+      "loss": 0.3081,
+      "step": 2074
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9516219244153666e-05,
+      "loss": 0.2991,
+      "step": 2076
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9515006016252816e-05,
+      "loss": 0.2934,
+      "step": 2078
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.951379130679277e-05,
+      "loss": 0.3083,
+      "step": 2080
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.951257511596267e-05,
+      "loss": 0.3318,
+      "step": 2082
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9511357443951885e-05,
+      "loss": 0.3098,
+      "step": 2084
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.951013829095002e-05,
+      "loss": 0.3099,
+      "step": 2086
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.95089176571469e-05,
+      "loss": 0.3306,
+      "step": 2088
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.950769554273259e-05,
+      "loss": 0.2969,
+      "step": 2090
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9506471947897378e-05,
+      "loss": 0.2976,
+      "step": 2092
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9505246872831792e-05,
+      "loss": 0.3151,
+      "step": 2094
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9504020317726585e-05,
+      "loss": 0.3159,
+      "step": 2096
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.950279228277274e-05,
+      "loss": 0.3117,
+      "step": 2098
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9501562768161477e-05,
+      "loss": 0.2997,
+      "step": 2100
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9500331774084232e-05,
+      "loss": 0.2662,
+      "step": 2102
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9499099300732687e-05,
+      "loss": 0.3126,
+      "step": 2104
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9497865348298738e-05,
+      "loss": 0.3011,
+      "step": 2106
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9496629916974533e-05,
+      "loss": 0.3095,
+      "step": 2108
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9495393006952433e-05,
+      "loss": 0.3251,
+      "step": 2110
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.949415461842503e-05,
+      "loss": 0.3057,
+      "step": 2112
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9492914751585157e-05,
+      "loss": 0.3205,
+      "step": 2114
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.949167340662587e-05,
+      "loss": 0.2909,
+      "step": 2116
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.949043058374045e-05,
+      "loss": 0.3098,
+      "step": 2118
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9489186283122417e-05,
+      "loss": 0.2958,
+      "step": 2120
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9487940504965525e-05,
+      "loss": 0.3054,
+      "step": 2122
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.948669324946374e-05,
+      "loss": 0.3021,
+      "step": 2124
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.948544451681128e-05,
+      "loss": 0.2918,
+      "step": 2126
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9484194307202573e-05,
+      "loss": 0.2922,
+      "step": 2128
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.948294262083229e-05,
+      "loss": 0.3322,
+      "step": 2130
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.948168945789533e-05,
+      "loss": 0.311,
+      "step": 2132
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9480434818586817e-05,
+      "loss": 0.3313,
+      "step": 2134
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9479178703102108e-05,
+      "loss": 0.2972,
+      "step": 2136
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9477921111636794e-05,
+      "loss": 0.3031,
+      "step": 2138
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9476662044386687e-05,
+      "loss": 0.2851,
+      "step": 2140
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9475401501547835e-05,
+      "loss": 0.2924,
+      "step": 2142
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9474139483316512e-05,
+      "loss": 0.2908,
+      "step": 2144
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9472875989889227e-05,
+      "loss": 0.2872,
+      "step": 2146
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.947161102146272e-05,
+      "loss": 0.297,
+      "step": 2148
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9470344578233942e-05,
+      "loss": 0.2732,
+      "step": 2150
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9469076660400104e-05,
+      "loss": 0.3101,
+      "step": 2152
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.946780726815862e-05,
+      "loss": 0.3093,
+      "step": 2154
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9466536401707146e-05,
+      "loss": 0.3066,
+      "step": 2156
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.946526406124357e-05,
+      "loss": 0.3087,
+      "step": 2158
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9463990246966e-05,
+      "loss": 0.3039,
+      "step": 2160
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.946271495907278e-05,
+      "loss": 0.3063,
+      "step": 2162
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9461438197762485e-05,
+      "loss": 0.2598,
+      "step": 2164
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9460159963233916e-05,
+      "loss": 0.3244,
+      "step": 2166
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.94588802556861e-05,
+      "loss": 0.2994,
+      "step": 2168
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9457599075318297e-05,
+      "loss": 0.3135,
+      "step": 2170
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9456316422330003e-05,
+      "loss": 0.3108,
+      "step": 2172
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9455032296920932e-05,
+      "loss": 0.3212,
+      "step": 2174
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.945374669929103e-05,
+      "loss": 0.2943,
+      "step": 2176
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.945245962964048e-05,
+      "loss": 0.2851,
+      "step": 2178
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9451171088169684e-05,
+      "loss": 0.292,
+      "step": 2180
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9449881075079282e-05,
+      "loss": 0.2911,
+      "step": 2182
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9448589590570136e-05,
+      "loss": 0.3014,
+      "step": 2184
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.944729663484334e-05,
+      "loss": 0.3048,
+      "step": 2186
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.944600220810022e-05,
+      "loss": 0.281,
+      "step": 2188
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9444706310542326e-05,
+      "loss": 0.3162,
+      "step": 2190
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9443408942371435e-05,
+      "loss": 0.3266,
+      "step": 2192
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.944211010378956e-05,
+      "loss": 0.3273,
+      "step": 2194
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9440809794998947e-05,
+      "loss": 0.2907,
+      "step": 2196
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9439508016202057e-05,
+      "loss": 0.3187,
+      "step": 2198
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9438204767601586e-05,
+      "loss": 0.2982,
+      "step": 2200
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9436900049400466e-05,
+      "loss": 0.305,
+      "step": 2202
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9435593861801846e-05,
+      "loss": 0.2946,
+      "step": 2204
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.943428620500911e-05,
+      "loss": 0.3075,
+      "step": 2206
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.943297707922587e-05,
+      "loss": 0.2992,
+      "step": 2208
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.943166648465597e-05,
+      "loss": 0.2891,
+      "step": 2210
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9430354421503476e-05,
+      "loss": 0.2712,
+      "step": 2212
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.942904088997269e-05,
+      "loss": 0.2869,
+      "step": 2214
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9427725890268133e-05,
+      "loss": 0.2955,
+      "step": 2216
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.942640942259457e-05,
+      "loss": 0.2933,
+      "step": 2218
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9425091487156976e-05,
+      "loss": 0.3002,
+      "step": 2220
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9423772084160565e-05,
+      "loss": 0.3053,
+      "step": 2222
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9422451213810786e-05,
+      "loss": 0.3091,
+      "step": 2224
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9421128876313297e-05,
+      "loss": 0.3076,
+      "step": 2226
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.941980507187401e-05,
+      "loss": 0.2878,
+      "step": 2228
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9418479800699035e-05,
+      "loss": 0.3039,
+      "step": 2230
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9417153062994737e-05,
+      "loss": 0.3059,
+      "step": 2232
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.94158248589677e-05,
+      "loss": 0.2913,
+      "step": 2234
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.941449518882473e-05,
+      "loss": 0.2824,
+      "step": 2236
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.941316405277287e-05,
+      "loss": 0.3135,
+      "step": 2238
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.941183145101939e-05,
+      "loss": 0.2874,
+      "step": 2240
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9410497383771782e-05,
+      "loss": 0.3155,
+      "step": 2242
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9409161851237775e-05,
+      "loss": 0.3064,
+      "step": 2244
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9407824853625316e-05,
+      "loss": 0.3088,
+      "step": 2246
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9406486391142593e-05,
+      "loss": 0.2959,
+      "step": 2248
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.940514646399801e-05,
+      "loss": 0.2983,
+      "step": 2250
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.94038050724002e-05,
+      "loss": 0.2917,
+      "step": 2252
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.940246221655804e-05,
+      "loss": 0.3225,
+      "step": 2254
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9401117896680607e-05,
+      "loss": 0.313,
+      "step": 2256
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9399772112977234e-05,
+      "loss": 0.2962,
+      "step": 2258
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.939842486565747e-05,
+      "loss": 0.3086,
+      "step": 2260
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9397076154931086e-05,
+      "loss": 0.2956,
+      "step": 2262
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9395725981008084e-05,
+      "loss": 0.3017,
+      "step": 2264
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9394374344098706e-05,
+      "loss": 0.3099,
+      "step": 2266
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9393021244413406e-05,
+      "loss": 0.3061,
+      "step": 2268
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.939166668216287e-05,
+      "loss": 0.2888,
+      "step": 2270
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.939031065755802e-05,
+      "loss": 0.2944,
+      "step": 2272
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9388953170809996e-05,
+      "loss": 0.3022,
+      "step": 2274
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.938759422213017e-05,
+      "loss": 0.3085,
+      "step": 2276
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9386233811730136e-05,
+      "loss": 0.2693,
+      "step": 2278
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.938487193982173e-05,
+      "loss": 0.3024,
+      "step": 2280
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9383508606616994e-05,
+      "loss": 0.3012,
+      "step": 2282
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9382143812328218e-05,
+      "loss": 0.3065,
+      "step": 2284
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9380777557167908e-05,
+      "loss": 0.3001,
+      "step": 2286
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9379409841348805e-05,
+      "loss": 0.2973,
+      "step": 2288
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9378040665083862e-05,
+      "loss": 0.3029,
+      "step": 2290
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9376670028586283e-05,
+      "loss": 0.3005,
+      "step": 2292
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9375297932069477e-05,
+      "loss": 0.3084,
+      "step": 2294
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9373924375747095e-05,
+      "loss": 0.289,
+      "step": 2296
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9372549359833005e-05,
+      "loss": 0.3064,
+      "step": 2298
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9371172884541316e-05,
+      "loss": 0.2915,
+      "step": 2300
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9369794950086346e-05,
+      "loss": 0.2964,
+      "step": 2302
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.936841555668266e-05,
+      "loss": 0.2922,
+      "step": 2304
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.936703470454503e-05,
+      "loss": 0.2912,
+      "step": 2306
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9365652393888476e-05,
+      "loss": 0.3013,
+      "step": 2308
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9364268624928223e-05,
+      "loss": 0.3095,
+      "step": 2310
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9362883397879738e-05,
+      "loss": 0.327,
+      "step": 2312
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9361496712958718e-05,
+      "loss": 0.3151,
+      "step": 2314
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9360108570381073e-05,
+      "loss": 0.2809,
+      "step": 2316
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9358718970362954e-05,
+      "loss": 0.3027,
+      "step": 2318
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9357327913120723e-05,
+      "loss": 0.31,
+      "step": 2320
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9355935398870987e-05,
+      "loss": 0.3004,
+      "step": 2322
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9354541427830566e-05,
+      "loss": 0.311,
+      "step": 2324
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9353146000216517e-05,
+      "loss": 0.2999,
+      "step": 2326
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.935174911624611e-05,
+      "loss": 0.303,
+      "step": 2328
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9350350776136862e-05,
+      "loss": 0.2858,
+      "step": 2330
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9348950980106496e-05,
+      "loss": 0.3085,
+      "step": 2332
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9347549728372977e-05,
+      "loss": 0.3017,
+      "step": 2334
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9346147021154485e-05,
+      "loss": 0.3166,
+      "step": 2336
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9344742858669435e-05,
+      "loss": 0.3032,
+      "step": 2338
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9343337241136466e-05,
+      "loss": 0.2859,
+      "step": 2340
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9341930168774446e-05,
+      "loss": 0.3238,
+      "step": 2342
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9340521641802464e-05,
+      "loss": 0.2853,
+      "step": 2344
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9339111660439835e-05,
+      "loss": 0.3151,
+      "step": 2346
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.933770022490611e-05,
+      "loss": 0.2785,
+      "step": 2348
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9336287335421057e-05,
+      "loss": 0.3166,
+      "step": 2350
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9334872992204677e-05,
+      "loss": 0.3239,
+      "step": 2352
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.933345719547719e-05,
+      "loss": 0.3006,
+      "step": 2354
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9332039945459044e-05,
+      "loss": 0.2962,
+      "step": 2356
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9330621242370924e-05,
+      "loss": 0.3114,
+      "step": 2358
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.932920108643373e-05,
+      "loss": 0.2896,
+      "step": 2360
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9327779477868586e-05,
+      "loss": 0.3102,
+      "step": 2362
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.932635641689685e-05,
+      "loss": 0.2954,
+      "step": 2364
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.932493190374011e-05,
+      "loss": 0.2728,
+      "step": 2366
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9323505938620163e-05,
+      "loss": 0.2819,
+      "step": 2368
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.932207852175905e-05,
+      "loss": 0.2837,
+      "step": 2370
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.932064965337903e-05,
+      "loss": 0.3101,
+      "step": 2372
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9319219333702584e-05,
+      "loss": 0.3016,
+      "step": 2374
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.931778756295243e-05,
+      "loss": 0.2733,
+      "step": 2376
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9316354341351497e-05,
+      "loss": 0.3171,
+      "step": 2378
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.931491966912296e-05,
+      "loss": 0.2806,
+      "step": 2380
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.93134835464902e-05,
+      "loss": 0.3011,
+      "step": 2382
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9312045973676835e-05,
+      "loss": 0.3103,
+      "step": 2384
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9310606950906706e-05,
+      "loss": 0.2665,
+      "step": 2386
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9309166478403882e-05,
+      "loss": 0.305,
+      "step": 2388
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9307724556392654e-05,
+      "loss": 0.3093,
+      "step": 2390
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9306281185097532e-05,
+      "loss": 0.268,
+      "step": 2392
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9304836364743277e-05,
+      "loss": 0.2995,
+      "step": 2394
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9303390095554842e-05,
+      "loss": 0.2973,
+      "step": 2396
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9301942377757433e-05,
+      "loss": 0.3085,
+      "step": 2398
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9300493211576467e-05,
+      "loss": 0.287,
+      "step": 2400
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.929904259723759e-05,
+      "loss": 0.2907,
+      "step": 2402
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9297590534966676e-05,
+      "loss": 0.288,
+      "step": 2404
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9296137024989818e-05,
+      "loss": 0.2706,
+      "step": 2406
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9294682067533338e-05,
+      "loss": 0.2976,
+      "step": 2408
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9293225662823788e-05,
+      "loss": 0.2883,
+      "step": 2410
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9291767811087942e-05,
+      "loss": 0.3068,
+      "step": 2412
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.92903085125528e-05,
+      "loss": 0.3056,
+      "step": 2414
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.928884776744558e-05,
+      "loss": 0.3094,
+      "step": 2416
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.928738557599373e-05,
+      "loss": 0.2917,
+      "step": 2418
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9285921938424934e-05,
+      "loss": 0.3092,
+      "step": 2420
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9284456854967086e-05,
+      "loss": 0.3042,
+      "step": 2422
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9282990325848307e-05,
+      "loss": 0.2973,
+      "step": 2424
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9281522351296952e-05,
+      "loss": 0.3205,
+      "step": 2426
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9280052931541594e-05,
+      "loss": 0.2851,
+      "step": 2428
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9278582066811034e-05,
+      "loss": 0.3056,
+      "step": 2430
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9277109757334296e-05,
+      "loss": 0.3115,
+      "step": 2432
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.927563600334063e-05,
+      "loss": 0.2996,
+      "step": 2434
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9274160805059513e-05,
+      "loss": 0.297,
+      "step": 2436
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.927268416272064e-05,
+      "loss": 0.3132,
+      "step": 2438
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9271206076553936e-05,
+      "loss": 0.296,
+      "step": 2440
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9269726546789553e-05,
+      "loss": 0.3107,
+      "step": 2442
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9268245573657867e-05,
+      "loss": 0.3028,
+      "step": 2444
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.926676315738947e-05,
+      "loss": 0.3078,
+      "step": 2446
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9265279298215197e-05,
+      "loss": 0.2909,
+      "step": 2448
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9263793996366083e-05,
+      "loss": 0.3159,
+      "step": 2450
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.926230725207341e-05,
+      "loss": 0.2733,
+      "step": 2452
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9260819065568667e-05,
+      "loss": 0.291,
+      "step": 2454
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9259329437083586e-05,
+      "loss": 0.3063,
+      "step": 2456
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9257838366850106e-05,
+      "loss": 0.2884,
+      "step": 2458
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.92563458551004e-05,
+      "loss": 0.2785,
+      "step": 2460
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9254851902066867e-05,
+      "loss": 0.3104,
+      "step": 2462
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9253356507982125e-05,
+      "loss": 0.2991,
+      "step": 2464
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9251859673079016e-05,
+      "loss": 0.2842,
+      "step": 2466
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.925036139759061e-05,
+      "loss": 0.2965,
+      "step": 2468
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.92488616817502e-05,
+      "loss": 0.2986,
+      "step": 2470
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9247360525791303e-05,
+      "loss": 0.2956,
+      "step": 2472
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.924585792994766e-05,
+      "loss": 0.2855,
+      "step": 2474
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.924435389445324e-05,
+      "loss": 0.2867,
+      "step": 2476
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9242848419542232e-05,
+      "loss": 0.3022,
+      "step": 2478
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9241341505449043e-05,
+      "loss": 0.3224,
+      "step": 2480
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.923983315240832e-05,
+      "loss": 0.2938,
+      "step": 2482
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.923832336065492e-05,
+      "loss": 0.3167,
+      "step": 2484
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.923681213042393e-05,
+      "loss": 0.3131,
+      "step": 2486
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.923529946195066e-05,
+      "loss": 0.2955,
+      "step": 2488
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9233785355470645e-05,
+      "loss": 0.3071,
+      "step": 2490
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9232269811219643e-05,
+      "loss": 0.3017,
+      "step": 2492
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.923075282943364e-05,
+      "loss": 0.2925,
+      "step": 2494
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.922923441034883e-05,
+      "loss": 0.3116,
+      "step": 2496
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9227714554201654e-05,
+      "loss": 0.3121,
+      "step": 2498
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.922619326122876e-05,
+      "loss": 0.2915,
+      "step": 2500
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9224670531667023e-05,
+      "loss": 0.2796,
+      "step": 2502
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9223146365753548e-05,
+      "loss": 0.2623,
+      "step": 2504
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9221620763725658e-05,
+      "loss": 0.2923,
+      "step": 2506
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9220093725820903e-05,
+      "loss": 0.3046,
+      "step": 2508
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9218565252277047e-05,
+      "loss": 0.3085,
+      "step": 2510
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9217035343332094e-05,
+      "loss": 0.2772,
+      "step": 2512
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.921550399922426e-05,
+      "loss": 0.2952,
+      "step": 2514
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.921397122019198e-05,
+      "loss": 0.2839,
+      "step": 2516
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9212437006473927e-05,
+      "loss": 0.3098,
+      "step": 2518
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.921090135830899e-05,
+      "loss": 0.2951,
+      "step": 2520
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9209364275936278e-05,
+      "loss": 0.3081,
+      "step": 2522
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9207825759595126e-05,
+      "loss": 0.2788,
+      "step": 2524
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9206285809525097e-05,
+      "loss": 0.2964,
+      "step": 2526
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9204744425965967e-05,
+      "loss": 0.2875,
+      "step": 2528
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9203201609157746e-05,
+      "loss": 0.2919,
+      "step": 2530
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9201657359340662e-05,
+      "loss": 0.3252,
+      "step": 2532
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9200111676755164e-05,
+      "loss": 0.3178,
+      "step": 2534
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9198564561641927e-05,
+      "loss": 0.3143,
+      "step": 2536
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9197016014241853e-05,
+      "loss": 0.3072,
+      "step": 2538
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9195466034796056e-05,
+      "loss": 0.3021,
+      "step": 2540
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.919391462354588e-05,
+      "loss": 0.3162,
+      "step": 2542
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9192361780732896e-05,
+      "loss": 0.301,
+      "step": 2544
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9190807506598895e-05,
+      "loss": 0.2746,
+      "step": 2546
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.918925180138589e-05,
+      "loss": 0.2919,
+      "step": 2548
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9187694665336106e-05,
+      "loss": 0.3085,
+      "step": 2550
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9186136098692008e-05,
+      "loss": 0.302,
+      "step": 2552
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9184576101696277e-05,
+      "loss": 0.3078,
+      "step": 2554
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.918301467459182e-05,
+      "loss": 0.2958,
+      "step": 2556
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9181451817621753e-05,
+      "loss": 0.2922,
+      "step": 2558
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.917988753102943e-05,
+      "loss": 0.3217,
+      "step": 2560
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9178321815058426e-05,
+      "loss": 0.285,
+      "step": 2562
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9176754669952534e-05,
+      "loss": 0.3077,
+      "step": 2564
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9175186095955766e-05,
+      "loss": 0.308,
+      "step": 2566
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9173616093312364e-05,
+      "loss": 0.2936,
+      "step": 2568
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9172044662266787e-05,
+      "loss": 0.2961,
+      "step": 2570
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.917047180306372e-05,
+      "loss": 0.2918,
+      "step": 2572
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9168897515948074e-05,
+      "loss": 0.314,
+      "step": 2574
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.916732180116497e-05,
+      "loss": 0.3068,
+      "step": 2576
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9165744658959763e-05,
+      "loss": 0.2924,
+      "step": 2578
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9164166089578026e-05,
+      "loss": 0.2987,
+      "step": 2580
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9162586093265555e-05,
+      "loss": 0.315,
+      "step": 2582
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9161004670268365e-05,
+      "loss": 0.2955,
+      "step": 2584
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9159421820832696e-05,
+      "loss": 0.3077,
+      "step": 2586
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.915783754520501e-05,
+      "loss": 0.3106,
+      "step": 2588
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9156251843631996e-05,
+      "loss": 0.3177,
+      "step": 2590
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.915466471636055e-05,
+      "loss": 0.3211,
+      "step": 2592
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9153076163637806e-05,
+      "loss": 0.2919,
+      "step": 2594
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9151486185711113e-05,
+      "loss": 0.2859,
+      "step": 2596
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9149894782828044e-05,
+      "loss": 0.3109,
+      "step": 2598
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.914830195523639e-05,
+      "loss": 0.2911,
+      "step": 2600
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9146707703184168e-05,
+      "loss": 0.2569,
+      "step": 2602
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9145112026919615e-05,
+      "loss": 0.3109,
+      "step": 2604
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.914351492669119e-05,
+      "loss": 0.3088,
+      "step": 2606
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9141916402747576e-05,
+      "loss": 0.2975,
+      "step": 2608
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.914031645533767e-05,
+      "loss": 0.2903,
+      "step": 2610
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.91387150847106e-05,
+      "loss": 0.3029,
+      "step": 2612
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9137112291115713e-05,
+      "loss": 0.2918,
+      "step": 2614
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.913550807480257e-05,
+      "loss": 0.3048,
+      "step": 2616
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9133902436020966e-05,
+      "loss": 0.2966,
+      "step": 2618
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.913229537502091e-05,
+      "loss": 0.2864,
+      "step": 2620
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9130686892052628e-05,
+      "loss": 0.3041,
+      "step": 2622
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.912907698736658e-05,
+      "loss": 0.2901,
+      "step": 2624
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.912746566121344e-05,
+      "loss": 0.3059,
+      "step": 2626
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9125852913844102e-05,
+      "loss": 0.2821,
+      "step": 2628
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9124238745509682e-05,
+      "loss": 0.3068,
+      "step": 2630
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.912262315646152e-05,
+      "loss": 0.3024,
+      "step": 2632
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9121006146951172e-05,
+      "loss": 0.2663,
+      "step": 2634
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9119387717230427e-05,
+      "loss": 0.283,
+      "step": 2636
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.911776786755128e-05,
+      "loss": 0.2837,
+      "step": 2638
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9116146598165954e-05,
+      "loss": 0.3077,
+      "step": 2640
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9114523909326892e-05,
+      "loss": 0.274,
+      "step": 2642
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9112899801286766e-05,
+      "loss": 0.2965,
+      "step": 2644
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9111274274298457e-05,
+      "loss": 0.3009,
+      "step": 2646
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9109647328615074e-05,
+      "loss": 0.2753,
+      "step": 2648
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9108018964489942e-05,
+      "loss": 0.2705,
+      "step": 2650
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.910638918217661e-05,
+      "loss": 0.2892,
+      "step": 2652
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9104757981928852e-05,
+      "loss": 0.3046,
+      "step": 2654
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.910312536400065e-05,
+      "loss": 0.2998,
+      "step": 2656
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9101491328646228e-05,
+      "loss": 0.2896,
+      "step": 2658
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9099855876120005e-05,
+      "loss": 0.2855,
+      "step": 2660
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.909821900667664e-05,
+      "loss": 0.2924,
+      "step": 2662
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9096580720571004e-05,
+      "loss": 0.2856,
+      "step": 2664
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.909494101805819e-05,
+      "loss": 0.2791,
+      "step": 2666
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.909329989939352e-05,
+      "loss": 0.2848,
+      "step": 2668
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9091657364832518e-05,
+      "loss": 0.3059,
+      "step": 2670
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9090013414630943e-05,
+      "loss": 0.2832,
+      "step": 2672
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.908836804904477e-05,
+      "loss": 0.2779,
+      "step": 2674
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.90867212683302e-05,
+      "loss": 0.2676,
+      "step": 2676
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9085073072743644e-05,
+      "loss": 0.3032,
+      "step": 2678
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.908342346254174e-05,
+      "loss": 0.3164,
+      "step": 2680
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9081772437981344e-05,
+      "loss": 0.2889,
+      "step": 2682
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9080119999319536e-05,
+      "loss": 0.2923,
+      "step": 2684
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.907846614681361e-05,
+      "loss": 0.2842,
+      "step": 2686
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9076810880721085e-05,
+      "loss": 0.3022,
+      "step": 2688
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.90751542012997e-05,
+      "loss": 0.2779,
+      "step": 2690
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.907349610880741e-05,
+      "loss": 0.2785,
+      "step": 2692
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9071836603502397e-05,
+      "loss": 0.2906,
+      "step": 2694
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9070175685643054e-05,
+      "loss": 0.2717,
+      "step": 2696
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9068513355487996e-05,
+      "loss": 0.2896,
+      "step": 2698
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.906684961329607e-05,
+      "loss": 0.2838,
+      "step": 2700
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9065184459326328e-05,
+      "loss": 0.3062,
+      "step": 2702
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9063517893838048e-05,
+      "loss": 0.3351,
+      "step": 2704
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.906184991709072e-05,
+      "loss": 0.3178,
+      "step": 2706
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.906018052934407e-05,
+      "loss": 0.313,
+      "step": 2708
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9058509730858037e-05,
+      "loss": 0.2599,
+      "step": 2710
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9056837521892765e-05,
+      "loss": 0.2795,
+      "step": 2712
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9055163902708642e-05,
+      "loss": 0.289,
+      "step": 2714
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.905348887356625e-05,
+      "loss": 0.2986,
+      "step": 2716
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9051812434726418e-05,
+      "loss": 0.2873,
+      "step": 2718
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9050134586450175e-05,
+      "loss": 0.3006,
+      "step": 2720
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.904845532899877e-05,
+      "loss": 0.3107,
+      "step": 2722
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.904677466263368e-05,
+      "loss": 0.2941,
+      "step": 2724
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9045092587616598e-05,
+      "loss": 0.3029,
+      "step": 2726
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9043409104209436e-05,
+      "loss": 0.2769,
+      "step": 2728
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9041724212674323e-05,
+      "loss": 0.2872,
+      "step": 2730
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9040037913273616e-05,
+      "loss": 0.287,
+      "step": 2732
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9038350206269874e-05,
+      "loss": 0.2954,
+      "step": 2734
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9036661091925894e-05,
+      "loss": 0.3083,
+      "step": 2736
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9034970570504684e-05,
+      "loss": 0.2589,
+      "step": 2738
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9033278642269467e-05,
+      "loss": 0.3008,
+      "step": 2740
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9031585307483688e-05,
+      "loss": 0.2895,
+      "step": 2742
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.902989056641102e-05,
+      "loss": 0.3067,
+      "step": 2744
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9028194419315338e-05,
+      "loss": 0.2846,
+      "step": 2746
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9026496866460753e-05,
+      "loss": 0.2877,
+      "step": 2748
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.902479790811158e-05,
+      "loss": 0.3036,
+      "step": 2750
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9023097544532368e-05,
+      "loss": 0.3012,
+      "step": 2752
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.902139577598787e-05,
+      "loss": 0.2911,
+      "step": 2754
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9019692602743067e-05,
+      "loss": 0.2929,
+      "step": 2756
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9017988025063153e-05,
+      "loss": 0.283,
+      "step": 2758
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9016282043213544e-05,
+      "loss": 0.2908,
+      "step": 2760
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.901457465745988e-05,
+      "loss": 0.2935,
+      "step": 2762
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.901286586806801e-05,
+      "loss": 0.3021,
+      "step": 2764
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9011155675304006e-05,
+      "loss": 0.2932,
+      "step": 2766
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9009444079434162e-05,
+      "loss": 0.3196,
+      "step": 2768
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9007731080724977e-05,
+      "loss": 0.2999,
+      "step": 2770
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9006016679443187e-05,
+      "loss": 0.2936,
+      "step": 2772
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9004300875855734e-05,
+      "loss": 0.285,
+      "step": 2774
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9002583670229778e-05,
+      "loss": 0.2927,
+      "step": 2776
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9000865062832708e-05,
+      "loss": 0.2814,
+      "step": 2778
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.899914505393212e-05,
+      "loss": 0.3099,
+      "step": 2780
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8997423643795836e-05,
+      "loss": 0.3047,
+      "step": 2782
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.899570083269189e-05,
+      "loss": 0.2742,
+      "step": 2784
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8993976620888536e-05,
+      "loss": 0.3233,
+      "step": 2786
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8992251008654248e-05,
+      "loss": 0.3003,
+      "step": 2788
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.899052399625772e-05,
+      "loss": 0.3147,
+      "step": 2790
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8988795583967855e-05,
+      "loss": 0.2998,
+      "step": 2792
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8987065772053785e-05,
+      "loss": 0.2711,
+      "step": 2794
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8985334560784853e-05,
+      "loss": 0.3152,
+      "step": 2796
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8983601950430623e-05,
+      "loss": 0.3011,
+      "step": 2798
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8981867941260875e-05,
+      "loss": 0.2822,
+      "step": 2800
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8980132533545605e-05,
+      "loss": 0.2903,
+      "step": 2802
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8978395727555034e-05,
+      "loss": 0.3006,
+      "step": 2804
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8976657523559592e-05,
+      "loss": 0.318,
+      "step": 2806
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8974917921829934e-05,
+      "loss": 0.2895,
+      "step": 2808
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8973176922636927e-05,
+      "loss": 0.3041,
+      "step": 2810
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.897143452625166e-05,
+      "loss": 0.3035,
+      "step": 2812
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.896969073294543e-05,
+      "loss": 0.2651,
+      "step": 2814
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.896794554298977e-05,
+      "loss": 0.2857,
+      "step": 2816
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.896619895665641e-05,
+      "loss": 0.2826,
+      "step": 2818
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8964450974217317e-05,
+      "loss": 0.2946,
+      "step": 2820
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8962701595944657e-05,
+      "loss": 0.295,
+      "step": 2822
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8960950822110822e-05,
+      "loss": 0.3086,
+      "step": 2824
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8959198652988423e-05,
+      "loss": 0.2817,
+      "step": 2826
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8957445088850286e-05,
+      "loss": 0.2671,
+      "step": 2828
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8955690129969455e-05,
+      "loss": 0.2744,
+      "step": 2830
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8953933776619192e-05,
+      "loss": 0.2972,
+      "step": 2832
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8952176029072968e-05,
+      "loss": 0.3035,
+      "step": 2834
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8950416887604485e-05,
+      "loss": 0.3069,
+      "step": 2836
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8948656352487652e-05,
+      "loss": 0.2702,
+      "step": 2838
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8946894423996592e-05,
+      "loss": 0.299,
+      "step": 2840
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8945131102405664e-05,
+      "loss": 0.2958,
+      "step": 2842
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.894336638798942e-05,
+      "loss": 0.2513,
+      "step": 2844
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.894160028102264e-05,
+      "loss": 0.316,
+      "step": 2846
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8939832781780326e-05,
+      "loss": 0.2811,
+      "step": 2848
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8938063890537687e-05,
+      "loss": 0.2888,
+      "step": 2850
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8936293607570154e-05,
+      "loss": 0.2695,
+      "step": 2852
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8934521933153376e-05,
+      "loss": 0.299,
+      "step": 2854
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.893274886756321e-05,
+      "loss": 0.2679,
+      "step": 2856
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8930974411075743e-05,
+      "loss": 0.2949,
+      "step": 2858
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8929198563967265e-05,
+      "loss": 0.2841,
+      "step": 2860
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8927421326514296e-05,
+      "loss": 0.2693,
+      "step": 2862
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.892564269899356e-05,
+      "loss": 0.2988,
+      "step": 2864
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8923862681682005e-05,
+      "loss": 0.3143,
+      "step": 2866
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.892208127485679e-05,
+      "loss": 0.2889,
+      "step": 2868
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8920298478795298e-05,
+      "loss": 0.2918,
+      "step": 2870
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8918514293775123e-05,
+      "loss": 0.2983,
+      "step": 2872
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8916728720074077e-05,
+      "loss": 0.2775,
+      "step": 2874
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8914941757970184e-05,
+      "loss": 0.2812,
+      "step": 2876
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.891315340774169e-05,
+      "loss": 0.2889,
+      "step": 2878
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8911363669667052e-05,
+      "loss": 0.2653,
+      "step": 2880
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8909572544024955e-05,
+      "loss": 0.2936,
+      "step": 2882
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8907780031094282e-05,
+      "loss": 0.2886,
+      "step": 2884
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8905986131154138e-05,
+      "loss": 0.2668,
+      "step": 2886
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.890419084448386e-05,
+      "loss": 0.3083,
+      "step": 2888
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8902394171362975e-05,
+      "loss": 0.2789,
+      "step": 2890
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.8900596112071247e-05,
+      "loss": 0.2871,
+      "step": 2892
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8898796666888645e-05,
+      "loss": 0.276,
+      "step": 2894
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.889699583609535e-05,
+      "loss": 0.2866,
+      "step": 2896
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.889519361997178e-05,
+      "loss": 0.2888,
+      "step": 2898
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8893390018798538e-05,
+      "loss": 0.2905,
+      "step": 2900
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.889158503285647e-05,
+      "loss": 0.3007,
+      "step": 2902
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.888977866242662e-05,
+      "loss": 0.2915,
+      "step": 2904
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8887970907790255e-05,
+      "loss": 0.308,
+      "step": 2906
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8886161769228857e-05,
+      "loss": 0.2867,
+      "step": 2908
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.888435124702412e-05,
+      "loss": 0.2805,
+      "step": 2910
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.888253934145796e-05,
+      "loss": 0.2939,
+      "step": 2912
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8880726052812502e-05,
+      "loss": 0.298,
+      "step": 2914
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.887891138137009e-05,
+      "loss": 0.2902,
+      "step": 2916
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8877095327413283e-05,
+      "loss": 0.2854,
+      "step": 2918
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8875277891224854e-05,
+      "loss": 0.2569,
+      "step": 2920
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.887345907308779e-05,
+      "loss": 0.3057,
+      "step": 2922
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8871638873285295e-05,
+      "loss": 0.2779,
+      "step": 2924
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8869817292100792e-05,
+      "loss": 0.2838,
+      "step": 2926
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.886799432981791e-05,
+      "loss": 0.3311,
+      "step": 2928
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.88661699867205e-05,
+      "loss": 0.2978,
+      "step": 2930
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.8864344263092627e-05,
+      "loss": 0.2899,
+      "step": 2932
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.886251715921857e-05,
+      "loss": 0.2747,
+      "step": 2934
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8860688675382823e-05,
+      "loss": 0.2994,
+      "step": 2936
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8858858811870094e-05,
+      "loss": 0.2803,
+      "step": 2938
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.885702756896531e-05,
+      "loss": 0.2816,
+      "step": 2940
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8855194946953605e-05,
+      "loss": 0.2603,
+      "step": 2942
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.885336094612033e-05,
+      "loss": 0.3026,
+      "step": 2944
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8851525566751062e-05,
+      "loss": 0.2952,
+      "step": 2946
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8849688809131578e-05,
+      "loss": 0.2844,
+      "step": 2948
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8847850673547877e-05,
+      "loss": 0.2769,
+      "step": 2950
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8846011160286168e-05,
+      "loss": 0.3053,
+      "step": 2952
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8844170269632877e-05,
+      "loss": 0.2921,
+      "step": 2954
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.884232800187465e-05,
+      "loss": 0.2964,
+      "step": 2956
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8840484357298338e-05,
+      "loss": 0.2978,
+      "step": 2958
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8838639336191012e-05,
+      "loss": 0.2862,
+      "step": 2960
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8836792938839955e-05,
+      "loss": 0.245,
+      "step": 2962
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8834945165532662e-05,
+      "loss": 0.2917,
+      "step": 2964
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8833096016556855e-05,
+      "loss": 0.2604,
+      "step": 2966
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.883124549220045e-05,
+      "loss": 0.2775,
+      "step": 2968
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8829393592751594e-05,
+      "loss": 0.2993,
+      "step": 2970
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8827540318498644e-05,
+      "loss": 0.2926,
+      "step": 2972
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.882568566973016e-05,
+      "loss": 0.3071,
+      "step": 2974
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.882382964673493e-05,
+      "loss": 0.2846,
+      "step": 2976
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8821972249801956e-05,
+      "loss": 0.2873,
+      "step": 2978
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8820113479220442e-05,
+      "loss": 0.2839,
+      "step": 2980
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8818253335279815e-05,
+      "loss": 0.2907,
+      "step": 2982
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.881639181826971e-05,
+      "loss": 0.2969,
+      "step": 2984
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.881452892847999e-05,
+      "loss": 0.2998,
+      "step": 2986
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.881266466620071e-05,
+      "loss": 0.3013,
+      "step": 2988
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8810799031722153e-05,
+      "loss": 0.2812,
+      "step": 2990
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8808932025334816e-05,
+      "loss": 0.3009,
+      "step": 2992
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8807063647329402e-05,
+      "loss": 0.287,
+      "step": 2994
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8805193897996834e-05,
+      "loss": 0.2658,
+      "step": 2996
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8803322777628247e-05,
+      "loss": 0.2631,
+      "step": 2998
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.880145028651499e-05,
+      "loss": 0.2893,
+      "step": 3000
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8799576424948616e-05,
+      "loss": 0.3094,
+      "step": 3002
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8797701193220908e-05,
+      "loss": 0.2807,
+      "step": 3004
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8795824591623853e-05,
+      "loss": 0.3014,
+      "step": 3006
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8793946620449647e-05,
+      "loss": 0.2875,
+      "step": 3008
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.879206727999071e-05,
+      "loss": 0.2879,
+      "step": 3010
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.879018657053967e-05,
+      "loss": 0.2772,
+      "step": 3012
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.878830449238936e-05,
+      "loss": 0.275,
+      "step": 3014
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8786421045832842e-05,
+      "loss": 0.2864,
+      "step": 3016
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.878453623116338e-05,
+      "loss": 0.3137,
+      "step": 3018
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8782650048674456e-05,
+      "loss": 0.2784,
+      "step": 3020
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8780762498659762e-05,
+      "loss": 0.2983,
+      "step": 3022
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.87788735814132e-05,
+      "loss": 0.2735,
+      "step": 3024
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8776983297228896e-05,
+      "loss": 0.2899,
+      "step": 3026
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8775091646401177e-05,
+      "loss": 0.3009,
+      "step": 3028
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8773198629224588e-05,
+      "loss": 0.2586,
+      "step": 3030
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8771304245993888e-05,
+      "loss": 0.2723,
+      "step": 3032
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.876940849700404e-05,
+      "loss": 0.278,
+      "step": 3034
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.876751138255024e-05,
+      "loss": 0.2593,
+      "step": 3036
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8765612902927874e-05,
+      "loss": 0.2761,
+      "step": 3038
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.876371305843255e-05,
+      "loss": 0.3046,
+      "step": 3040
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8761811849360093e-05,
+      "loss": 0.2837,
+      "step": 3042
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8759909276006527e-05,
+      "loss": 0.2996,
+      "step": 3044
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8758005338668105e-05,
+      "loss": 0.2737,
+      "step": 3046
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8756100037641282e-05,
+      "loss": 0.267,
+      "step": 3048
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8754193373222732e-05,
+      "loss": 0.2913,
+      "step": 3050
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.875228534570933e-05,
+      "loss": 0.2832,
+      "step": 3052
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.8750375955398176e-05,
+      "loss": 0.3088,
+      "step": 3054
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.874846520258657e-05,
+      "loss": 0.2808,
+      "step": 3056
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8746553087572042e-05,
+      "loss": 0.2693,
+      "step": 3058
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.874463961065231e-05,
+      "loss": 0.2798,
+      "step": 3060
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8742724772125326e-05,
+      "loss": 0.2919,
+      "step": 3062
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8740808572289246e-05,
+      "loss": 0.2757,
+      "step": 3064
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.873889101144243e-05,
+      "loss": 0.2987,
+      "step": 3066
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.873697208988346e-05,
+      "loss": 0.2629,
+      "step": 3068
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8735051807911127e-05,
+      "loss": 0.2788,
+      "step": 3070
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8733130165824437e-05,
+      "loss": 0.2753,
+      "step": 3072
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8731207163922597e-05,
+      "loss": 0.3008,
+      "step": 3074
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8729282802505036e-05,
+      "loss": 0.2813,
+      "step": 3076
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8727357081871398e-05,
+      "loss": 0.2807,
+      "step": 3078
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.872543000232152e-05,
+      "loss": 0.2905,
+      "step": 3080
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8723501564155477e-05,
+      "loss": 0.2691,
+      "step": 3082
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8721571767673534e-05,
+      "loss": 0.2876,
+      "step": 3084
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8719640613176175e-05,
+      "loss": 0.302,
+      "step": 3086
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8717708100964095e-05,
+      "loss": 0.2913,
+      "step": 3088
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8715774231338203e-05,
+      "loss": 0.2932,
+      "step": 3090
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.871383900459962e-05,
+      "loss": 0.3084,
+      "step": 3092
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8711902421049668e-05,
+      "loss": 0.2886,
+      "step": 3094
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8709964480989896e-05,
+      "loss": 0.3087,
+      "step": 3096
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.8708025184722046e-05,
+      "loss": 0.2581,
+      "step": 3098
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.8706084532548093e-05,
+      "loss": 0.2853,
+      "step": 3100
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.8704142524770207e-05,
+      "loss": 0.2844,
+      "step": 3102
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.870219916169077e-05,
+      "loss": 0.2671,
+      "step": 3104
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.8700254443612376e-05,
+      "loss": 0.2533,
+      "step": 3106
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.8698308370837847e-05,
+      "loss": 0.2754,
+      "step": 3108
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.869636094367018e-05,
+      "loss": 0.2697,
+      "step": 3110
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.8694412162412626e-05,
+      "loss": 0.3111,
+      "step": 3112
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.869246202736861e-05,
+      "loss": 0.2928,
+      "step": 3114
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.8690510538841792e-05,
+      "loss": 0.2738,
+      "step": 3116
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.868855769713603e-05,
+      "loss": 0.2643,
+      "step": 3118
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.868660350255539e-05,
+      "loss": 0.2714,
+      "step": 3120
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.8684647955404168e-05,
+      "loss": 0.2977,
+      "step": 3122
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.8682691055986847e-05,
+      "loss": 0.2676,
+      "step": 3124
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.8680732804608136e-05,
+      "loss": 0.27,
+      "step": 3126
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.8678773201572948e-05,
+      "loss": 0.2979,
+      "step": 3128
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.867681224718641e-05,
+      "loss": 0.2846,
+      "step": 3130
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.8674849941753857e-05,
+      "loss": 0.2798,
+      "step": 3132
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.8672886285580835e-05,
+      "loss": 0.3039,
+      "step": 3134
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.8670921278973098e-05,
+      "loss": 0.2681,
+      "step": 3136
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.8668954922236613e-05,
+      "loss": 0.3153,
+      "step": 3138
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.866698721567756e-05,
+      "loss": 0.3031,
+      "step": 3140
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.8665018159602323e-05,
+      "loss": 0.2815,
+      "step": 3142
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.8663047754317503e-05,
+      "loss": 0.2812,
+      "step": 3144
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.8661076000129902e-05,
+      "loss": 0.2441,
+      "step": 3146
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.8659102897346543e-05,
+      "loss": 0.2887,
+      "step": 3148
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.8657128446274644e-05,
+      "loss": 0.2726,
+      "step": 3150
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.8655152647221652e-05,
+      "loss": 0.3039,
+      "step": 3152
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.865317550049521e-05,
+      "loss": 0.2891,
+      "step": 3154
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.8651197006403173e-05,
+      "loss": 0.2501,
+      "step": 3156
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.8649217165253608e-05,
+      "loss": 0.3005,
+      "step": 3158
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.8647235977354796e-05,
+      "loss": 0.2675,
+      "step": 3160
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.864525344301522e-05,
+      "loss": 0.2938,
+      "step": 3162
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.8643269562543573e-05,
+      "loss": 0.2947,
+      "step": 3164
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.8641284336248763e-05,
+      "loss": 0.2873,
+      "step": 3166
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.8639297764439908e-05,
+      "loss": 0.2959,
+      "step": 3168
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.863730984742633e-05,
+      "loss": 0.2689,
+      "step": 3170
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.8635320585517558e-05,
+      "loss": 0.2744,
+      "step": 3172
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.863332997902334e-05,
+      "loss": 0.2702,
+      "step": 3174
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.863133802825363e-05,
+      "loss": 0.2874,
+      "step": 3176
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.8629344733518587e-05,
+      "loss": 0.2631,
+      "step": 3178
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.8627350095128583e-05,
+      "loss": 0.3093,
+      "step": 3180
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8625354113394202e-05,
+      "loss": 0.2725,
+      "step": 3182
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.862335678862623e-05,
+      "loss": 0.2925,
+      "step": 3184
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8621358121135665e-05,
+      "loss": 0.2886,
+      "step": 3186
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8619358111233717e-05,
+      "loss": 0.2929,
+      "step": 3188
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8617356759231798e-05,
+      "loss": 0.2846,
+      "step": 3190
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.861535406544154e-05,
+      "loss": 0.2764,
+      "step": 3192
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8613350030174777e-05,
+      "loss": 0.2613,
+      "step": 3194
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8611344653743552e-05,
+      "loss": 0.2757,
+      "step": 3196
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.860933793646011e-05,
+      "loss": 0.2532,
+      "step": 3198
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8607329878636925e-05,
+      "loss": 0.2976,
+      "step": 3200
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8605320480586658e-05,
+      "loss": 0.2663,
+      "step": 3202
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.860330974262219e-05,
+      "loss": 0.2917,
+      "step": 3204
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.860129766505661e-05,
+      "loss": 0.2856,
+      "step": 3206
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8599284248203215e-05,
+      "loss": 0.2747,
+      "step": 3208
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.85972694923755e-05,
+      "loss": 0.2939,
+      "step": 3210
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8595253397887187e-05,
+      "loss": 0.2551,
+      "step": 3212
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.85932359650522e-05,
+      "loss": 0.2922,
+      "step": 3214
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.859121719418466e-05,
+      "loss": 0.2691,
+      "step": 3216
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8589197085598907e-05,
+      "loss": 0.2693,
+      "step": 3218
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.8587175639609492e-05,
+      "loss": 0.2956,
+      "step": 3220
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.8585152856531166e-05,
+      "loss": 0.2831,
+      "step": 3222
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.858312873667889e-05,
+      "loss": 0.2593,
+      "step": 3224
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.858110328036784e-05,
+      "loss": 0.2913,
+      "step": 3226
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.8579076487913394e-05,
+      "loss": 0.2999,
+      "step": 3228
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.8577048359631134e-05,
+      "loss": 0.2493,
+      "step": 3230
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.857501889583686e-05,
+      "loss": 0.2561,
+      "step": 3232
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.857298809684658e-05,
+      "loss": 0.2595,
+      "step": 3234
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.857095596297649e-05,
+      "loss": 0.2918,
+      "step": 3236
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.8568922494543023e-05,
+      "loss": 0.2856,
+      "step": 3238
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.8566887691862797e-05,
+      "loss": 0.3029,
+      "step": 3240
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.856485155525265e-05,
+      "loss": 0.2972,
+      "step": 3242
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.8562814085029622e-05,
+      "loss": 0.3018,
+      "step": 3244
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.8560775281510968e-05,
+      "loss": 0.2929,
+      "step": 3246
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.855873514501414e-05,
+      "loss": 0.2733,
+      "step": 3248
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.85566936758568e-05,
+      "loss": 0.3046,
+      "step": 3250
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.855465087435683e-05,
+      "loss": 0.3002,
+      "step": 3252
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.85526067408323e-05,
+      "loss": 0.2803,
+      "step": 3254
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.8550561275601504e-05,
+      "loss": 0.266,
+      "step": 3256
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.8548514478982934e-05,
+      "loss": 0.2744,
+      "step": 3258
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.8546466351295293e-05,
+      "loss": 0.3024,
+      "step": 3260
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.854441689285749e-05,
+      "loss": 0.2852,
+      "step": 3262
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8542366103988637e-05,
+      "loss": 0.2911,
+      "step": 3264
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8540313985008063e-05,
+      "loss": 0.3099,
+      "step": 3266
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.85382605362353e-05,
+      "loss": 0.2841,
+      "step": 3268
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8536205757990077e-05,
+      "loss": 0.307,
+      "step": 3270
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.853414965059235e-05,
+      "loss": 0.2702,
+      "step": 3272
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.853209221436226e-05,
+      "loss": 0.2847,
+      "step": 3274
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.853003344962017e-05,
+      "loss": 0.2993,
+      "step": 3276
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8527973356686647e-05,
+      "loss": 0.2662,
+      "step": 3278
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8525911935882463e-05,
+      "loss": 0.2886,
+      "step": 3280
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8523849187528596e-05,
+      "loss": 0.2778,
+      "step": 3282
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8521785111946227e-05,
+      "loss": 0.2657,
+      "step": 3284
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.851971970945676e-05,
+      "loss": 0.285,
+      "step": 3286
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.851765298038178e-05,
+      "loss": 0.2591,
+      "step": 3288
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8515584925043105e-05,
+      "loss": 0.2755,
+      "step": 3290
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.851351554376274e-05,
+      "loss": 0.2671,
+      "step": 3292
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8511444836862905e-05,
+      "loss": 0.2477,
+      "step": 3294
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8509372804666022e-05,
+      "loss": 0.2592,
+      "step": 3296
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8507299447494728e-05,
+      "loss": 0.2712,
+      "step": 3298
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.8505224765671857e-05,
+      "loss": 0.2794,
+      "step": 3300
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.850314875952045e-05,
+      "loss": 0.2924,
+      "step": 3302
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8501071429363768e-05,
+      "loss": 0.2983,
+      "step": 3304
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8498992775525255e-05,
+      "loss": 0.3002,
+      "step": 3306
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.849691279832858e-05,
+      "loss": 0.2934,
+      "step": 3308
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.849483149809761e-05,
+      "loss": 0.2765,
+      "step": 3310
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8492748875156414e-05,
+      "loss": 0.3018,
+      "step": 3312
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.849066492982928e-05,
+      "loss": 0.2974,
+      "step": 3314
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.848857966244069e-05,
+      "loss": 0.2681,
+      "step": 3316
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.848649307331534e-05,
+      "loss": 0.314,
+      "step": 3318
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8484405162778122e-05,
+      "loss": 0.2714,
+      "step": 3320
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8482315931154146e-05,
+      "loss": 0.2594,
+      "step": 3322
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8480225378768717e-05,
+      "loss": 0.2724,
+      "step": 3324
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.847813350594735e-05,
+      "loss": 0.2667,
+      "step": 3326
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8476040313015767e-05,
+      "loss": 0.2719,
+      "step": 3328
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8473945800299896e-05,
+      "loss": 0.2878,
+      "step": 3330
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8471849968125867e-05,
+      "loss": 0.2531,
+      "step": 3332
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8469752816820013e-05,
+      "loss": 0.2869,
+      "step": 3334
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.846765434670888e-05,
+      "loss": 0.2731,
+      "step": 3336
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.846555455811922e-05,
+      "loss": 0.2727,
+      "step": 3338
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.846345345137798e-05,
+      "loss": 0.278,
+      "step": 3340
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8461351026812318e-05,
+      "loss": 0.2972,
+      "step": 3342
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8459247284749607e-05,
+      "loss": 0.2508,
+      "step": 3344
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8457142225517404e-05,
+      "loss": 0.295,
+      "step": 3346
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.845503584944349e-05,
+      "loss": 0.2679,
+      "step": 3348
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.845292815685584e-05,
+      "loss": 0.2655,
+      "step": 3350
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8450819148082638e-05,
+      "loss": 0.2707,
+      "step": 3352
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8448708823452275e-05,
+      "loss": 0.2818,
+      "step": 3354
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8446597183293345e-05,
+      "loss": 0.2954,
+      "step": 3356
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8444484227934644e-05,
+      "loss": 0.2882,
+      "step": 3358
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8442369957705178e-05,
+      "loss": 0.2732,
+      "step": 3360
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.844025437293415e-05,
+      "loss": 0.2784,
+      "step": 3362
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8438137473950984e-05,
+      "loss": 0.2874,
+      "step": 3364
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.843601926108528e-05,
+      "loss": 0.2865,
+      "step": 3366
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.843389973466688e-05,
+      "loss": 0.2693,
+      "step": 3368
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8431778895025795e-05,
+      "loss": 0.2997,
+      "step": 3370
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8429656742492263e-05,
+      "loss": 0.2787,
+      "step": 3372
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8427533277396713e-05,
+      "loss": 0.2914,
+      "step": 3374
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8425408500069794e-05,
+      "loss": 0.261,
+      "step": 3376
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8423282410842344e-05,
+      "loss": 0.2572,
+      "step": 3378
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8421155010045414e-05,
+      "loss": 0.2966,
+      "step": 3380
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.8419026298010256e-05,
+      "loss": 0.2618,
+      "step": 3382
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.841689627506832e-05,
+      "loss": 0.2907,
+      "step": 3384
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.8414764941551277e-05,
+      "loss": 0.2892,
+      "step": 3386
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.8412632297790985e-05,
+      "loss": 0.2807,
+      "step": 3388
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.841049834411952e-05,
+      "loss": 0.2901,
+      "step": 3390
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.8408363080869153e-05,
+      "loss": 0.2693,
+      "step": 3392
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.840622650837235e-05,
+      "loss": 0.2874,
+      "step": 3394
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.8404088626961803e-05,
+      "loss": 0.2841,
+      "step": 3396
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.8401949436970397e-05,
+      "loss": 0.2967,
+      "step": 3398
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.8399808938731218e-05,
+      "loss": 0.2746,
+      "step": 3400
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.839766713257755e-05,
+      "loss": 0.2864,
+      "step": 3402
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.8395524018842904e-05,
+      "loss": 0.2806,
+      "step": 3404
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.8393379597860965e-05,
+      "loss": 0.2732,
+      "step": 3406
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.839123386996564e-05,
+      "loss": 0.303,
+      "step": 3408
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.838908683549104e-05,
+      "loss": 0.278,
+      "step": 3410
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.8386938494771473e-05,
+      "loss": 0.298,
+      "step": 3412
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.8384788848141447e-05,
+      "loss": 0.2732,
+      "step": 3414
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.838263789593568e-05,
+      "loss": 0.2543,
+      "step": 3416
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.8380485638489096e-05,
+      "loss": 0.2586,
+      "step": 3418
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.8378332076136818e-05,
+      "loss": 0.2696,
+      "step": 3420
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.8376177209214166e-05,
+      "loss": 0.3002,
+      "step": 3422
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.8374021038056673e-05,
+      "loss": 0.2638,
+      "step": 3424
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.837186356300007e-05,
+      "loss": 0.283,
+      "step": 3426
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.8369704784380298e-05,
+      "loss": 0.2841,
+      "step": 3428
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.8367544702533487e-05,
+      "loss": 0.3038,
+      "step": 3430
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.8365383317795983e-05,
+      "loss": 0.2626,
+      "step": 3432
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.836322063050433e-05,
+      "loss": 0.2508,
+      "step": 3434
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.8361056640995275e-05,
+      "loss": 0.2866,
+      "step": 3436
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.8358891349605764e-05,
+      "loss": 0.2871,
+      "step": 3438
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.8356724756672955e-05,
+      "loss": 0.2821,
+      "step": 3440
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.83545568625342e-05,
+      "loss": 0.2614,
+      "step": 3442
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.8352387667527057e-05,
+      "loss": 0.2998,
+      "step": 3444
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.8350217171989287e-05,
+      "loss": 0.2834,
+      "step": 3446
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.834804537625885e-05,
+      "loss": 0.2654,
+      "step": 3448
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.834587228067391e-05,
+      "loss": 0.2774,
+      "step": 3450
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.8343697885572845e-05,
+      "loss": 0.2714,
+      "step": 3452
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.8341522191294216e-05,
+      "loss": 0.273,
+      "step": 3454
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.8339345198176798e-05,
+      "loss": 0.2867,
+      "step": 3456
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.833716690655956e-05,
+      "loss": 0.2783,
+      "step": 3458
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.833498731678169e-05,
+      "loss": 0.2828,
+      "step": 3460
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.8332806429182556e-05,
+      "loss": 0.2855,
+      "step": 3462
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.8330624244101742e-05,
+      "loss": 0.2501,
+      "step": 3464
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.8328440761879033e-05,
+      "loss": 0.263,
+      "step": 3466
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.832625598285441e-05,
+      "loss": 0.2848,
+      "step": 3468
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8324069907368068e-05,
+      "loss": 0.2734,
+      "step": 3470
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8321882535760386e-05,
+      "loss": 0.2838,
+      "step": 3472
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.831969386837196e-05,
+      "loss": 0.2597,
+      "step": 3474
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.831750390554358e-05,
+      "loss": 0.2614,
+      "step": 3476
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8315312647616245e-05,
+      "loss": 0.2845,
+      "step": 3478
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.831312009493114e-05,
+      "loss": 0.2673,
+      "step": 3480
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8310926247829672e-05,
+      "loss": 0.28,
+      "step": 3482
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8308731106653436e-05,
+      "loss": 0.2865,
+      "step": 3484
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8306534671744234e-05,
+      "loss": 0.2754,
+      "step": 3486
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8304336943444066e-05,
+      "loss": 0.2809,
+      "step": 3488
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8302137922095133e-05,
+      "loss": 0.2749,
+      "step": 3490
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8299937608039845e-05,
+      "loss": 0.2943,
+      "step": 3492
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8297736001620805e-05,
+      "loss": 0.2843,
+      "step": 3494
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.829553310318082e-05,
+      "loss": 0.2729,
+      "step": 3496
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8293328913062892e-05,
+      "loss": 0.2629,
+      "step": 3498
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.829112343161024e-05,
+      "loss": 0.2886,
+      "step": 3500
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.828891665916627e-05,
+      "loss": 0.2669,
+      "step": 3502
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8286708596074596e-05,
+      "loss": 0.2594,
+      "step": 3504
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8284499242679027e-05,
+      "loss": 0.2645,
+      "step": 3506
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8282288599323576e-05,
+      "loss": 0.2906,
+      "step": 3508
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8280076666352457e-05,
+      "loss": 0.2944,
+      "step": 3510
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.827786344411009e-05,
+      "loss": 0.2694,
+      "step": 3512
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8275648932941085e-05,
+      "loss": 0.2701,
+      "step": 3514
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.827343313319026e-05,
+      "loss": 0.2529,
+      "step": 3516
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8271216045202627e-05,
+      "loss": 0.2737,
+      "step": 3518
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8268997669323418e-05,
+      "loss": 0.2881,
+      "step": 3520
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8266778005898035e-05,
+      "loss": 0.2758,
+      "step": 3522
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.826455705527211e-05,
+      "loss": 0.2675,
+      "step": 3524
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8262334817791448e-05,
+      "loss": 0.2757,
+      "step": 3526
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.826011129380208e-05,
+      "loss": 0.279,
+      "step": 3528
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.825788648365022e-05,
+      "loss": 0.2403,
+      "step": 3530
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8255660387682287e-05,
+      "loss": 0.263,
+      "step": 3532
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8253433006244907e-05,
+      "loss": 0.2824,
+      "step": 3534
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8251204339684894e-05,
+      "loss": 0.2693,
+      "step": 3536
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8248974388349275e-05,
+      "loss": 0.2676,
+      "step": 3538
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8246743152585264e-05,
+      "loss": 0.2564,
+      "step": 3540
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8244510632740285e-05,
+      "loss": 0.2905,
+      "step": 3542
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.824227682916196e-05,
+      "loss": 0.2779,
+      "step": 3544
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8240041742198102e-05,
+      "loss": 0.2716,
+      "step": 3546
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8237805372196737e-05,
+      "loss": 0.2858,
+      "step": 3548
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8235567719506084e-05,
+      "loss": 0.2716,
+      "step": 3550
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8233328784474564e-05,
+      "loss": 0.2582,
+      "step": 3552
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8231088567450793e-05,
+      "loss": 0.2474,
+      "step": 3554
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8228847068783598e-05,
+      "loss": 0.2637,
+      "step": 3556
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8226604288821982e-05,
+      "loss": 0.2809,
+      "step": 3558
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.822436022791518e-05,
+      "loss": 0.2794,
+      "step": 3560
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8222114886412598e-05,
+      "loss": 0.2815,
+      "step": 3562
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8219868264663854e-05,
+      "loss": 0.2772,
+      "step": 3564
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8217620363018773e-05,
+      "loss": 0.2892,
+      "step": 3566
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8215371181827358e-05,
+      "loss": 0.2637,
+      "step": 3568
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8213120721439837e-05,
+      "loss": 0.2556,
+      "step": 3570
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.821086898220661e-05,
+      "loss": 0.246,
+      "step": 3572
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.82086159644783e-05,
+      "loss": 0.2754,
+      "step": 3574
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8206361668605717e-05,
+      "loss": 0.2892,
+      "step": 3576
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8204106094939868e-05,
+      "loss": 0.2541,
+      "step": 3578
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8201849243831966e-05,
+      "loss": 0.261,
+      "step": 3580
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8199591115633424e-05,
+      "loss": 0.2728,
+      "step": 3582
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8197331710695843e-05,
+      "loss": 0.2745,
+      "step": 3584
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8195071029371032e-05,
+      "loss": 0.2728,
+      "step": 3586
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8192809072010998e-05,
+      "loss": 0.2754,
+      "step": 3588
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8190545838967945e-05,
+      "loss": 0.2568,
+      "step": 3590
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.818828133059427e-05,
+      "loss": 0.2508,
+      "step": 3592
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8186015547242582e-05,
+      "loss": 0.2784,
+      "step": 3594
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8183748489265683e-05,
+      "loss": 0.2851,
+      "step": 3596
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8181480157016558e-05,
+      "loss": 0.2759,
+      "step": 3598
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8179210550848413e-05,
+      "loss": 0.2819,
+      "step": 3600
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8176939671114645e-05,
+      "loss": 0.2839,
+      "step": 3602
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.817466751816884e-05,
+      "loss": 0.2627,
+      "step": 3604
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.81723940923648e-05,
+      "loss": 0.2747,
+      "step": 3606
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.81701193940565e-05,
+      "loss": 0.2798,
+      "step": 3608
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.816784342359814e-05,
+      "loss": 0.2587,
+      "step": 3610
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8165566181344104e-05,
+      "loss": 0.2338,
+      "step": 3612
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8163287667648974e-05,
+      "loss": 0.2786,
+      "step": 3614
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.816100788286753e-05,
+      "loss": 0.2696,
+      "step": 3616
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8158726827354756e-05,
+      "loss": 0.2438,
+      "step": 3618
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8156444501465827e-05,
+      "loss": 0.2606,
+      "step": 3620
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.815416090555612e-05,
+      "loss": 0.2681,
+      "step": 3622
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8151876039981208e-05,
+      "loss": 0.2857,
+      "step": 3624
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8149589905096866e-05,
+      "loss": 0.2803,
+      "step": 3626
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8147302501259055e-05,
+      "loss": 0.2647,
+      "step": 3628
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8145013828823943e-05,
+      "loss": 0.278,
+      "step": 3630
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8142723888147897e-05,
+      "loss": 0.2824,
+      "step": 3632
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.814043267958748e-05,
+      "loss": 0.2832,
+      "step": 3634
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8138140203499443e-05,
+      "loss": 0.2856,
+      "step": 3636
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8135846460240752e-05,
+      "loss": 0.2615,
+      "step": 3638
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8133551450168553e-05,
+      "loss": 0.2731,
+      "step": 3640
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8131255173640196e-05,
+      "loss": 0.268,
+      "step": 3642
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8128957631013235e-05,
+      "loss": 0.2652,
+      "step": 3644
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.812665882264541e-05,
+      "loss": 0.2669,
+      "step": 3646
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8124358748894667e-05,
+      "loss": 0.2823,
+      "step": 3648
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.812205741011914e-05,
+      "loss": 0.2815,
+      "step": 3650
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8119754806677167e-05,
+      "loss": 0.2726,
+      "step": 3652
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.811745093892728e-05,
+      "loss": 0.2767,
+      "step": 3654
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8115145807228215e-05,
+      "loss": 0.2743,
+      "step": 3656
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8112839411938892e-05,
+      "loss": 0.2806,
+      "step": 3658
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.811053175341844e-05,
+      "loss": 0.2809,
+      "step": 3660
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.810822283202617e-05,
+      "loss": 0.2689,
+      "step": 3662
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.810591264812161e-05,
+      "loss": 0.2431,
+      "step": 3664
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8103601202064463e-05,
+      "loss": 0.2743,
+      "step": 3666
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8101288494214647e-05,
+      "loss": 0.2466,
+      "step": 3668
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.809897452493226e-05,
+      "loss": 0.2607,
+      "step": 3670
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8096659294577612e-05,
+      "loss": 0.2422,
+      "step": 3672
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.80943428035112e-05,
+      "loss": 0.2809,
+      "step": 3674
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.809202505209372e-05,
+      "loss": 0.2642,
+      "step": 3676
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.808970604068606e-05,
+      "loss": 0.2734,
+      "step": 3678
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.808738576964931e-05,
+      "loss": 0.2741,
+      "step": 3680
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8085064239344757e-05,
+      "loss": 0.2887,
+      "step": 3682
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8082741450133875e-05,
+      "loss": 0.2625,
+      "step": 3684
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8080417402378347e-05,
+      "loss": 0.27,
+      "step": 3686
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8078092096440038e-05,
+      "loss": 0.2485,
+      "step": 3688
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.807576553268102e-05,
+      "loss": 0.2768,
+      "step": 3690
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.807343771146356e-05,
+      "loss": 0.2637,
+      "step": 3692
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.807110863315011e-05,
+      "loss": 0.2496,
+      "step": 3694
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8068778298103327e-05,
+      "loss": 0.2774,
+      "step": 3696
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8066446706686066e-05,
+      "loss": 0.2708,
+      "step": 3698
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8064113859261373e-05,
+      "loss": 0.2626,
+      "step": 3700
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8061779756192485e-05,
+      "loss": 0.2615,
+      "step": 3702
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8059444397842846e-05,
+      "loss": 0.2869,
+      "step": 3704
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8057107784576088e-05,
+      "loss": 0.2622,
+      "step": 3706
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.805476991675603e-05,
+      "loss": 0.2866,
+      "step": 3708
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8052430794746714e-05,
+      "loss": 0.2508,
+      "step": 3710
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8050090418912343e-05,
+      "loss": 0.2642,
+      "step": 3712
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8047748789617342e-05,
+      "loss": 0.2361,
+      "step": 3714
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8045405907226312e-05,
+      "loss": 0.2506,
+      "step": 3716
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8043061772104064e-05,
+      "loss": 0.2407,
+      "step": 3718
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8040716384615596e-05,
+      "loss": 0.266,
+      "step": 3720
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.80383697451261e-05,
+      "loss": 0.25,
+      "step": 3722
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8036021854000968e-05,
+      "loss": 0.2429,
+      "step": 3724
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.803367271160579e-05,
+      "loss": 0.262,
+      "step": 3726
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8031322318306333e-05,
+      "loss": 0.2723,
+      "step": 3728
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8028970674468583e-05,
+      "loss": 0.2677,
+      "step": 3730
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8026617780458703e-05,
+      "loss": 0.2422,
+      "step": 3732
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8024263636643057e-05,
+      "loss": 0.2729,
+      "step": 3734
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8021908243388205e-05,
+      "loss": 0.2511,
+      "step": 3736
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8019551601060897e-05,
+      "loss": 0.2589,
+      "step": 3738
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8017193710028084e-05,
+      "loss": 0.2637,
+      "step": 3740
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8014834570656903e-05,
+      "loss": 0.2846,
+      "step": 3742
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8012474183314697e-05,
+      "loss": 0.2552,
+      "step": 3744
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.801011254836899e-05,
+      "loss": 0.2708,
+      "step": 3746
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.800774966618751e-05,
+      "loss": 0.2664,
+      "step": 3748
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8005385537138172e-05,
+      "loss": 0.2536,
+      "step": 3750
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8003020161589094e-05,
+      "loss": 0.252,
+      "step": 3752
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8000653539908575e-05,
+      "loss": 0.2624,
+      "step": 3754
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7998285672465128e-05,
+      "loss": 0.2758,
+      "step": 3756
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7995916559627437e-05,
+      "loss": 0.2431,
+      "step": 3758
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.79935462017644e-05,
+      "loss": 0.2682,
+      "step": 3760
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7991174599245094e-05,
+      "loss": 0.273,
+      "step": 3762
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7988801752438797e-05,
+      "loss": 0.2567,
+      "step": 3764
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.798642766171498e-05,
+      "loss": 0.279,
+      "step": 3766
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7984052327443305e-05,
+      "loss": 0.2389,
+      "step": 3768
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7981675749993633e-05,
+      "loss": 0.25,
+      "step": 3770
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7979297929736012e-05,
+      "loss": 0.2561,
+      "step": 3772
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.797691886704069e-05,
+      "loss": 0.245,
+      "step": 3774
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.79745385622781e-05,
+      "loss": 0.2615,
+      "step": 3776
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7972157015818882e-05,
+      "loss": 0.2782,
+      "step": 3778
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7969774228033857e-05,
+      "loss": 0.2681,
+      "step": 3780
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7967390199294038e-05,
+      "loss": 0.2712,
+      "step": 3782
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7965004929970645e-05,
+      "loss": 0.2597,
+      "step": 3784
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7962618420435075e-05,
+      "loss": 0.2627,
+      "step": 3786
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7960230671058935e-05,
+      "loss": 0.2557,
+      "step": 3788
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7957841682214005e-05,
+      "loss": 0.2708,
+      "step": 3790
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7955451454272278e-05,
+      "loss": 0.2446,
+      "step": 3792
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7953059987605926e-05,
+      "loss": 0.2617,
+      "step": 3794
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.795066728258732e-05,
+      "loss": 0.2812,
+      "step": 3796
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.794827333958902e-05,
+      "loss": 0.256,
+      "step": 3798
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.794587815898378e-05,
+      "loss": 0.2581,
+      "step": 3800
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.794348174114456e-05,
+      "loss": 0.2626,
+      "step": 3802
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7941084086444486e-05,
+      "loss": 0.2621,
+      "step": 3804
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.79386851952569e-05,
+      "loss": 0.269,
+      "step": 3806
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7936285067955324e-05,
+      "loss": 0.2676,
+      "step": 3808
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7933883704913475e-05,
+      "loss": 0.2468,
+      "step": 3810
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7931481106505266e-05,
+      "loss": 0.2564,
+      "step": 3812
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.79290772731048e-05,
+      "loss": 0.2806,
+      "step": 3814
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7926672205086374e-05,
+      "loss": 0.2671,
+      "step": 3816
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7924265902824468e-05,
+      "loss": 0.2566,
+      "step": 3818
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7921858366693774e-05,
+      "loss": 0.2603,
+      "step": 3820
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.791944959706915e-05,
+      "loss": 0.2804,
+      "step": 3822
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.791703959432567e-05,
+      "loss": 0.2598,
+      "step": 3824
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7914628358838586e-05,
+      "loss": 0.2566,
+      "step": 3826
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7912215890983344e-05,
+      "loss": 0.2532,
+      "step": 3828
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7909802191135588e-05,
+      "loss": 0.2795,
+      "step": 3830
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.790738725967115e-05,
+      "loss": 0.2441,
+      "step": 3832
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7904971096966044e-05,
+      "loss": 0.2764,
+      "step": 3834
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.7902553703396493e-05,
+      "loss": 0.2491,
+      "step": 3836
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7900135079338907e-05,
+      "loss": 0.2561,
+      "step": 3838
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7897715225169876e-05,
+      "loss": 0.2609,
+      "step": 3840
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.789529414126619e-05,
+      "loss": 0.2748,
+      "step": 3842
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7892871828004843e-05,
+      "loss": 0.2354,
+      "step": 3844
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7890448285762992e-05,
+      "loss": 0.2548,
+      "step": 3846
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.788802351491801e-05,
+      "loss": 0.2576,
+      "step": 3848
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7885597515847448e-05,
+      "loss": 0.2618,
+      "step": 3850
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7883170288929054e-05,
+      "loss": 0.2448,
+      "step": 3852
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.788074183454077e-05,
+      "loss": 0.2654,
+      "step": 3854
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7878312153060714e-05,
+      "loss": 0.2575,
+      "step": 3856
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.787588124486722e-05,
+      "loss": 0.2793,
+      "step": 3858
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.787344911033879e-05,
+      "loss": 0.2445,
+      "step": 3860
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7871015749854124e-05,
+      "loss": 0.2633,
+      "step": 3862
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.786858116379212e-05,
+      "loss": 0.2619,
+      "step": 3864
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7866145352531865e-05,
+      "loss": 0.2529,
+      "step": 3866
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7863708316452625e-05,
+      "loss": 0.2612,
+      "step": 3868
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.786127005593387e-05,
+      "loss": 0.2554,
+      "step": 3870
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7858830571355258e-05,
+      "loss": 0.2319,
+      "step": 3872
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7856389863096626e-05,
+      "loss": 0.2341,
+      "step": 3874
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7853947931538023e-05,
+      "loss": 0.2753,
+      "step": 3876
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.785150477705967e-05,
+      "loss": 0.2645,
+      "step": 3878
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.784906040004198e-05,
+      "loss": 0.2741,
+      "step": 3880
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.784661480086557e-05,
+      "loss": 0.235,
+      "step": 3882
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.7844167979911238e-05,
+      "loss": 0.2461,
+      "step": 3884
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.784171993755997e-05,
+      "loss": 0.2653,
+      "step": 3886
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.7839270674192942e-05,
+      "loss": 0.2591,
+      "step": 3888
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.7836820190191524e-05,
+      "loss": 0.2519,
+      "step": 3890
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.7834368485937278e-05,
+      "loss": 0.2516,
+      "step": 3892
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.7831915561811955e-05,
+      "loss": 0.2179,
+      "step": 3894
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.7829461418197492e-05,
+      "loss": 0.2696,
+      "step": 3896
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.7827006055476013e-05,
+      "loss": 0.2476,
+      "step": 3898
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.782454947402984e-05,
+      "loss": 0.2823,
+      "step": 3900
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.7822091674241487e-05,
+      "loss": 0.2498,
+      "step": 3902
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.7819632656493647e-05,
+      "loss": 0.2632,
+      "step": 3904
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.7817172421169206e-05,
+      "loss": 0.2461,
+      "step": 3906
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.7814710968651244e-05,
+      "loss": 0.2687,
+      "step": 3908
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.7812248299323026e-05,
+      "loss": 0.2754,
+      "step": 3910
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.780978441356801e-05,
+      "loss": 0.259,
+      "step": 3912
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.780731931176985e-05,
+      "loss": 0.2468,
+      "step": 3914
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.7804852994312365e-05,
+      "loss": 0.2661,
+      "step": 3916
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.7802385461579593e-05,
+      "loss": 0.2485,
+      "step": 3918
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.779991671395574e-05,
+      "loss": 0.2621,
+      "step": 3920
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.779744675182521e-05,
+      "loss": 0.2483,
+      "step": 3922
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7794975575572596e-05,
+      "loss": 0.255,
+      "step": 3924
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7792503185582684e-05,
+      "loss": 0.2348,
+      "step": 3926
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7790029582240436e-05,
+      "loss": 0.2623,
+      "step": 3928
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7787554765931013e-05,
+      "loss": 0.2423,
+      "step": 3930
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7785078737039766e-05,
+      "loss": 0.2559,
+      "step": 3932
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7782601495952232e-05,
+      "loss": 0.2484,
+      "step": 3934
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7780123043054135e-05,
+      "loss": 0.2284,
+      "step": 3936
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7777643378731384e-05,
+      "loss": 0.2216,
+      "step": 3938
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.777516250337009e-05,
+      "loss": 0.2606,
+      "step": 3940
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7772680417356537e-05,
+      "loss": 0.2244,
+      "step": 3942
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7770197121077207e-05,
+      "loss": 0.279,
+      "step": 3944
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7767712614918773e-05,
+      "loss": 0.2562,
+      "step": 3946
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7765226899268085e-05,
+      "loss": 0.2466,
+      "step": 3948
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7762739974512194e-05,
+      "loss": 0.2496,
+      "step": 3950
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7760251841038325e-05,
+      "loss": 0.2598,
+      "step": 3952
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.775776249923391e-05,
+      "loss": 0.2678,
+      "step": 3954
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.775527194948655e-05,
+      "loss": 0.2676,
+      "step": 3956
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7752780192184043e-05,
+      "loss": 0.2677,
+      "step": 3958
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.7750287227714382e-05,
+      "loss": 0.2566,
+      "step": 3960
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7747793056465734e-05,
+      "loss": 0.2087,
+      "step": 3962
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7745297678826464e-05,
+      "loss": 0.2527,
+      "step": 3964
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.774280109518512e-05,
+      "loss": 0.2456,
+      "step": 3966
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7740303305930437e-05,
+      "loss": 0.2669,
+      "step": 3968
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.773780431145134e-05,
+      "loss": 0.2549,
+      "step": 3970
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7735304112136947e-05,
+      "loss": 0.205,
+      "step": 3972
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7732802708376554e-05,
+      "loss": 0.2572,
+      "step": 3974
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7730300100559645e-05,
+      "loss": 0.2716,
+      "step": 3976
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7727796289075905e-05,
+      "loss": 0.2565,
+      "step": 3978
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7725291274315188e-05,
+      "loss": 0.2251,
+      "step": 3980
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.772278505666755e-05,
+      "loss": 0.2422,
+      "step": 3982
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.772027763652322e-05,
+      "loss": 0.2535,
+      "step": 3984
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.771776901427263e-05,
+      "loss": 0.2694,
+      "step": 3986
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7715259190306393e-05,
+      "loss": 0.2421,
+      "step": 3988
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.77127481650153e-05,
+      "loss": 0.2569,
+      "step": 3990
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7710235938790344e-05,
+      "loss": 0.2304,
+      "step": 3992
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7707722512022692e-05,
+      "loss": 0.2402,
+      "step": 3994
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7705207885103707e-05,
+      "loss": 0.211,
+      "step": 3996
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.770269205842494e-05,
+      "loss": 0.265,
+      "step": 3998
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7700175032378114e-05,
+      "loss": 0.2543,
+      "step": 4000
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7697656807355157e-05,
+      "loss": 0.2455,
+      "step": 4002
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7695137383748172e-05,
+      "loss": 0.226,
+      "step": 4004
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7692616761949458e-05,
+      "loss": 0.2697,
+      "step": 4006
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7690094942351488e-05,
+      "loss": 0.2476,
+      "step": 4008
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7687571925346934e-05,
+      "loss": 0.2375,
+      "step": 4010
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7685047711328643e-05,
+      "loss": 0.2509,
+      "step": 4012
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7682522300689665e-05,
+      "loss": 0.264,
+      "step": 4014
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7679995693823217e-05,
+      "loss": 0.2397,
+      "step": 4016
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.767746789112271e-05,
+      "loss": 0.2345,
+      "step": 4018
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7674938892981747e-05,
+      "loss": 0.256,
+      "step": 4020
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.767240869979411e-05,
+      "loss": 0.2561,
+      "step": 4022
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7669877311953773e-05,
+      "loss": 0.2506,
+      "step": 4024
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.766734472985489e-05,
+      "loss": 0.2642,
+      "step": 4026
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.76648109538918e-05,
+      "loss": 0.235,
+      "step": 4028
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7662275984459036e-05,
+      "loss": 0.2313,
+      "step": 4030
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.765973982195131e-05,
+      "loss": 0.2288,
+      "step": 4032
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7657202466763523e-05,
+      "loss": 0.2154,
+      "step": 4034
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.765466391929076e-05,
+      "loss": 0.247,
+      "step": 4036
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.765212417992829e-05,
+      "loss": 0.2406,
+      "step": 4038
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7649583249071574e-05,
+      "loss": 0.234,
+      "step": 4040
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.764704112711625e-05,
+      "loss": 0.2357,
+      "step": 4042
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7644497814458153e-05,
+      "loss": 0.2465,
+      "step": 4044
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.764195331149329e-05,
+      "loss": 0.2609,
+      "step": 4046
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.763940761861786e-05,
+      "loss": 0.2645,
+      "step": 4048
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.763686073622825e-05,
+      "loss": 0.2594,
+      "step": 4050
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.763431266472102e-05,
+      "loss": 0.2444,
+      "step": 4052
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7631763404492935e-05,
+      "loss": 0.2453,
+      "step": 4054
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7629212955940934e-05,
+      "loss": 0.2682,
+      "step": 4056
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7626661319462134e-05,
+      "loss": 0.2622,
+      "step": 4058
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7624108495453848e-05,
+      "loss": 0.2333,
+      "step": 4060
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7621554484313573e-05,
+      "loss": 0.2378,
+      "step": 4062
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7618999286438986e-05,
+      "loss": 0.2577,
+      "step": 4064
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.761644290222795e-05,
+      "loss": 0.238,
+      "step": 4066
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7613885332078508e-05,
+      "loss": 0.233,
+      "step": 4068
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7611326576388905e-05,
+      "loss": 0.2374,
+      "step": 4070
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7608766635557552e-05,
+      "loss": 0.2344,
+      "step": 4072
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7606205509983048e-05,
+      "loss": 0.2529,
+      "step": 4074
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7603643200064188e-05,
+      "loss": 0.217,
+      "step": 4076
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.760107970619994e-05,
+      "loss": 0.2563,
+      "step": 4078
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.759851502878946e-05,
+      "loss": 0.2288,
+      "step": 4080
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7595949168232083e-05,
+      "loss": 0.2457,
+      "step": 4082
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.759338212492734e-05,
+      "loss": 0.2308,
+      "step": 4084
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.759081389927493e-05,
+      "loss": 0.2734,
+      "step": 4086
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.758824449167476e-05,
+      "loss": 0.2532,
+      "step": 4088
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7585673902526893e-05,
+      "loss": 0.2262,
+      "step": 4090
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7583102132231594e-05,
+      "loss": 0.2583,
+      "step": 4092
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.758052918118931e-05,
+      "loss": 0.2481,
+      "step": 4094
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.757795504980067e-05,
+      "loss": 0.2223,
+      "step": 4096
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7575379738466475e-05,
+      "loss": 0.2158,
+      "step": 4098
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7572803247587734e-05,
+      "loss": 0.2527,
+      "step": 4100
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7570225577565622e-05,
+      "loss": 0.2227,
+      "step": 4102
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7567646728801497e-05,
+      "loss": 0.2587,
+      "step": 4104
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.756506670169691e-05,
+      "loss": 0.1924,
+      "step": 4106
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.756248549665359e-05,
+      "loss": 0.191,
+      "step": 4108
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.755990311407345e-05,
+      "loss": 0.2099,
+      "step": 4110
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.755731955435859e-05,
+      "loss": 0.1711,
+      "step": 4112
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7554734817911283e-05,
+      "loss": 0.2189,
+      "step": 4114
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7552148905133995e-05,
+      "loss": 0.2072,
+      "step": 4116
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.754956181642938e-05,
+      "loss": 0.196,
+      "step": 4118
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7546973552200256e-05,
+      "loss": 0.2036,
+      "step": 4120
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7544384112849648e-05,
+      "loss": 0.2007,
+      "step": 4122
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7541793498780738e-05,
+      "loss": 0.208,
+      "step": 4124
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7539201710396913e-05,
+      "loss": 0.2203,
+      "step": 4126
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7536608748101728e-05,
+      "loss": 0.209,
+      "step": 4128
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.753401461229894e-05,
+      "loss": 0.2173,
+      "step": 4130
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.753141930339246e-05,
+      "loss": 0.2103,
+      "step": 4132
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7528822821786403e-05,
+      "loss": 0.2167,
+      "step": 4134
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7526225167885068e-05,
+      "loss": 0.2148,
+      "step": 4136
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.752362634209292e-05,
+      "loss": 0.1986,
+      "step": 4138
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.752102634481462e-05,
+      "loss": 0.1869,
+      "step": 4140
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7518425176455008e-05,
+      "loss": 0.2136,
+      "step": 4142
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.751582283741911e-05,
+      "loss": 0.1973,
+      "step": 4144
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.751321932811212e-05,
+      "loss": 0.2078,
+      "step": 4146
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7510614648939432e-05,
+      "loss": 0.2088,
+      "step": 4148
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.750800880030661e-05,
+      "loss": 0.1904,
+      "step": 4150
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7505401782619408e-05,
+      "loss": 0.2362,
+      "step": 4152
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7502793596283756e-05,
+      "loss": 0.2193,
+      "step": 4154
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.750018424170577e-05,
+      "loss": 0.2163,
+      "step": 4156
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.749757371929175e-05,
+      "loss": 0.1923,
+      "step": 4158
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7494962029448165e-05,
+      "loss": 0.226,
+      "step": 4160
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.749234917258168e-05,
+      "loss": 0.1786,
+      "step": 4162
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.7489735149099143e-05,
+      "loss": 0.235,
+      "step": 4164
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.748711995940757e-05,
+      "loss": 0.228,
+      "step": 4166
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.748450360391417e-05,
+      "loss": 0.2078,
+      "step": 4168
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.7481886083026323e-05,
+      "loss": 0.2359,
+      "step": 4170
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.7479267397151604e-05,
+      "loss": 0.2137,
+      "step": 4172
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.7476647546697758e-05,
+      "loss": 0.2122,
+      "step": 4174
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.747402653207272e-05,
+      "loss": 0.1959,
+      "step": 4176
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.74714043536846e-05,
+      "loss": 0.2113,
+      "step": 4178
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.746878101194169e-05,
+      "loss": 0.2128,
+      "step": 4180
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.7466156507252462e-05,
+      "loss": 0.1976,
+      "step": 4182
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.7463530840025578e-05,
+      "loss": 0.2239,
+      "step": 4184
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.746090401066987e-05,
+      "loss": 0.219,
+      "step": 4186
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.745827601959436e-05,
+      "loss": 0.2199,
+      "step": 4188
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.7455646867208237e-05,
+      "loss": 0.1982,
+      "step": 4190
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.7453016553920893e-05,
+      "loss": 0.1771,
+      "step": 4192
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.7450385080141874e-05,
+      "loss": 0.2305,
+      "step": 4194
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.7447752446280933e-05,
+      "loss": 0.2265,
+      "step": 4196
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.7445118652747983e-05,
+      "loss": 0.186,
+      "step": 4198
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.7442483699953128e-05,
+      "loss": 0.1697,
+      "step": 4200
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.7439847588306653e-05,
+      "loss": 0.2082,
+      "step": 4202
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.7437210318219017e-05,
+      "loss": 0.2284,
+      "step": 4204
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.7434571890100864e-05,
+      "loss": 0.1919,
+      "step": 4206
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.743193230436302e-05,
+      "loss": 0.1843,
+      "step": 4208
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.7429291561416485e-05,
+      "loss": 0.1933,
+      "step": 4210
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.7426649661672446e-05,
+      "loss": 0.2136,
+      "step": 4212
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.7424006605542266e-05,
+      "loss": 0.2131,
+      "step": 4214
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.742136239343749e-05,
+      "loss": 0.1739,
+      "step": 4216
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.7418717025769835e-05,
+      "loss": 0.212,
+      "step": 4218
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.7416070502951215e-05,
+      "loss": 0.2331,
+      "step": 4220
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.741342282539371e-05,
+      "loss": 0.2165,
+      "step": 4222
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.7410773993509585e-05,
+      "loss": 0.2241,
+      "step": 4224
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.740812400771128e-05,
+      "loss": 0.2422,
+      "step": 4226
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.7405472868411416e-05,
+      "loss": 0.1957,
+      "step": 4228
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.74028205760228e-05,
+      "loss": 0.2168,
+      "step": 4230
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.740016713095842e-05,
+      "loss": 0.1938,
+      "step": 4232
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.7397512533631424e-05,
+      "loss": 0.1813,
+      "step": 4234
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.7394856784455164e-05,
+      "loss": 0.1872,
+      "step": 4236
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.7392199883843156e-05,
+      "loss": 0.2159,
+      "step": 4238
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.73895418322091e-05,
+      "loss": 0.1855,
+      "step": 4240
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.7386882629966875e-05,
+      "loss": 0.2032,
+      "step": 4242
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.7384222277530544e-05,
+      "loss": 0.2146,
+      "step": 4244
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.7381560775314336e-05,
+      "loss": 0.2007,
+      "step": 4246
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.737889812373267e-05,
+      "loss": 0.2126,
+      "step": 4248
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.7376234323200144e-05,
+      "loss": 0.1904,
+      "step": 4250
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.737356937413153e-05,
+      "loss": 0.2157,
+      "step": 4252
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.737090327694178e-05,
+      "loss": 0.175,
+      "step": 4254
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.7368236032046027e-05,
+      "loss": 0.1881,
+      "step": 4256
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.7365567639859583e-05,
+      "loss": 0.1764,
+      "step": 4258
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.7362898100797938e-05,
+      "loss": 0.2059,
+      "step": 4260
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.7360227415276755e-05,
+      "loss": 0.2129,
+      "step": 4262
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.735755558371188e-05,
+      "loss": 0.2048,
+      "step": 4264
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.735488260651934e-05,
+      "loss": 0.1988,
+      "step": 4266
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.735220848411534e-05,
+      "loss": 0.2093,
+      "step": 4268
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.734953321691626e-05,
+      "loss": 0.1806,
+      "step": 4270
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.734685680533866e-05,
+      "loss": 0.2161,
+      "step": 4272
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.7344179249799274e-05,
+      "loss": 0.1984,
+      "step": 4274
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.734150055071502e-05,
+      "loss": 0.2153,
+      "step": 4276
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.7338820708502994e-05,
+      "loss": 0.2043,
+      "step": 4278
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.733613972358046e-05,
+      "loss": 0.1908,
+      "step": 4280
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.7333457596364885e-05,
+      "loss": 0.195,
+      "step": 4282
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.7330774327273882e-05,
+      "loss": 0.205,
+      "step": 4284
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.7328089916725257e-05,
+      "loss": 0.2049,
+      "step": 4286
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.7325404365136997e-05,
+      "loss": 0.2158,
+      "step": 4288
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.7322717672927266e-05,
+      "loss": 0.1865,
+      "step": 4290
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.7320029840514395e-05,
+      "loss": 0.2017,
+      "step": 4292
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.7317340868316907e-05,
+      "loss": 0.1783,
+      "step": 4294
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.7314650756753494e-05,
+      "loss": 0.1964,
+      "step": 4296
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.731195950624302e-05,
+      "loss": 0.2124,
+      "step": 4298
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.7309267117204545e-05,
+      "loss": 0.2162,
+      "step": 4300
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.7306573590057284e-05,
+      "loss": 0.1979,
+      "step": 4302
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.7303878925220646e-05,
+      "loss": 0.1782,
+      "step": 4304
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.730118312311421e-05,
+      "loss": 0.2197,
+      "step": 4306
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.729848618415773e-05,
+      "loss": 0.1779,
+      "step": 4308
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.7295788108771145e-05,
+      "loss": 0.2027,
+      "step": 4310
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.729308889737456e-05,
+      "loss": 0.1925,
+      "step": 4312
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.7290388550388267e-05,
+      "loss": 0.2162,
+      "step": 4314
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.728768706823273e-05,
+      "loss": 0.1943,
+      "step": 4316
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.728498445132859e-05,
+      "loss": 0.1996,
+      "step": 4318
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.728228070009667e-05,
+      "loss": 0.1945,
+      "step": 4320
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.7279575814957952e-05,
+      "loss": 0.2198,
+      "step": 4322
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.7276869796333616e-05,
+      "loss": 0.2125,
+      "step": 4324
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.7274162644645013e-05,
+      "loss": 0.2211,
+      "step": 4326
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.7271454360313663e-05,
+      "loss": 0.2254,
+      "step": 4328
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.7268744943761264e-05,
+      "loss": 0.2025,
+      "step": 4330
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.7266034395409698e-05,
+      "loss": 0.2073,
+      "step": 4332
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.7263322715681015e-05,
+      "loss": 0.215,
+      "step": 4334
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.7260609904997445e-05,
+      "loss": 0.247,
+      "step": 4336
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.725789596378139e-05,
+      "loss": 0.1902,
+      "step": 4338
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.725518089245544e-05,
+      "loss": 0.1993,
+      "step": 4340
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.725246469144235e-05,
+      "loss": 0.1685,
+      "step": 4342
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.7249747361165044e-05,
+      "loss": 0.1995,
+      "step": 4344
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.724702890204664e-05,
+      "loss": 0.2143,
+      "step": 4346
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.724430931451042e-05,
+      "loss": 0.2039,
+      "step": 4348
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.7241588598979845e-05,
+      "loss": 0.2046,
+      "step": 4350
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.723886675587855e-05,
+      "loss": 0.2055,
+      "step": 4352
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.7236143785630347e-05,
+      "loss": 0.174,
+      "step": 4354
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.7233419688659228e-05,
+      "loss": 0.1643,
+      "step": 4356
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.7230694465389352e-05,
+      "loss": 0.2063,
+      "step": 4358
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.7227968116245058e-05,
+      "loss": 0.2005,
+      "step": 4360
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.7225240641650854e-05,
+      "loss": 0.2329,
+      "step": 4362
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.7222512042031438e-05,
+      "loss": 0.219,
+      "step": 4364
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.7219782317811663e-05,
+      "loss": 0.2296,
+      "step": 4366
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.721705146941658e-05,
+      "loss": 0.1989,
+      "step": 4368
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.721431949727139e-05,
+      "loss": 0.2158,
+      "step": 4370
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.7211586401801498e-05,
+      "loss": 0.2015,
+      "step": 4372
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.7208852183432455e-05,
+      "loss": 0.2065,
+      "step": 4374
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.720611684259e-05,
+      "loss": 0.1761,
+      "step": 4376
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.7203380379700053e-05,
+      "loss": 0.1877,
+      "step": 4378
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.7200642795188697e-05,
+      "loss": 0.1953,
+      "step": 4380
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.71979040894822e-05,
+      "loss": 0.1861,
+      "step": 4382
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.7195164263006995e-05,
+      "loss": 0.2139,
+      "step": 4384
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.7192423316189696e-05,
+      "loss": 0.1917,
+      "step": 4386
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.718968124945709e-05,
+      "loss": 0.187,
+      "step": 4388
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.7186938063236133e-05,
+      "loss": 0.2014,
+      "step": 4390
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.7184193757953963e-05,
+      "loss": 0.2031,
+      "step": 4392
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.7181448334037894e-05,
+      "loss": 0.1763,
+      "step": 4394
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.7178701791915404e-05,
+      "loss": 0.196,
+      "step": 4396
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.717595413201415e-05,
+      "loss": 0.2152,
+      "step": 4398
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.717320535476197e-05,
+      "loss": 0.1932,
+      "step": 4400
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.7170455460586858e-05,
+      "loss": 0.2013,
+      "step": 4402
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.7167704449917006e-05,
+      "loss": 0.2072,
+      "step": 4404
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.716495232318076e-05,
+      "loss": 0.1828,
+      "step": 4406
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.7162199080806652e-05,
+      "loss": 0.1907,
+      "step": 4408
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.715944472322338e-05,
+      "loss": 0.2076,
+      "step": 4410
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.7156689250859817e-05,
+      "loss": 0.193,
+      "step": 4412
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.7153932664145015e-05,
+      "loss": 0.1934,
+      "step": 4414
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.7151174963508192e-05,
+      "loss": 0.1976,
+      "step": 4416
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.7148416149378743e-05,
+      "loss": 0.1774,
+      "step": 4418
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.714565622218624e-05,
+      "loss": 0.196,
+      "step": 4420
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.714289518236042e-05,
+      "loss": 0.2089,
+      "step": 4422
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.7140133030331205e-05,
+      "loss": 0.1858,
+      "step": 4424
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.713736976652868e-05,
+      "loss": 0.1916,
+      "step": 4426
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.71346053913831e-05,
+      "loss": 0.2183,
+      "step": 4428
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.713183990532491e-05,
+      "loss": 0.2112,
+      "step": 4430
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.712907330878471e-05,
+      "loss": 0.1865,
+      "step": 4432
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.7126305602193282e-05,
+      "loss": 0.1848,
+      "step": 4434
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.7123536785981582e-05,
+      "loss": 0.2214,
+      "step": 4436
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.712076686058073e-05,
+      "loss": 0.1895,
+      "step": 4438
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.7117995826422028e-05,
+      "loss": 0.2154,
+      "step": 4440
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.7115223683936948e-05,
+      "loss": 0.201,
+      "step": 4442
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.711245043355713e-05,
+      "loss": 0.2141,
+      "step": 4444
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.7109676075714396e-05,
+      "loss": 0.1983,
+      "step": 4446
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.710690061084073e-05,
+      "loss": 0.2037,
+      "step": 4448
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.7104124039368296e-05,
+      "loss": 0.1779,
+      "step": 4450
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.7101346361729426e-05,
+      "loss": 0.208,
+      "step": 4452
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.7098567578356626e-05,
+      "loss": 0.2189,
+      "step": 4454
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.709578768968258e-05,
+      "loss": 0.1942,
+      "step": 4456
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.7093006696140123e-05,
+      "loss": 0.1858,
+      "step": 4458
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.7090224598162287e-05,
+      "loss": 0.1859,
+      "step": 4460
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.7087441396182267e-05,
+      "loss": 0.1605,
+      "step": 4462
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.708465709063342e-05,
+      "loss": 0.2249,
+      "step": 4464
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.70818716819493e-05,
+      "loss": 0.2052,
+      "step": 4466
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.7079085170563602e-05,
+      "loss": 0.1763,
+      "step": 4468
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.707629755691021e-05,
+      "loss": 0.1703,
+      "step": 4470
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.707350884142318e-05,
+      "loss": 0.1929,
+      "step": 4472
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.7070719024536736e-05,
+      "loss": 0.1979,
+      "step": 4474
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.7067928106685273e-05,
+      "loss": 0.1613,
+      "step": 4476
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.7065136088303356e-05,
+      "loss": 0.192,
+      "step": 4478
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.7062342969825724e-05,
+      "loss": 0.1922,
+      "step": 4480
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.705954875168729e-05,
+      "loss": 0.1291,
+      "step": 4482
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.7056753434323134e-05,
+      "loss": 0.2087,
+      "step": 4484
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.7053957018168506e-05,
+      "loss": 0.2033,
+      "step": 4486
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.7051159503658833e-05,
+      "loss": 0.1272,
+      "step": 4488
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.704836089122971e-05,
+      "loss": 0.1954,
+      "step": 4490
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.7045561181316894e-05,
+      "loss": 0.1574,
+      "step": 4492
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.7042760374356327e-05,
+      "loss": 0.1813,
+      "step": 4494
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.7039958470784125e-05,
+      "loss": 0.1815,
+      "step": 4496
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.7037155471036547e-05,
+      "loss": 0.1988,
+      "step": 4498
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.7034351375550055e-05,
+      "loss": 0.1873,
+      "step": 4500
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.7031546184761264e-05,
+      "loss": 0.1639,
+      "step": 4502
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.7028739899106966e-05,
+      "loss": 0.1751,
+      "step": 4504
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.702593251902412e-05,
+      "loss": 0.1922,
+      "step": 4506
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.7023124044949855e-05,
+      "loss": 0.1884,
+      "step": 4508
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.7020314477321472e-05,
+      "loss": 0.2247,
+      "step": 4510
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.7017503816576444e-05,
+      "loss": 0.1862,
+      "step": 4512
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.7014692063152408e-05,
+      "loss": 0.2017,
+      "step": 4514
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.7011879217487183e-05,
+      "loss": 0.2115,
+      "step": 4516
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.7009065280018745e-05,
+      "loss": 0.2063,
+      "step": 4518
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.700625025118525e-05,
+      "loss": 0.1763,
+      "step": 4520
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.700343413142501e-05,
+      "loss": 0.1738,
+      "step": 4522
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.7000616921176528e-05,
+      "loss": 0.1802,
+      "step": 4524
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.6997798620878457e-05,
+      "loss": 0.2135,
+      "step": 4526
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.6994979230969634e-05,
+      "loss": 0.185,
+      "step": 4528
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.6992158751889055e-05,
+      "loss": 0.2059,
+      "step": 4530
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.698933718407589e-05,
+      "loss": 0.1831,
+      "step": 4532
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.6986514527969486e-05,
+      "loss": 0.1846,
+      "step": 4534
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.6983690784009343e-05,
+      "loss": 0.1925,
+      "step": 4536
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.6980865952635143e-05,
+      "loss": 0.1735,
+      "step": 4538
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.697804003428673e-05,
+      "loss": 0.1924,
+      "step": 4540
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.697521302940413e-05,
+      "loss": 0.162,
+      "step": 4542
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.697238493842752e-05,
+      "loss": 0.2009,
+      "step": 4544
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.6969555761797266e-05,
+      "loss": 0.2054,
+      "step": 4546
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.6966725499953885e-05,
+      "loss": 0.1751,
+      "step": 4548
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.696389415333807e-05,
+      "loss": 0.1728,
+      "step": 4550
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.6961061722390683e-05,
+      "loss": 0.1744,
+      "step": 4552
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.6958228207552755e-05,
+      "loss": 0.2016,
+      "step": 4554
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.6955393609265494e-05,
+      "loss": 0.1812,
+      "step": 4556
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.695255792797026e-05,
+      "loss": 0.1804,
+      "step": 4558
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.6949721164108594e-05,
+      "loss": 0.1967,
+      "step": 4560
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.69468833181222e-05,
+      "loss": 0.1877,
+      "step": 4562
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.694404439045295e-05,
+      "loss": 0.1795,
+      "step": 4564
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.694120438154289e-05,
+      "loss": 0.232,
+      "step": 4566
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.6938363291834234e-05,
+      "loss": 0.2077,
+      "step": 4568
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.693552112176936e-05,
+      "loss": 0.1965,
+      "step": 4570
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.6932677871790807e-05,
+      "loss": 0.1888,
+      "step": 4572
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.6929833542341303e-05,
+      "loss": 0.163,
+      "step": 4574
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.6926988133863728e-05,
+      "loss": 0.1926,
+      "step": 4576
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.6924141646801126e-05,
+      "loss": 0.2078,
+      "step": 4578
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.6921294081596727e-05,
+      "loss": 0.202,
+      "step": 4580
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.6918445438693912e-05,
+      "loss": 0.1746,
+      "step": 4582
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.6915595718536244e-05,
+      "loss": 0.2046,
+      "step": 4584
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.6912744921567437e-05,
+      "loss": 0.162,
+      "step": 4586
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.690989304823139e-05,
+      "loss": 0.2115,
+      "step": 4588
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.6907040098972153e-05,
+      "loss": 0.1968,
+      "step": 4590
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.6904186074233962e-05,
+      "loss": 0.1774,
+      "step": 4592
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.6901330974461203e-05,
+      "loss": 0.1958,
+      "step": 4594
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.6898474800098438e-05,
+      "loss": 0.2087,
+      "step": 4596
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.6895617551590398e-05,
+      "loss": 0.1676,
+      "step": 4598
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.689275922938198e-05,
+      "loss": 0.1793,
+      "step": 4600
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.6889899833918237e-05,
+      "loss": 0.184,
+      "step": 4602
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.6887039365644407e-05,
+      "loss": 0.1909,
+      "step": 4604
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.6884177825005886e-05,
+      "loss": 0.2137,
+      "step": 4606
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.6881315212448238e-05,
+      "loss": 0.1776,
+      "step": 4608
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.687845152841719e-05,
+      "loss": 0.182,
+      "step": 4610
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.6875586773358645e-05,
+      "loss": 0.1627,
+      "step": 4612
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.6872720947718663e-05,
+      "loss": 0.194,
+      "step": 4614
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.6869854051943476e-05,
+      "loss": 0.1551,
+      "step": 4616
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.686698608647948e-05,
+      "loss": 0.1795,
+      "step": 4618
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.6864117051773242e-05,
+      "loss": 0.2038,
+      "step": 4620
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.6861246948271494e-05,
+      "loss": 0.1884,
+      "step": 4622
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.685837577642113e-05,
+      "loss": 0.2006,
+      "step": 4624
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.685550353666921e-05,
+      "loss": 0.2014,
+      "step": 4626
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.685263022946297e-05,
+      "loss": 0.2005,
+      "step": 4628
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.6849755855249803e-05,
+      "loss": 0.1806,
+      "step": 4630
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.6846880414477275e-05,
+      "loss": 0.1985,
+      "step": 4632
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.6844003907593104e-05,
+      "loss": 0.2016,
+      "step": 4634
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.68411263350452e-05,
+      "loss": 0.1854,
+      "step": 4636
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.6838247697281604e-05,
+      "loss": 0.2079,
+      "step": 4638
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.6835367994750558e-05,
+      "loss": 0.1965,
+      "step": 4640
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.6832487227900448e-05,
+      "loss": 0.1885,
+      "step": 4642
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.682960539717983e-05,
+      "loss": 0.1874,
+      "step": 4644
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.6826722503037424e-05,
+      "loss": 0.176,
+      "step": 4646
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.682383854592213e-05,
+      "loss": 0.2045,
+      "step": 4648
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.682095352628299e-05,
+      "loss": 0.1989,
+      "step": 4650
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.6818067444569227e-05,
+      "loss": 0.1943,
+      "step": 4652
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.681518030123023e-05,
+      "loss": 0.1746,
+      "step": 4654
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.6812292096715546e-05,
+      "loss": 0.1709,
+      "step": 4656
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.680940283147489e-05,
+      "loss": 0.1619,
+      "step": 4658
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.6806512505958148e-05,
+      "loss": 0.2099,
+      "step": 4660
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.6803621120615357e-05,
+      "loss": 0.1606,
+      "step": 4662
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.6800728675896737e-05,
+      "loss": 0.1682,
+      "step": 4664
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.679783517225266e-05,
+      "loss": 0.2007,
+      "step": 4666
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.6794940610133665e-05,
+      "loss": 0.1717,
+      "step": 4668
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.6792044989990453e-05,
+      "loss": 0.1863,
+      "step": 4670
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.6789148312273905e-05,
+      "loss": 0.1652,
+      "step": 4672
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.678625057743505e-05,
+      "loss": 0.1698,
+      "step": 4674
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.678335178592509e-05,
+      "loss": 0.184,
+      "step": 4676
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.6780451938195382e-05,
+      "loss": 0.163,
+      "step": 4678
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.677755103469746e-05,
+      "loss": 0.1996,
+      "step": 4680
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.677464907588302e-05,
+      "loss": 0.2059,
+      "step": 4682
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.6771746062203914e-05,
+      "loss": 0.1709,
+      "step": 4684
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.6768841994112165e-05,
+      "loss": 0.1587,
+      "step": 4686
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.676593687205996e-05,
+      "loss": 0.1776,
+      "step": 4688
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.676303069649964e-05,
+      "loss": 0.1927,
+      "step": 4690
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.6760123467883733e-05,
+      "loss": 0.1881,
+      "step": 4692
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.6757215186664907e-05,
+      "loss": 0.1775,
+      "step": 4694
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.675430585329601e-05,
+      "loss": 0.1765,
+      "step": 4696
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.6751395468230042e-05,
+      "loss": 0.2045,
+      "step": 4698
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.6748484031920173e-05,
+      "loss": 0.1908,
+      "step": 4700
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.6745571544819735e-05,
+      "loss": 0.1479,
+      "step": 4702
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.674265800738223e-05,
+      "loss": 0.1763,
+      "step": 4704
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.6739743420061314e-05,
+      "loss": 0.1842,
+      "step": 4706
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.673682778331081e-05,
+      "loss": 0.1549,
+      "step": 4708
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.6733911097584707e-05,
+      "loss": 0.1494,
+      "step": 4710
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.673099336333715e-05,
+      "loss": 0.1831,
+      "step": 4712
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.6728074581022462e-05,
+      "loss": 0.1716,
+      "step": 4714
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.672515475109511e-05,
+      "loss": 0.1693,
+      "step": 4716
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.6722233874009745e-05,
+      "loss": 0.1735,
+      "step": 4718
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.6719311950221158e-05,
+      "loss": 0.2006,
+      "step": 4720
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.671638898018432e-05,
+      "loss": 0.1862,
+      "step": 4722
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.6713464964354363e-05,
+      "loss": 0.1965,
+      "step": 4724
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.671053990318657e-05,
+      "loss": 0.1649,
+      "step": 4726
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.670761379713641e-05,
+      "loss": 0.1991,
+      "step": 4728
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.670468664665948e-05,
+      "loss": 0.1855,
+      "step": 4730
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.670175845221158e-05,
+      "loss": 0.1344,
+      "step": 4732
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.669882921424864e-05,
+      "loss": 0.1736,
+      "step": 4734
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.6695898933226768e-05,
+      "loss": 0.163,
+      "step": 4736
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.669296760960223e-05,
+      "loss": 0.2212,
+      "step": 4738
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.6690035243831455e-05,
+      "loss": 0.1596,
+      "step": 4740
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.668710183637104e-05,
+      "loss": 0.1738,
+      "step": 4742
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.668416738767773e-05,
+      "loss": 0.1881,
+      "step": 4744
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.668123189820845e-05,
+      "loss": 0.1818,
+      "step": 4746
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.6678295368420273e-05,
+      "loss": 0.208,
+      "step": 4748
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.667535779877044e-05,
+      "loss": 0.1328,
+      "step": 4750
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.6672419189716354e-05,
+      "loss": 0.1954,
+      "step": 4752
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.6669479541715577e-05,
+      "loss": 0.2125,
+      "step": 4754
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.666653885522584e-05,
+      "loss": 0.1761,
+      "step": 4756
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.6663597130705023e-05,
+      "loss": 0.1873,
+      "step": 4758
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.6660654368611183e-05,
+      "loss": 0.1924,
+      "step": 4760
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.6657710569402522e-05,
+      "loss": 0.1692,
+      "step": 4762
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.665476573353742e-05,
+      "loss": 0.1395,
+      "step": 4764
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.6651819861474402e-05,
+      "loss": 0.1405,
+      "step": 4766
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.6648872953672172e-05,
+      "loss": 0.1608,
+      "step": 4768
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.6645925010589576e-05,
+      "loss": 0.173,
+      "step": 4770
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.664297603268564e-05,
+      "loss": 0.1847,
+      "step": 4772
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.664002602041954e-05,
+      "loss": 0.1991,
+      "step": 4774
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.6637074974250614e-05,
+      "loss": 0.1877,
+      "step": 4776
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.663412289463836e-05,
+      "loss": 0.1613,
+      "step": 4778
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.6631169782042446e-05,
+      "loss": 0.1652,
+      "step": 4780
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.662821563692269e-05,
+      "loss": 0.2009,
+      "step": 4782
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.6625260459739077e-05,
+      "loss": 0.1781,
+      "step": 4784
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.6622304250951745e-05,
+      "loss": 0.1937,
+      "step": 4786
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.661934701102101e-05,
+      "loss": 0.19,
+      "step": 4788
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.6616388740407325e-05,
+      "loss": 0.1968,
+      "step": 4790
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.661342943957132e-05,
+      "loss": 0.171,
+      "step": 4792
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.661046910897378e-05,
+      "loss": 0.1429,
+      "step": 4794
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.660750774907565e-05,
+      "loss": 0.1749,
+      "step": 4796
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.660454536033804e-05,
+      "loss": 0.1703,
+      "step": 4798
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.6601581943222217e-05,
+      "loss": 0.1988,
+      "step": 4800
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.6598617498189604e-05,
+      "loss": 0.1599,
+      "step": 4802
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.6595652025701788e-05,
+      "loss": 0.18,
+      "step": 4804
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.6592685526220516e-05,
+      "loss": 0.1812,
+      "step": 4806
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.65897180002077e-05,
+      "loss": 0.1677,
+      "step": 4808
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.65867494481254e-05,
+      "loss": 0.169,
+      "step": 4810
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.6583779870435845e-05,
+      "loss": 0.1723,
+      "step": 4812
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.658080926760142e-05,
+      "loss": 0.1933,
+      "step": 4814
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.6577837640084672e-05,
+      "loss": 0.1681,
+      "step": 4816
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.65748649883483e-05,
+      "loss": 0.2008,
+      "step": 4818
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.657189131285518e-05,
+      "loss": 0.1612,
+      "step": 4820
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.6568916614068323e-05,
+      "loss": 0.1622,
+      "step": 4822
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.6565940892450925e-05,
+      "loss": 0.1727,
+      "step": 4824
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.6562964148466318e-05,
+      "loss": 0.1719,
+      "step": 4826
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.655998638257801e-05,
+      "loss": 0.1945,
+      "step": 4828
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.6557007595249663e-05,
+      "loss": 0.1706,
+      "step": 4830
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.655402778694509e-05,
+      "loss": 0.1743,
+      "step": 4832
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.6551046958128275e-05,
+      "loss": 0.19,
+      "step": 4834
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.654806510926336e-05,
+      "loss": 0.1869,
+      "step": 4836
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.654508224081463e-05,
+      "loss": 0.1662,
+      "step": 4838
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.654209835324655e-05,
+      "loss": 0.1768,
+      "step": 4840
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.6539113447023733e-05,
+      "loss": 0.1614,
+      "step": 4842
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.6536127522610946e-05,
+      "loss": 0.169,
+      "step": 4844
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.6533140580473128e-05,
+      "loss": 0.1682,
+      "step": 4846
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.6530152621075364e-05,
+      "loss": 0.1838,
+      "step": 4848
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.6527163644882902e-05,
+      "loss": 0.1891,
+      "step": 4850
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.652417365236115e-05,
+      "loss": 0.1951,
+      "step": 4852
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.6521182643975675e-05,
+      "loss": 0.1473,
+      "step": 4854
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.651819062019219e-05,
+      "loss": 0.1976,
+      "step": 4856
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.651519758147659e-05,
+      "loss": 0.1624,
+      "step": 4858
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.6512203528294902e-05,
+      "loss": 0.1665,
+      "step": 4860
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.650920846111333e-05,
+      "loss": 0.1799,
+      "step": 4862
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.650621238039823e-05,
+      "loss": 0.181,
+      "step": 4864
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.650321528661611e-05,
+      "loss": 0.1849,
+      "step": 4866
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.650021718023364e-05,
+      "loss": 0.1286,
+      "step": 4868
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.6497218061717654e-05,
+      "loss": 0.1518,
+      "step": 4870
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.649421793153513e-05,
+      "loss": 0.1738,
+      "step": 4872
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.6491216790153214e-05,
+      "loss": 0.1603,
+      "step": 4874
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.6488214638039213e-05,
+      "loss": 0.1952,
+      "step": 4876
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.6485211475660577e-05,
+      "loss": 0.1931,
+      "step": 4878
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.6482207303484924e-05,
+      "loss": 0.1886,
+      "step": 4880
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.6479202121980024e-05,
+      "loss": 0.1882,
+      "step": 4882
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.6476195931613815e-05,
+      "loss": 0.165,
+      "step": 4884
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.6473188732854374e-05,
+      "loss": 0.1808,
+      "step": 4886
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.647018052616995e-05,
+      "loss": 0.1829,
+      "step": 4888
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.6467171312028945e-05,
+      "loss": 0.1655,
+      "step": 4890
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.6464161090899914e-05,
+      "loss": 0.1568,
+      "step": 4892
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.646114986325157e-05,
+      "loss": 0.1608,
+      "step": 4894
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.6458137629552785e-05,
+      "loss": 0.1738,
+      "step": 4896
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.645512439027259e-05,
+      "loss": 0.1875,
+      "step": 4898
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.6452110145880167e-05,
+      "loss": 0.1516,
+      "step": 4900
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.6449094896844856e-05,
+      "loss": 0.1574,
+      "step": 4902
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.644607864363616e-05,
+      "loss": 0.1701,
+      "step": 4904
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.6443061386723722e-05,
+      "loss": 0.1677,
+      "step": 4906
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.6440043126577364e-05,
+      "loss": 0.2029,
+      "step": 4908
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.6437023863667043e-05,
+      "loss": 0.162,
+      "step": 4910
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.6434003598462886e-05,
+      "loss": 0.1602,
+      "step": 4912
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.643098233143517e-05,
+      "loss": 0.1662,
+      "step": 4914
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.642796006305433e-05,
+      "loss": 0.163,
+      "step": 4916
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.6424936793790954e-05,
+      "loss": 0.1472,
+      "step": 4918
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.6421912524115794e-05,
+      "loss": 0.175,
+      "step": 4920
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.6418887254499744e-05,
+      "loss": 0.1747,
+      "step": 4922
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.6415860985413865e-05,
+      "loss": 0.15,
+      "step": 4924
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.641283371732937e-05,
+      "loss": 0.1647,
+      "step": 4926
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.6409805450717633e-05,
+      "loss": 0.1705,
+      "step": 4928
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.6406776186050168e-05,
+      "loss": 0.1574,
+      "step": 4930
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.6403745923798666e-05,
+      "loss": 0.1969,
+      "step": 4932
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.6400714664434957e-05,
+      "loss": 0.1738,
+      "step": 4934
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.6397682408431025e-05,
+      "loss": 0.1752,
+      "step": 4936
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.6394649156259027e-05,
+      "loss": 0.1621,
+      "step": 4938
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.6391614908391254e-05,
+      "loss": 0.1694,
+      "step": 4940
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.6388579665300165e-05,
+      "loss": 0.1404,
+      "step": 4942
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.6385543427458374e-05,
+      "loss": 0.1475,
+      "step": 4944
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.638250619533864e-05,
+      "loss": 0.1527,
+      "step": 4946
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.6379467969413888e-05,
+      "loss": 0.1312,
+      "step": 4948
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.637642875015719e-05,
+      "loss": 0.1693,
+      "step": 4950
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.6373388538041778e-05,
+      "loss": 0.1595,
+      "step": 4952
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.637034733354104e-05,
+      "loss": 0.1671,
+      "step": 4954
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.6367305137128507e-05,
+      "loss": 0.1916,
+      "step": 4956
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.6364261949277876e-05,
+      "loss": 0.1269,
+      "step": 4958
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.636121777046299e-05,
+      "loss": 0.211,
+      "step": 4960
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.635817260115786e-05,
+      "loss": 0.178,
+      "step": 4962
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.6355126441836637e-05,
+      "loss": 0.1628,
+      "step": 4964
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.635207929297363e-05,
+      "loss": 0.1891,
+      "step": 4966
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.6349031155043303e-05,
+      "loss": 0.1496,
+      "step": 4968
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.6345982028520284e-05,
+      "loss": 0.1683,
+      "step": 4970
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.6342931913879328e-05,
+      "loss": 0.1483,
+      "step": 4972
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.6339880811595377e-05,
+      "loss": 0.1297,
+      "step": 4974
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.63368287221435e-05,
+      "loss": 0.1338,
+      "step": 4976
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.6333775645998938e-05,
+      "loss": 0.1618,
+      "step": 4978
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.6330721583637078e-05,
+      "loss": 0.1563,
+      "step": 4980
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.632766653553346e-05,
+      "loss": 0.168,
+      "step": 4982
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.632461050216377e-05,
+      "loss": 0.1493,
+      "step": 4984
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.632155348400387e-05,
+      "loss": 0.1942,
+      "step": 4986
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.6318495481529746e-05,
+      "loss": 0.16,
+      "step": 4988
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.6315436495217567e-05,
+      "loss": 0.1347,
+      "step": 4990
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.6312376525543633e-05,
+      "loss": 0.1761,
+      "step": 4992
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.63093155729844e-05,
+      "loss": 0.1847,
+      "step": 4994
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.630625363801649e-05,
+      "loss": 0.1803,
+      "step": 4996
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.630319072111667e-05,
+      "loss": 0.1562,
+      "step": 4998
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.6300126822761853e-05,
+      "loss": 0.1568,
+      "step": 5000
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.629706194342912e-05,
+      "loss": 0.1618,
+      "step": 5002
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.6293996083595683e-05,
+      "loss": 0.1238,
+      "step": 5004
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.629092924373893e-05,
+      "loss": 0.1588,
+      "step": 5006
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.6287861424336392e-05,
+      "loss": 0.1736,
+      "step": 5008
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.628479262586575e-05,
+      "loss": 0.1463,
+      "step": 5010
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.6281722848804836e-05,
+      "loss": 0.1548,
+      "step": 5012
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.627865209363164e-05,
+      "loss": 0.1759,
+      "step": 5014
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.62755803608243e-05,
+      "loss": 0.1406,
+      "step": 5016
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.627250765086112e-05,
+      "loss": 0.1887,
+      "step": 5018
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.6269433964220525e-05,
+      "loss": 0.1939,
+      "step": 5020
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.6266359301381128e-05,
+      "loss": 0.1574,
+      "step": 5022
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.6263283662821666e-05,
+      "loss": 0.1493,
+      "step": 5024
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.626020704902105e-05,
+      "loss": 0.1571,
+      "step": 5026
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.625712946045832e-05,
+      "loss": 0.1496,
+      "step": 5028
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.6254050897612694e-05,
+      "loss": 0.1453,
+      "step": 5030
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.6250971360963515e-05,
+      "loss": 0.1798,
+      "step": 5032
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.62478908509903e-05,
+      "loss": 0.1395,
+      "step": 5034
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.62448093681727e-05,
+      "loss": 0.167,
+      "step": 5036
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.6241726912990534e-05,
+      "loss": 0.1424,
+      "step": 5038
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.623864348592376e-05,
+      "loss": 0.1638,
+      "step": 5040
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.6235559087452485e-05,
+      "loss": 0.1719,
+      "step": 5042
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.6232473718056986e-05,
+      "loss": 0.1934,
+      "step": 5044
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.6229387378217665e-05,
+      "loss": 0.186,
+      "step": 5046
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.6226300068415096e-05,
+      "loss": 0.1315,
+      "step": 5048
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.622321178913e-05,
+      "loss": 0.1764,
+      "step": 5050
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.622012254084324e-05,
+      "loss": 0.1671,
+      "step": 5052
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.6217032324035837e-05,
+      "loss": 0.1717,
+      "step": 5054
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.6213941139188963e-05,
+      "loss": 0.1749,
+      "step": 5056
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.6210848986783933e-05,
+      "loss": 0.149,
+      "step": 5058
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.620775586730223e-05,
+      "loss": 0.1757,
+      "step": 5060
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.6204661781225464e-05,
+      "loss": 0.1637,
+      "step": 5062
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.6201566729035418e-05,
+      "loss": 0.1762,
+      "step": 5064
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.6198470711214008e-05,
+      "loss": 0.14,
+      "step": 5066
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.6195373728243312e-05,
+      "loss": 0.146,
+      "step": 5068
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.6192275780605556e-05,
+      "loss": 0.1846,
+      "step": 5070
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.6189176868783105e-05,
+      "loss": 0.1927,
+      "step": 5072
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.6186076993258493e-05,
+      "loss": 0.1685,
+      "step": 5074
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.618297615451439e-05,
+      "loss": 0.1297,
+      "step": 5076
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.6179874353033618e-05,
+      "loss": 0.1731,
+      "step": 5078
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.6176771589299156e-05,
+      "loss": 0.1626,
+      "step": 5080
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.6173667863794124e-05,
+      "loss": 0.2089,
+      "step": 5082
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.61705631770018e-05,
+      "loss": 0.1399,
+      "step": 5084
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.6167457529405607e-05,
+      "loss": 0.1658,
+      "step": 5086
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.6164350921489112e-05,
+      "loss": 0.1587,
+      "step": 5088
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.6161243353736044e-05,
+      "loss": 0.1445,
+      "step": 5090
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.615813482663027e-05,
+      "loss": 0.1835,
+      "step": 5092
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.615502534065582e-05,
+      "loss": 0.1539,
+      "step": 5094
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.6151914896296856e-05,
+      "loss": 0.157,
+      "step": 5096
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.61488034940377e-05,
+      "loss": 0.1584,
+      "step": 5098
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.6145691134362823e-05,
+      "loss": 0.1792,
+      "step": 5100
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.6142577817756847e-05,
+      "loss": 0.1999,
+      "step": 5102
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.613946354470453e-05,
+      "loss": 0.1591,
+      "step": 5104
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.6136348315690794e-05,
+      "loss": 0.1248,
+      "step": 5106
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.61332321312007e-05,
+      "loss": 0.1505,
+      "step": 5108
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.6130114991719468e-05,
+      "loss": 0.1611,
+      "step": 5110
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.6126996897732456e-05,
+      "loss": 0.1611,
+      "step": 5112
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.6123877849725174e-05,
+      "loss": 0.1464,
+      "step": 5114
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.6120757848183285e-05,
+      "loss": 0.1447,
+      "step": 5116
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.6117636893592595e-05,
+      "loss": 0.1368,
+      "step": 5118
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.611451498643906e-05,
+      "loss": 0.1685,
+      "step": 5120
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.6111392127208788e-05,
+      "loss": 0.1594,
+      "step": 5122
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.610826831638803e-05,
+      "loss": 0.1472,
+      "step": 5124
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.610514355446319e-05,
+      "loss": 0.1431,
+      "step": 5126
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.610201784192081e-05,
+      "loss": 0.139,
+      "step": 5128
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.60988911792476e-05,
+      "loss": 0.1428,
+      "step": 5130
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.6095763566930388e-05,
+      "loss": 0.1593,
+      "step": 5132
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.6092635005456178e-05,
+      "loss": 0.1303,
+      "step": 5134
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.608950549531211e-05,
+      "loss": 0.1821,
+      "step": 5136
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.6086375036985477e-05,
+      "loss": 0.1179,
+      "step": 5138
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.6083243630963705e-05,
+      "loss": 0.1674,
+      "step": 5140
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.6080111277734384e-05,
+      "loss": 0.1468,
+      "step": 5142
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.6076977977785245e-05,
+      "loss": 0.1852,
+      "step": 5144
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.6073843731604172e-05,
+      "loss": 0.1528,
+      "step": 5146
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.6070708539679175e-05,
+      "loss": 0.1183,
+      "step": 5148
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.6067572402498447e-05,
+      "loss": 0.1815,
+      "step": 5150
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.6064435320550292e-05,
+      "loss": 0.1327,
+      "step": 5152
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.6061297294323187e-05,
+      "loss": 0.1351,
+      "step": 5154
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.6058158324305747e-05,
+      "loss": 0.1657,
+      "step": 5156
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.6055018410986732e-05,
+      "loss": 0.1488,
+      "step": 5158
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.605187755485505e-05,
+      "loss": 0.1529,
+      "step": 5160
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.6048735756399755e-05,
+      "loss": 0.1373,
+      "step": 5162
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.6045593016110047e-05,
+      "loss": 0.1778,
+      "step": 5164
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.604244933447528e-05,
+      "loss": 0.1656,
+      "step": 5166
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.6039304711984947e-05,
+      "loss": 0.1701,
+      "step": 5168
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.603615914912869e-05,
+      "loss": 0.1554,
+      "step": 5170
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.6033012646396297e-05,
+      "loss": 0.1599,
+      "step": 5172
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.6029865204277704e-05,
+      "loss": 0.1491,
+      "step": 5174
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.6026716823262988e-05,
+      "loss": 0.1581,
+      "step": 5176
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.602356750384238e-05,
+      "loss": 0.1128,
+      "step": 5178
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.6020417246506252e-05,
+      "loss": 0.1509,
+      "step": 5180
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.601726605174512e-05,
+      "loss": 0.1501,
+      "step": 5182
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.6014113920049656e-05,
+      "loss": 0.1542,
+      "step": 5184
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.6010960851910665e-05,
+      "loss": 0.1658,
+      "step": 5186
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.60078068478191e-05,
+      "loss": 0.1627,
+      "step": 5188
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.600465190826608e-05,
+      "loss": 0.1456,
+      "step": 5190
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.6001496033742833e-05,
+      "loss": 0.1601,
+      "step": 5192
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.5998339224740767e-05,
+      "loss": 0.1657,
+      "step": 5194
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.5995181481751415e-05,
+      "loss": 0.1278,
+      "step": 5196
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.5992022805266466e-05,
+      "loss": 0.1563,
+      "step": 5198
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.5988863195777745e-05,
+      "loss": 0.1191,
+      "step": 5200
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.5985702653777228e-05,
+      "loss": 0.1368,
+      "step": 5202
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.5982541179757038e-05,
+      "loss": 0.1869,
+      "step": 5204
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.5979378774209443e-05,
+      "loss": 0.149,
+      "step": 5206
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.5976215437626845e-05,
+      "loss": 0.166,
+      "step": 5208
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.5973051170501814e-05,
+      "loss": 0.1343,
+      "step": 5210
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.5969885973327034e-05,
+      "loss": 0.1548,
+      "step": 5212
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.596671984659536e-05,
+      "loss": 0.1496,
+      "step": 5214
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.596355279079978e-05,
+      "loss": 0.2045,
+      "step": 5216
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.596038480643343e-05,
+      "loss": 0.1596,
+      "step": 5218
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.595721589398959e-05,
+      "loss": 0.1504,
+      "step": 5220
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.5954046053961683e-05,
+      "loss": 0.1384,
+      "step": 5222
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.5950875286843274e-05,
+      "loss": 0.1435,
+      "step": 5224
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.5947703593128084e-05,
+      "loss": 0.1179,
+      "step": 5226
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.594453097330996e-05,
+      "loss": 0.1327,
+      "step": 5228
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.594135742788291e-05,
+      "loss": 0.1396,
+      "step": 5230
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.5938182957341072e-05,
+      "loss": 0.1433,
+      "step": 5232
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.5935007562178747e-05,
+      "loss": 0.1705,
+      "step": 5234
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.5931831242890357e-05,
+      "loss": 0.1546,
+      "step": 5236
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.5928653999970486e-05,
+      "loss": 0.141,
+      "step": 5238
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.592547583391385e-05,
+      "loss": 0.1632,
+      "step": 5240
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.592229674521532e-05,
+      "loss": 0.1296,
+      "step": 5242
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.59191167343699e-05,
+      "loss": 0.1326,
+      "step": 5244
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.5915935801872742e-05,
+      "loss": 0.1522,
+      "step": 5246
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.5912753948219143e-05,
+      "loss": 0.159,
+      "step": 5248
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.5909571173904543e-05,
+      "loss": 0.1287,
+      "step": 5250
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.5906387479424523e-05,
+      "loss": 0.1781,
+      "step": 5252
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.5903202865274807e-05,
+      "loss": 0.1649,
+      "step": 5254
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.5900017331951267e-05,
+      "loss": 0.129,
+      "step": 5256
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.589683087994991e-05,
+      "loss": 0.1417,
+      "step": 5258
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.5893643509766895e-05,
+      "loss": 0.141,
+      "step": 5260
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.589045522189852e-05,
+      "loss": 0.1519,
+      "step": 5262
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.588726601684122e-05,
+      "loss": 0.1618,
+      "step": 5264
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.588407589509159e-05,
+      "loss": 0.168,
+      "step": 5266
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.5880884857146343e-05,
+      "loss": 0.1387,
+      "step": 5268
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.587769290350236e-05,
+      "loss": 0.1507,
+      "step": 5270
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.5874500034656646e-05,
+      "loss": 0.1274,
+      "step": 5272
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.5871306251106354e-05,
+      "loss": 0.168,
+      "step": 5274
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.5868111553348787e-05,
+      "loss": 0.1014,
+      "step": 5276
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.5864915941881378e-05,
+      "loss": 0.1572,
+      "step": 5278
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.5861719417201707e-05,
+      "loss": 0.1601,
+      "step": 5280
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.5858521979807508e-05,
+      "loss": 0.1349,
+      "step": 5282
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.585532363019663e-05,
+      "loss": 0.1456,
+      "step": 5284
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.5852124368867093e-05,
+      "loss": 0.1686,
+      "step": 5286
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.5848924196317042e-05,
+      "loss": 0.1299,
+      "step": 5288
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.5845723113044773e-05,
+      "loss": 0.1532,
+      "step": 5290
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.5842521119548706e-05,
+      "loss": 0.16,
+      "step": 5292
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.583931821632743e-05,
+      "loss": 0.1419,
+      "step": 5294
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.5836114403879655e-05,
+      "loss": 0.1427,
+      "step": 5296
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.5832909682704246e-05,
+      "loss": 0.1639,
+      "step": 5298
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.582970405330019e-05,
+      "loss": 0.1237,
+      "step": 5300
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.5826497516166642e-05,
+      "loss": 0.1219,
+      "step": 5302
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.5823290071802872e-05,
+      "loss": 0.1622,
+      "step": 5304
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.5820081720708313e-05,
+      "loss": 0.1287,
+      "step": 5306
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.581687246338252e-05,
+      "loss": 0.1344,
+      "step": 5308
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.581366230032521e-05,
+      "loss": 0.1493,
+      "step": 5310
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.5810451232036226e-05,
+      "loss": 0.1333,
+      "step": 5312
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.5807239259015553e-05,
+      "loss": 0.1301,
+      "step": 5314
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.5804026381763324e-05,
+      "loss": 0.1138,
+      "step": 5316
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.58008126007798e-05,
+      "loss": 0.112,
+      "step": 5318
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.5797597916565402e-05,
+      "loss": 0.1443,
+      "step": 5320
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.5794382329620672e-05,
+      "loss": 0.1454,
+      "step": 5322
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.579116584044631e-05,
+      "loss": 0.1356,
+      "step": 5324
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.578794844954314e-05,
+      "loss": 0.119,
+      "step": 5326
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.578473015741214e-05,
+      "loss": 0.1512,
+      "step": 5328
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.578151096455442e-05,
+      "loss": 0.132,
+      "step": 5330
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.5778290871471234e-05,
+      "loss": 0.146,
+      "step": 5332
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.577506987866398e-05,
+      "loss": 0.1439,
+      "step": 5334
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.5771847986634178e-05,
+      "loss": 0.1358,
+      "step": 5336
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.576862519588351e-05,
+      "loss": 0.1098,
+      "step": 5338
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.576540150691379e-05,
+      "loss": 0.123,
+      "step": 5340
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.576217692022697e-05,
+      "loss": 0.1648,
+      "step": 5342
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.5758951436325142e-05,
+      "loss": 0.1709,
+      "step": 5344
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.5755725055710537e-05,
+      "loss": 0.149,
+      "step": 5346
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.5752497778885532e-05,
+      "loss": 0.1494,
+      "step": 5348
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.5749269606352632e-05,
+      "loss": 0.1305,
+      "step": 5350
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.5746040538614493e-05,
+      "loss": 0.1163,
+      "step": 5352
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.5742810576173907e-05,
+      "loss": 0.1354,
+      "step": 5354
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.5739579719533797e-05,
+      "loss": 0.1426,
+      "step": 5356
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.5736347969197235e-05,
+      "loss": 0.1567,
+      "step": 5358
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.5733115325667435e-05,
+      "loss": 0.1307,
+      "step": 5360
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.572988178944774e-05,
+      "loss": 0.1419,
+      "step": 5362
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.5726647361041632e-05,
+      "loss": 0.1363,
+      "step": 5364
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.572341204095274e-05,
+      "loss": 0.1426,
+      "step": 5366
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.572017582968483e-05,
+      "loss": 0.1329,
+      "step": 5368
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.5716938727741803e-05,
+      "loss": 0.1406,
+      "step": 5370
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.5713700735627703e-05,
+      "loss": 0.1342,
+      "step": 5372
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.5710461853846702e-05,
+      "loss": 0.172,
+      "step": 5374
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.570722208290313e-05,
+      "loss": 0.1423,
+      "step": 5376
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.5703981423301438e-05,
+      "loss": 0.1524,
+      "step": 5378
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.570073987554622e-05,
+      "loss": 0.1668,
+      "step": 5380
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.5697497440142214e-05,
+      "loss": 0.1692,
+      "step": 5382
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.5694254117594286e-05,
+      "loss": 0.1663,
+      "step": 5384
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.5691009908407454e-05,
+      "loss": 0.11,
+      "step": 5386
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.5687764813086856e-05,
+      "loss": 0.1256,
+      "step": 5388
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.5684518832137793e-05,
+      "loss": 0.0977,
+      "step": 5390
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.5681271966065673e-05,
+      "loss": 0.1296,
+      "step": 5392
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.5678024215376067e-05,
+      "loss": 0.1265,
+      "step": 5394
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.5674775580574674e-05,
+      "loss": 0.1173,
+      "step": 5396
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.567152606216733e-05,
+      "loss": 0.1524,
+      "step": 5398
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.5668275660660012e-05,
+      "loss": 0.1154,
+      "step": 5400
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.5665024376558826e-05,
+      "loss": 0.1707,
+      "step": 5402
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.5661772210370028e-05,
+      "loss": 0.1331,
+      "step": 5404
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.5658519162600003e-05,
+      "loss": 0.1561,
+      "step": 5406
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.5655265233755276e-05,
+      "loss": 0.1632,
+      "step": 5408
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.565201042434251e-05,
+      "loss": 0.1741,
+      "step": 5410
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.5648754734868498e-05,
+      "loss": 0.1444,
+      "step": 5412
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.5645498165840183e-05,
+      "loss": 0.1654,
+      "step": 5414
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.5642240717764634e-05,
+      "loss": 0.1294,
+      "step": 5416
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.563898239114906e-05,
+      "loss": 0.1418,
+      "step": 5418
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.563572318650081e-05,
+      "loss": 0.158,
+      "step": 5420
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.5632463104327363e-05,
+      "loss": 0.1268,
+      "step": 5422
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.5629202145136343e-05,
+      "loss": 0.1379,
+      "step": 5424
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.5625940309435503e-05,
+      "loss": 0.1185,
+      "step": 5426
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.5622677597732735e-05,
+      "loss": 0.1172,
+      "step": 5428
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.5619414010536077e-05,
+      "loss": 0.1336,
+      "step": 5430
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.5616149548353682e-05,
+      "loss": 0.1179,
+      "step": 5432
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.5612884211693863e-05,
+      "loss": 0.1264,
+      "step": 5434
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.560961800106505e-05,
+      "loss": 0.1161,
+      "step": 5436
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.5606350916975815e-05,
+      "loss": 0.1395,
+      "step": 5438
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.5603082959934877e-05,
+      "loss": 0.1206,
+      "step": 5440
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.5599814130451078e-05,
+      "loss": 0.1248,
+      "step": 5442
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.5596544429033395e-05,
+      "loss": 0.1335,
+      "step": 5444
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.5593273856190957e-05,
+      "loss": 0.1212,
+      "step": 5446
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.5590002412433004e-05,
+      "loss": 0.145,
+      "step": 5448
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.558673009826893e-05,
+      "loss": 0.1463,
+      "step": 5450
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.558345691420826e-05,
+      "loss": 0.15,
+      "step": 5452
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.558018286076066e-05,
+      "loss": 0.11,
+      "step": 5454
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.557690793843591e-05,
+      "loss": 0.176,
+      "step": 5456
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.5573632147743953e-05,
+      "loss": 0.1572,
+      "step": 5458
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.5570355489194852e-05,
+      "loss": 0.1118,
+      "step": 5460
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.5567077963298806e-05,
+      "loss": 0.1183,
+      "step": 5462
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.5563799570566152e-05,
+      "loss": 0.1191,
+      "step": 5464
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.556052031150736e-05,
+      "loss": 0.1051,
+      "step": 5466
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.5557240186633037e-05,
+      "loss": 0.1357,
+      "step": 5468
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.5553959196453922e-05,
+      "loss": 0.1501,
+      "step": 5470
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.555067734148089e-05,
+      "loss": 0.1298,
+      "step": 5472
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.554739462222495e-05,
+      "loss": 0.1183,
+      "step": 5474
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.5544111039197255e-05,
+      "loss": 0.1113,
+      "step": 5476
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.554082659290907e-05,
+      "loss": 0.1249,
+      "step": 5478
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.553754128387182e-05,
+      "loss": 0.1607,
+      "step": 5480
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.5534255112597047e-05,
+      "loss": 0.1129,
+      "step": 5482
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.5530968079596438e-05,
+      "loss": 0.1558,
+      "step": 5484
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.5527680185381803e-05,
+      "loss": 0.1407,
+      "step": 5486
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.5524391430465094e-05,
+      "loss": 0.1204,
+      "step": 5488
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.5521101815358396e-05,
+      "loss": 0.1346,
+      "step": 5490
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.5517811340573926e-05,
+      "loss": 0.1014,
+      "step": 5492
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.5514520006624038e-05,
+      "loss": 0.1221,
+      "step": 5494
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.5511227814021223e-05,
+      "loss": 0.1233,
+      "step": 5496
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.550793476327809e-05,
+      "loss": 0.1165,
+      "step": 5498
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.55046408549074e-05,
+      "loss": 0.1417,
+      "step": 5500
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.550134608942203e-05,
+      "loss": 0.1238,
+      "step": 5502
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.5498050467335012e-05,
+      "loss": 0.1351,
+      "step": 5504
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.5494753989159493e-05,
+      "loss": 0.1548,
+      "step": 5506
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.5491456655408767e-05,
+      "loss": 0.1529,
+      "step": 5508
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.548815846659624e-05,
+      "loss": 0.1242,
+      "step": 5510
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.548485942323548e-05,
+      "loss": 0.1164,
+      "step": 5512
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.5481559525840167e-05,
+      "loss": 0.1456,
+      "step": 5514
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.5478258774924117e-05,
+      "loss": 0.1564,
+      "step": 5516
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.5474957171001287e-05,
+      "loss": 0.14,
+      "step": 5518
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.547165471458576e-05,
+      "loss": 0.1094,
+      "step": 5520
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.546835140619175e-05,
+      "loss": 0.1484,
+      "step": 5522
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.546504724633362e-05,
+      "loss": 0.1461,
+      "step": 5524
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.546174223552584e-05,
+      "loss": 0.1358,
+      "step": 5526
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.545843637428303e-05,
+      "loss": 0.1604,
+      "step": 5528
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.5455129663119936e-05,
+      "loss": 0.139,
+      "step": 5530
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.545182210255144e-05,
+      "loss": 0.1456,
+      "step": 5532
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.5448513693092558e-05,
+      "loss": 0.1082,
+      "step": 5534
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.5445204435258427e-05,
+      "loss": 0.1419,
+      "step": 5536
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.544189432956433e-05,
+      "loss": 0.1547,
+      "step": 5538
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.5438583376525676e-05,
+      "loss": 0.1146,
+      "step": 5540
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.5435271576658e-05,
+      "loss": 0.1392,
+      "step": 5542
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.543195893047698e-05,
+      "loss": 0.1439,
+      "step": 5544
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.5428645438498416e-05,
+      "loss": 0.1478,
+      "step": 5546
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.5425331101238246e-05,
+      "loss": 0.1426,
+      "step": 5548
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.5422015919212538e-05,
+      "loss": 0.1713,
+      "step": 5550
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.5418699892937494e-05,
+      "loss": 0.1137,
+      "step": 5552
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.541538302292944e-05,
+      "loss": 0.1055,
+      "step": 5554
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.541206530970484e-05,
+      "loss": 0.1285,
+      "step": 5556
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.5408746753780288e-05,
+      "loss": 0.1351,
+      "step": 5558
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.5405427355672506e-05,
+      "loss": 0.1664,
+      "step": 5560
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.540210711589835e-05,
+      "loss": 0.1539,
+      "step": 5562
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.539878603497481e-05,
+      "loss": 0.1294,
+      "step": 5564
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.5395464113419003e-05,
+      "loss": 0.1472,
+      "step": 5566
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.5392141351748175e-05,
+      "loss": 0.1132,
+      "step": 5568
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.5388817750479706e-05,
+      "loss": 0.1146,
+      "step": 5570
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.5385493310131106e-05,
+      "loss": 0.1115,
+      "step": 5572
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.538216803122002e-05,
+      "loss": 0.1531,
+      "step": 5574
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.5378841914264217e-05,
+      "loss": 0.1595,
+      "step": 5576
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.537551495978159e-05,
+      "loss": 0.0974,
+      "step": 5578
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.5372187168290186e-05,
+      "loss": 0.1615,
+      "step": 5580
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.5368858540308158e-05,
+      "loss": 0.1167,
+      "step": 5582
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.5365529076353803e-05,
+      "loss": 0.1464,
+      "step": 5584
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.536219877694554e-05,
+      "loss": 0.1424,
+      "step": 5586
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.5358867642601928e-05,
+      "loss": 0.1117,
+      "step": 5588
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.5355535673841643e-05,
+      "loss": 0.1558,
+      "step": 5590
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.5352202871183504e-05,
+      "loss": 0.1282,
+      "step": 5592
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.534886923514645e-05,
+      "loss": 0.1326,
+      "step": 5594
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.534553476624956e-05,
+      "loss": 0.1486,
+      "step": 5596
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.5342199465012026e-05,
+      "loss": 0.1142,
+      "step": 5598
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.5338863331953183e-05,
+      "loss": 0.117,
+      "step": 5600
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.5335526367592503e-05,
+      "loss": 0.1362,
+      "step": 5602
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.533218857244956e-05,
+      "loss": 0.1052,
+      "step": 5604
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.5328849947044088e-05,
+      "loss": 0.0941,
+      "step": 5606
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.5325510491895934e-05,
+      "loss": 0.1076,
+      "step": 5608
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.532217020752507e-05,
+      "loss": 0.1427,
+      "step": 5610
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.531882909445161e-05,
+      "loss": 0.1112,
+      "step": 5612
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.531548715319579e-05,
+      "loss": 0.1277,
+      "step": 5614
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.5312144384277966e-05,
+      "loss": 0.0868,
+      "step": 5616
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.530880078821865e-05,
+      "loss": 0.1269,
+      "step": 5618
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.5305456365538455e-05,
+      "loss": 0.1594,
+      "step": 5620
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.5302111116758132e-05,
+      "loss": 0.1352,
+      "step": 5622
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.5298765042398567e-05,
+      "loss": 0.1203,
+      "step": 5624
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.5295418142980766e-05,
+      "loss": 0.1415,
+      "step": 5626
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.5292070419025868e-05,
+      "loss": 0.1195,
+      "step": 5628
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.528872187105514e-05,
+      "loss": 0.1074,
+      "step": 5630
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.5285372499589978e-05,
+      "loss": 0.1317,
+      "step": 5632
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.5282022305151897e-05,
+      "loss": 0.1093,
+      "step": 5634
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.5278671288262558e-05,
+      "loss": 0.1401,
+      "step": 5636
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.5275319449443733e-05,
+      "loss": 0.1387,
+      "step": 5638
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.527196678921733e-05,
+      "loss": 0.0884,
+      "step": 5640
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.526861330810539e-05,
+      "loss": 0.1308,
+      "step": 5642
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.5265259006630065e-05,
+      "loss": 0.0974,
+      "step": 5644
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.526190388531365e-05,
+      "loss": 0.1652,
+      "step": 5646
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.525854794467857e-05,
+      "loss": 0.1436,
+      "step": 5648
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.5255191185247362e-05,
+      "loss": 0.1153,
+      "step": 5650
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.5251833607542703e-05,
+      "loss": 0.1627,
+      "step": 5652
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.524847521208739e-05,
+      "loss": 0.1208,
+      "step": 5654
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.5245115999404356e-05,
+      "loss": 0.1362,
+      "step": 5656
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.5241755970016651e-05,
+      "loss": 0.1521,
+      "step": 5658
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.523839512444746e-05,
+      "loss": 0.119,
+      "step": 5660
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.5235033463220092e-05,
+      "loss": 0.1285,
+      "step": 5662
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.5231670986857987e-05,
+      "loss": 0.0939,
+      "step": 5664
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.5228307695884699e-05,
+      "loss": 0.1077,
+      "step": 5666
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.5224943590823926e-05,
+      "loss": 0.1158,
+      "step": 5668
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.5221578672199484e-05,
+      "loss": 0.1278,
+      "step": 5670
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.5218212940535312e-05,
+      "loss": 0.1345,
+      "step": 5672
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.5214846396355484e-05,
+      "loss": 0.1321,
+      "step": 5674
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.5211479040184198e-05,
+      "loss": 0.1213,
+      "step": 5676
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.5208110872545776e-05,
+      "loss": 0.1339,
+      "step": 5678
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.5204741893964663e-05,
+      "loss": 0.1485,
+      "step": 5680
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.5201372104965439e-05,
+      "loss": 0.1245,
+      "step": 5682
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.5198001506072806e-05,
+      "loss": 0.1426,
+      "step": 5684
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.5194630097811591e-05,
+      "loss": 0.1319,
+      "step": 5686
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.519125788070675e-05,
+      "loss": 0.1286,
+      "step": 5688
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.5187884855283358e-05,
+      "loss": 0.1152,
+      "step": 5690
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.518451102206663e-05,
+      "loss": 0.1348,
+      "step": 5692
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.5181136381581884e-05,
+      "loss": 0.1375,
+      "step": 5694
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.5177760934354591e-05,
+      "loss": 0.1217,
+      "step": 5696
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.5174384680910323e-05,
+      "loss": 0.1631,
+      "step": 5698
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.5171007621774798e-05,
+      "loss": 0.127,
+      "step": 5700
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.5167629757473843e-05,
+      "loss": 0.1128,
+      "step": 5702
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.516425108853342e-05,
+      "loss": 0.0953,
+      "step": 5704
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.5160871615479613e-05,
+      "loss": 0.1366,
+      "step": 5706
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.5157491338838633e-05,
+      "loss": 0.1395,
+      "step": 5708
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.5154110259136813e-05,
+      "loss": 0.1358,
+      "step": 5710
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.5150728376900614e-05,
+      "loss": 0.1308,
+      "step": 5712
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.5147345692656622e-05,
+      "loss": 0.1258,
+      "step": 5714
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.5143962206931545e-05,
+      "loss": 0.1314,
+      "step": 5716
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.5140577920252217e-05,
+      "loss": 0.1048,
+      "step": 5718
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.5137192833145604e-05,
+      "loss": 0.1118,
+      "step": 5720
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.5133806946138779e-05,
+      "loss": 0.0972,
+      "step": 5722
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.5130420259758958e-05,
+      "loss": 0.1365,
+      "step": 5724
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.5127032774533468e-05,
+      "loss": 0.1101,
+      "step": 5726
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.5123644490989777e-05,
+      "loss": 0.1262,
+      "step": 5728
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.5120255409655456e-05,
+      "loss": 0.1442,
+      "step": 5730
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.5116865531058216e-05,
+      "loss": 0.1052,
+      "step": 5732
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.5113474855725886e-05,
+      "loss": 0.1113,
+      "step": 5734
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.511008338418642e-05,
+      "loss": 0.1176,
+      "step": 5736
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.5106691116967895e-05,
+      "loss": 0.1197,
+      "step": 5738
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.5103298054598515e-05,
+      "loss": 0.1444,
+      "step": 5740
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.5099904197606605e-05,
+      "loss": 0.0986,
+      "step": 5742
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.5096509546520614e-05,
+      "loss": 0.1298,
+      "step": 5744
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.5093114101869115e-05,
+      "loss": 0.1464,
+      "step": 5746
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.5089717864180804e-05,
+      "loss": 0.0886,
+      "step": 5748
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.5086320833984504e-05,
+      "loss": 0.1282,
+      "step": 5750
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.508292301180916e-05,
+      "loss": 0.1222,
+      "step": 5752
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.5079524398183826e-05,
+      "loss": 0.1261,
+      "step": 5754
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.5076124993637713e-05,
+      "loss": 0.126,
+      "step": 5756
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.5072724798700119e-05,
+      "loss": 0.078,
+      "step": 5758
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.506932381390048e-05,
+      "loss": 0.1408,
+      "step": 5760
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.5065922039768364e-05,
+      "loss": 0.1541,
+      "step": 5762
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.5062519476833451e-05,
+      "loss": 0.1278,
+      "step": 5764
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.5059116125625538e-05,
+      "loss": 0.1084,
+      "step": 5766
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.5055711986674564e-05,
+      "loss": 0.0804,
+      "step": 5768
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.5052307060510574e-05,
+      "loss": 0.0902,
+      "step": 5770
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.504890134766374e-05,
+      "loss": 0.102,
+      "step": 5772
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.5045494848664359e-05,
+      "loss": 0.1267,
+      "step": 5774
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.5042087564042849e-05,
+      "loss": 0.1299,
+      "step": 5776
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.503867949432975e-05,
+      "loss": 0.1404,
+      "step": 5778
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.5035270640055726e-05,
+      "loss": 0.1019,
+      "step": 5780
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.5031861001751558e-05,
+      "loss": 0.1227,
+      "step": 5782
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.502845057994816e-05,
+      "loss": 0.1436,
+      "step": 5784
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.5025039375176552e-05,
+      "loss": 0.1125,
+      "step": 5786
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.5021627387967889e-05,
+      "loss": 0.1081,
+      "step": 5788
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.5018214618853443e-05,
+      "loss": 0.0876,
+      "step": 5790
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.5014801068364609e-05,
+      "loss": 0.1656,
+      "step": 5792
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.50113867370329e-05,
+      "loss": 0.1332,
+      "step": 5794
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.5007971625389956e-05,
+      "loss": 0.144,
+      "step": 5796
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.500455573396754e-05,
+      "loss": 0.0871,
+      "step": 5798
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.5001139063297525e-05,
+      "loss": 0.1082,
+      "step": 5800
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.4997721613911916e-05,
+      "loss": 0.101,
+      "step": 5802
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.4994303386342837e-05,
+      "loss": 0.1037,
+      "step": 5804
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.4990884381122527e-05,
+      "loss": 0.1269,
+      "step": 5806
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.498746459878336e-05,
+      "loss": 0.1443,
+      "step": 5808
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.4984044039857814e-05,
+      "loss": 0.1405,
+      "step": 5810
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.4980622704878505e-05,
+      "loss": 0.1277,
+      "step": 5812
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.4977200594378152e-05,
+      "loss": 0.1255,
+      "step": 5814
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.4973777708889608e-05,
+      "loss": 0.1199,
+      "step": 5816
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.497035404894584e-05,
+      "loss": 0.1337,
+      "step": 5818
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.4966929615079946e-05,
+      "loss": 0.1169,
+      "step": 5820
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.4963504407825125e-05,
+      "loss": 0.1119,
+      "step": 5822
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.4960078427714715e-05,
+      "loss": 0.1142,
+      "step": 5824
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.4956651675282165e-05,
+      "loss": 0.0938,
+      "step": 5826
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.495322415106105e-05,
+      "loss": 0.1182,
+      "step": 5828
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.4949795855585056e-05,
+      "loss": 0.1009,
+      "step": 5830
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.4946366789387998e-05,
+      "loss": 0.1079,
+      "step": 5832
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.494293695300381e-05,
+      "loss": 0.1111,
+      "step": 5834
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.4939506346966541e-05,
+      "loss": 0.1,
+      "step": 5836
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.4936074971810361e-05,
+      "loss": 0.0966,
+      "step": 5838
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.4932642828069568e-05,
+      "loss": 0.1247,
+      "step": 5840
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.4929209916278564e-05,
+      "loss": 0.108,
+      "step": 5842
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.4925776236971885e-05,
+      "loss": 0.0985,
+      "step": 5844
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.4922341790684182e-05,
+      "loss": 0.1109,
+      "step": 5846
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.4918906577950223e-05,
+      "loss": 0.1093,
+      "step": 5848
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.4915470599304893e-05,
+      "loss": 0.1403,
+      "step": 5850
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.4912033855283205e-05,
+      "loss": 0.1137,
+      "step": 5852
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.4908596346420288e-05,
+      "loss": 0.1273,
+      "step": 5854
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.4905158073251385e-05,
+      "loss": 0.1217,
+      "step": 5856
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.4901719036311857e-05,
+      "loss": 0.1224,
+      "step": 5858
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.4898279236137199e-05,
+      "loss": 0.1389,
+      "step": 5860
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.4894838673263006e-05,
+      "loss": 0.1226,
+      "step": 5862
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.4891397348225004e-05,
+      "loss": 0.1578,
+      "step": 5864
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.4887955261559028e-05,
+      "loss": 0.1173,
+      "step": 5866
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.4884512413801049e-05,
+      "loss": 0.0972,
+      "step": 5868
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.488106880548713e-05,
+      "loss": 0.1213,
+      "step": 5870
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.4877624437153477e-05,
+      "loss": 0.1422,
+      "step": 5872
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.4874179309336398e-05,
+      "loss": 0.1236,
+      "step": 5874
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.4870733422572337e-05,
+      "loss": 0.1105,
+      "step": 5876
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.4867286777397829e-05,
+      "loss": 0.0917,
+      "step": 5878
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.4863839374349555e-05,
+      "loss": 0.1134,
+      "step": 5880
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.48603912139643e-05,
+      "loss": 0.1181,
+      "step": 5882
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.4856942296778967e-05,
+      "loss": 0.1018,
+      "step": 5884
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.4853492623330574e-05,
+      "loss": 0.1018,
+      "step": 5886
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.485004219415627e-05,
+      "loss": 0.1067,
+      "step": 5888
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.4846591009793313e-05,
+      "loss": 0.1152,
+      "step": 5890
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.4843139070779073e-05,
+      "loss": 0.0953,
+      "step": 5892
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.4839686377651044e-05,
+      "loss": 0.1135,
+      "step": 5894
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.4836232930946844e-05,
+      "loss": 0.0802,
+      "step": 5896
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.483277873120419e-05,
+      "loss": 0.1216,
+      "step": 5898
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.4829323778960933e-05,
+      "loss": 0.1143,
+      "step": 5900
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.4825868074755037e-05,
+      "loss": 0.1223,
+      "step": 5902
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.4822411619124581e-05,
+      "loss": 0.1322,
+      "step": 5904
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.481895441260776e-05,
+      "loss": 0.0934,
+      "step": 5906
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.4815496455742889e-05,
+      "loss": 0.106,
+      "step": 5908
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.4812037749068396e-05,
+      "loss": 0.0994,
+      "step": 5910
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.480857829312283e-05,
+      "loss": 0.1217,
+      "step": 5912
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.4805118088444852e-05,
+      "loss": 0.1033,
+      "step": 5914
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.480165713557325e-05,
+      "loss": 0.1132,
+      "step": 5916
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.4798195435046911e-05,
+      "loss": 0.1144,
+      "step": 5918
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.4794732987404853e-05,
+      "loss": 0.1049,
+      "step": 5920
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.4791269793186204e-05,
+      "loss": 0.0806,
+      "step": 5922
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.4787805852930216e-05,
+      "loss": 0.132,
+      "step": 5924
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.4784341167176242e-05,
+      "loss": 0.1218,
+      "step": 5926
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.4780875736463764e-05,
+      "loss": 0.1071,
+      "step": 5928
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.4777409561332376e-05,
+      "loss": 0.111,
+      "step": 5930
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.4773942642321792e-05,
+      "loss": 0.1218,
+      "step": 5932
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.477047497997183e-05,
+      "loss": 0.1157,
+      "step": 5934
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.4767006574822437e-05,
+      "loss": 0.0967,
+      "step": 5936
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.4763537427413669e-05,
+      "loss": 0.1028,
+      "step": 5938
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.4760067538285698e-05,
+      "loss": 0.1493,
+      "step": 5940
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.475659690797881e-05,
+      "loss": 0.1018,
+      "step": 5942
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.4753125537033415e-05,
+      "loss": 0.1384,
+      "step": 5944
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.4749653425990027e-05,
+      "loss": 0.0771,
+      "step": 5946
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.4746180575389284e-05,
+      "loss": 0.1299,
+      "step": 5948
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.4742706985771928e-05,
+      "loss": 0.1114,
+      "step": 5950
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.4739232657678832e-05,
+      "loss": 0.0903,
+      "step": 5952
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.4735757591650972e-05,
+      "loss": 0.1031,
+      "step": 5954
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.4732281788229443e-05,
+      "loss": 0.1015,
+      "step": 5956
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.4728805247955448e-05,
+      "loss": 0.1063,
+      "step": 5958
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.4725327971370325e-05,
+      "loss": 0.1123,
+      "step": 5960
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.4721849959015498e-05,
+      "loss": 0.0838,
+      "step": 5962
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.4718371211432527e-05,
+      "loss": 0.1304,
+      "step": 5964
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.471489172916308e-05,
+      "loss": 0.0669,
+      "step": 5966
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.4711411512748936e-05,
+      "loss": 0.1024,
+      "step": 5968
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.4707930562731994e-05,
+      "loss": 0.132,
+      "step": 5970
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.4704448879654264e-05,
+      "loss": 0.1174,
+      "step": 5972
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.4700966464057868e-05,
+      "loss": 0.0985,
+      "step": 5974
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.4697483316485048e-05,
+      "loss": 0.1103,
+      "step": 5976
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.4693999437478158e-05,
+      "loss": 0.1277,
+      "step": 5978
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.4690514827579658e-05,
+      "loss": 0.1202,
+      "step": 5980
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.4687029487332137e-05,
+      "loss": 0.1218,
+      "step": 5982
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.4683543417278281e-05,
+      "loss": 0.1117,
+      "step": 5984
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.4680056617960903e-05,
+      "loss": 0.1059,
+      "step": 5986
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.4676569089922923e-05,
+      "loss": 0.1184,
+      "step": 5988
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.4673080833707376e-05,
+      "loss": 0.093,
+      "step": 5990
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.4669591849857407e-05,
+      "loss": 0.1105,
+      "step": 5992
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.4666102138916282e-05,
+      "loss": 0.1102,
+      "step": 5994
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.4662611701427376e-05,
+      "loss": 0.12,
+      "step": 5996
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.4659120537934173e-05,
+      "loss": 0.0972,
+      "step": 5998
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.4655628648980273e-05,
+      "loss": 0.1231,
+      "step": 6000
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.4652136035109394e-05,
+      "loss": 0.1563,
+      "step": 6002
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.4648642696865363e-05,
+      "loss": 0.088,
+      "step": 6004
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.4645148634792116e-05,
+      "loss": 0.1007,
+      "step": 6006
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.4641653849433709e-05,
+      "loss": 0.0866,
+      "step": 6008
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.4638158341334303e-05,
+      "loss": 0.0987,
+      "step": 6010
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.4634662111038176e-05,
+      "loss": 0.1255,
+      "step": 6012
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.463116515908972e-05,
+      "loss": 0.0836,
+      "step": 6014
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.4627667486033435e-05,
+      "loss": 0.1185,
+      "step": 6016
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.4624169092413935e-05,
+      "loss": 0.1086,
+      "step": 6018
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.462066997877595e-05,
+      "loss": 0.1047,
+      "step": 6020
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.4617170145664314e-05,
+      "loss": 0.0858,
+      "step": 6022
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.4613669593623985e-05,
+      "loss": 0.112,
+      "step": 6024
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.4610168323200017e-05,
+      "loss": 0.1033,
+      "step": 6026
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.4606666334937589e-05,
+      "loss": 0.1323,
+      "step": 6028
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.4603163629381988e-05,
+      "loss": 0.1033,
+      "step": 6030
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.4599660207078614e-05,
+      "loss": 0.1045,
+      "step": 6032
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.459615606857297e-05,
+      "loss": 0.1361,
+      "step": 6034
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.4592651214410683e-05,
+      "loss": 0.0967,
+      "step": 6036
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.4589145645137483e-05,
+      "loss": 0.1178,
+      "step": 6038
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.4585639361299213e-05,
+      "loss": 0.1041,
+      "step": 6040
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.4582132363441832e-05,
+      "loss": 0.1004,
+      "step": 6042
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.4578624652111403e-05,
+      "loss": 0.1071,
+      "step": 6044
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.4575116227854105e-05,
+      "loss": 0.1068,
+      "step": 6046
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.4571607091216223e-05,
+      "loss": 0.0852,
+      "step": 6048
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.4568097242744161e-05,
+      "loss": 0.106,
+      "step": 6050
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.456458668298443e-05,
+      "loss": 0.0901,
+      "step": 6052
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.4561075412483648e-05,
+      "loss": 0.1098,
+      "step": 6054
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.4557563431788545e-05,
+      "loss": 0.123,
+      "step": 6056
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.4554050741445967e-05,
+      "loss": 0.0801,
+      "step": 6058
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.4550537342002865e-05,
+      "loss": 0.0853,
+      "step": 6060
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.4547023234006304e-05,
+      "loss": 0.1047,
+      "step": 6062
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.4543508418003452e-05,
+      "loss": 0.1323,
+      "step": 6064
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.4539992894541602e-05,
+      "loss": 0.0731,
+      "step": 6066
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.4536476664168137e-05,
+      "loss": 0.132,
+      "step": 6068
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.4532959727430571e-05,
+      "loss": 0.1445,
+      "step": 6070
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.4529442084876513e-05,
+      "loss": 0.0888,
+      "step": 6072
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.4525923737053687e-05,
+      "loss": 0.1083,
+      "step": 6074
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.4522404684509923e-05,
+      "loss": 0.0778,
+      "step": 6076
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.4518884927793174e-05,
+      "loss": 0.1409,
+      "step": 6078
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.4515364467451485e-05,
+      "loss": 0.112,
+      "step": 6080
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.4511843304033022e-05,
+      "loss": 0.0801,
+      "step": 6082
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.4508321438086052e-05,
+      "loss": 0.0965,
+      "step": 6084
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.4504798870158964e-05,
+      "loss": 0.0899,
+      "step": 6086
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.4501275600800244e-05,
+      "loss": 0.0907,
+      "step": 6088
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.4497751630558497e-05,
+      "loss": 0.0885,
+      "step": 6090
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.4494226959982423e-05,
+      "loss": 0.0759,
+      "step": 6092
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.449070158962085e-05,
+      "loss": 0.1384,
+      "step": 6094
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.4487175520022699e-05,
+      "loss": 0.0863,
+      "step": 6096
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.4483648751737007e-05,
+      "loss": 0.1147,
+      "step": 6098
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.4480121285312917e-05,
+      "loss": 0.1112,
+      "step": 6100
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.447659312129969e-05,
+      "loss": 0.1479,
+      "step": 6102
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.447306426024668e-05,
+      "loss": 0.1094,
+      "step": 6104
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.4469534702703363e-05,
+      "loss": 0.1542,
+      "step": 6106
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.4466004449219315e-05,
+      "loss": 0.0746,
+      "step": 6108
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.4462473500344222e-05,
+      "loss": 0.0985,
+      "step": 6110
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.4458941856627884e-05,
+      "loss": 0.1136,
+      "step": 6112
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.4455409518620203e-05,
+      "loss": 0.0918,
+      "step": 6114
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.445187648687119e-05,
+      "loss": 0.1103,
+      "step": 6116
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.4448342761930966e-05,
+      "loss": 0.0859,
+      "step": 6118
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.4444808344349758e-05,
+      "loss": 0.0783,
+      "step": 6120
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.4441273234677908e-05,
+      "loss": 0.1161,
+      "step": 6122
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.443773743346585e-05,
+      "loss": 0.0838,
+      "step": 6124
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.4434200941264141e-05,
+      "loss": 0.1051,
+      "step": 6126
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.4430663758623437e-05,
+      "loss": 0.0745,
+      "step": 6128
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.4427125886094512e-05,
+      "loss": 0.1197,
+      "step": 6130
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.4423587324228225e-05,
+      "loss": 0.1153,
+      "step": 6132
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.4420048073575573e-05,
+      "loss": 0.1036,
+      "step": 6134
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.4416508134687634e-05,
+      "loss": 0.1269,
+      "step": 6136
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.441296750811561e-05,
+      "loss": 0.0646,
+      "step": 6138
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.4409426194410796e-05,
+      "loss": 0.1054,
+      "step": 6140
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.4405884194124608e-05,
+      "loss": 0.0993,
+      "step": 6142
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.440234150780856e-05,
+      "loss": 0.0915,
+      "step": 6144
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.439879813601428e-05,
+      "loss": 0.0799,
+      "step": 6146
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.4395254079293488e-05,
+      "loss": 0.0987,
+      "step": 6148
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.4391709338198032e-05,
+      "loss": 0.1367,
+      "step": 6150
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.4388163913279849e-05,
+      "loss": 0.0956,
+      "step": 6152
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.438461780509099e-05,
+      "loss": 0.095,
+      "step": 6154
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.438107101418361e-05,
+      "loss": 0.1123,
+      "step": 6156
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.4377523541109975e-05,
+      "loss": 0.074,
+      "step": 6158
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.4373975386422448e-05,
+      "loss": 0.1176,
+      "step": 6160
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.4370426550673507e-05,
+      "loss": 0.1224,
+      "step": 6162
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.4366877034415736e-05,
+      "loss": 0.1145,
+      "step": 6164
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.4363326838201817e-05,
+      "loss": 0.0839,
+      "step": 6166
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.435977596258454e-05,
+      "loss": 0.1347,
+      "step": 6168
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.4356224408116814e-05,
+      "loss": 0.1113,
+      "step": 6170
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.4352672175351638e-05,
+      "loss": 0.1176,
+      "step": 6172
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.4349119264842117e-05,
+      "loss": 0.1102,
+      "step": 6174
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.4345565677141472e-05,
+      "loss": 0.0892,
+      "step": 6176
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.434201141280302e-05,
+      "loss": 0.1004,
+      "step": 6178
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.4338456472380193e-05,
+      "loss": 0.0913,
+      "step": 6180
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.4334900856426516e-05,
+      "loss": 0.0929,
+      "step": 6182
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.4331344565495628e-05,
+      "loss": 0.0892,
+      "step": 6184
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.4327787600141274e-05,
+      "loss": 0.0961,
+      "step": 6186
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.4324229960917293e-05,
+      "loss": 0.0726,
+      "step": 6188
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.4320671648377645e-05,
+      "loss": 0.086,
+      "step": 6190
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.4317112663076382e-05,
+      "loss": 0.0861,
+      "step": 6192
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.4313553005567664e-05,
+      "loss": 0.1012,
+      "step": 6194
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.430999267640576e-05,
+      "loss": 0.0886,
+      "step": 6196
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.4306431676145038e-05,
+      "loss": 0.0944,
+      "step": 6198
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.4302870005339975e-05,
+      "loss": 0.1189,
+      "step": 6200
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.4299307664545152e-05,
+      "loss": 0.0685,
+      "step": 6202
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.4295744654315247e-05,
+      "loss": 0.0749,
+      "step": 6204
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.4292180975205052e-05,
+      "loss": 0.092,
+      "step": 6206
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.4288616627769458e-05,
+      "loss": 0.061,
+      "step": 6208
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.4285051612563462e-05,
+      "loss": 0.0825,
+      "step": 6210
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.4281485930142163e-05,
+      "loss": 0.0969,
+      "step": 6212
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.4277919581060771e-05,
+      "loss": 0.0764,
+      "step": 6214
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.4274352565874581e-05,
+      "loss": 0.1221,
+      "step": 6216
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.4270784885139015e-05,
+      "loss": 0.1147,
+      "step": 6218
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.4267216539409588e-05,
+      "loss": 0.1137,
+      "step": 6220
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.4263647529241914e-05,
+      "loss": 0.1143,
+      "step": 6222
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.4260077855191714e-05,
+      "loss": 0.1059,
+      "step": 6224
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.4256507517814818e-05,
+      "loss": 0.0878,
+      "step": 6226
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.4252936517667156e-05,
+      "loss": 0.1016,
+      "step": 6228
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.4249364855304755e-05,
+      "loss": 0.1202,
+      "step": 6230
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.4245792531283754e-05,
+      "loss": 0.1082,
+      "step": 6232
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.4242219546160388e-05,
+      "loss": 0.1111,
+      "step": 6234
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.4238645900491e-05,
+      "loss": 0.1022,
+      "step": 6236
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.4235071594832036e-05,
+      "loss": 0.0812,
+      "step": 6238
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.4231496629740038e-05,
+      "loss": 0.0813,
+      "step": 6240
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.4227921005771661e-05,
+      "loss": 0.0885,
+      "step": 6242
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.4224344723483652e-05,
+      "loss": 0.1035,
+      "step": 6244
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.4220767783432868e-05,
+      "loss": 0.1093,
+      "step": 6246
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.4217190186176266e-05,
+      "loss": 0.1197,
+      "step": 6248
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.4213611932270903e-05,
+      "loss": 0.1098,
+      "step": 6250
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.421003302227394e-05,
+      "loss": 0.1052,
+      "step": 6252
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.4206453456742649e-05,
+      "loss": 0.0923,
+      "step": 6254
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.4202873236234384e-05,
+      "loss": 0.116,
+      "step": 6256
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.4199292361306623e-05,
+      "loss": 0.0929,
+      "step": 6258
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.4195710832516926e-05,
+      "loss": 0.0794,
+      "step": 6260
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.4192128650422974e-05,
+      "loss": 0.0726,
+      "step": 6262
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.4188545815582532e-05,
+      "loss": 0.1018,
+      "step": 6264
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.418496232855348e-05,
+      "loss": 0.088,
+      "step": 6266
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.4181378189893788e-05,
+      "loss": 0.1082,
+      "step": 6268
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.4177793400161545e-05,
+      "loss": 0.0993,
+      "step": 6270
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.4174207959914919e-05,
+      "loss": 0.0931,
+      "step": 6272
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.4170621869712197e-05,
+      "loss": 0.1065,
+      "step": 6274
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.4167035130111758e-05,
+      "loss": 0.0877,
+      "step": 6276
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.4163447741672087e-05,
+      "loss": 0.1107,
+      "step": 6278
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.4159859704951762e-05,
+      "loss": 0.0969,
+      "step": 6280
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.4156271020509476e-05,
+      "loss": 0.1076,
+      "step": 6282
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.4152681688904009e-05,
+      "loss": 0.0982,
+      "step": 6284
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.4149091710694246e-05,
+      "loss": 0.1082,
+      "step": 6286
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.414550108643918e-05,
+      "loss": 0.0902,
+      "step": 6288
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.4141909816697896e-05,
+      "loss": 0.1011,
+      "step": 6290
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.4138317902029582e-05,
+      "loss": 0.0917,
+      "step": 6292
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.4134725342993524e-05,
+      "loss": 0.1049,
+      "step": 6294
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.4131132140149114e-05,
+      "loss": 0.1184,
+      "step": 6296
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.4127538294055845e-05,
+      "loss": 0.1051,
+      "step": 6298
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.4123943805273298e-05,
+      "loss": 0.1399,
+      "step": 6300
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.4120348674361167e-05,
+      "loss": 0.118,
+      "step": 6302
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.4116752901879236e-05,
+      "loss": 0.1346,
+      "step": 6304
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.4113156488387405e-05,
+      "loss": 0.1015,
+      "step": 6306
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.4109559434445652e-05,
+      "loss": 0.0904,
+      "step": 6308
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.4105961740614076e-05,
+      "loss": 0.0995,
+      "step": 6310
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.4102363407452857e-05,
+      "loss": 0.0938,
+      "step": 6312
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.4098764435522288e-05,
+      "loss": 0.0961,
+      "step": 6314
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.409516482538275e-05,
+      "loss": 0.1001,
+      "step": 6316
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.4091564577594739e-05,
+      "loss": 0.0802,
+      "step": 6318
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.4087963692718833e-05,
+      "loss": 0.0751,
+      "step": 6320
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.4084362171315723e-05,
+      "loss": 0.0737,
+      "step": 6322
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.408076001394619e-05,
+      "loss": 0.1007,
+      "step": 6324
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.4077157221171121e-05,
+      "loss": 0.0879,
+      "step": 6326
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.4073553793551495e-05,
+      "loss": 0.079,
+      "step": 6328
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.4069949731648394e-05,
+      "loss": 0.1108,
+      "step": 6330
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.4066345036022998e-05,
+      "loss": 0.066,
+      "step": 6332
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.4062739707236588e-05,
+      "loss": 0.1281,
+      "step": 6334
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.4059133745850534e-05,
+      "loss": 0.1125,
+      "step": 6336
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.4055527152426323e-05,
+      "loss": 0.0769,
+      "step": 6338
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.4051919927525521e-05,
+      "loss": 0.1046,
+      "step": 6340
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.4048312071709803e-05,
+      "loss": 0.0938,
+      "step": 6342
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.4044703585540935e-05,
+      "loss": 0.0938,
+      "step": 6344
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.4041094469580796e-05,
+      "loss": 0.0689,
+      "step": 6346
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.4037484724391345e-05,
+      "loss": 0.0901,
+      "step": 6348
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.4033874350534648e-05,
+      "loss": 0.0418,
+      "step": 6350
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.403026334857287e-05,
+      "loss": 0.0833,
+      "step": 6352
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.4026651719068271e-05,
+      "loss": 0.0939,
+      "step": 6354
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.402303946258321e-05,
+      "loss": 0.1081,
+      "step": 6356
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.4019426579680143e-05,
+      "loss": 0.0756,
+      "step": 6358
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.4015813070921617e-05,
+      "loss": 0.1146,
+      "step": 6360
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.4012198936870294e-05,
+      "loss": 0.1008,
+      "step": 6362
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.4008584178088914e-05,
+      "loss": 0.0906,
+      "step": 6364
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.4004968795140324e-05,
+      "loss": 0.0532,
+      "step": 6366
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.400135278858747e-05,
+      "loss": 0.0909,
+      "step": 6368
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.3997736158993388e-05,
+      "loss": 0.0986,
+      "step": 6370
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.3994118906921219e-05,
+      "loss": 0.1116,
+      "step": 6372
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.3990501032934193e-05,
+      "loss": 0.0664,
+      "step": 6374
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.3986882537595646e-05,
+      "loss": 0.0952,
+      "step": 6376
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.3983263421468998e-05,
+      "loss": 0.1138,
+      "step": 6378
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.3979643685117775e-05,
+      "loss": 0.0898,
+      "step": 6380
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.3976023329105601e-05,
+      "loss": 0.0842,
+      "step": 6382
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.3972402353996193e-05,
+      "loss": 0.0735,
+      "step": 6384
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.396878076035336e-05,
+      "loss": 0.1043,
+      "step": 6386
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.3965158548741016e-05,
+      "loss": 0.1062,
+      "step": 6388
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.3961535719723168e-05,
+      "loss": 0.0908,
+      "step": 6390
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.3957912273863912e-05,
+      "loss": 0.0975,
+      "step": 6392
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.3954288211727454e-05,
+      "loss": 0.0786,
+      "step": 6394
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.3950663533878084e-05,
+      "loss": 0.0687,
+      "step": 6396
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.3947038240880191e-05,
+      "loss": 0.1054,
+      "step": 6398
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.3943412333298261e-05,
+      "loss": 0.0816,
+      "step": 6400
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.3939785811696878e-05,
+      "loss": 0.1035,
+      "step": 6402
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.3936158676640719e-05,
+      "loss": 0.0964,
+      "step": 6404
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.3932530928694555e-05,
+      "loss": 0.0817,
+      "step": 6406
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.3928902568423252e-05,
+      "loss": 0.0969,
+      "step": 6408
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.392527359639178e-05,
+      "loss": 0.1049,
+      "step": 6410
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.3921644013165192e-05,
+      "loss": 0.0704,
+      "step": 6412
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.3918013819308644e-05,
+      "loss": 0.0889,
+      "step": 6414
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.3914383015387382e-05,
+      "loss": 0.1163,
+      "step": 6416
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.391075160196676e-05,
+      "loss": 0.0687,
+      "step": 6418
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.3907119579612203e-05,
+      "loss": 0.1149,
+      "step": 6420
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.390348694888925e-05,
+      "loss": 0.1164,
+      "step": 6422
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.3899853710363536e-05,
+      "loss": 0.091,
+      "step": 6424
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.3896219864600778e-05,
+      "loss": 0.059,
+      "step": 6426
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.3892585412166788e-05,
+      "loss": 0.0787,
+      "step": 6428
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.3888950353627489e-05,
+      "loss": 0.1243,
+      "step": 6430
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.388531468954888e-05,
+      "loss": 0.0921,
+      "step": 6432
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.3881678420497067e-05,
+      "loss": 0.1013,
+      "step": 6434
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.3878041547038238e-05,
+      "loss": 0.0861,
+      "step": 6436
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.3874404069738692e-05,
+      "loss": 0.1073,
+      "step": 6438
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.38707659891648e-05,
+      "loss": 0.0689,
+      "step": 6440
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.386712730588305e-05,
+      "loss": 0.0693,
+      "step": 6442
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.3863488020460004e-05,
+      "loss": 0.108,
+      "step": 6444
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.3859848133462335e-05,
+      "loss": 0.0884,
+      "step": 6446
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.3856207645456792e-05,
+      "loss": 0.0763,
+      "step": 6448
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.3852566557010234e-05,
+      "loss": 0.0818,
+      "step": 6450
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.3848924868689604e-05,
+      "loss": 0.1001,
+      "step": 6452
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.384528258106194e-05,
+      "loss": 0.0767,
+      "step": 6454
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.3841639694694373e-05,
+      "loss": 0.1017,
+      "step": 6456
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.3837996210154132e-05,
+      "loss": 0.0771,
+      "step": 6458
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.3834352128008533e-05,
+      "loss": 0.1086,
+      "step": 6460
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.383070744882499e-05,
+      "loss": 0.0926,
+      "step": 6462
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.3827062173170998e-05,
+      "loss": 0.0757,
+      "step": 6464
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.3823416301614168e-05,
+      "loss": 0.0725,
+      "step": 6466
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.381976983472218e-05,
+      "loss": 0.1125,
+      "step": 6468
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.3816122773062824e-05,
+      "loss": 0.1036,
+      "step": 6470
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.3812475117203968e-05,
+      "loss": 0.099,
+      "step": 6472
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.3808826867713587e-05,
+      "loss": 0.1099,
+      "step": 6474
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.3805178025159734e-05,
+      "loss": 0.1017,
+      "step": 6476
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.3801528590110566e-05,
+      "loss": 0.0859,
+      "step": 6478
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.3797878563134327e-05,
+      "loss": 0.0944,
+      "step": 6480
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.3794227944799362e-05,
+      "loss": 0.1067,
+      "step": 6482
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.3790576735674085e-05,
+      "loss": 0.0733,
+      "step": 6484
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.378692493632703e-05,
+      "loss": 0.0918,
+      "step": 6486
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.37832725473268e-05,
+      "loss": 0.0913,
+      "step": 6488
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.3779619569242109e-05,
+      "loss": 0.1443,
+      "step": 6490
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.3775966002641747e-05,
+      "loss": 0.0941,
+      "step": 6492
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.3772311848094607e-05,
+      "loss": 0.1119,
+      "step": 6494
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.3768657106169668e-05,
+      "loss": 0.0961,
+      "step": 6496
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.3765001777435996e-05,
+      "loss": 0.1158,
+      "step": 6498
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.376134586246276e-05,
+      "loss": 0.0841,
+      "step": 6500
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.3757689361819209e-05,
+      "loss": 0.0953,
+      "step": 6502
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.3754032276074695e-05,
+      "loss": 0.0758,
+      "step": 6504
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.3750374605798647e-05,
+      "loss": 0.1309,
+      "step": 6506
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.3746716351560597e-05,
+      "loss": 0.1379,
+      "step": 6508
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.3743057513930161e-05,
+      "loss": 0.0891,
+      "step": 6510
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.3739398093477045e-05,
+      "loss": 0.0682,
+      "step": 6512
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.3735738090771053e-05,
+      "loss": 0.0714,
+      "step": 6514
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.3732077506382075e-05,
+      "loss": 0.0855,
+      "step": 6516
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.3728416340880093e-05,
+      "loss": 0.0952,
+      "step": 6518
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.3724754594835178e-05,
+      "loss": 0.0899,
+      "step": 6520
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.3721092268817484e-05,
+      "loss": 0.0974,
+      "step": 6522
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.3717429363397276e-05,
+      "loss": 0.075,
+      "step": 6524
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.3713765879144886e-05,
+      "loss": 0.103,
+      "step": 6526
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.3710101816630755e-05,
+      "loss": 0.0908,
+      "step": 6528
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.3706437176425399e-05,
+      "loss": 0.0942,
+      "step": 6530
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.3702771959099434e-05,
+      "loss": 0.1166,
+      "step": 6532
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.3699106165223558e-05,
+      "loss": 0.0825,
+      "step": 6534
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.369543979536857e-05,
+      "loss": 0.0648,
+      "step": 6536
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.3691772850105348e-05,
+      "loss": 0.115,
+      "step": 6538
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.3688105330004862e-05,
+      "loss": 0.0745,
+      "step": 6540
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.3684437235638173e-05,
+      "loss": 0.111,
+      "step": 6542
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.3680768567576434e-05,
+      "loss": 0.1024,
+      "step": 6544
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.3677099326390886e-05,
+      "loss": 0.1281,
+      "step": 6546
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.3673429512652854e-05,
+      "loss": 0.0773,
+      "step": 6548
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.3669759126933754e-05,
+      "loss": 0.0829,
+      "step": 6550
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.3666088169805102e-05,
+      "loss": 0.0926,
+      "step": 6552
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.3662416641838486e-05,
+      "loss": 0.077,
+      "step": 6554
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.3658744543605594e-05,
+      "loss": 0.0866,
+      "step": 6556
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.3655071875678197e-05,
+      "loss": 0.0599,
+      "step": 6558
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.3651398638628164e-05,
+      "loss": 0.1072,
+      "step": 6560
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.3647724833027437e-05,
+      "loss": 0.1026,
+      "step": 6562
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.3644050459448066e-05,
+      "loss": 0.1246,
+      "step": 6564
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.364037551846217e-05,
+      "loss": 0.1003,
+      "step": 6566
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.3636700010641972e-05,
+      "loss": 0.0968,
+      "step": 6568
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.363302393655977e-05,
+      "loss": 0.0981,
+      "step": 6570
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.3629347296787963e-05,
+      "loss": 0.09,
+      "step": 6572
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.362567009189903e-05,
+      "loss": 0.0695,
+      "step": 6574
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.362199232246554e-05,
+      "loss": 0.1228,
+      "step": 6576
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.3618313989060146e-05,
+      "loss": 0.0869,
+      "step": 6578
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.3614635092255597e-05,
+      "loss": 0.0724,
+      "step": 6580
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.3610955632624726e-05,
+      "loss": 0.1333,
+      "step": 6582
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.3607275610740451e-05,
+      "loss": 0.0725,
+      "step": 6584
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.3603595027175777e-05,
+      "loss": 0.0734,
+      "step": 6586
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.3599913882503808e-05,
+      "loss": 0.0788,
+      "step": 6588
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.3596232177297714e-05,
+      "loss": 0.0734,
+      "step": 6590
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.3592549912130775e-05,
+      "loss": 0.0769,
+      "step": 6592
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.358886708757634e-05,
+      "loss": 0.1177,
+      "step": 6594
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.3585183704207861e-05,
+      "loss": 0.0876,
+      "step": 6596
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.358149976259886e-05,
+      "loss": 0.0744,
+      "step": 6598
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.3577815263322962e-05,
+      "loss": 0.0821,
+      "step": 6600
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.357413020695387e-05,
+      "loss": 0.0667,
+      "step": 6602
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.3570444594065375e-05,
+      "loss": 0.0631,
+      "step": 6604
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.356675842523135e-05,
+      "loss": 0.0886,
+      "step": 6606
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.356307170102577e-05,
+      "loss": 0.0796,
+      "step": 6608
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.3559384422022677e-05,
+      "loss": 0.0657,
+      "step": 6610
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.3555696588796214e-05,
+      "loss": 0.0681,
+      "step": 6612
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.3552008201920602e-05,
+      "loss": 0.0894,
+      "step": 6614
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.3548319261970152e-05,
+      "loss": 0.0864,
+      "step": 6616
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.354462976951926e-05,
+      "loss": 0.0782,
+      "step": 6618
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.354093972514241e-05,
+      "loss": 0.0908,
+      "step": 6620
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.3537249129414166e-05,
+      "loss": 0.0792,
+      "step": 6622
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.353355798290919e-05,
+      "loss": 0.0967,
+      "step": 6624
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.3529866286202209e-05,
+      "loss": 0.1142,
+      "step": 6626
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.3526174039868059e-05,
+      "loss": 0.0679,
+      "step": 6628
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.3522481244481647e-05,
+      "loss": 0.0894,
+      "step": 6630
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.351878790061797e-05,
+      "loss": 0.0971,
+      "step": 6632
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.351509400885211e-05,
+      "loss": 0.0876,
+      "step": 6634
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.3511399569759234e-05,
+      "loss": 0.064,
+      "step": 6636
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.3507704583914594e-05,
+      "loss": 0.0719,
+      "step": 6638
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.3504009051893529e-05,
+      "loss": 0.101,
+      "step": 6640
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.350031297427146e-05,
+      "loss": 0.0776,
+      "step": 6642
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.3496616351623898e-05,
+      "loss": 0.0993,
+      "step": 6644
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.349291918452643e-05,
+      "loss": 0.0617,
+      "step": 6646
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.3489221473554735e-05,
+      "loss": 0.1,
+      "step": 6648
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.3485523219284578e-05,
+      "loss": 0.0901,
+      "step": 6650
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.34818244222918e-05,
+      "loss": 0.088,
+      "step": 6652
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.3478125083152337e-05,
+      "loss": 0.0493,
+      "step": 6654
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.3474425202442204e-05,
+      "loss": 0.0586,
+      "step": 6656
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.3470724780737498e-05,
+      "loss": 0.0762,
+      "step": 6658
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.3467023818614404e-05,
+      "loss": 0.0733,
+      "step": 6660
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.346332231664919e-05,
+      "loss": 0.0833,
+      "step": 6662
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.345962027541821e-05,
+      "loss": 0.053,
+      "step": 6664
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.3455917695497898e-05,
+      "loss": 0.097,
+      "step": 6666
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.3452214577464772e-05,
+      "loss": 0.065,
+      "step": 6668
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.3448510921895441e-05,
+      "loss": 0.0837,
+      "step": 6670
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.3444806729366593e-05,
+      "loss": 0.0551,
+      "step": 6672
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.344110200045499e-05,
+      "loss": 0.0591,
+      "step": 6674
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.3437396735737495e-05,
+      "loss": 0.0876,
+      "step": 6676
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.3433690935791045e-05,
+      "loss": 0.084,
+      "step": 6678
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.342998460119266e-05,
+      "loss": 0.0704,
+      "step": 6680
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.3426277732519442e-05,
+      "loss": 0.0588,
+      "step": 6682
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.3422570330348583e-05,
+      "loss": 0.0687,
+      "step": 6684
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.3418862395257353e-05,
+      "loss": 0.0762,
+      "step": 6686
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.3415153927823105e-05,
+      "loss": 0.0834,
+      "step": 6688
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.3411444928623274e-05,
+      "loss": 0.0674,
+      "step": 6690
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.3407735398235384e-05,
+      "loss": 0.0903,
+      "step": 6692
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.3404025337237033e-05,
+      "loss": 0.0852,
+      "step": 6694
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.3400314746205909e-05,
+      "loss": 0.0731,
+      "step": 6696
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.3396603625719777e-05,
+      "loss": 0.0913,
+      "step": 6698
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.339289197635649e-05,
+      "loss": 0.0917,
+      "step": 6700
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.3389179798693974e-05,
+      "loss": 0.0653,
+      "step": 6702
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.3385467093310251e-05,
+      "loss": 0.1114,
+      "step": 6704
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.338175386078341e-05,
+      "loss": 0.1016,
+      "step": 6706
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.337804010169164e-05,
+      "loss": 0.0744,
+      "step": 6708
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.3374325816613189e-05,
+      "loss": 0.0955,
+      "step": 6710
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.337061100612641e-05,
+      "loss": 0.1176,
+      "step": 6712
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.3366895670809723e-05,
+      "loss": 0.0606,
+      "step": 6714
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.3363179811241635e-05,
+      "loss": 0.0663,
+      "step": 6716
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.3359463428000732e-05,
+      "loss": 0.0803,
+      "step": 6718
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.3355746521665685e-05,
+      "loss": 0.0724,
+      "step": 6720
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.3352029092815244e-05,
+      "loss": 0.0864,
+      "step": 6722
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.3348311142028242e-05,
+      "loss": 0.0666,
+      "step": 6724
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.3344592669883591e-05,
+      "loss": 0.1026,
+      "step": 6726
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.334087367696029e-05,
+      "loss": 0.0513,
+      "step": 6728
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.3337154163837406e-05,
+      "loss": 0.0928,
+      "step": 6730
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.3333434131094103e-05,
+      "loss": 0.0912,
+      "step": 6732
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.3329713579309615e-05,
+      "loss": 0.0679,
+      "step": 6734
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.3325992509063266e-05,
+      "loss": 0.0704,
+      "step": 6736
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.3322270920934443e-05,
+      "loss": 0.0817,
+      "step": 6738
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.3318548815502638e-05,
+      "loss": 0.0806,
+      "step": 6740
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.3314826193347408e-05,
+      "loss": 0.0747,
+      "step": 6742
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.331110305504839e-05,
+      "loss": 0.068,
+      "step": 6744
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.3307379401185307e-05,
+      "loss": 0.0744,
+      "step": 6746
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.3303655232337962e-05,
+      "loss": 0.0704,
+      "step": 6748
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.3299930549086235e-05,
+      "loss": 0.0685,
+      "step": 6750
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.329620535201009e-05,
+      "loss": 0.0694,
+      "step": 6752
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.3292479641689565e-05,
+      "loss": 0.0921,
+      "step": 6754
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.3288753418704783e-05,
+      "loss": 0.0766,
+      "step": 6756
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.3285026683635947e-05,
+      "loss": 0.0638,
+      "step": 6758
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.3281299437063336e-05,
+      "loss": 0.0723,
+      "step": 6760
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.3277571679567309e-05,
+      "loss": 0.0767,
+      "step": 6762
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.3273843411728314e-05,
+      "loss": 0.0923,
+      "step": 6764
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.3270114634126862e-05,
+      "loss": 0.0905,
+      "step": 6766
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.3266385347343557e-05,
+      "loss": 0.0921,
+      "step": 6768
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.3262655551959079e-05,
+      "loss": 0.0729,
+      "step": 6770
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.3258925248554183e-05,
+      "loss": 0.0848,
+      "step": 6772
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.3255194437709702e-05,
+      "loss": 0.067,
+      "step": 6774
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.3251463120006558e-05,
+      "loss": 0.0976,
+      "step": 6776
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.3247731296025743e-05,
+      "loss": 0.0812,
+      "step": 6778
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.3243998966348333e-05,
+      "loss": 0.093,
+      "step": 6780
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.3240266131555475e-05,
+      "loss": 0.0781,
+      "step": 6782
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.3236532792228405e-05,
+      "loss": 0.1145,
+      "step": 6784
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.323279894894843e-05,
+      "loss": 0.0821,
+      "step": 6786
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.322906460229694e-05,
+      "loss": 0.0943,
+      "step": 6788
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.3225329752855395e-05,
+      "loss": 0.115,
+      "step": 6790
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.3221594401205354e-05,
+      "loss": 0.1004,
+      "step": 6792
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.3217858547928422e-05,
+      "loss": 0.0497,
+      "step": 6794
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.3214122193606312e-05,
+      "loss": 0.0828,
+      "step": 6796
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.3210385338820798e-05,
+      "loss": 0.0554,
+      "step": 6798
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.3206647984153738e-05,
+      "loss": 0.0439,
+      "step": 6800
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.3202910130187066e-05,
+      "loss": 0.0718,
+      "step": 6802
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.3199171777502796e-05,
+      "loss": 0.0786,
+      "step": 6804
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.3195432926683016e-05,
+      "loss": 0.0559,
+      "step": 6806
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.3191693578309898e-05,
+      "loss": 0.0879,
+      "step": 6808
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.3187953732965681e-05,
+      "loss": 0.0609,
+      "step": 6810
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.3184213391232693e-05,
+      "loss": 0.075,
+      "step": 6812
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.3180472553693329e-05,
+      "loss": 0.0882,
+      "step": 6814
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.317673122093007e-05,
+      "loss": 0.1017,
+      "step": 6816
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.3172989393525469e-05,
+      "loss": 0.0492,
+      "step": 6818
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.3169247072062158e-05,
+      "loss": 0.0848,
+      "step": 6820
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.3165504257122838e-05,
+      "loss": 0.079,
+      "step": 6822
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.3161760949290306e-05,
+      "loss": 0.0513,
+      "step": 6824
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.3158017149147415e-05,
+      "loss": 0.0672,
+      "step": 6826
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.3154272857277107e-05,
+      "loss": 0.069,
+      "step": 6828
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.315052807426239e-05,
+      "loss": 0.1033,
+      "step": 6830
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.3146782800686366e-05,
+      "loss": 0.0862,
+      "step": 6832
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3143037037132195e-05,
+      "loss": 0.0774,
+      "step": 6834
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3139290784183127e-05,
+      "loss": 0.0861,
+      "step": 6836
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3135544042422475e-05,
+      "loss": 0.099,
+      "step": 6838
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3131796812433641e-05,
+      "loss": 0.0701,
+      "step": 6840
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3128049094800095e-05,
+      "loss": 0.1021,
+      "step": 6842
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3124300890105388e-05,
+      "loss": 0.0735,
+      "step": 6844
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3120552198933138e-05,
+      "loss": 0.0844,
+      "step": 6846
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3116803021867054e-05,
+      "loss": 0.0757,
+      "step": 6848
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3113053359490901e-05,
+      "loss": 0.0762,
+      "step": 6850
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3109303212388538e-05,
+      "loss": 0.0775,
+      "step": 6852
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.310555258114389e-05,
+      "loss": 0.065,
+      "step": 6854
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3101801466340958e-05,
+      "loss": 0.0842,
+      "step": 6856
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3098049868563818e-05,
+      "loss": 0.0922,
+      "step": 6858
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3094297788396623e-05,
+      "loss": 0.0922,
+      "step": 6860
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3090545226423604e-05,
+      "loss": 0.0887,
+      "step": 6862
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.308679218322906e-05,
+      "loss": 0.0852,
+      "step": 6864
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3083038659397367e-05,
+      "loss": 0.0745,
+      "step": 6866
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.307928465551298e-05,
+      "loss": 0.0971,
+      "step": 6868
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3075530172160428e-05,
+      "loss": 0.075,
+      "step": 6870
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3071775209924313e-05,
+      "loss": 0.0622,
+      "step": 6872
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3068019769389304e-05,
+      "loss": 0.0791,
+      "step": 6874
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3064263851140163e-05,
+      "loss": 0.06,
+      "step": 6876
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3060507455761702e-05,
+      "loss": 0.0656,
+      "step": 6878
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3056750583838831e-05,
+      "loss": 0.0621,
+      "step": 6880
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3052993235956519e-05,
+      "loss": 0.0763,
+      "step": 6882
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3049235412699818e-05,
+      "loss": 0.0601,
+      "step": 6884
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3045477114653844e-05,
+      "loss": 0.0569,
+      "step": 6886
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3041718342403796e-05,
+      "loss": 0.0619,
+      "step": 6888
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3037959096534943e-05,
+      "loss": 0.0899,
+      "step": 6890
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.303419937763263e-05,
+      "loss": 0.0981,
+      "step": 6892
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3030439186282269e-05,
+      "loss": 0.0686,
+      "step": 6894
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3026678523069355e-05,
+      "loss": 0.072,
+      "step": 6896
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3022917388579455e-05,
+      "loss": 0.0743,
+      "step": 6898
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3019155783398199e-05,
+      "loss": 0.0568,
+      "step": 6900
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3015393708111299e-05,
+      "loss": 0.0667,
+      "step": 6902
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3011631163304548e-05,
+      "loss": 0.0629,
+      "step": 6904
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.300786814956379e-05,
+      "loss": 0.0757,
+      "step": 6906
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3004104667474962e-05,
+      "loss": 0.066,
+      "step": 6908
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3000340717624064e-05,
+      "loss": 0.0786,
+      "step": 6910
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.2996576300597181e-05,
+      "loss": 0.0548,
+      "step": 6912
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.2992811416980446e-05,
+      "loss": 0.0579,
+      "step": 6914
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2989046067360093e-05,
+      "loss": 0.0517,
+      "step": 6916
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.298528025232241e-05,
+      "loss": 0.0682,
+      "step": 6918
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2981513972453766e-05,
+      "loss": 0.0604,
+      "step": 6920
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2977747228340594e-05,
+      "loss": 0.0646,
+      "step": 6922
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2973980020569413e-05,
+      "loss": 0.085,
+      "step": 6924
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2970212349726798e-05,
+      "loss": 0.0795,
+      "step": 6926
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.296644421639941e-05,
+      "loss": 0.062,
+      "step": 6928
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2962675621173972e-05,
+      "loss": 0.0899,
+      "step": 6930
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2958906564637287e-05,
+      "loss": 0.0663,
+      "step": 6932
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2955137047376227e-05,
+      "loss": 0.0601,
+      "step": 6934
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.295136706997773e-05,
+      "loss": 0.0444,
+      "step": 6936
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2947596633028808e-05,
+      "loss": 0.0382,
+      "step": 6938
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2943825737116558e-05,
+      "loss": 0.0671,
+      "step": 6940
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2940054382828124e-05,
+      "loss": 0.0548,
+      "step": 6942
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2936282570750745e-05,
+      "loss": 0.0785,
+      "step": 6944
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2932510301471714e-05,
+      "loss": 0.0671,
+      "step": 6946
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2928737575578407e-05,
+      "loss": 0.0715,
+      "step": 6948
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2924964393658263e-05,
+      "loss": 0.0892,
+      "step": 6950
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2921190756298798e-05,
+      "loss": 0.0792,
+      "step": 6952
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2917416664087596e-05,
+      "loss": 0.0751,
+      "step": 6954
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.2913642117612311e-05,
+      "loss": 0.0708,
+      "step": 6956
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.2909867117460664e-05,
+      "loss": 0.1013,
+      "step": 6958
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.2906091664220461e-05,
+      "loss": 0.0618,
+      "step": 6960
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.2902315758479562e-05,
+      "loss": 0.0884,
+      "step": 6962
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.289853940082591e-05,
+      "loss": 0.0489,
+      "step": 6964
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.2894762591847502e-05,
+      "loss": 0.0689,
+      "step": 6966
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.2890985332132432e-05,
+      "loss": 0.0461,
+      "step": 6968
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.2887207622268831e-05,
+      "loss": 0.0751,
+      "step": 6970
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.288342946284493e-05,
+      "loss": 0.0822,
+      "step": 6972
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.2879650854449013e-05,
+      "loss": 0.0637,
+      "step": 6974
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.287587179766944e-05,
+      "loss": 0.0525,
+      "step": 6976
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.2872092293094631e-05,
+      "loss": 0.0633,
+      "step": 6978
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.2868312341313096e-05,
+      "loss": 0.0648,
+      "step": 6980
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.2864531942913394e-05,
+      "loss": 0.0624,
+      "step": 6982
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.2860751098484165e-05,
+      "loss": 0.0744,
+      "step": 6984
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.2856969808614115e-05,
+      "loss": 0.0659,
+      "step": 6986
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.285318807389202e-05,
+      "loss": 0.0635,
+      "step": 6988
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.2849405894906724e-05,
+      "loss": 0.0604,
+      "step": 6990
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.2845623272247142e-05,
+      "loss": 0.0527,
+      "step": 6992
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.2841840206502254e-05,
+      "loss": 0.0505,
+      "step": 6994
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.2838056698261122e-05,
+      "loss": 0.0698,
+      "step": 6996
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.2834272748112855e-05,
+      "loss": 0.0681,
+      "step": 6998
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.283048835664665e-05,
+      "loss": 0.073,
+      "step": 7000
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.2826703524451764e-05,
+      "loss": 0.0798,
+      "step": 7002
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.2822918252117525e-05,
+      "loss": 0.0707,
+      "step": 7004
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.2819132540233326e-05,
+      "loss": 0.0698,
+      "step": 7006
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.2815346389388637e-05,
+      "loss": 0.0702,
+      "step": 7008
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.2811559800172986e-05,
+      "loss": 0.0673,
+      "step": 7010
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.2807772773175975e-05,
+      "loss": 0.0444,
+      "step": 7012
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.280398530898727e-05,
+      "loss": 0.0665,
+      "step": 7014
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.2800197408196616e-05,
+      "loss": 0.052,
+      "step": 7016
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.279640907139381e-05,
+      "loss": 0.0917,
+      "step": 7018
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.279262029916873e-05,
+      "loss": 0.0769,
+      "step": 7020
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.2788831092111316e-05,
+      "loss": 0.0858,
+      "step": 7022
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.2785041450811574e-05,
+      "loss": 0.0873,
+      "step": 7024
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.2781251375859585e-05,
+      "loss": 0.0654,
+      "step": 7026
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.2777460867845485e-05,
+      "loss": 0.0692,
+      "step": 7028
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.2773669927359494e-05,
+      "loss": 0.0733,
+      "step": 7030
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.2769878554991882e-05,
+      "loss": 0.0602,
+      "step": 7032
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.2766086751333e-05,
+      "loss": 0.0904,
+      "step": 7034
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.276229451697326e-05,
+      "loss": 0.0866,
+      "step": 7036
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.275850185250314e-05,
+      "loss": 0.065,
+      "step": 7038
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.2754708758513192e-05,
+      "loss": 0.0648,
+      "step": 7040
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.2750915235594023e-05,
+      "loss": 0.0751,
+      "step": 7042
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.2747121284336317e-05,
+      "loss": 0.0602,
+      "step": 7044
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.2743326905330822e-05,
+      "loss": 0.0571,
+      "step": 7046
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.2739532099168347e-05,
+      "loss": 0.0627,
+      "step": 7048
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.273573686643978e-05,
+      "loss": 0.0493,
+      "step": 7050
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.2731941207736063e-05,
+      "loss": 0.0748,
+      "step": 7052
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.272814512364821e-05,
+      "loss": 0.0552,
+      "step": 7054
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.2724348614767296e-05,
+      "loss": 0.0575,
+      "step": 7056
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.2720551681684475e-05,
+      "loss": 0.0828,
+      "step": 7058
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.2716754324990952e-05,
+      "loss": 0.0615,
+      "step": 7060
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.2712956545278008e-05,
+      "loss": 0.0633,
+      "step": 7062
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.2709158343136983e-05,
+      "loss": 0.0567,
+      "step": 7064
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.2705359719159288e-05,
+      "loss": 0.0764,
+      "step": 7066
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.2701560673936399e-05,
+      "loss": 0.0633,
+      "step": 7068
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.2697761208059856e-05,
+      "loss": 0.0733,
+      "step": 7070
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.2693961322121262e-05,
+      "loss": 0.0731,
+      "step": 7072
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.2690161016712295e-05,
+      "loss": 0.0748,
+      "step": 7074
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.268636029242468e-05,
+      "loss": 0.0657,
+      "step": 7076
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.2682559149850229e-05,
+      "loss": 0.0762,
+      "step": 7078
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.2678757589580805e-05,
+      "loss": 0.0656,
+      "step": 7080
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.2674955612208343e-05,
+      "loss": 0.0784,
+      "step": 7082
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.2671153218324834e-05,
+      "loss": 0.0761,
+      "step": 7084
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.2667350408522347e-05,
+      "loss": 0.0839,
+      "step": 7086
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.2663547183393006e-05,
+      "loss": 0.0711,
+      "step": 7088
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.2659743543529e-05,
+      "loss": 0.0582,
+      "step": 7090
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.2655939489522582e-05,
+      "loss": 0.0774,
+      "step": 7092
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.2652135021966081e-05,
+      "loss": 0.0714,
+      "step": 7094
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.2648330141451877e-05,
+      "loss": 0.0388,
+      "step": 7096
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.2644524848572419e-05,
+      "loss": 0.0422,
+      "step": 7098
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.2640719143920216e-05,
+      "loss": 0.0827,
+      "step": 7100
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.2636913028087856e-05,
+      "loss": 0.0647,
+      "step": 7102
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.2633106501667971e-05,
+      "loss": 0.0638,
+      "step": 7104
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.2629299565253268e-05,
+      "loss": 0.061,
+      "step": 7106
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.2625492219436517e-05,
+      "loss": 0.0665,
+      "step": 7108
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.2621684464810556e-05,
+      "loss": 0.0547,
+      "step": 7110
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.2617876301968273e-05,
+      "loss": 0.0812,
+      "step": 7112
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.261406773150263e-05,
+      "loss": 0.0571,
+      "step": 7114
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.2610258754006656e-05,
+      "loss": 0.075,
+      "step": 7116
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.2606449370073435e-05,
+      "loss": 0.0742,
+      "step": 7118
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.2602639580296113e-05,
+      "loss": 0.0594,
+      "step": 7120
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.2598829385267908e-05,
+      "loss": 0.0501,
+      "step": 7122
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.2595018785582096e-05,
+      "loss": 0.0565,
+      "step": 7124
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.2591207781832015e-05,
+      "loss": 0.044,
+      "step": 7126
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.258739637461107e-05,
+      "loss": 0.0562,
+      "step": 7128
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.2583584564512723e-05,
+      "loss": 0.0556,
+      "step": 7130
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.2579772352130503e-05,
+      "loss": 0.0764,
+      "step": 7132
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.2575959738058004e-05,
+      "loss": 0.0508,
+      "step": 7134
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.2572146722888871e-05,
+      "loss": 0.0608,
+      "step": 7136
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.256833330721683e-05,
+      "loss": 0.0632,
+      "step": 7138
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.2564519491635651e-05,
+      "loss": 0.0377,
+      "step": 7140
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.256070527673918e-05,
+      "loss": 0.043,
+      "step": 7142
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.2556890663121314e-05,
+      "loss": 0.0714,
+      "step": 7144
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.2553075651376023e-05,
+      "loss": 0.0483,
+      "step": 7146
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.2549260242097328e-05,
+      "loss": 0.0759,
+      "step": 7148
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.254544443587932e-05,
+      "loss": 0.0351,
+      "step": 7150
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.254162823331615e-05,
+      "loss": 0.0651,
+      "step": 7152
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.253781163500203e-05,
+      "loss": 0.0596,
+      "step": 7154
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.253399464153123e-05,
+      "loss": 0.0418,
+      "step": 7156
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.253017725349809e-05,
+      "loss": 0.0747,
+      "step": 7158
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.2526359471497004e-05,
+      "loss": 0.0731,
+      "step": 7160
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.252254129612243e-05,
+      "loss": 0.0658,
+      "step": 7162
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2518722727968886e-05,
+      "loss": 0.0795,
+      "step": 7164
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2514903767630957e-05,
+      "loss": 0.0771,
+      "step": 7166
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2511084415703276e-05,
+      "loss": 0.0422,
+      "step": 7168
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2507264672780553e-05,
+      "loss": 0.052,
+      "step": 7170
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.250344453945755e-05,
+      "loss": 0.0968,
+      "step": 7172
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2499624016329086e-05,
+      "loss": 0.057,
+      "step": 7174
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2495803103990047e-05,
+      "loss": 0.0577,
+      "step": 7176
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2491981803035384e-05,
+      "loss": 0.0737,
+      "step": 7178
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2488160114060099e-05,
+      "loss": 0.0838,
+      "step": 7180
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2484338037659258e-05,
+      "loss": 0.0762,
+      "step": 7182
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2480515574427985e-05,
+      "loss": 0.0482,
+      "step": 7184
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.247669272496147e-05,
+      "loss": 0.0824,
+      "step": 7186
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2472869489854961e-05,
+      "loss": 0.0583,
+      "step": 7188
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2469045869703764e-05,
+      "loss": 0.0522,
+      "step": 7190
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2465221865103243e-05,
+      "loss": 0.071,
+      "step": 7192
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2461397476648828e-05,
+      "loss": 0.0606,
+      "step": 7194
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2457572704936004e-05,
+      "loss": 0.0634,
+      "step": 7196
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2453747550560317e-05,
+      "loss": 0.0957,
+      "step": 7198
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2449922014117376e-05,
+      "loss": 0.0689,
+      "step": 7200
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2446096096202843e-05,
+      "loss": 0.0602,
+      "step": 7202
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.244226979741244e-05,
+      "loss": 0.0644,
+      "step": 7204
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.2438443118341957e-05,
+      "loss": 0.0556,
+      "step": 7206
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.2434616059587235e-05,
+      "loss": 0.0592,
+      "step": 7208
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.2430788621744174e-05,
+      "loss": 0.0664,
+      "step": 7210
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.2426960805408739e-05,
+      "loss": 0.0707,
+      "step": 7212
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.2423132611176947e-05,
+      "loss": 0.0505,
+      "step": 7214
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.241930403964488e-05,
+      "loss": 0.0452,
+      "step": 7216
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.2415475091408675e-05,
+      "loss": 0.057,
+      "step": 7218
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.2411645767064524e-05,
+      "loss": 0.0425,
+      "step": 7220
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.2407816067208692e-05,
+      "loss": 0.0651,
+      "step": 7222
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.2403985992437482e-05,
+      "loss": 0.073,
+      "step": 7224
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.2400155543347272e-05,
+      "loss": 0.0842,
+      "step": 7226
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.2396324720534491e-05,
+      "loss": 0.0768,
+      "step": 7228
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.2392493524595629e-05,
+      "loss": 0.089,
+      "step": 7230
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.2388661956127225e-05,
+      "loss": 0.0947,
+      "step": 7232
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.2384830015725893e-05,
+      "loss": 0.0511,
+      "step": 7234
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.2380997703988294e-05,
+      "loss": 0.0574,
+      "step": 7236
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.2377165021511142e-05,
+      "loss": 0.0552,
+      "step": 7238
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.2373331968891217e-05,
+      "loss": 0.059,
+      "step": 7240
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.236949854672536e-05,
+      "loss": 0.0623,
+      "step": 7242
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.2365664755610461e-05,
+      "loss": 0.0496,
+      "step": 7244
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.236183059614347e-05,
+      "loss": 0.0766,
+      "step": 7246
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.2357996068921392e-05,
+      "loss": 0.0592,
+      "step": 7248
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.23541611745413e-05,
+      "loss": 0.0564,
+      "step": 7250
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.2350325913600307e-05,
+      "loss": 0.0574,
+      "step": 7252
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.2346490286695601e-05,
+      "loss": 0.0581,
+      "step": 7254
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.2342654294424413e-05,
+      "loss": 0.0536,
+      "step": 7256
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.2338817937384038e-05,
+      "loss": 0.054,
+      "step": 7258
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.2334981216171823e-05,
+      "loss": 0.0912,
+      "step": 7260
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.2331144131385183e-05,
+      "loss": 0.043,
+      "step": 7262
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.2327306683621573e-05,
+      "loss": 0.0451,
+      "step": 7264
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.2323468873478518e-05,
+      "loss": 0.0478,
+      "step": 7266
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.2319630701553589e-05,
+      "loss": 0.0463,
+      "step": 7268
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.2315792168444424e-05,
+      "loss": 0.0596,
+      "step": 7270
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.2311953274748714e-05,
+      "loss": 0.053,
+      "step": 7272
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.2308114021064197e-05,
+      "loss": 0.0691,
+      "step": 7274
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.2304274407988676e-05,
+      "loss": 0.0751,
+      "step": 7276
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.2300434436120017e-05,
+      "loss": 0.0597,
+      "step": 7278
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.2296594106056118e-05,
+      "loss": 0.0776,
+      "step": 7280
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.2292753418394958e-05,
+      "loss": 0.0516,
+      "step": 7282
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.2288912373734558e-05,
+      "loss": 0.0512,
+      "step": 7284
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.2285070972673004e-05,
+      "loss": 0.0703,
+      "step": 7286
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.228122921580842e-05,
+      "loss": 0.0541,
+      "step": 7288
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.2277387103739006e-05,
+      "loss": 0.048,
+      "step": 7290
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.2273544637063006e-05,
+      "loss": 0.0695,
+      "step": 7292
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.2269701816378721e-05,
+      "loss": 0.0592,
+      "step": 7294
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.2265858642284505e-05,
+      "loss": 0.0675,
+      "step": 7296
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.2262015115378774e-05,
+      "loss": 0.0494,
+      "step": 7298
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.2258171236259993e-05,
+      "loss": 0.0649,
+      "step": 7300
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.2254327005526683e-05,
+      "loss": 0.0679,
+      "step": 7302
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.2250482423777419e-05,
+      "loss": 0.0423,
+      "step": 7304
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.2246637491610835e-05,
+      "loss": 0.0632,
+      "step": 7306
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.2242792209625613e-05,
+      "loss": 0.0735,
+      "step": 7308
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.2238946578420493e-05,
+      "loss": 0.0536,
+      "step": 7310
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.2235100598594271e-05,
+      "loss": 0.0353,
+      "step": 7312
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.2231254270745798e-05,
+      "loss": 0.0562,
+      "step": 7314
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.222740759547397e-05,
+      "loss": 0.0477,
+      "step": 7316
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.2223560573377749e-05,
+      "loss": 0.0357,
+      "step": 7318
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.2219713205056143e-05,
+      "loss": 0.0444,
+      "step": 7320
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.2215865491108216e-05,
+      "loss": 0.0584,
+      "step": 7322
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.221201743213309e-05,
+      "loss": 0.0737,
+      "step": 7324
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.2208169028729934e-05,
+      "loss": 0.0562,
+      "step": 7326
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.2204320281497977e-05,
+      "loss": 0.0666,
+      "step": 7328
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.2200471191036496e-05,
+      "loss": 0.0591,
+      "step": 7330
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.2196621757944822e-05,
+      "loss": 0.0802,
+      "step": 7332
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.2192771982822346e-05,
+      "loss": 0.0904,
+      "step": 7334
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.2188921866268503e-05,
+      "loss": 0.0593,
+      "step": 7336
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.2185071408882792e-05,
+      "loss": 0.0706,
+      "step": 7338
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.2181220611264748e-05,
+      "loss": 0.0537,
+      "step": 7340
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.217736947401398e-05,
+      "loss": 0.0791,
+      "step": 7342
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.2173517997730133e-05,
+      "loss": 0.0722,
+      "step": 7344
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.2169666183012915e-05,
+      "loss": 0.0609,
+      "step": 7346
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.2165814030462083e-05,
+      "loss": 0.0408,
+      "step": 7348
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.2161961540677442e-05,
+      "loss": 0.0463,
+      "step": 7350
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.2158108714258859e-05,
+      "loss": 0.0581,
+      "step": 7352
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.2154255551806246e-05,
+      "loss": 0.051,
+      "step": 7354
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.2150402053919571e-05,
+      "loss": 0.0894,
+      "step": 7356
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.2146548221198855e-05,
+      "loss": 0.0349,
+      "step": 7358
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.2142694054244164e-05,
+      "loss": 0.0638,
+      "step": 7360
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.2138839553655625e-05,
+      "loss": 0.0809,
+      "step": 7362
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.2134984720033414e-05,
+      "loss": 0.0432,
+      "step": 7364
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.2131129553977756e-05,
+      "loss": 0.0694,
+      "step": 7366
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.212727405608893e-05,
+      "loss": 0.0583,
+      "step": 7368
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.212341822696727e-05,
+      "loss": 0.0403,
+      "step": 7370
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.2119562067213148e-05,
+      "loss": 0.0793,
+      "step": 7372
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.211570557742701e-05,
+      "loss": 0.0586,
+      "step": 7374
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.2111848758209335e-05,
+      "loss": 0.067,
+      "step": 7376
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.2107991610160662e-05,
+      "loss": 0.0443,
+      "step": 7378
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.2104134133881568e-05,
+      "loss": 0.0592,
+      "step": 7380
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.2100276329972707e-05,
+      "loss": 0.0659,
+      "step": 7382
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.2096418199034761e-05,
+      "loss": 0.0783,
+      "step": 7384
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.2092559741668469e-05,
+      "loss": 0.0566,
+      "step": 7386
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.2088700958474622e-05,
+      "loss": 0.0612,
+      "step": 7388
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.2084841850054067e-05,
+      "loss": 0.0438,
+      "step": 7390
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.2080982417007694e-05,
+      "loss": 0.0786,
+      "step": 7392
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.2077122659936446e-05,
+      "loss": 0.0555,
+      "step": 7394
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.2073262579441317e-05,
+      "loss": 0.0824,
+      "step": 7396
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.2069402176123351e-05,
+      "loss": 0.0639,
+      "step": 7398
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.206554145058364e-05,
+      "loss": 0.0706,
+      "step": 7400
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.2061680403423333e-05,
+      "loss": 0.0571,
+      "step": 7402
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.2057819035243622e-05,
+      "loss": 0.0457,
+      "step": 7404
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.2053957346645751e-05,
+      "loss": 0.0595,
+      "step": 7406
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.2050095338231015e-05,
+      "loss": 0.0435,
+      "step": 7408
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.2046233010600758e-05,
+      "loss": 0.0591,
+      "step": 7410
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.2042370364356375e-05,
+      "loss": 0.0569,
+      "step": 7412
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.203850740009931e-05,
+      "loss": 0.0637,
+      "step": 7414
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.2034644118431054e-05,
+      "loss": 0.0537,
+      "step": 7416
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.2030780519953149e-05,
+      "loss": 0.0588,
+      "step": 7418
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.2026916605267191e-05,
+      "loss": 0.0919,
+      "step": 7420
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.202305237497482e-05,
+      "loss": 0.0683,
+      "step": 7422
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.201918782967772e-05,
+      "loss": 0.0617,
+      "step": 7424
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.2015322969977638e-05,
+      "loss": 0.0796,
+      "step": 7426
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.2011457796476359e-05,
+      "loss": 0.0273,
+      "step": 7428
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.200759230977572e-05,
+      "loss": 0.0583,
+      "step": 7430
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.2003726510477605e-05,
+      "loss": 0.0635,
+      "step": 7432
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.1999860399183954e-05,
+      "loss": 0.0606,
+      "step": 7434
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.1995993976496742e-05,
+      "loss": 0.0623,
+      "step": 7436
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.1992127243018009e-05,
+      "loss": 0.0387,
+      "step": 7438
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.198826019934983e-05,
+      "loss": 0.0544,
+      "step": 7440
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.1984392846094335e-05,
+      "loss": 0.0673,
+      "step": 7442
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.1980525183853697e-05,
+      "loss": 0.056,
+      "step": 7444
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.1976657213230147e-05,
+      "loss": 0.0527,
+      "step": 7446
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.1972788934825953e-05,
+      "loss": 0.0665,
+      "step": 7448
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.1968920349243435e-05,
+      "loss": 0.0574,
+      "step": 7450
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.1965051457084965e-05,
+      "loss": 0.046,
+      "step": 7452
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.1961182258952958e-05,
+      "loss": 0.082,
+      "step": 7454
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.1957312755449874e-05,
+      "loss": 0.0669,
+      "step": 7456
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.1953442947178228e-05,
+      "loss": 0.0371,
+      "step": 7458
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.194957283474058e-05,
+      "loss": 0.0458,
+      "step": 7460
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.1945702418739533e-05,
+      "loss": 0.0513,
+      "step": 7462
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.1941831699777738e-05,
+      "loss": 0.0445,
+      "step": 7464
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.1937960678457902e-05,
+      "loss": 0.0513,
+      "step": 7466
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.193408935538277e-05,
+      "loss": 0.0386,
+      "step": 7468
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.1930217731155133e-05,
+      "loss": 0.0375,
+      "step": 7470
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.1926345806377837e-05,
+      "loss": 0.0555,
+      "step": 7472
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.192247358165377e-05,
+      "loss": 0.0475,
+      "step": 7474
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.1918601057585866e-05,
+      "loss": 0.0672,
+      "step": 7476
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.1914728234777104e-05,
+      "loss": 0.0506,
+      "step": 7478
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.1910855113830515e-05,
+      "loss": 0.0644,
+      "step": 7480
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.1906981695349178e-05,
+      "loss": 0.0561,
+      "step": 7482
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.1903107979936203e-05,
+      "loss": 0.0377,
+      "step": 7484
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.1899233968194766e-05,
+      "loss": 0.0492,
+      "step": 7486
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.1895359660728073e-05,
+      "loss": 0.0442,
+      "step": 7488
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.1891485058139396e-05,
+      "loss": 0.082,
+      "step": 7490
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.1887610161032026e-05,
+      "loss": 0.0474,
+      "step": 7492
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.1883734970009321e-05,
+      "loss": 0.062,
+      "step": 7494
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.1879859485674678e-05,
+      "loss": 0.0484,
+      "step": 7496
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.1875983708631538e-05,
+      "loss": 0.067,
+      "step": 7498
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.1872107639483389e-05,
+      "loss": 0.0289,
+      "step": 7500
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.1868231278833764e-05,
+      "loss": 0.0618,
+      "step": 7502
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.1864354627286245e-05,
+      "loss": 0.0663,
+      "step": 7504
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.1860477685444456e-05,
+      "loss": 0.0584,
+      "step": 7506
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.1856600453912062e-05,
+      "loss": 0.0609,
+      "step": 7508
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.1852722933292781e-05,
+      "loss": 0.054,
+      "step": 7510
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.1848845124190374e-05,
+      "loss": 0.06,
+      "step": 7512
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.1844967027208642e-05,
+      "loss": 0.0598,
+      "step": 7514
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.1841088642951434e-05,
+      "loss": 0.0688,
+      "step": 7516
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.1837209972022652e-05,
+      "loss": 0.0468,
+      "step": 7518
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.1833331015026224e-05,
+      "loss": 0.0393,
+      "step": 7520
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.182945177256614e-05,
+      "loss": 0.0479,
+      "step": 7522
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.1825572245246426e-05,
+      "loss": 0.0501,
+      "step": 7524
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.1821692433671154e-05,
+      "loss": 0.046,
+      "step": 7526
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.1817812338444437e-05,
+      "loss": 0.0462,
+      "step": 7528
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.1813931960170442e-05,
+      "loss": 0.0494,
+      "step": 7530
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.181005129945337e-05,
+      "loss": 0.069,
+      "step": 7532
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.1806170356897471e-05,
+      "loss": 0.0546,
+      "step": 7534
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.1802289133107032e-05,
+      "loss": 0.0531,
+      "step": 7536
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.1798407628686396e-05,
+      "loss": 0.0584,
+      "step": 7538
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.1794525844239941e-05,
+      "loss": 0.0441,
+      "step": 7540
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.179064378037209e-05,
+      "loss": 0.0678,
+      "step": 7542
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.1786761437687311e-05,
+      "loss": 0.0746,
+      "step": 7544
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.1782878816790114e-05,
+      "loss": 0.041,
+      "step": 7546
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.1778995918285047e-05,
+      "loss": 0.0991,
+      "step": 7548
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.1775112742776715e-05,
+      "loss": 0.0414,
+      "step": 7550
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.1771229290869757e-05,
+      "loss": 0.0695,
+      "step": 7552
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.1767345563168852e-05,
+      "loss": 0.0457,
+      "step": 7554
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.176346156027873e-05,
+      "loss": 0.0418,
+      "step": 7556
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.1759577282804157e-05,
+      "loss": 0.0333,
+      "step": 7558
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.1755692731349947e-05,
+      "loss": 0.0486,
+      "step": 7560
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.1751807906520956e-05,
+      "loss": 0.0484,
+      "step": 7562
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.1747922808922074e-05,
+      "loss": 0.0518,
+      "step": 7564
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.1744037439158247e-05,
+      "loss": 0.0293,
+      "step": 7566
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.1740151797834459e-05,
+      "loss": 0.0558,
+      "step": 7568
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.1736265885555722e-05,
+      "loss": 0.0539,
+      "step": 7570
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.1732379702927114e-05,
+      "loss": 0.0306,
+      "step": 7572
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.172849325055374e-05,
+      "loss": 0.0461,
+      "step": 7574
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.1724606529040749e-05,
+      "loss": 0.0704,
+      "step": 7576
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.1720719538993332e-05,
+      "loss": 0.0465,
+      "step": 7578
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.1716832281016726e-05,
+      "loss": 0.0553,
+      "step": 7580
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.1712944755716207e-05,
+      "loss": 0.0388,
+      "step": 7582
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.1709056963697091e-05,
+      "loss": 0.0304,
+      "step": 7584
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.1705168905564737e-05,
+      "loss": 0.0426,
+      "step": 7586
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.1701280581924546e-05,
+      "loss": 0.0628,
+      "step": 7588
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.1697391993381956e-05,
+      "loss": 0.0755,
+      "step": 7590
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.169350314054246e-05,
+      "loss": 0.0762,
+      "step": 7592
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.1689614024011569e-05,
+      "loss": 0.0569,
+      "step": 7594
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.1685724644394858e-05,
+      "loss": 0.0615,
+      "step": 7596
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.1681835002297927e-05,
+      "loss": 0.0571,
+      "step": 7598
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.1677945098326429e-05,
+      "loss": 0.0297,
+      "step": 7600
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.1674054933086048e-05,
+      "loss": 0.0459,
+      "step": 7602
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.167016450718251e-05,
+      "loss": 0.0535,
+      "step": 7604
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.1666273821221588e-05,
+      "loss": 0.0368,
+      "step": 7606
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.1662382875809094e-05,
+      "loss": 0.0658,
+      "step": 7608
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.165849167155087e-05,
+      "loss": 0.0469,
+      "step": 7610
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.1654600209052815e-05,
+      "loss": 0.0301,
+      "step": 7612
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.1650708488920851e-05,
+      "loss": 0.0431,
+      "step": 7614
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.1646816511760956e-05,
+      "loss": 0.0431,
+      "step": 7616
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.1642924278179137e-05,
+      "loss": 0.0605,
+      "step": 7618
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.1639031788781446e-05,
+      "loss": 0.0425,
+      "step": 7620
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.1635139044173969e-05,
+      "loss": 0.0449,
+      "step": 7622
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.1631246044962846e-05,
+      "loss": 0.0392,
+      "step": 7624
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.1627352791754232e-05,
+      "loss": 0.0475,
+      "step": 7626
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.1623459285154347e-05,
+      "loss": 0.0607,
+      "step": 7628
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.1619565525769441e-05,
+      "loss": 0.0473,
+      "step": 7630
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.1615671514205798e-05,
+      "loss": 0.0601,
+      "step": 7632
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.161177725106974e-05,
+      "loss": 0.0459,
+      "step": 7634
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.1607882736967643e-05,
+      "loss": 0.0704,
+      "step": 7636
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.160398797250591e-05,
+      "loss": 0.0571,
+      "step": 7638
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.1600092958290985e-05,
+      "loss": 0.0279,
+      "step": 7640
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.1596197694929348e-05,
+      "loss": 0.0464,
+      "step": 7642
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.1592302183027526e-05,
+      "loss": 0.0503,
+      "step": 7644
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.1588406423192077e-05,
+      "loss": 0.0279,
+      "step": 7646
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.1584510416029607e-05,
+      "loss": 0.0315,
+      "step": 7648
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.1580614162146742e-05,
+      "loss": 0.0345,
+      "step": 7650
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.157671766215017e-05,
+      "loss": 0.0339,
+      "step": 7652
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.15728209166466e-05,
+      "loss": 0.0618,
+      "step": 7654
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.1568923926242786e-05,
+      "loss": 0.0564,
+      "step": 7656
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.156502669154552e-05,
+      "loss": 0.068,
+      "step": 7658
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.1561129213161633e-05,
+      "loss": 0.0486,
+      "step": 7660
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.1557231491697987e-05,
+      "loss": 0.0585,
+      "step": 7662
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.1553333527761493e-05,
+      "loss": 0.0451,
+      "step": 7664
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.1549435321959087e-05,
+      "loss": 0.0424,
+      "step": 7666
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.1545536874897758e-05,
+      "loss": 0.0484,
+      "step": 7668
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.1541638187184514e-05,
+      "loss": 0.0546,
+      "step": 7670
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.153773925942642e-05,
+      "loss": 0.0494,
+      "step": 7672
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.1533840092230564e-05,
+      "loss": 0.0472,
+      "step": 7674
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.1529940686204078e-05,
+      "loss": 0.051,
+      "step": 7676
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.1526041041954126e-05,
+      "loss": 0.0327,
+      "step": 7678
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.1522141160087916e-05,
+      "loss": 0.0485,
+      "step": 7680
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.1518241041212686e-05,
+      "loss": 0.0634,
+      "step": 7682
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.1514340685935719e-05,
+      "loss": 0.0518,
+      "step": 7684
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.1510440094864328e-05,
+      "loss": 0.0507,
+      "step": 7686
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.1506539268605863e-05,
+      "loss": 0.0374,
+      "step": 7688
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.1502638207767712e-05,
+      "loss": 0.0627,
+      "step": 7690
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.1498736912957304e-05,
+      "loss": 0.0582,
+      "step": 7692
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.1494835384782099e-05,
+      "loss": 0.0506,
+      "step": 7694
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.1490933623849594e-05,
+      "loss": 0.0514,
+      "step": 7696
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.148703163076732e-05,
+      "loss": 0.038,
+      "step": 7698
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.1483129406142855e-05,
+      "loss": 0.0517,
+      "step": 7700
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.1479226950583797e-05,
+      "loss": 0.045,
+      "step": 7702
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.1475324264697793e-05,
+      "loss": 0.0536,
+      "step": 7704
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.1471421349092517e-05,
+      "loss": 0.0401,
+      "step": 7706
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.1467518204375692e-05,
+      "loss": 0.0395,
+      "step": 7708
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.1463614831155054e-05,
+      "loss": 0.0255,
+      "step": 7710
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.14597112300384e-05,
+      "loss": 0.0377,
+      "step": 7712
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.145580740163354e-05,
+      "loss": 0.0363,
+      "step": 7714
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.1451903346548343e-05,
+      "loss": 0.0424,
+      "step": 7716
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.1447999065390686e-05,
+      "loss": 0.0533,
+      "step": 7718
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.1444094558768506e-05,
+      "loss": 0.0512,
+      "step": 7720
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.144018982728976e-05,
+      "loss": 0.0423,
+      "step": 7722
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.1436284871562446e-05,
+      "loss": 0.0439,
+      "step": 7724
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.1432379692194593e-05,
+      "loss": 0.0202,
+      "step": 7726
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.1428474289794269e-05,
+      "loss": 0.0439,
+      "step": 7728
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.1424568664969578e-05,
+      "loss": 0.0575,
+      "step": 7730
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.1420662818328649e-05,
+      "loss": 0.0481,
+      "step": 7732
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.1416756750479657e-05,
+      "loss": 0.0439,
+      "step": 7734
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.1412850462030806e-05,
+      "loss": 0.0269,
+      "step": 7736
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.1408943953590335e-05,
+      "loss": 0.0288,
+      "step": 7738
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.1405037225766518e-05,
+      "loss": 0.0515,
+      "step": 7740
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.1401130279167655e-05,
+      "loss": 0.0356,
+      "step": 7742
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.13972231144021e-05,
+      "loss": 0.0463,
+      "step": 7744
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.1393315732078219e-05,
+      "loss": 0.0463,
+      "step": 7746
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.1389408132804426e-05,
+      "loss": 0.036,
+      "step": 7748
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.1385500317189163e-05,
+      "loss": 0.0508,
+      "step": 7750
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.1381592285840903e-05,
+      "loss": 0.0498,
+      "step": 7752
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.137768403936816e-05,
+      "loss": 0.0297,
+      "step": 7754
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.137377557837948e-05,
+      "loss": 0.0464,
+      "step": 7756
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.1369866903483437e-05,
+      "loss": 0.052,
+      "step": 7758
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.136595801528864e-05,
+      "loss": 0.0511,
+      "step": 7760
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.1362048914403736e-05,
+      "loss": 0.0447,
+      "step": 7762
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.1358139601437402e-05,
+      "loss": 0.0638,
+      "step": 7764
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.1354230076998347e-05,
+      "loss": 0.0534,
+      "step": 7766
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.1350320341695314e-05,
+      "loss": 0.0392,
+      "step": 7768
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.1346410396137075e-05,
+      "loss": 0.0487,
+      "step": 7770
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.1342500240932445e-05,
+      "loss": 0.0457,
+      "step": 7772
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.1338589876690261e-05,
+      "loss": 0.0435,
+      "step": 7774
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.1334679304019397e-05,
+      "loss": 0.0728,
+      "step": 7776
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.1330768523528761e-05,
+      "loss": 0.0512,
+      "step": 7778
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.132685753582729e-05,
+      "loss": 0.0334,
+      "step": 7780
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.132294634152395e-05,
+      "loss": 0.055,
+      "step": 7782
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.1319034941227754e-05,
+      "loss": 0.052,
+      "step": 7784
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.131512333554773e-05,
+      "loss": 0.0687,
+      "step": 7786
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.1311211525092947e-05,
+      "loss": 0.0563,
+      "step": 7788
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.13072995104725e-05,
+      "loss": 0.0289,
+      "step": 7790
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.1303387292295524e-05,
+      "loss": 0.0318,
+      "step": 7792
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.1299474871171184e-05,
+      "loss": 0.0357,
+      "step": 7794
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.1295562247708665e-05,
+      "loss": 0.0283,
+      "step": 7796
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.1291649422517196e-05,
+      "loss": 0.0341,
+      "step": 7798
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.1287736396206043e-05,
+      "loss": 0.0473,
+      "step": 7800
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.128382316938448e-05,
+      "loss": 0.0694,
+      "step": 7802
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.1279909742661834e-05,
+      "loss": 0.0444,
+      "step": 7804
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.1275996116647454e-05,
+      "loss": 0.0306,
+      "step": 7806
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.1272082291950723e-05,
+      "loss": 0.0405,
+      "step": 7808
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.1268168269181047e-05,
+      "loss": 0.051,
+      "step": 7810
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.126425404894788e-05,
+      "loss": 0.051,
+      "step": 7812
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.1260339631860688e-05,
+      "loss": 0.0346,
+      "step": 7814
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.125642501852898e-05,
+      "loss": 0.0586,
+      "step": 7816
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.1252510209562284e-05,
+      "loss": 0.0491,
+      "step": 7818
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1248595205570174e-05,
+      "loss": 0.0461,
+      "step": 7820
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1244680007162246e-05,
+      "loss": 0.0478,
+      "step": 7822
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.124076461494812e-05,
+      "loss": 0.0301,
+      "step": 7824
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1236849029537453e-05,
+      "loss": 0.0567,
+      "step": 7826
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1232933251539941e-05,
+      "loss": 0.0397,
+      "step": 7828
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1229017281565288e-05,
+      "loss": 0.0522,
+      "step": 7830
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.122510112022325e-05,
+      "loss": 0.0357,
+      "step": 7832
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1221184768123598e-05,
+      "loss": 0.0452,
+      "step": 7834
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.121726822587614e-05,
+      "loss": 0.0441,
+      "step": 7836
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.121335149409071e-05,
+      "loss": 0.0346,
+      "step": 7838
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1209434573377176e-05,
+      "loss": 0.0482,
+      "step": 7840
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.120551746434543e-05,
+      "loss": 0.0434,
+      "step": 7842
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1201600167605397e-05,
+      "loss": 0.0564,
+      "step": 7844
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1197682683767028e-05,
+      "loss": 0.0537,
+      "step": 7846
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.119376501344031e-05,
+      "loss": 0.0479,
+      "step": 7848
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1189847157235249e-05,
+      "loss": 0.0449,
+      "step": 7850
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1185929115761889e-05,
+      "loss": 0.0575,
+      "step": 7852
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1182010889630295e-05,
+      "loss": 0.0479,
+      "step": 7854
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.117809247945057e-05,
+      "loss": 0.0384,
+      "step": 7856
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.1174173885832835e-05,
+      "loss": 0.0481,
+      "step": 7858
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.1170255109387248e-05,
+      "loss": 0.0357,
+      "step": 7860
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.1166336150723992e-05,
+      "loss": 0.0313,
+      "step": 7862
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.1162417010453281e-05,
+      "loss": 0.0547,
+      "step": 7864
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.1158497689185347e-05,
+      "loss": 0.0415,
+      "step": 7866
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.115457818753047e-05,
+      "loss": 0.0375,
+      "step": 7868
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.1150658506098938e-05,
+      "loss": 0.0392,
+      "step": 7870
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.114673864550108e-05,
+      "loss": 0.0445,
+      "step": 7872
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.1142818606347243e-05,
+      "loss": 0.0353,
+      "step": 7874
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.1138898389247812e-05,
+      "loss": 0.0281,
+      "step": 7876
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.1134977994813191e-05,
+      "loss": 0.0302,
+      "step": 7878
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.1131057423653822e-05,
+      "loss": 0.0335,
+      "step": 7880
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.112713667638016e-05,
+      "loss": 0.0299,
+      "step": 7882
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.11232157536027e-05,
+      "loss": 0.0306,
+      "step": 7884
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.1119294655931956e-05,
+      "loss": 0.0358,
+      "step": 7886
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.1115373383978478e-05,
+      "loss": 0.0605,
+      "step": 7888
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.1111451938352833e-05,
+      "loss": 0.0521,
+      "step": 7890
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.1107530319665625e-05,
+      "loss": 0.0637,
+      "step": 7892
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.1103608528527475e-05,
+      "loss": 0.0534,
+      "step": 7894
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.109968656554904e-05,
+      "loss": 0.0549,
+      "step": 7896
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.1095764431340996e-05,
+      "loss": 0.0276,
+      "step": 7898
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.1091842126514052e-05,
+      "loss": 0.0486,
+      "step": 7900
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.1087919651678938e-05,
+      "loss": 0.0407,
+      "step": 7902
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.1083997007446418e-05,
+      "loss": 0.0588,
+      "step": 7904
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.1080074194427275e-05,
+      "loss": 0.0533,
+      "step": 7906
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.107615121323232e-05,
+      "loss": 0.0421,
+      "step": 7908
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.1072228064472391e-05,
+      "loss": 0.0633,
+      "step": 7910
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.1068304748758356e-05,
+      "loss": 0.0433,
+      "step": 7912
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.10643812667011e-05,
+      "loss": 0.074,
+      "step": 7914
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.1060457618911544e-05,
+      "loss": 0.0198,
+      "step": 7916
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.1056533806000625e-05,
+      "loss": 0.0433,
+      "step": 7918
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.1052609828579318e-05,
+      "loss": 0.0421,
+      "step": 7920
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.1048685687258607e-05,
+      "loss": 0.0387,
+      "step": 7922
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.1044761382649519e-05,
+      "loss": 0.0298,
+      "step": 7924
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.1040836915363093e-05,
+      "loss": 0.0515,
+      "step": 7926
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.1036912286010402e-05,
+      "loss": 0.0444,
+      "step": 7928
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.1032987495202536e-05,
+      "loss": 0.0399,
+      "step": 7930
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.1029062543550619e-05,
+      "loss": 0.055,
+      "step": 7932
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.1025137431665798e-05,
+      "loss": 0.0416,
+      "step": 7934
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.1021212160159238e-05,
+      "loss": 0.0374,
+      "step": 7936
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.1017286729642133e-05,
+      "loss": 0.0323,
+      "step": 7938
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.1013361140725712e-05,
+      "loss": 0.0488,
+      "step": 7940
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.1009435394021208e-05,
+      "loss": 0.0487,
+      "step": 7942
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.1005509490139897e-05,
+      "loss": 0.0384,
+      "step": 7944
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.1001583429693063e-05,
+      "loss": 0.0501,
+      "step": 7946
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.099765721329204e-05,
+      "loss": 0.0436,
+      "step": 7948
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.0993730841548153e-05,
+      "loss": 0.0422,
+      "step": 7950
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.0989804315072779e-05,
+      "loss": 0.038,
+      "step": 7952
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.0985877634477301e-05,
+      "loss": 0.045,
+      "step": 7954
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.098195080037314e-05,
+      "loss": 0.0301,
+      "step": 7956
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.0978023813371728e-05,
+      "loss": 0.0396,
+      "step": 7958
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.0974096674084531e-05,
+      "loss": 0.0443,
+      "step": 7960
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.0970169383123035e-05,
+      "loss": 0.0487,
+      "step": 7962
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.0966241941098745e-05,
+      "loss": 0.0412,
+      "step": 7964
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.0962314348623196e-05,
+      "loss": 0.0621,
+      "step": 7966
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.0958386606307947e-05,
+      "loss": 0.0469,
+      "step": 7968
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.0954458714764573e-05,
+      "loss": 0.0413,
+      "step": 7970
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.095053067460468e-05,
+      "loss": 0.0447,
+      "step": 7972
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.094660248643989e-05,
+      "loss": 0.0226,
+      "step": 7974
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.0942674150881859e-05,
+      "loss": 0.0208,
+      "step": 7976
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.093874566854225e-05,
+      "loss": 0.043,
+      "step": 7978
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.0934817040032763e-05,
+      "loss": 0.0375,
+      "step": 7980
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.0930888265965116e-05,
+      "loss": 0.0679,
+      "step": 7982
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.0926959346951046e-05,
+      "loss": 0.0555,
+      "step": 7984
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.0923030283602318e-05,
+      "loss": 0.0534,
+      "step": 7986
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.0919101076530719e-05,
+      "loss": 0.072,
+      "step": 7988
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.0915171726348053e-05,
+      "loss": 0.034,
+      "step": 7990
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.0911242233666152e-05,
+      "loss": 0.0437,
+      "step": 7992
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.0907312599096864e-05,
+      "loss": 0.0326,
+      "step": 7994
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.0903382823252069e-05,
+      "loss": 0.0421,
+      "step": 7996
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.0899452906743662e-05,
+      "loss": 0.0288,
+      "step": 7998
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.0895522850183557e-05,
+      "loss": 0.0327,
+      "step": 8000
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.0891592654183695e-05,
+      "loss": 0.0469,
+      "step": 8002
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.0887662319356045e-05,
+      "loss": 0.0251,
+      "step": 8004
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.088373184631258e-05,
+      "loss": 0.0282,
+      "step": 8006
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.0879801235665311e-05,
+      "loss": 0.0486,
+      "step": 8008
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.087587048802626e-05,
+      "loss": 0.0395,
+      "step": 8010
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.0871939604007477e-05,
+      "loss": 0.0366,
+      "step": 8012
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.086800858422103e-05,
+      "loss": 0.0515,
+      "step": 8014
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.086407742927901e-05,
+      "loss": 0.0379,
+      "step": 8016
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.0860146139793525e-05,
+      "loss": 0.0424,
+      "step": 8018
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.0856214716376712e-05,
+      "loss": 0.0439,
+      "step": 8020
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.0852283159640718e-05,
+      "loss": 0.0494,
+      "step": 8022
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.084835147019772e-05,
+      "loss": 0.0484,
+      "step": 8024
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0844419648659912e-05,
+      "loss": 0.0313,
+      "step": 8026
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0840487695639506e-05,
+      "loss": 0.052,
+      "step": 8028
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0836555611748739e-05,
+      "loss": 0.0334,
+      "step": 8030
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0832623397599869e-05,
+      "loss": 0.0577,
+      "step": 8032
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0828691053805165e-05,
+      "loss": 0.0274,
+      "step": 8034
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0824758580976929e-05,
+      "loss": 0.0532,
+      "step": 8036
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0820825979727477e-05,
+      "loss": 0.0234,
+      "step": 8038
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.081689325066914e-05,
+      "loss": 0.05,
+      "step": 8040
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0812960394414278e-05,
+      "loss": 0.0279,
+      "step": 8042
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0809027411575267e-05,
+      "loss": 0.0333,
+      "step": 8044
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0805094302764505e-05,
+      "loss": 0.0405,
+      "step": 8046
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0801161068594401e-05,
+      "loss": 0.0307,
+      "step": 8048
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.079722770967739e-05,
+      "loss": 0.0375,
+      "step": 8050
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0793294226625932e-05,
+      "loss": 0.0394,
+      "step": 8052
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0789360620052496e-05,
+      "loss": 0.0318,
+      "step": 8054
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0785426890569575e-05,
+      "loss": 0.0292,
+      "step": 8056
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0781493038789682e-05,
+      "loss": 0.0332,
+      "step": 8058
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0777559065325347e-05,
+      "loss": 0.0598,
+      "step": 8060
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0773624970789118e-05,
+      "loss": 0.0342,
+      "step": 8062
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.0769690755793567e-05,
+      "loss": 0.0371,
+      "step": 8064
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.076575642095128e-05,
+      "loss": 0.0434,
+      "step": 8066
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0761821966874862e-05,
+      "loss": 0.0435,
+      "step": 8068
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.075788739417694e-05,
+      "loss": 0.0477,
+      "step": 8070
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0753952703470152e-05,
+      "loss": 0.0393,
+      "step": 8072
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0750017895367165e-05,
+      "loss": 0.0469,
+      "step": 8074
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0746082970480658e-05,
+      "loss": 0.0254,
+      "step": 8076
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0742147929423326e-05,
+      "loss": 0.0331,
+      "step": 8078
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0738212772807883e-05,
+      "loss": 0.0347,
+      "step": 8080
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0734277501247073e-05,
+      "loss": 0.0454,
+      "step": 8082
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0730342115353635e-05,
+      "loss": 0.052,
+      "step": 8084
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.072640661574035e-05,
+      "loss": 0.036,
+      "step": 8086
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0722471003019998e-05,
+      "loss": 0.0569,
+      "step": 8088
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0718535277805388e-05,
+      "loss": 0.049,
+      "step": 8090
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.071459944070934e-05,
+      "loss": 0.0353,
+      "step": 8092
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0710663492344693e-05,
+      "loss": 0.0394,
+      "step": 8094
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0706727433324308e-05,
+      "loss": 0.0485,
+      "step": 8096
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0702791264261056e-05,
+      "loss": 0.0335,
+      "step": 8098
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0698854985767831e-05,
+      "loss": 0.0302,
+      "step": 8100
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.069491859845754e-05,
+      "loss": 0.0371,
+      "step": 8102
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.0690982102943113e-05,
+      "loss": 0.0345,
+      "step": 8104
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.0687045499837486e-05,
+      "loss": 0.0313,
+      "step": 8106
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.0683108789753618e-05,
+      "loss": 0.0341,
+      "step": 8108
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.0679171973304494e-05,
+      "loss": 0.0459,
+      "step": 8110
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.0675235051103097e-05,
+      "loss": 0.0301,
+      "step": 8112
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.067129802376244e-05,
+      "loss": 0.0269,
+      "step": 8114
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.0667360891895547e-05,
+      "loss": 0.0452,
+      "step": 8116
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.0663423656115461e-05,
+      "loss": 0.0285,
+      "step": 8118
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.0659486317035237e-05,
+      "loss": 0.0586,
+      "step": 8120
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.0655548875267951e-05,
+      "loss": 0.0384,
+      "step": 8122
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.0651611331426694e-05,
+      "loss": 0.0415,
+      "step": 8124
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.064767368612457e-05,
+      "loss": 0.0362,
+      "step": 8126
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.0643735939974698e-05,
+      "loss": 0.0274,
+      "step": 8128
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.0639798093590221e-05,
+      "loss": 0.0448,
+      "step": 8130
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.0635860147584286e-05,
+      "loss": 0.0221,
+      "step": 8132
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.0631922102570066e-05,
+      "loss": 0.0387,
+      "step": 8134
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.0627983959160739e-05,
+      "loss": 0.0381,
+      "step": 8136
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.0624045717969514e-05,
+      "loss": 0.0307,
+      "step": 8138
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.0620107379609592e-05,
+      "loss": 0.0443,
+      "step": 8140
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.0616168944694212e-05,
+      "loss": 0.0449,
+      "step": 8142
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.0612230413836615e-05,
+      "loss": 0.0474,
+      "step": 8144
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.0608291787650064e-05,
+      "loss": 0.0362,
+      "step": 8146
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.0604353066747825e-05,
+      "loss": 0.0477,
+      "step": 8148
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.0600414251743195e-05,
+      "loss": 0.032,
+      "step": 8150
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.0596475343249478e-05,
+      "loss": 0.0534,
+      "step": 8152
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.0592536341879986e-05,
+      "loss": 0.0288,
+      "step": 8154
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.0588597248248054e-05,
+      "loss": 0.0239,
+      "step": 8156
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.058465806296703e-05,
+      "loss": 0.0279,
+      "step": 8158
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.0580718786650275e-05,
+      "loss": 0.0401,
+      "step": 8160
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.0576779419911165e-05,
+      "loss": 0.0421,
+      "step": 8162
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.0572839963363088e-05,
+      "loss": 0.0413,
+      "step": 8164
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.056890041761945e-05,
+      "loss": 0.0301,
+      "step": 8166
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.0564960783293666e-05,
+      "loss": 0.0297,
+      "step": 8168
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.0561021060999168e-05,
+      "loss": 0.037,
+      "step": 8170
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.05570812513494e-05,
+      "loss": 0.0246,
+      "step": 8172
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.0553141354957823e-05,
+      "loss": 0.0507,
+      "step": 8174
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.0549201372437905e-05,
+      "loss": 0.0384,
+      "step": 8176
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.0545261304403137e-05,
+      "loss": 0.0504,
+      "step": 8178
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.0541321151467012e-05,
+      "loss": 0.0367,
+      "step": 8180
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.0537380914243045e-05,
+      "loss": 0.0399,
+      "step": 8182
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.053344059334476e-05,
+      "loss": 0.0241,
+      "step": 8184
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.0529500189385696e-05,
+      "loss": 0.0202,
+      "step": 8186
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.0525559702979402e-05,
+      "loss": 0.0375,
+      "step": 8188
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.0521619134739447e-05,
+      "loss": 0.0425,
+      "step": 8190
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.05176784852794e-05,
+      "loss": 0.0536,
+      "step": 8192
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.0513737755212856e-05,
+      "loss": 0.0386,
+      "step": 8194
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.0509796945153416e-05,
+      "loss": 0.0288,
+      "step": 8196
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.0505856055714693e-05,
+      "loss": 0.0298,
+      "step": 8198
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.0501915087510314e-05,
+      "loss": 0.0213,
+      "step": 8200
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.049797404115392e-05,
+      "loss": 0.0503,
+      "step": 8202
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.0494032917259159e-05,
+      "loss": 0.0498,
+      "step": 8204
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.0490091716439693e-05,
+      "loss": 0.0372,
+      "step": 8206
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.04861504393092e-05,
+      "loss": 0.0302,
+      "step": 8208
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.0482209086481368e-05,
+      "loss": 0.0288,
+      "step": 8210
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.047826765856989e-05,
+      "loss": 0.0146,
+      "step": 8212
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.0474326156188483e-05,
+      "loss": 0.0127,
+      "step": 8214
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.0470384579950867e-05,
+      "loss": 0.0208,
+      "step": 8216
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.0466442930470775e-05,
+      "loss": 0.0144,
+      "step": 8218
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.0462501208361948e-05,
+      "loss": 0.0075,
+      "step": 8220
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.0458559414238152e-05,
+      "loss": 0.0204,
+      "step": 8222
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.0454617548713147e-05,
+      "loss": 0.0187,
+      "step": 8224
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.045067561240071e-05,
+      "loss": 0.0132,
+      "step": 8226
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.0446733605914636e-05,
+      "loss": 0.0206,
+      "step": 8228
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.0442791529868727e-05,
+      "loss": 0.0268,
+      "step": 8230
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.0438849384876789e-05,
+      "loss": 0.0134,
+      "step": 8232
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.0434907171552644e-05,
+      "loss": 0.014,
+      "step": 8234
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.0430964890510129e-05,
+      "loss": 0.0223,
+      "step": 8236
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.0427022542363082e-05,
+      "loss": 0.0163,
+      "step": 8238
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.0423080127725362e-05,
+      "loss": 0.0229,
+      "step": 8240
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.041913764721083e-05,
+      "loss": 0.0164,
+      "step": 8242
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.0415195101433364e-05,
+      "loss": 0.0242,
+      "step": 8244
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.0411252491006844e-05,
+      "loss": 0.0133,
+      "step": 8246
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.0407309816545166e-05,
+      "loss": 0.0088,
+      "step": 8248
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.0403367078662236e-05,
+      "loss": 0.0196,
+      "step": 8250
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.0399424277971968e-05,
+      "loss": 0.0146,
+      "step": 8252
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.0395481415088284e-05,
+      "loss": 0.0186,
+      "step": 8254
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.039153849062512e-05,
+      "loss": 0.0188,
+      "step": 8256
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.0387595505196421e-05,
+      "loss": 0.0091,
+      "step": 8258
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.0383652459416134e-05,
+      "loss": 0.0209,
+      "step": 8260
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.0379709353898229e-05,
+      "loss": 0.0091,
+      "step": 8262
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.0375766189256673e-05,
+      "loss": 0.014,
+      "step": 8264
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.0371822966105449e-05,
+      "loss": 0.012,
+      "step": 8266
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.0367879685058542e-05,
+      "loss": 0.0188,
+      "step": 8268
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.036393634672996e-05,
+      "loss": 0.0137,
+      "step": 8270
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.0359992951733704e-05,
+      "loss": 0.0175,
+      "step": 8272
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.0356049500683792e-05,
+      "loss": 0.0181,
+      "step": 8274
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.0352105994194248e-05,
+      "loss": 0.0202,
+      "step": 8276
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.0348162432879111e-05,
+      "loss": 0.0104,
+      "step": 8278
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.0344218817352422e-05,
+      "loss": 0.0095,
+      "step": 8280
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.034027514822823e-05,
+      "loss": 0.0235,
+      "step": 8282
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.0336331426120595e-05,
+      "loss": 0.0153,
+      "step": 8284
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.0332387651643587e-05,
+      "loss": 0.0207,
+      "step": 8286
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.0328443825411276e-05,
+      "loss": 0.015,
+      "step": 8288
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.0324499948037753e-05,
+      "loss": 0.0144,
+      "step": 8290
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.0320556020137104e-05,
+      "loss": 0.0096,
+      "step": 8292
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.0316612042323438e-05,
+      "loss": 0.0229,
+      "step": 8294
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.0312668015210848e-05,
+      "loss": 0.0082,
+      "step": 8296
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.030872393941346e-05,
+      "loss": 0.0131,
+      "step": 8298
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.0304779815545396e-05,
+      "loss": 0.0146,
+      "step": 8300
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.0300835644220783e-05,
+      "loss": 0.016,
+      "step": 8302
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.0296891426053759e-05,
+      "loss": 0.0113,
+      "step": 8304
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.029294716165847e-05,
+      "loss": 0.0298,
+      "step": 8306
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.0289002851649068e-05,
+      "loss": 0.0168,
+      "step": 8308
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.0285058496639713e-05,
+      "loss": 0.0171,
+      "step": 8310
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.0281114097244568e-05,
+      "loss": 0.0072,
+      "step": 8312
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.0277169654077813e-05,
+      "loss": 0.0126,
+      "step": 8314
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.0273225167753618e-05,
+      "loss": 0.0237,
+      "step": 8316
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.0269280638886178e-05,
+      "loss": 0.0141,
+      "step": 8318
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.0265336068089682e-05,
+      "loss": 0.015,
+      "step": 8320
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.0261391455978335e-05,
+      "loss": 0.0087,
+      "step": 8322
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.0257446803166336e-05,
+      "loss": 0.0226,
+      "step": 8324
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.02535021102679e-05,
+      "loss": 0.0223,
+      "step": 8326
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.0249557377897251e-05,
+      "loss": 0.0152,
+      "step": 8328
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.0245612606668609e-05,
+      "loss": 0.0081,
+      "step": 8330
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.0241667797196202e-05,
+      "loss": 0.0234,
+      "step": 8332
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.0237722950094275e-05,
+      "loss": 0.0155,
+      "step": 8334
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.0233778065977067e-05,
+      "loss": 0.013,
+      "step": 8336
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.0229833145458825e-05,
+      "loss": 0.0254,
+      "step": 8338
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.0225888189153803e-05,
+      "loss": 0.0144,
+      "step": 8340
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.0221943197676265e-05,
+      "loss": 0.0224,
+      "step": 8342
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.0217998171640475e-05,
+      "loss": 0.0247,
+      "step": 8344
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.0214053111660702e-05,
+      "loss": 0.019,
+      "step": 8346
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.0210108018351222e-05,
+      "loss": 0.0122,
+      "step": 8348
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.020616289232632e-05,
+      "loss": 0.0255,
+      "step": 8350
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.0202217734200273e-05,
+      "loss": 0.0143,
+      "step": 8352
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.0198272544587382e-05,
+      "loss": 0.0308,
+      "step": 8354
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.019432732410194e-05,
+      "loss": 0.0123,
+      "step": 8356
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.019038207335825e-05,
+      "loss": 0.014,
+      "step": 8358
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.0186436792970608e-05,
+      "loss": 0.018,
+      "step": 8360
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.0182491483553339e-05,
+      "loss": 0.0066,
+      "step": 8362
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.0178546145720746e-05,
+      "loss": 0.0208,
+      "step": 8364
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.0174600780087155e-05,
+      "loss": 0.0116,
+      "step": 8366
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.0170655387266888e-05,
+      "loss": 0.0172,
+      "step": 8368
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.016670996787427e-05,
+      "loss": 0.0105,
+      "step": 8370
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.0162764522523638e-05,
+      "loss": 0.0158,
+      "step": 8372
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.0158819051829325e-05,
+      "loss": 0.0128,
+      "step": 8374
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.015487355640567e-05,
+      "loss": 0.008,
+      "step": 8376
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.0150928036867019e-05,
+      "loss": 0.0125,
+      "step": 8378
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.0146982493827717e-05,
+      "loss": 0.014,
+      "step": 8380
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.0143036927902116e-05,
+      "loss": 0.0122,
+      "step": 8382
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.0139091339704574e-05,
+      "loss": 0.0234,
+      "step": 8384
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.0135145729849448e-05,
+      "loss": 0.0164,
+      "step": 8386
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.0131200098951092e-05,
+      "loss": 0.0187,
+      "step": 8388
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.0127254447623885e-05,
+      "loss": 0.0103,
+      "step": 8390
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.0123308776482184e-05,
+      "loss": 0.0194,
+      "step": 8392
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.0119363086140363e-05,
+      "loss": 0.0128,
+      "step": 8394
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.0115417377212797e-05,
+      "loss": 0.0118,
+      "step": 8396
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.0111471650313862e-05,
+      "loss": 0.0204,
+      "step": 8398
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.0107525906057943e-05,
+      "loss": 0.0154,
+      "step": 8400
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.0103580145059415e-05,
+      "loss": 0.0073,
+      "step": 8402
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.0099634367932664e-05,
+      "loss": 0.024,
+      "step": 8404
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.0095688575292087e-05,
+      "loss": 0.0142,
+      "step": 8406
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.009174276775206e-05,
+      "loss": 0.0155,
+      "step": 8408
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.0087796945926984e-05,
+      "loss": 0.0136,
+      "step": 8410
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.0083851110431256e-05,
+      "loss": 0.0253,
+      "step": 8412
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.0079905261879265e-05,
+      "loss": 0.0132,
+      "step": 8414
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.0075959400885413e-05,
+      "loss": 0.0079,
+      "step": 8416
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.0072013528064107e-05,
+      "loss": 0.0107,
+      "step": 8418
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.006806764402974e-05,
+      "loss": 0.0183,
+      "step": 8420
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.006412174939672e-05,
+      "loss": 0.0249,
+      "step": 8422
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.0060175844779454e-05,
+      "loss": 0.0171,
+      "step": 8424
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.0056229930792349e-05,
+      "loss": 0.0104,
+      "step": 8426
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.0052284008049813e-05,
+      "loss": 0.0132,
+      "step": 8428
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.0048338077166259e-05,
+      "loss": 0.0159,
+      "step": 8430
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.0044392138756094e-05,
+      "loss": 0.0111,
+      "step": 8432
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.0040446193433737e-05,
+      "loss": 0.0086,
+      "step": 8434
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.0036500241813596e-05,
+      "loss": 0.0108,
+      "step": 8436
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.003255428451009e-05,
+      "loss": 0.0134,
+      "step": 8438
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.0028608322137632e-05,
+      "loss": 0.0162,
+      "step": 8440
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.002466235531064e-05,
+      "loss": 0.0192,
+      "step": 8442
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.0020716384643527e-05,
+      "loss": 0.0276,
+      "step": 8444
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.0016770410750718e-05,
+      "loss": 0.0091,
+      "step": 8446
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.0012824434246628e-05,
+      "loss": 0.012,
+      "step": 8448
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.0008878455745676e-05,
+      "loss": 0.0144,
+      "step": 8450
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.0004932475862277e-05,
+      "loss": 0.0139,
+      "step": 8452
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.0000986495210858e-05,
+      "loss": 0.0213,
+      "step": 8454
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.997040514405832e-06,
+      "loss": 0.0126,
+      "step": 8456
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.99309453406162e-06,
+      "loss": 0.0117,
+      "step": 8458
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.989148554792645e-06,
+      "loss": 0.0072,
+      "step": 8460
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.98520257721332e-06,
+      "loss": 0.0191,
+      "step": 8462
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.981256601938067e-06,
+      "loss": 0.0095,
+      "step": 8464
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.977310629581303e-06,
+      "loss": 0.0122,
+      "step": 8466
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.973364660757453e-06,
+      "loss": 0.0201,
+      "step": 8468
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.969418696080926e-06,
+      "loss": 0.0141,
+      "step": 8470
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.96547273616614e-06,
+      "loss": 0.0156,
+      "step": 8472
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.961526781627512e-06,
+      "loss": 0.0204,
+      "step": 8474
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.957580833079463e-06,
+      "loss": 0.0163,
+      "step": 8476
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.953634891136401e-06,
+      "loss": 0.013,
+      "step": 8478
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.949688956412739e-06,
+      "loss": 0.0126,
+      "step": 8480
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.945743029522893e-06,
+      "loss": 0.0136,
+      "step": 8482
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.941797111081273e-06,
+      "loss": 0.0182,
+      "step": 8484
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.937851201702285e-06,
+      "loss": 0.0131,
+      "step": 8486
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.933905302000341e-06,
+      "loss": 0.0114,
+      "step": 8488
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.929959412589848e-06,
+      "loss": 0.0122,
+      "step": 8490
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.926013534085212e-06,
+      "loss": 0.0173,
+      "step": 8492
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.922067667100829e-06,
+      "loss": 0.0238,
+      "step": 8494
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.91812181225111e-06,
+      "loss": 0.0137,
+      "step": 8496
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.914175970150449e-06,
+      "loss": 0.0096,
+      "step": 8498
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.910230141413249e-06,
+      "loss": 0.011,
+      "step": 8500
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.9062843266539e-06,
+      "loss": 0.0176,
+      "step": 8502
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.902338526486799e-06,
+      "loss": 0.0095,
+      "step": 8504
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.898392741526333e-06,
+      "loss": 0.0195,
+      "step": 8506
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.8944469723869e-06,
+      "loss": 0.0152,
+      "step": 8508
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.890501219682874e-06,
+      "loss": 0.0199,
+      "step": 8510
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.88655548402865e-06,
+      "loss": 0.0102,
+      "step": 8512
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.882609766038602e-06,
+      "loss": 0.0204,
+      "step": 8514
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.878664066327114e-06,
+      "loss": 0.021,
+      "step": 8516
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.874718385508555e-06,
+      "loss": 0.0066,
+      "step": 8518
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.870772724197304e-06,
+      "loss": 0.0185,
+      "step": 8520
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.866827083007725e-06,
+      "loss": 0.0136,
+      "step": 8522
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.862881462554192e-06,
+      "loss": 0.0178,
+      "step": 8524
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.858935863451062e-06,
+      "loss": 0.02,
+      "step": 8526
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.854990286312695e-06,
+      "loss": 0.0274,
+      "step": 8528
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.851044731753448e-06,
+      "loss": 0.0165,
+      "step": 8530
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.847099200387682e-06,
+      "loss": 0.0165,
+      "step": 8532
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.843153692829734e-06,
+      "loss": 0.0119,
+      "step": 8534
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.839208209693956e-06,
+      "loss": 0.0188,
+      "step": 8536
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.83526275159469e-06,
+      "loss": 0.0159,
+      "step": 8538
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.831317319146277e-06,
+      "loss": 0.015,
+      "step": 8540
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.827371912963042e-06,
+      "loss": 0.0186,
+      "step": 8542
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.82342653365932e-06,
+      "loss": 0.0152,
+      "step": 8544
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.819481181849439e-06,
+      "loss": 0.017,
+      "step": 8546
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.81553585814772e-06,
+      "loss": 0.0277,
+      "step": 8548
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.811590563168475e-06,
+      "loss": 0.0112,
+      "step": 8550
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.807645297526019e-06,
+      "loss": 0.0134,
+      "step": 8552
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.803700061834665e-06,
+      "loss": 0.0073,
+      "step": 8554
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.799754856708713e-06,
+      "loss": 0.0141,
+      "step": 8556
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.795809682762457e-06,
+      "loss": 0.0192,
+      "step": 8558
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.791864540610198e-06,
+      "loss": 0.0137,
+      "step": 8560
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.787919430866218e-06,
+      "loss": 0.02,
+      "step": 8562
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.78397435414481e-06,
+      "loss": 0.0134,
+      "step": 8564
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.780029311060241e-06,
+      "loss": 0.0227,
+      "step": 8566
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.776084302226794e-06,
+      "loss": 0.0155,
+      "step": 8568
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.77213932825873e-06,
+      "loss": 0.0134,
+      "step": 8570
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.76819438977032e-06,
+      "loss": 0.0065,
+      "step": 8572
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.764249487375812e-06,
+      "loss": 0.0187,
+      "step": 8574
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.760304621689462e-06,
+      "loss": 0.0065,
+      "step": 8576
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.756359793325515e-06,
+      "loss": 0.0164,
+      "step": 8578
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.752415002898214e-06,
+      "loss": 0.0157,
+      "step": 8580
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.74847025102179e-06,
+      "loss": 0.016,
+      "step": 8582
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.744525538310468e-06,
+      "loss": 0.0155,
+      "step": 8584
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.740580865378478e-06,
+      "loss": 0.0143,
+      "step": 8586
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.736636232840026e-06,
+      "loss": 0.016,
+      "step": 8588
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.732691641309328e-06,
+      "loss": 0.0172,
+      "step": 8590
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.728747091400585e-06,
+      "loss": 0.0132,
+      "step": 8592
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.724802583727996e-06,
+      "loss": 0.0129,
+      "step": 8594
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.720858118905747e-06,
+      "loss": 0.0088,
+      "step": 8596
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.716913697548025e-06,
+      "loss": 0.0079,
+      "step": 8598
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.712969320269002e-06,
+      "loss": 0.0179,
+      "step": 8600
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.709024987682855e-06,
+      "loss": 0.0148,
+      "step": 8602
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.705080700403741e-06,
+      "loss": 0.0107,
+      "step": 8604
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.701136459045814e-06,
+      "loss": 0.0142,
+      "step": 8606
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.697192264223226e-06,
+      "loss": 0.0241,
+      "step": 8608
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.693248116550123e-06,
+      "loss": 0.0194,
+      "step": 8610
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.689304016640631e-06,
+      "loss": 0.0123,
+      "step": 8612
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.685359965108878e-06,
+      "loss": 0.0185,
+      "step": 8614
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.681415962568986e-06,
+      "loss": 0.0157,
+      "step": 8616
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.677472009635064e-06,
+      "loss": 0.0108,
+      "step": 8618
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.673528106921218e-06,
+      "loss": 0.0156,
+      "step": 8620
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.669584255041538e-06,
+      "loss": 0.011,
+      "step": 8622
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.66564045461012e-06,
+      "loss": 0.0096,
+      "step": 8624
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.661696706241042e-06,
+      "loss": 0.0235,
+      "step": 8626
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.657753010548368e-06,
+      "loss": 0.0199,
+      "step": 8628
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.653809368146169e-06,
+      "loss": 0.0106,
+      "step": 8630
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.6498657796485e-06,
+      "loss": 0.0185,
+      "step": 8632
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.645922245669408e-06,
+      "loss": 0.0255,
+      "step": 8634
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.641978766822925e-06,
+      "loss": 0.0211,
+      "step": 8636
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.63803534372309e-06,
+      "loss": 0.0173,
+      "step": 8638
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.634091976983916e-06,
+      "loss": 0.0089,
+      "step": 8640
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.630148667219424e-06,
+      "loss": 0.012,
+      "step": 8642
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.62620541504361e-06,
+      "loss": 0.0126,
+      "step": 8644
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.62226222107047e-06,
+      "loss": 0.0112,
+      "step": 8646
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.61831908591399e-06,
+      "loss": 0.0138,
+      "step": 8648
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.614376010188151e-06,
+      "loss": 0.0267,
+      "step": 8650
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.610432994506912e-06,
+      "loss": 0.0218,
+      "step": 8652
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.606490039484237e-06,
+      "loss": 0.013,
+      "step": 8654
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.602547145734069e-06,
+      "loss": 0.0159,
+      "step": 8656
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.598604313870355e-06,
+      "loss": 0.0112,
+      "step": 8658
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.594661544507015e-06,
+      "loss": 0.0212,
+      "step": 8660
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.590718838257973e-06,
+      "loss": 0.013,
+      "step": 8662
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.586776195737133e-06,
+      "loss": 0.0105,
+      "step": 8664
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.582833617558406e-06,
+      "loss": 0.0126,
+      "step": 8666
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.578891104335673e-06,
+      "loss": 0.0173,
+      "step": 8668
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.574948656682813e-06,
+      "loss": 0.0141,
+      "step": 8670
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.571006275213695e-06,
+      "loss": 0.018,
+      "step": 8672
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.567063960542187e-06,
+      "loss": 0.0111,
+      "step": 8674
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.563121713282126e-06,
+      "loss": 0.0166,
+      "step": 8676
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.559179534047355e-06,
+      "loss": 0.012,
+      "step": 8678
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.555237423451704e-06,
+      "loss": 0.0143,
+      "step": 8680
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.551295382108988e-06,
+      "loss": 0.0153,
+      "step": 8682
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.547353410633009e-06,
+      "loss": 0.0134,
+      "step": 8684
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.543411509637564e-06,
+      "loss": 0.0213,
+      "step": 8686
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.53946967973644e-06,
+      "loss": 0.012,
+      "step": 8688
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.535527921543411e-06,
+      "loss": 0.0155,
+      "step": 8690
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.531586235672232e-06,
+      "loss": 0.0325,
+      "step": 8692
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.52764462273666e-06,
+      "loss": 0.0147,
+      "step": 8694
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.52370308335043e-06,
+      "loss": 0.0074,
+      "step": 8696
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.519761618127277e-06,
+      "loss": 0.0156,
+      "step": 8698
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.515820227680909e-06,
+      "loss": 0.0173,
+      "step": 8700
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.511878912625035e-06,
+      "loss": 0.0114,
+      "step": 8702
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.507937673573343e-06,
+      "loss": 0.0156,
+      "step": 8704
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.503996511139523e-06,
+      "loss": 0.017,
+      "step": 8706
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.500055425937235e-06,
+      "loss": 0.0145,
+      "step": 8708
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.496114418580144e-06,
+      "loss": 0.0165,
+      "step": 8710
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.492173489681888e-06,
+      "loss": 0.0136,
+      "step": 8712
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.488232639856106e-06,
+      "loss": 0.0187,
+      "step": 8714
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.484291869716414e-06,
+      "loss": 0.018,
+      "step": 8716
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.480351179876417e-06,
+      "loss": 0.0174,
+      "step": 8718
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.476410570949718e-06,
+      "loss": 0.0121,
+      "step": 8720
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.4724700435499e-06,
+      "loss": 0.0255,
+      "step": 8722
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.468529598290526e-06,
+      "loss": 0.0123,
+      "step": 8724
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.464589235785157e-06,
+      "loss": 0.0186,
+      "step": 8726
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.460648956647339e-06,
+      "loss": 0.0174,
+      "step": 8728
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.456708761490606e-06,
+      "loss": 0.0144,
+      "step": 8730
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.452768650928472e-06,
+      "loss": 0.0173,
+      "step": 8732
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.44882862557444e-06,
+      "loss": 0.0058,
+      "step": 8734
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.444888686042007e-06,
+      "loss": 0.017,
+      "step": 8736
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.440948832944654e-06,
+      "loss": 0.0162,
+      "step": 8738
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.437009066895837e-06,
+      "loss": 0.0137,
+      "step": 8740
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.433069388509015e-06,
+      "loss": 0.0128,
+      "step": 8742
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.429129798397625e-06,
+      "loss": 0.0152,
+      "step": 8744
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.425190297175096e-06,
+      "loss": 0.0218,
+      "step": 8746
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.421250885454825e-06,
+      "loss": 0.012,
+      "step": 8748
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.417311563850218e-06,
+      "loss": 0.0131,
+      "step": 8750
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.41337233297466e-06,
+      "loss": 0.0191,
+      "step": 8752
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.409433193441516e-06,
+      "loss": 0.0172,
+      "step": 8754
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.405494145864134e-06,
+      "loss": 0.014,
+      "step": 8756
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.401555190855863e-06,
+      "loss": 0.013,
+      "step": 8758
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.397616329030023e-06,
+      "loss": 0.0126,
+      "step": 8760
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.39367756099993e-06,
+      "loss": 0.0152,
+      "step": 8762
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.389738887378873e-06,
+      "loss": 0.0269,
+      "step": 8764
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.385800308780141e-06,
+      "loss": 0.0092,
+      "step": 8766
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.381861825816995e-06,
+      "loss": 0.012,
+      "step": 8768
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.377923439102693e-06,
+      "loss": 0.015,
+      "step": 8770
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.373985149250464e-06,
+      "loss": 0.0181,
+      "step": 8772
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.370046956873536e-06,
+      "loss": 0.0339,
+      "step": 8774
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.366108862585113e-06,
+      "loss": 0.0125,
+      "step": 8776
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.362170866998391e-06,
+      "loss": 0.0288,
+      "step": 8778
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.358232970726541e-06,
+      "loss": 0.0241,
+      "step": 8780
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.354295174382721e-06,
+      "loss": 0.006,
+      "step": 8782
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.350357478580082e-06,
+      "loss": 0.0203,
+      "step": 8784
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.346419883931752e-06,
+      "loss": 0.0129,
+      "step": 8786
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.342482391050844e-06,
+      "loss": 0.0073,
+      "step": 8788
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.338545000550453e-06,
+      "loss": 0.0185,
+      "step": 8790
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.334607713043666e-06,
+      "loss": 0.0166,
+      "step": 8792
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.330670529143545e-06,
+      "loss": 0.0145,
+      "step": 8794
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.326733449463143e-06,
+      "loss": 0.0123,
+      "step": 8796
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.322796474615487e-06,
+      "loss": 0.013,
+      "step": 8798
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.3188596052136e-06,
+      "loss": 0.0127,
+      "step": 8800
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.314922841870484e-06,
+      "loss": 0.0127,
+      "step": 8802
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.310986185199115e-06,
+      "loss": 0.0096,
+      "step": 8804
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.307049635812466e-06,
+      "loss": 0.0285,
+      "step": 8806
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.303113194323491e-06,
+      "loss": 0.0219,
+      "step": 8808
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.29917686134512e-06,
+      "loss": 0.0225,
+      "step": 8810
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.295240637490269e-06,
+      "loss": 0.025,
+      "step": 8812
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.29130452337184e-06,
+      "loss": 0.0174,
+      "step": 8814
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.287368519602715e-06,
+      "loss": 0.0128,
+      "step": 8816
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.283432626795764e-06,
+      "loss": 0.0231,
+      "step": 8818
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.279496845563828e-06,
+      "loss": 0.0112,
+      "step": 8820
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.275561176519747e-06,
+      "loss": 0.0259,
+      "step": 8822
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.271625620276326e-06,
+      "loss": 0.019,
+      "step": 8824
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.26769017744637e-06,
+      "loss": 0.0207,
+      "step": 8826
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.263754848642649e-06,
+      "loss": 0.0091,
+      "step": 8828
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.25981963447793e-06,
+      "loss": 0.0089,
+      "step": 8830
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.255884535564952e-06,
+      "loss": 0.0198,
+      "step": 8832
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.251949552516447e-06,
+      "loss": 0.0227,
+      "step": 8834
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.248014685945113e-06,
+      "loss": 0.012,
+      "step": 8836
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.24407993646364e-06,
+      "loss": 0.0074,
+      "step": 8838
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.240145304684701e-06,
+      "loss": 0.0172,
+      "step": 8840
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.236210791220956e-06,
+      "loss": 0.0239,
+      "step": 8842
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.232276396685025e-06,
+      "loss": 0.0112,
+      "step": 8844
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.228342121689531e-06,
+      "loss": 0.0156,
+      "step": 8846
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.224407966847067e-06,
+      "loss": 0.0159,
+      "step": 8848
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.220473932770217e-06,
+      "loss": 0.0163,
+      "step": 8850
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.216540020071537e-06,
+      "loss": 0.0195,
+      "step": 8852
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.212606229363563e-06,
+      "loss": 0.0209,
+      "step": 8854
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.20867256125882e-06,
+      "loss": 0.0172,
+      "step": 8856
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.204739016369817e-06,
+      "loss": 0.0261,
+      "step": 8858
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.200805595309023e-06,
+      "loss": 0.0099,
+      "step": 8860
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.196872298688908e-06,
+      "loss": 0.0131,
+      "step": 8862
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.192939127121922e-06,
+      "loss": 0.0156,
+      "step": 8864
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.189006081220486e-06,
+      "loss": 0.0161,
+      "step": 8866
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.185073161597e-06,
+      "loss": 0.0178,
+      "step": 8868
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.181140368863857e-06,
+      "loss": 0.0088,
+      "step": 8870
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.177207703633417e-06,
+      "loss": 0.0244,
+      "step": 8872
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.173275166518035e-06,
+      "loss": 0.014,
+      "step": 8874
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.169342758130026e-06,
+      "loss": 0.0125,
+      "step": 8876
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.165410479081704e-06,
+      "loss": 0.014,
+      "step": 8878
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.161478329985349e-06,
+      "loss": 0.0164,
+      "step": 8880
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.157546311453235e-06,
+      "loss": 0.0166,
+      "step": 8882
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.153614424097597e-06,
+      "loss": 0.0131,
+      "step": 8884
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.14968266853067e-06,
+      "loss": 0.0255,
+      "step": 8886
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.145751045364648e-06,
+      "loss": 0.0166,
+      "step": 8888
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.141819555211728e-06,
+      "loss": 0.0094,
+      "step": 8890
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.137888198684061e-06,
+      "loss": 0.0128,
+      "step": 8892
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.133956976393791e-06,
+      "loss": 0.0218,
+      "step": 8894
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.130025888953044e-06,
+      "loss": 0.0141,
+      "step": 8896
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.126094936973922e-06,
+      "loss": 0.0085,
+      "step": 8898
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.122164121068498e-06,
+      "loss": 0.0211,
+      "step": 8900
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.118233441848832e-06,
+      "loss": 0.0129,
+      "step": 8902
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.11430289992696e-06,
+      "loss": 0.0174,
+      "step": 8904
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.110372495914906e-06,
+      "loss": 0.0088,
+      "step": 8906
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.106442230424654e-06,
+      "loss": 0.0166,
+      "step": 8908
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.102512104068175e-06,
+      "loss": 0.0204,
+      "step": 8910
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.098582117457429e-06,
+      "loss": 0.0136,
+      "step": 8912
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.09465227120434e-06,
+      "loss": 0.0101,
+      "step": 8914
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.090722565920813e-06,
+      "loss": 0.0125,
+      "step": 8916
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.086793002218735e-06,
+      "loss": 0.0211,
+      "step": 8918
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.08286358070997e-06,
+      "loss": 0.0104,
+      "step": 8920
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.078934302006363e-06,
+      "loss": 0.0131,
+      "step": 8922
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.075005166719724e-06,
+      "loss": 0.0114,
+      "step": 8924
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.071076175461851e-06,
+      "loss": 0.0162,
+      "step": 8926
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.067147328844526e-06,
+      "loss": 0.0186,
+      "step": 8928
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.063218627479494e-06,
+      "loss": 0.0189,
+      "step": 8930
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.059290071978482e-06,
+      "loss": 0.0134,
+      "step": 8932
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.055361662953202e-06,
+      "loss": 0.0118,
+      "step": 8934
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.051433401015333e-06,
+      "loss": 0.0183,
+      "step": 8936
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.04750528677654e-06,
+      "loss": 0.0157,
+      "step": 8938
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.043577320848452e-06,
+      "loss": 0.0164,
+      "step": 8940
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.039649503842691e-06,
+      "loss": 0.0108,
+      "step": 8942
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.035721836370845e-06,
+      "loss": 0.01,
+      "step": 8944
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.031794319044486e-06,
+      "loss": 0.0088,
+      "step": 8946
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.02786695247515e-06,
+      "loss": 0.015,
+      "step": 8948
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.023939737274366e-06,
+      "loss": 0.0224,
+      "step": 8950
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.020012674053627e-06,
+      "loss": 0.0158,
+      "step": 8952
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.016085763424411e-06,
+      "loss": 0.0248,
+      "step": 8954
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.012159005998165e-06,
+      "loss": 0.0136,
+      "step": 8956
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.00823240238631e-06,
+      "loss": 0.0182,
+      "step": 8958
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.004305953200258e-06,
+      "loss": 0.0144,
+      "step": 8960
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.000379659051383e-06,
+      "loss": 0.0132,
+      "step": 8962
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.99645352055104e-06,
+      "loss": 0.0215,
+      "step": 8964
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 8.992527538310554e-06,
+      "loss": 0.0095,
+      "step": 8966
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.988601712941237e-06,
+      "loss": 0.0113,
+      "step": 8968
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.984676045054365e-06,
+      "loss": 0.0118,
+      "step": 8970
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.980750535261199e-06,
+      "loss": 0.0148,
+      "step": 8972
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.976825184172966e-06,
+      "loss": 0.0191,
+      "step": 8974
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.972899992400878e-06,
+      "loss": 0.0156,
+      "step": 8976
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.968974960556117e-06,
+      "loss": 0.0088,
+      "step": 8978
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.965050089249832e-06,
+      "loss": 0.0155,
+      "step": 8980
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.961125379093165e-06,
+      "loss": 0.0198,
+      "step": 8982
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.95720083069722e-06,
+      "loss": 0.0202,
+      "step": 8984
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.953276444673084e-06,
+      "loss": 0.0103,
+      "step": 8986
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.949352221631805e-06,
+      "loss": 0.0183,
+      "step": 8988
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.94542816218442e-06,
+      "loss": 0.0138,
+      "step": 8990
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.941504266941932e-06,
+      "loss": 0.011,
+      "step": 8992
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.937580536515328e-06,
+      "loss": 0.0115,
+      "step": 8994
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.933656971515554e-06,
+      "loss": 0.0231,
+      "step": 8996
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.929733572553546e-06,
+      "loss": 0.0164,
+      "step": 8998
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.925810340240203e-06,
+      "loss": 0.0102,
+      "step": 9000
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.921887275186408e-06,
+      "loss": 0.016,
+      "step": 9002
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.917964378003003e-06,
+      "loss": 0.0266,
+      "step": 9004
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.914041649300822e-06,
+      "loss": 0.0162,
+      "step": 9006
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 8.910119089690657e-06,
+      "loss": 0.0125,
+      "step": 9008
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.906196699783286e-06,
+      "loss": 0.0257,
+      "step": 9010
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.902274480189454e-06,
+      "loss": 0.0151,
+      "step": 9012
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.898352431519876e-06,
+      "loss": 0.009,
+      "step": 9014
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.894430554385249e-06,
+      "loss": 0.0126,
+      "step": 9016
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.890508849396243e-06,
+      "loss": 0.0139,
+      "step": 9018
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.88658731716349e-06,
+      "loss": 0.0221,
+      "step": 9020
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.882665958297603e-06,
+      "loss": 0.0121,
+      "step": 9022
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.878744773409171e-06,
+      "loss": 0.0136,
+      "step": 9024
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.874823763108755e-06,
+      "loss": 0.0203,
+      "step": 9026
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.870902928006882e-06,
+      "loss": 0.0075,
+      "step": 9028
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.866982268714054e-06,
+      "loss": 0.0207,
+      "step": 9030
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.863061785840753e-06,
+      "loss": 0.0118,
+      "step": 9032
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.859141479997427e-06,
+      "loss": 0.0199,
+      "step": 9034
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.855221351794493e-06,
+      "loss": 0.0201,
+      "step": 9036
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.851301401842348e-06,
+      "loss": 0.0138,
+      "step": 9038
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.84738163075136e-06,
+      "loss": 0.024,
+      "step": 9040
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.843462039131869e-06,
+      "loss": 0.0081,
+      "step": 9042
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.839542627594177e-06,
+      "loss": 0.0243,
+      "step": 9044
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.835623396748574e-06,
+      "loss": 0.0228,
+      "step": 9046
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 8.83170434720531e-06,
+      "loss": 0.0102,
+      "step": 9048
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.827785479574616e-06,
+      "loss": 0.0072,
+      "step": 9050
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.823866794466683e-06,
+      "loss": 0.0144,
+      "step": 9052
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.819948292491686e-06,
+      "loss": 0.0083,
+      "step": 9054
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.81602997425976e-06,
+      "loss": 0.0149,
+      "step": 9056
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.812111840381025e-06,
+      "loss": 0.005,
+      "step": 9058
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.808193891465555e-06,
+      "loss": 0.0089,
+      "step": 9060
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.804276128123412e-06,
+      "loss": 0.0104,
+      "step": 9062
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.800358550964618e-06,
+      "loss": 0.0129,
+      "step": 9064
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.796441160599175e-06,
+      "loss": 0.0119,
+      "step": 9066
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.792523957637043e-06,
+      "loss": 0.0098,
+      "step": 9068
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.788606942688163e-06,
+      "loss": 0.0156,
+      "step": 9070
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.784690116362444e-06,
+      "loss": 0.0075,
+      "step": 9072
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.780773479269773e-06,
+      "loss": 0.0176,
+      "step": 9074
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.776857032019991e-06,
+      "loss": 0.017,
+      "step": 9076
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.772940775222918e-06,
+      "loss": 0.0137,
+      "step": 9078
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.769024709488353e-06,
+      "loss": 0.0114,
+      "step": 9080
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.765108835426057e-06,
+      "loss": 0.0167,
+      "step": 9082
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.761193153645753e-06,
+      "loss": 0.014,
+      "step": 9084
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.757277664757148e-06,
+      "loss": 0.0091,
+      "step": 9086
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.753362369369915e-06,
+      "loss": 0.0123,
+      "step": 9088
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 8.749447268093697e-06,
+      "loss": 0.0146,
+      "step": 9090
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.745532361538094e-06,
+      "loss": 0.0148,
+      "step": 9092
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.741617650312699e-06,
+      "loss": 0.0202,
+      "step": 9094
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.737703135027058e-06,
+      "loss": 0.0158,
+      "step": 9096
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.733788816290694e-06,
+      "loss": 0.0101,
+      "step": 9098
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.729874694713089e-06,
+      "loss": 0.0253,
+      "step": 9100
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.725960770903706e-06,
+      "loss": 0.017,
+      "step": 9102
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.722047045471977e-06,
+      "loss": 0.0148,
+      "step": 9104
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.718133519027295e-06,
+      "loss": 0.0163,
+      "step": 9106
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.714220192179023e-06,
+      "loss": 0.0191,
+      "step": 9108
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.7103070655365e-06,
+      "loss": 0.009,
+      "step": 9110
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.706394139709032e-06,
+      "loss": 0.0096,
+      "step": 9112
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.702481415305885e-06,
+      "loss": 0.0175,
+      "step": 9114
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.698568892936301e-06,
+      "loss": 0.022,
+      "step": 9116
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.694656573209494e-06,
+      "loss": 0.0122,
+      "step": 9118
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.690744456734644e-06,
+      "loss": 0.0156,
+      "step": 9120
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.68683254412089e-06,
+      "loss": 0.0205,
+      "step": 9122
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.682920835977347e-06,
+      "loss": 0.018,
+      "step": 9124
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.679009332913106e-06,
+      "loss": 0.0079,
+      "step": 9126
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.675098035537214e-06,
+      "loss": 0.014,
+      "step": 9128
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 8.671186944458685e-06,
+      "loss": 0.0166,
+      "step": 9130
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.667276060286511e-06,
+      "loss": 0.0133,
+      "step": 9132
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.663365383629643e-06,
+      "loss": 0.0172,
+      "step": 9134
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.659454915097011e-06,
+      "loss": 0.0172,
+      "step": 9136
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.655544655297494e-06,
+      "loss": 0.0196,
+      "step": 9138
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.651634604839958e-06,
+      "loss": 0.0117,
+      "step": 9140
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.647724764333223e-06,
+      "loss": 0.012,
+      "step": 9142
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.643815134386085e-06,
+      "loss": 0.012,
+      "step": 9144
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.6399057156073e-06,
+      "loss": 0.0103,
+      "step": 9146
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.635996508605593e-06,
+      "loss": 0.008,
+      "step": 9148
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.63208751398966e-06,
+      "loss": 0.0147,
+      "step": 9150
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.628178732368164e-06,
+      "loss": 0.0117,
+      "step": 9152
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.624270164349726e-06,
+      "loss": 0.0149,
+      "step": 9154
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.620361810542944e-06,
+      "loss": 0.0156,
+      "step": 9156
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.616453671556375e-06,
+      "loss": 0.0113,
+      "step": 9158
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.612545747998555e-06,
+      "loss": 0.0112,
+      "step": 9160
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.60863804047797e-06,
+      "loss": 0.0132,
+      "step": 9162
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.604730549603077e-06,
+      "loss": 0.0187,
+      "step": 9164
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.60082327598231e-06,
+      "loss": 0.0096,
+      "step": 9166
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.59691622022406e-06,
+      "loss": 0.006,
+      "step": 9168
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.593009382936679e-06,
+      "loss": 0.0238,
+      "step": 9170
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 8.589102764728495e-06,
+      "loss": 0.0159,
+      "step": 9172
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 8.585196366207803e-06,
+      "loss": 0.026,
+      "step": 9174
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 8.581290187982855e-06,
+      "loss": 0.0217,
+      "step": 9176
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 8.57738423066187e-06,
+      "loss": 0.014,
+      "step": 9178
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 8.57347849485304e-06,
+      "loss": 0.0229,
+      "step": 9180
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 8.569572981164516e-06,
+      "loss": 0.0201,
+      "step": 9182
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 8.56566769020442e-06,
+      "loss": 0.0182,
+      "step": 9184
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 8.56176262258083e-06,
+      "loss": 0.0146,
+      "step": 9186
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 8.557857778901798e-06,
+      "loss": 0.0101,
+      "step": 9188
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 8.553953159775335e-06,
+      "loss": 0.0198,
+      "step": 9190
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 8.550048765809427e-06,
+      "loss": 0.0132,
+      "step": 9192
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 8.54614459761201e-06,
+      "loss": 0.0035,
+      "step": 9194
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 8.542240655790997e-06,
+      "loss": 0.0085,
+      "step": 9196
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 8.538336940954259e-06,
+      "loss": 0.0122,
+      "step": 9198
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 8.534433453709642e-06,
+      "loss": 0.0296,
+      "step": 9200
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 8.530530194664937e-06,
+      "loss": 0.0117,
+      "step": 9202
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 8.526627164427919e-06,
+      "loss": 0.0095,
+      "step": 9204
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 8.522724363606317e-06,
+      "loss": 0.0139,
+      "step": 9206
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 8.518821792807832e-06,
+      "loss": 0.0198,
+      "step": 9208
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 8.514919452640117e-06,
+      "loss": 0.0138,
+      "step": 9210
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 8.511017343710796e-06,
+      "loss": 0.0154,
+      "step": 9212
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 8.507115466627462e-06,
+      "loss": 0.0152,
+      "step": 9214
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 8.50321382199767e-06,
+      "loss": 0.0068,
+      "step": 9216
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 8.499312410428928e-06,
+      "loss": 0.0157,
+      "step": 9218
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 8.495411232528718e-06,
+      "loss": 0.0143,
+      "step": 9220
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 8.491510288904485e-06,
+      "loss": 0.0238,
+      "step": 9222
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 8.487609580163636e-06,
+      "loss": 0.0176,
+      "step": 9224
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 8.48370910691354e-06,
+      "loss": 0.0176,
+      "step": 9226
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 8.479808869761528e-06,
+      "loss": 0.0179,
+      "step": 9228
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 8.475908869314901e-06,
+      "loss": 0.0164,
+      "step": 9230
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 8.47200910618092e-06,
+      "loss": 0.0071,
+      "step": 9232
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 8.4681095809668e-06,
+      "loss": 0.0126,
+      "step": 9234
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 8.464210294279734e-06,
+      "loss": 0.0111,
+      "step": 9236
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 8.46031124672687e-06,
+      "loss": 0.0146,
+      "step": 9238
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 8.45641243891532e-06,
+      "loss": 0.0155,
+      "step": 9240
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 8.452513871452153e-06,
+      "loss": 0.0194,
+      "step": 9242
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 8.448615544944412e-06,
+      "loss": 0.0065,
+      "step": 9244
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 8.44471745999909e-06,
+      "loss": 0.009,
+      "step": 9246
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 8.440819617223157e-06,
+      "loss": 0.0129,
+      "step": 9248
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 8.436922017223528e-06,
+      "loss": 0.0103,
+      "step": 9250
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 8.433024660607094e-06,
+      "loss": 0.0141,
+      "step": 9252
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 8.429127547980702e-06,
+      "loss": 0.009,
+      "step": 9254
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 8.425230679951166e-06,
+      "loss": 0.0178,
+      "step": 9256
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 8.42133405712525e-06,
+      "loss": 0.0091,
+      "step": 9258
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 8.417437680109697e-06,
+      "loss": 0.015,
+      "step": 9260
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 8.413541549511195e-06,
+      "loss": 0.0134,
+      "step": 9262
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 8.40964566593641e-06,
+      "loss": 0.0182,
+      "step": 9264
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 8.405750029991953e-06,
+      "loss": 0.0218,
+      "step": 9266
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 8.401854642284405e-06,
+      "loss": 0.0159,
+      "step": 9268
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 8.39795950342031e-06,
+      "loss": 0.0115,
+      "step": 9270
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 8.394064614006174e-06,
+      "loss": 0.0147,
+      "step": 9272
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 8.390169974648456e-06,
+      "loss": 0.0139,
+      "step": 9274
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 8.38627558595358e-06,
+      "loss": 0.0148,
+      "step": 9276
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 8.382381448527934e-06,
+      "loss": 0.017,
+      "step": 9278
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 8.378487562977872e-06,
+      "loss": 0.0147,
+      "step": 9280
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 8.374593929909694e-06,
+      "loss": 0.0081,
+      "step": 9282
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 8.370700549929667e-06,
+      "loss": 0.0156,
+      "step": 9284
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 8.366807423644025e-06,
+      "loss": 0.0098,
+      "step": 9286
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 8.362914551658958e-06,
+      "loss": 0.0168,
+      "step": 9288
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 8.359021934580607e-06,
+      "loss": 0.0082,
+      "step": 9290
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 8.355129573015092e-06,
+      "loss": 0.0168,
+      "step": 9292
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 8.35123746756848e-06,
+      "loss": 0.0085,
+      "step": 9294
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 8.347345618846807e-06,
+      "loss": 0.0146,
+      "step": 9296
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 8.343454027456051e-06,
+      "loss": 0.0139,
+      "step": 9298
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 8.339562694002171e-06,
+      "loss": 0.0111,
+      "step": 9300
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 8.33567161909108e-06,
+      "loss": 0.0099,
+      "step": 9302
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 8.331780803328643e-06,
+      "loss": 0.0134,
+      "step": 9304
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 8.32789024732069e-06,
+      "loss": 0.0132,
+      "step": 9306
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 8.323999951673013e-06,
+      "loss": 0.0239,
+      "step": 9308
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 8.320109916991357e-06,
+      "loss": 0.0204,
+      "step": 9310
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 8.316220143881438e-06,
+      "loss": 0.0128,
+      "step": 9312
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 8.312330632948911e-06,
+      "loss": 0.0132,
+      "step": 9314
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 8.308441384799413e-06,
+      "loss": 0.0197,
+      "step": 9316
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 8.304552400038523e-06,
+      "loss": 0.0251,
+      "step": 9318
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 8.300663679271794e-06,
+      "loss": 0.0129,
+      "step": 9320
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 8.296775223104722e-06,
+      "loss": 0.0171,
+      "step": 9322
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 8.292887032142772e-06,
+      "loss": 0.0094,
+      "step": 9324
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 8.288999106991363e-06,
+      "loss": 0.0074,
+      "step": 9326
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 8.28511144825588e-06,
+      "loss": 0.0112,
+      "step": 9328
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 8.281224056541655e-06,
+      "loss": 0.0143,
+      "step": 9330
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 8.277336932453985e-06,
+      "loss": 0.0112,
+      "step": 9332
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 8.27345007659813e-06,
+      "loss": 0.0205,
+      "step": 9334
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 8.269563489579302e-06,
+      "loss": 0.0104,
+      "step": 9336
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 8.265677172002669e-06,
+      "loss": 0.0109,
+      "step": 9338
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 8.261791124473362e-06,
+      "loss": 0.0121,
+      "step": 9340
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 8.257905347596468e-06,
+      "loss": 0.0133,
+      "step": 9342
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 8.254019841977036e-06,
+      "loss": 0.0299,
+      "step": 9344
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 8.25013460822006e-06,
+      "loss": 0.0064,
+      "step": 9346
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 8.246249646930507e-06,
+      "loss": 0.0174,
+      "step": 9348
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 8.242364958713297e-06,
+      "loss": 0.0208,
+      "step": 9350
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 8.238480544173304e-06,
+      "loss": 0.0113,
+      "step": 9352
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 8.234596403915357e-06,
+      "loss": 0.0244,
+      "step": 9354
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 8.23071253854425e-06,
+      "loss": 0.0177,
+      "step": 9356
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 8.22682894866473e-06,
+      "loss": 0.0095,
+      "step": 9358
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 8.222945634881504e-06,
+      "loss": 0.0213,
+      "step": 9360
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 8.219062597799227e-06,
+      "loss": 0.027,
+      "step": 9362
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 8.215179838022524e-06,
+      "loss": 0.0194,
+      "step": 9364
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 8.211297356155966e-06,
+      "loss": 0.0125,
+      "step": 9366
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 8.207415152804091e-06,
+      "loss": 0.0077,
+      "step": 9368
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 8.20353322857138e-06,
+      "loss": 0.0225,
+      "step": 9370
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 8.199651584062285e-06,
+      "loss": 0.0196,
+      "step": 9372
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 8.195770219881203e-06,
+      "loss": 0.0177,
+      "step": 9374
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 8.191889136632498e-06,
+      "loss": 0.022,
+      "step": 9376
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 8.188008334920475e-06,
+      "loss": 0.0162,
+      "step": 9378
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 8.184127815349415e-06,
+      "loss": 0.0109,
+      "step": 9380
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 8.180247578523535e-06,
+      "loss": 0.0177,
+      "step": 9382
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 8.176367625047026e-06,
+      "loss": 0.0119,
+      "step": 9384
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 8.172487955524022e-06,
+      "loss": 0.0108,
+      "step": 9386
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 8.168608570558614e-06,
+      "loss": 0.015,
+      "step": 9388
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 8.164729470754855e-06,
+      "loss": 0.0097,
+      "step": 9390
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 8.160850656716757e-06,
+      "loss": 0.0104,
+      "step": 9392
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 8.156972129048273e-06,
+      "loss": 0.0126,
+      "step": 9394
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 8.153093888353318e-06,
+      "loss": 0.0094,
+      "step": 9396
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 8.14921593523577e-06,
+      "loss": 0.015,
+      "step": 9398
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 8.145338270299454e-06,
+      "loss": 0.0091,
+      "step": 9400
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 8.141460894148149e-06,
+      "loss": 0.0176,
+      "step": 9402
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 8.137583807385595e-06,
+      "loss": 0.0122,
+      "step": 9404
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 8.133707010615485e-06,
+      "loss": 0.0136,
+      "step": 9406
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 8.129830504441466e-06,
+      "loss": 0.0128,
+      "step": 9408
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 8.125954289467132e-06,
+      "loss": 0.0161,
+      "step": 9410
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 8.122078366296048e-06,
+      "loss": 0.015,
+      "step": 9412
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 8.118202735531724e-06,
+      "loss": 0.0145,
+      "step": 9414
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 8.114327397777624e-06,
+      "loss": 0.0213,
+      "step": 9416
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 8.110452353637165e-06,
+      "loss": 0.0215,
+      "step": 9418
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 8.106577603713724e-06,
+      "loss": 0.0125,
+      "step": 9420
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 8.102703148610627e-06,
+      "loss": 0.0101,
+      "step": 9422
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 8.098828988931162e-06,
+      "loss": 0.0123,
+      "step": 9424
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 8.094955125278555e-06,
+      "loss": 0.0113,
+      "step": 9426
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 8.091081558256006e-06,
+      "loss": 0.0085,
+      "step": 9428
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 8.087208288466653e-06,
+      "loss": 0.0138,
+      "step": 9430
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 8.0833353165136e-06,
+      "loss": 0.0117,
+      "step": 9432
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 8.07946264299989e-06,
+      "loss": 0.0185,
+      "step": 9434
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 8.075590268528535e-06,
+      "loss": 0.0208,
+      "step": 9436
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 8.071718193702486e-06,
+      "loss": 0.0065,
+      "step": 9438
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 8.067846419124665e-06,
+      "loss": 0.0176,
+      "step": 9440
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 8.06397494539793e-06,
+      "loss": 0.0116,
+      "step": 9442
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 8.060103773125097e-06,
+      "loss": 0.0127,
+      "step": 9444
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 8.056232902908941e-06,
+      "loss": 0.0172,
+      "step": 9446
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 8.052362335352189e-06,
+      "loss": 0.018,
+      "step": 9448
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 8.048492071057512e-06,
+      "loss": 0.0145,
+      "step": 9450
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 8.044622110627539e-06,
+      "loss": 0.0108,
+      "step": 9452
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 8.040752454664855e-06,
+      "loss": 0.0137,
+      "step": 9454
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 8.036883103772001e-06,
+      "loss": 0.0173,
+      "step": 9456
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 8.033014058551458e-06,
+      "loss": 0.0145,
+      "step": 9458
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 8.029145319605664e-06,
+      "loss": 0.0145,
+      "step": 9460
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 8.025276887537016e-06,
+      "loss": 0.0199,
+      "step": 9462
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 8.021408762947859e-06,
+      "loss": 0.0123,
+      "step": 9464
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 8.017540946440482e-06,
+      "loss": 0.0148,
+      "step": 9466
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 8.01367343861714e-06,
+      "loss": 0.0122,
+      "step": 9468
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 8.009806240080034e-06,
+      "loss": 0.0135,
+      "step": 9470
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 8.005939351431316e-06,
+      "loss": 0.0161,
+      "step": 9472
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 8.002072773273087e-06,
+      "loss": 0.0202,
+      "step": 9474
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.998206506207402e-06,
+      "loss": 0.0211,
+      "step": 9476
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.994340550836275e-06,
+      "loss": 0.0059,
+      "step": 9478
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.990474907761664e-06,
+      "loss": 0.0134,
+      "step": 9480
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.986609577585469e-06,
+      "loss": 0.0088,
+      "step": 9482
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.98274456090956e-06,
+      "loss": 0.0142,
+      "step": 9484
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.978879858335748e-06,
+      "loss": 0.0106,
+      "step": 9486
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.9750154704658e-06,
+      "loss": 0.0182,
+      "step": 9488
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.971151397901425e-06,
+      "loss": 0.019,
+      "step": 9490
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.967287641244292e-06,
+      "loss": 0.0097,
+      "step": 9492
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.963424201096014e-06,
+      "loss": 0.0113,
+      "step": 9494
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.959561078058168e-06,
+      "loss": 0.0183,
+      "step": 9496
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.955698272732261e-06,
+      "loss": 0.0228,
+      "step": 9498
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 7.95183578571976e-06,
+      "loss": 0.0149,
+      "step": 9500
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.947973617622093e-06,
+      "loss": 0.0146,
+      "step": 9502
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.944111769040628e-06,
+      "loss": 0.006,
+      "step": 9504
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.940250240576679e-06,
+      "loss": 0.0257,
+      "step": 9506
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.936389032831514e-06,
+      "loss": 0.0198,
+      "step": 9508
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.932528146406357e-06,
+      "loss": 0.0162,
+      "step": 9510
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.928667581902382e-06,
+      "loss": 0.0113,
+      "step": 9512
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.924807339920701e-06,
+      "loss": 0.0226,
+      "step": 9514
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.920947421062383e-06,
+      "loss": 0.0074,
+      "step": 9516
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.91708782592845e-06,
+      "loss": 0.0139,
+      "step": 9518
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.913228555119875e-06,
+      "loss": 0.0107,
+      "step": 9520
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.909369609237564e-06,
+      "loss": 0.0153,
+      "step": 9522
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.905510988882392e-06,
+      "loss": 0.0057,
+      "step": 9524
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.901652694655177e-06,
+      "loss": 0.0184,
+      "step": 9526
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.897794727156685e-06,
+      "loss": 0.0128,
+      "step": 9528
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.893937086987626e-06,
+      "loss": 0.0153,
+      "step": 9530
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.890079774748665e-06,
+      "loss": 0.0091,
+      "step": 9532
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.886222791040418e-06,
+      "loss": 0.0156,
+      "step": 9534
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.88236613646345e-06,
+      "loss": 0.0139,
+      "step": 9536
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.878509811618263e-06,
+      "loss": 0.0112,
+      "step": 9538
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 7.874653817105322e-06,
+      "loss": 0.0147,
+      "step": 9540
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.87079815352503e-06,
+      "loss": 0.0065,
+      "step": 9542
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.866942821477754e-06,
+      "loss": 0.0136,
+      "step": 9544
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.863087821563785e-06,
+      "loss": 0.0099,
+      "step": 9546
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.859233154383386e-06,
+      "loss": 0.0071,
+      "step": 9548
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.855378820536751e-06,
+      "loss": 0.0076,
+      "step": 9550
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.851524820624039e-06,
+      "loss": 0.0191,
+      "step": 9552
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.847671155245334e-06,
+      "loss": 0.0099,
+      "step": 9554
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.843817825000693e-06,
+      "loss": 0.0131,
+      "step": 9556
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.839964830490102e-06,
+      "loss": 0.0197,
+      "step": 9558
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.836112172313505e-06,
+      "loss": 0.017,
+      "step": 9560
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.83225985107079e-06,
+      "loss": 0.0148,
+      "step": 9562
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.828407867361789e-06,
+      "loss": 0.0177,
+      "step": 9564
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.82455622178629e-06,
+      "loss": 0.0087,
+      "step": 9566
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.820704914944024e-06,
+      "loss": 0.0184,
+      "step": 9568
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.816853947434666e-06,
+      "loss": 0.0102,
+      "step": 9570
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.81300331985784e-06,
+      "loss": 0.0165,
+      "step": 9572
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.809153032813124e-06,
+      "loss": 0.0221,
+      "step": 9574
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.805303086900032e-06,
+      "loss": 0.0121,
+      "step": 9576
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.801453482718032e-06,
+      "loss": 0.0091,
+      "step": 9578
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.797604220866532e-06,
+      "loss": 0.0097,
+      "step": 9580
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 7.7937553019449e-06,
+      "loss": 0.0174,
+      "step": 9582
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.789906726552439e-06,
+      "loss": 0.0077,
+      "step": 9584
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.786058495288396e-06,
+      "loss": 0.0095,
+      "step": 9586
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.782210608751975e-06,
+      "loss": 0.0087,
+      "step": 9588
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.778363067542325e-06,
+      "loss": 0.0143,
+      "step": 9590
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.774515872258533e-06,
+      "loss": 0.0138,
+      "step": 9592
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.770669023499633e-06,
+      "loss": 0.0096,
+      "step": 9594
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.766822521864617e-06,
+      "loss": 0.0193,
+      "step": 9596
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.762976367952406e-06,
+      "loss": 0.015,
+      "step": 9598
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.759130562361888e-06,
+      "loss": 0.0059,
+      "step": 9600
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.755285105691869e-06,
+      "loss": 0.0164,
+      "step": 9602
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.751439998541127e-06,
+      "loss": 0.0144,
+      "step": 9604
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.747595241508368e-06,
+      "loss": 0.0067,
+      "step": 9606
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.74375083519226e-06,
+      "loss": 0.0157,
+      "step": 9608
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.739906780191394e-06,
+      "loss": 0.0109,
+      "step": 9610
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.736063077104326e-06,
+      "loss": 0.0132,
+      "step": 9612
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.732219726529546e-06,
+      "loss": 0.0088,
+      "step": 9614
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.728376729065501e-06,
+      "loss": 0.0078,
+      "step": 9616
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.724534085310568e-06,
+      "loss": 0.0066,
+      "step": 9618
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.720691795863075e-06,
+      "loss": 0.0221,
+      "step": 9620
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 7.716849861321296e-06,
+      "loss": 0.0173,
+      "step": 9622
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.71300828228346e-06,
+      "loss": 0.0055,
+      "step": 9624
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.709167059347718e-06,
+      "loss": 0.0128,
+      "step": 9626
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.70532619311218e-06,
+      "loss": 0.0121,
+      "step": 9628
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.701485684174905e-06,
+      "loss": 0.0158,
+      "step": 9630
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.697645533133883e-06,
+      "loss": 0.0123,
+      "step": 9632
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.693805740587055e-06,
+      "loss": 0.0128,
+      "step": 9634
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.689966307132306e-06,
+      "loss": 0.0195,
+      "step": 9636
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.686127233367473e-06,
+      "loss": 0.0097,
+      "step": 9638
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.68228851989032e-06,
+      "loss": 0.0116,
+      "step": 9640
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.678450167298566e-06,
+      "loss": 0.011,
+      "step": 9642
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.674612176189872e-06,
+      "loss": 0.0098,
+      "step": 9644
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.67077454716185e-06,
+      "loss": 0.0153,
+      "step": 9646
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.666937280812038e-06,
+      "loss": 0.0067,
+      "step": 9648
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.66310037773793e-06,
+      "loss": 0.0141,
+      "step": 9650
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.659263838536963e-06,
+      "loss": 0.0088,
+      "step": 9652
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.655427663806517e-06,
+      "loss": 0.0082,
+      "step": 9654
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.651591854143911e-06,
+      "loss": 0.0061,
+      "step": 9656
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.64775641014641e-06,
+      "loss": 0.0111,
+      "step": 9658
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.643921332411223e-06,
+      "loss": 0.0082,
+      "step": 9660
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.640086621535504e-06,
+      "loss": 0.0145,
+      "step": 9662
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 7.636252278116338e-06,
+      "loss": 0.0108,
+      "step": 9664
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.632418302750767e-06,
+      "loss": 0.021,
+      "step": 9666
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.628584696035775e-06,
+      "loss": 0.0179,
+      "step": 9668
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.624751458568279e-06,
+      "loss": 0.0153,
+      "step": 9670
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.6209185909451414e-06,
+      "loss": 0.0088,
+      "step": 9672
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.6170860937631705e-06,
+      "loss": 0.0093,
+      "step": 9674
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.613253967619117e-06,
+      "loss": 0.0111,
+      "step": 9676
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.6094222131096725e-06,
+      "loss": 0.0097,
+      "step": 9678
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.6055908308314675e-06,
+      "loss": 0.0202,
+      "step": 9680
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.601759821381081e-06,
+      "loss": 0.0094,
+      "step": 9682
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.597929185355025e-06,
+      "loss": 0.0118,
+      "step": 9684
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.594098923349769e-06,
+      "loss": 0.019,
+      "step": 9686
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.590269035961701e-06,
+      "loss": 0.0196,
+      "step": 9688
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.586439523787176e-06,
+      "loss": 0.0096,
+      "step": 9690
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.582610387422468e-06,
+      "loss": 0.0165,
+      "step": 9692
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.578781627463814e-06,
+      "loss": 0.0173,
+      "step": 9694
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.574953244507371e-06,
+      "loss": 0.016,
+      "step": 9696
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.57112523914925e-06,
+      "loss": 0.0121,
+      "step": 9698
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.5672976119855e-06,
+      "loss": 0.0125,
+      "step": 9700
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.563470363612121e-06,
+      "loss": 0.0199,
+      "step": 9702
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 7.559643494625034e-06,
+      "loss": 0.0136,
+      "step": 9704
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 7.555817005620114e-06,
+      "loss": 0.015,
+      "step": 9706
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 7.551990897193175e-06,
+      "loss": 0.016,
+      "step": 9708
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 7.548165169939978e-06,
+      "loss": 0.0163,
+      "step": 9710
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 7.54433982445621e-06,
+      "loss": 0.0153,
+      "step": 9712
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 7.540514861337506e-06,
+      "loss": 0.0104,
+      "step": 9714
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 7.5366902811794465e-06,
+      "loss": 0.0131,
+      "step": 9716
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 7.53286608457755e-06,
+      "loss": 0.016,
+      "step": 9718
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 7.529042272127264e-06,
+      "loss": 0.0126,
+      "step": 9720
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 7.525218844423991e-06,
+      "loss": 0.0173,
+      "step": 9722
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 7.52139580206307e-06,
+      "loss": 0.0161,
+      "step": 9724
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 7.5175731456397785e-06,
+      "loss": 0.0143,
+      "step": 9726
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 7.513750875749326e-06,
+      "loss": 0.0215,
+      "step": 9728
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 7.509928992986872e-06,
+      "loss": 0.0105,
+      "step": 9730
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 7.5061074979475194e-06,
+      "loss": 0.01,
+      "step": 9732
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 7.502286391226299e-06,
+      "loss": 0.01,
+      "step": 9734
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 7.4984656734181825e-06,
+      "loss": 0.0057,
+      "step": 9736
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 7.494645345118092e-06,
+      "loss": 0.02,
+      "step": 9738
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 7.490825406920878e-06,
+      "loss": 0.0176,
+      "step": 9740
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 7.487005859421337e-06,
+      "loss": 0.0113,
+      "step": 9742
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 7.4831867032141955e-06,
+      "loss": 0.0071,
+      "step": 9744
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 7.479367938894133e-06,
+      "loss": 0.0069,
+      "step": 9746
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 7.475549567055754e-06,
+      "loss": 0.0039,
+      "step": 9748
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 7.471731588293616e-06,
+      "loss": 0.0118,
+      "step": 9750
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 7.467914003202197e-06,
+      "loss": 0.0216,
+      "step": 9752
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 7.464096812375932e-06,
+      "loss": 0.0217,
+      "step": 9754
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 7.460280016409183e-06,
+      "loss": 0.0105,
+      "step": 9756
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 7.45646361589626e-06,
+      "loss": 0.0128,
+      "step": 9758
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 7.4526476114313986e-06,
+      "loss": 0.0167,
+      "step": 9760
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 7.448832003608781e-06,
+      "loss": 0.0199,
+      "step": 9762
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 7.4450167930225295e-06,
+      "loss": 0.0041,
+      "step": 9764
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 7.4412019802667036e-06,
+      "loss": 0.006,
+      "step": 9766
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 7.437387565935294e-06,
+      "loss": 0.0146,
+      "step": 9768
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 7.433573550622233e-06,
+      "loss": 0.0118,
+      "step": 9770
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 7.429759934921397e-06,
+      "loss": 0.0041,
+      "step": 9772
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 7.4259467194265955e-06,
+      "loss": 0.0113,
+      "step": 9774
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 7.422133904731568e-06,
+      "loss": 0.0215,
+      "step": 9776
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 7.418321491430003e-06,
+      "loss": 0.0164,
+      "step": 9778
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 7.4145094801155255e-06,
+      "loss": 0.0062,
+      "step": 9780
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 7.410697871381695e-06,
+      "loss": 0.0106,
+      "step": 9782
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 7.406886665822001e-06,
+      "loss": 0.0196,
+      "step": 9784
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 7.403075864029881e-06,
+      "loss": 0.0233,
+      "step": 9786
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 7.3992654665987095e-06,
+      "loss": 0.0073,
+      "step": 9788
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 7.395455474121793e-06,
+      "loss": 0.0136,
+      "step": 9790
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 7.391645887192374e-06,
+      "loss": 0.02,
+      "step": 9792
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 7.387836706403635e-06,
+      "loss": 0.0122,
+      "step": 9794
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 7.384027932348692e-06,
+      "loss": 0.0195,
+      "step": 9796
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 7.38021956562061e-06,
+      "loss": 0.0178,
+      "step": 9798
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 7.376411606812368e-06,
+      "loss": 0.0148,
+      "step": 9800
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 7.372604056516904e-06,
+      "loss": 0.0127,
+      "step": 9802
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 7.368796915327076e-06,
+      "loss": 0.0118,
+      "step": 9804
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 7.364990183835694e-06,
+      "loss": 0.0137,
+      "step": 9806
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 7.361183862635484e-06,
+      "loss": 0.0136,
+      "step": 9808
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 7.357377952319127e-06,
+      "loss": 0.0158,
+      "step": 9810
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 7.353572453479228e-06,
+      "loss": 0.0131,
+      "step": 9812
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 7.349767366708338e-06,
+      "loss": 0.0174,
+      "step": 9814
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 7.345962692598934e-06,
+      "loss": 0.0109,
+      "step": 9816
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 7.342158431743429e-06,
+      "loss": 0.01,
+      "step": 9818
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 7.338354584734182e-06,
+      "loss": 0.0067,
+      "step": 9820
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 7.334551152163481e-06,
+      "loss": 0.0179,
+      "step": 9822
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 7.330748134623546e-06,
+      "loss": 0.0075,
+      "step": 9824
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 7.326945532706535e-06,
+      "loss": 0.0112,
+      "step": 9826
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 7.323143347004547e-06,
+      "loss": 0.0096,
+      "step": 9828
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 7.319341578109609e-06,
+      "loss": 0.0133,
+      "step": 9830
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 7.315540226613684e-06,
+      "loss": 0.0258,
+      "step": 9832
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 7.3117392931086726e-06,
+      "loss": 0.0186,
+      "step": 9834
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 7.30793877818641e-06,
+      "loss": 0.0155,
+      "step": 9836
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 7.304138682438669e-06,
+      "loss": 0.0241,
+      "step": 9838
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 7.3003390064571425e-06,
+      "loss": 0.0139,
+      "step": 9840
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 7.296539750833478e-06,
+      "loss": 0.0162,
+      "step": 9842
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 7.292740916159249e-06,
+      "loss": 0.0088,
+      "step": 9844
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 7.288942503025962e-06,
+      "loss": 0.0211,
+      "step": 9846
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 7.285144512025053e-06,
+      "loss": 0.0162,
+      "step": 9848
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 7.281346943747907e-06,
+      "loss": 0.0156,
+      "step": 9850
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 7.277549798785825e-06,
+      "loss": 0.0235,
+      "step": 9852
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 7.273753077730065e-06,
+      "loss": 0.0255,
+      "step": 9854
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 7.269956781171792e-06,
+      "loss": 0.0194,
+      "step": 9856
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 7.266160909702125e-06,
+      "loss": 0.0117,
+      "step": 9858
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 7.262365463912108e-06,
+      "loss": 0.0144,
+      "step": 9860
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 7.258570444392725e-06,
+      "loss": 0.0096,
+      "step": 9862
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 7.254775851734883e-06,
+      "loss": 0.0076,
+      "step": 9864
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 7.250981686529436e-06,
+      "loss": 0.0104,
+      "step": 9866
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 7.247187949367158e-06,
+      "loss": 0.0096,
+      "step": 9868
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 7.24339464083877e-06,
+      "loss": 0.0133,
+      "step": 9870
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 7.239601761534913e-06,
+      "loss": 0.022,
+      "step": 9872
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 7.235809312046169e-06,
+      "loss": 0.0146,
+      "step": 9874
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 7.232017292963049e-06,
+      "loss": 0.0084,
+      "step": 9876
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 7.22822570487601e-06,
+      "loss": 0.0192,
+      "step": 9878
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 7.224434548375419e-06,
+      "loss": 0.017,
+      "step": 9880
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 7.220643824051592e-06,
+      "loss": 0.0135,
+      "step": 9882
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 7.216853532494773e-06,
+      "loss": 0.021,
+      "step": 9884
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 7.213063674295146e-06,
+      "loss": 0.0179,
+      "step": 9886
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 7.2092742500428126e-06,
+      "loss": 0.0178,
+      "step": 9888
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 7.205485260327817e-06,
+      "loss": 0.0146,
+      "step": 9890
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 7.201696705740137e-06,
+      "loss": 0.0105,
+      "step": 9892
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 7.19790858686968e-06,
+      "loss": 0.012,
+      "step": 9894
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 7.194120904306277e-06,
+      "loss": 0.0092,
+      "step": 9896
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 7.190333658639705e-06,
+      "loss": 0.0071,
+      "step": 9898
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 7.1865468504596704e-06,
+      "loss": 0.0152,
+      "step": 9900
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 7.182760480355806e-06,
+      "loss": 0.0131,
+      "step": 9902
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 7.178974548917672e-06,
+      "loss": 0.0161,
+      "step": 9904
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 7.175189056734774e-06,
+      "loss": 0.0138,
+      "step": 9906
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 7.17140400439654e-06,
+      "loss": 0.0083,
+      "step": 9908
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 7.167619392492337e-06,
+      "loss": 0.0108,
+      "step": 9910
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 7.163835221611446e-06,
+      "loss": 0.0103,
+      "step": 9912
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 7.160051492343101e-06,
+      "loss": 0.009,
+      "step": 9914
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 7.156268205276453e-06,
+      "loss": 0.013,
+      "step": 9916
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 7.152485361000595e-06,
+      "loss": 0.0129,
+      "step": 9918
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 7.148702960104535e-06,
+      "loss": 0.0141,
+      "step": 9920
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 7.14492100317723e-06,
+      "loss": 0.0107,
+      "step": 9922
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 7.141139490807554e-06,
+      "loss": 0.0219,
+      "step": 9924
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 7.137358423584324e-06,
+      "loss": 0.0121,
+      "step": 9926
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 7.133577802096274e-06,
+      "loss": 0.0141,
+      "step": 9928
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 7.12979762693208e-06,
+      "loss": 0.0201,
+      "step": 9930
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 7.1260178986803415e-06,
+      "loss": 0.0046,
+      "step": 9932
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 7.122238617929596e-06,
+      "loss": 0.0123,
+      "step": 9934
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 7.118459785268301e-06,
+      "loss": 0.0125,
+      "step": 9936
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 7.114681401284848e-06,
+      "loss": 0.0191,
+      "step": 9938
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 7.110903466567567e-06,
+      "loss": 0.0147,
+      "step": 9940
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 7.10712598170471e-06,
+      "loss": 0.0215,
+      "step": 9942
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 7.1033489472844566e-06,
+      "loss": 0.0156,
+      "step": 9944
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 7.0995723638949195e-06,
+      "loss": 0.0105,
+      "step": 9946
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 7.095796232124148e-06,
+      "loss": 0.0089,
+      "step": 9948
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 7.092020552560111e-06,
+      "loss": 0.013,
+      "step": 9950
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 7.088245325790705e-06,
+      "loss": 0.0085,
+      "step": 9952
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 7.084470552403769e-06,
+      "loss": 0.0188,
+      "step": 9954
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 7.080696232987063e-06,
+      "loss": 0.0135,
+      "step": 9956
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 7.07692236812828e-06,
+      "loss": 0.0071,
+      "step": 9958
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 7.07314895841503e-06,
+      "loss": 0.0109,
+      "step": 9960
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 7.0693760044348695e-06,
+      "loss": 0.0116,
+      "step": 9962
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 7.065603506775276e-06,
+      "loss": 0.0155,
+      "step": 9964
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 7.061831466023656e-06,
+      "loss": 0.0096,
+      "step": 9966
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 7.058059882767341e-06,
+      "loss": 0.0105,
+      "step": 9968
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 7.054288757593599e-06,
+      "loss": 0.0113,
+      "step": 9970
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 7.050518091089621e-06,
+      "loss": 0.0111,
+      "step": 9972
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 7.046747883842534e-06,
+      "loss": 0.0068,
+      "step": 9974
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 7.04297813643938e-06,
+      "loss": 0.0163,
+      "step": 9976
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 7.039208849467143e-06,
+      "loss": 0.0243,
+      "step": 9978
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 7.0354400235127264e-06,
+      "loss": 0.01,
+      "step": 9980
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 7.0316716591629706e-06,
+      "loss": 0.0097,
+      "step": 9982
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 7.0279037570046325e-06,
+      "loss": 0.0091,
+      "step": 9984
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 7.024136317624407e-06,
+      "loss": 0.0141,
+      "step": 9986
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 7.02036934160891e-06,
+      "loss": 0.0088,
+      "step": 9988
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 7.016602829544696e-06,
+      "loss": 0.0174,
+      "step": 9990
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 7.012836782018232e-06,
+      "loss": 0.0346,
+      "step": 9992
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 7.00907119961592e-06,
+      "loss": 0.0124,
+      "step": 9994
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 7.005306082924094e-06,
+      "loss": 0.0128,
+      "step": 9996
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 7.001541432529013e-06,
+      "loss": 0.0114,
+      "step": 9998
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.9977772490168594e-06,
+      "loss": 0.0196,
+      "step": 10000
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.9940135329737404e-06,
+      "loss": 0.0138,
+      "step": 10002
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.9902502849857036e-06,
+      "loss": 0.0127,
+      "step": 10004
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.98648750563871e-06,
+      "loss": 0.0162,
+      "step": 10006
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.982725195518658e-06,
+      "loss": 0.016,
+      "step": 10008
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.978963355211361e-06,
+      "loss": 0.0138,
+      "step": 10010
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.975201985302573e-06,
+      "loss": 0.0205,
+      "step": 10012
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.971441086377968e-06,
+      "loss": 0.0087,
+      "step": 10014
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.96768065902314e-06,
+      "loss": 0.0135,
+      "step": 10016
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.963920703823619e-06,
+      "loss": 0.0127,
+      "step": 10018
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.960161221364864e-06,
+      "loss": 0.0092,
+      "step": 10020
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.956402212232254e-06,
+      "loss": 0.0209,
+      "step": 10022
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.952643677011086e-06,
+      "loss": 0.0194,
+      "step": 10024
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.948885616286605e-06,
+      "loss": 0.0136,
+      "step": 10026
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.945128030643959e-06,
+      "loss": 0.0159,
+      "step": 10028
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.941370920668244e-06,
+      "loss": 0.0192,
+      "step": 10030
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.937614286944461e-06,
+      "loss": 0.019,
+      "step": 10032
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 6.933858130057553e-06,
+      "loss": 0.0139,
+      "step": 10034
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.930102450592376e-06,
+      "loss": 0.0094,
+      "step": 10036
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.926347249133727e-06,
+      "loss": 0.0107,
+      "step": 10038
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.922592526266312e-06,
+      "loss": 0.0154,
+      "step": 10040
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.9188382825747715e-06,
+      "loss": 0.0082,
+      "step": 10042
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.915084518643671e-06,
+      "loss": 0.0134,
+      "step": 10044
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.9113312350575035e-06,
+      "loss": 0.0054,
+      "step": 10046
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.907578432400679e-06,
+      "loss": 0.0104,
+      "step": 10048
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.903826111257536e-06,
+      "loss": 0.0139,
+      "step": 10050
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.9000742722123445e-06,
+      "loss": 0.0105,
+      "step": 10052
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.896322915849298e-06,
+      "loss": 0.0087,
+      "step": 10054
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.892572042752501e-06,
+      "loss": 0.0124,
+      "step": 10056
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.8888216535059985e-06,
+      "loss": 0.0083,
+      "step": 10058
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.885071748693755e-06,
+      "loss": 0.0198,
+      "step": 10060
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.881322328899661e-06,
+      "loss": 0.0112,
+      "step": 10062
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.8775733947075265e-06,
+      "loss": 0.0087,
+      "step": 10064
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.873824946701089e-06,
+      "loss": 0.0084,
+      "step": 10066
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.8700769854640135e-06,
+      "loss": 0.0107,
+      "step": 10068
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.866329511579888e-06,
+      "loss": 0.0079,
+      "step": 10070
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.862582525632215e-06,
+      "loss": 0.0123,
+      "step": 10072
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 6.858836028204434e-06,
+      "loss": 0.0209,
+      "step": 10074
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.855090019879904e-06,
+      "loss": 0.0169,
+      "step": 10076
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.851344501241908e-06,
+      "loss": 0.0159,
+      "step": 10078
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.847599472873646e-06,
+      "loss": 0.0178,
+      "step": 10080
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.843854935358252e-06,
+      "loss": 0.0126,
+      "step": 10082
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.84011088927878e-06,
+      "loss": 0.0099,
+      "step": 10084
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.836367335218206e-06,
+      "loss": 0.0093,
+      "step": 10086
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.832624273759428e-06,
+      "loss": 0.0159,
+      "step": 10088
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.8288817054852685e-06,
+      "loss": 0.0127,
+      "step": 10090
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.8251396309784764e-06,
+      "loss": 0.0093,
+      "step": 10092
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.8213980508217235e-06,
+      "loss": 0.0197,
+      "step": 10094
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.817656965597597e-06,
+      "loss": 0.0064,
+      "step": 10096
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.813916375888617e-06,
+      "loss": 0.0117,
+      "step": 10098
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.8101762822772176e-06,
+      "loss": 0.0255,
+      "step": 10100
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.806436685345768e-06,
+      "loss": 0.0069,
+      "step": 10102
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.802697585676543e-06,
+      "loss": 0.0172,
+      "step": 10104
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.798958983851751e-06,
+      "loss": 0.0057,
+      "step": 10106
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.795220880453521e-06,
+      "loss": 0.0053,
+      "step": 10108
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.791483276063913e-06,
+      "loss": 0.0086,
+      "step": 10110
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.7877461712648885e-06,
+      "loss": 0.0133,
+      "step": 10112
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.784009566638348e-06,
+      "loss": 0.0176,
+      "step": 10114
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 6.780273462766107e-06,
+      "loss": 0.0153,
+      "step": 10116
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.776537860229915e-06,
+      "loss": 0.0103,
+      "step": 10118
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.772802759611423e-06,
+      "loss": 0.011,
+      "step": 10120
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.769068161492217e-06,
+      "loss": 0.0133,
+      "step": 10122
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.7653340664538055e-06,
+      "loss": 0.0054,
+      "step": 10124
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.761600475077618e-06,
+      "loss": 0.0069,
+      "step": 10126
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.757867387944994e-06,
+      "loss": 0.0104,
+      "step": 10128
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.7541348056372095e-06,
+      "loss": 0.0124,
+      "step": 10130
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.750402728735457e-06,
+      "loss": 0.0179,
+      "step": 10132
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.7466711578208524e-06,
+      "loss": 0.0147,
+      "step": 10134
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.74294009347442e-06,
+      "loss": 0.0161,
+      "step": 10136
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.73920953627712e-06,
+      "loss": 0.0139,
+      "step": 10138
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.735479486809832e-06,
+      "loss": 0.0109,
+      "step": 10140
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.731749945653352e-06,
+      "loss": 0.0101,
+      "step": 10142
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.728020913388393e-06,
+      "loss": 0.0125,
+      "step": 10144
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.7242923905956e-06,
+      "loss": 0.0073,
+      "step": 10146
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.720564377855527e-06,
+      "loss": 0.0079,
+      "step": 10148
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.716836875748663e-06,
+      "loss": 0.0185,
+      "step": 10150
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.713109884855397e-06,
+      "loss": 0.0111,
+      "step": 10152
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.709383405756058e-06,
+      "loss": 0.0103,
+      "step": 10154
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 6.705657439030888e-06,
+      "loss": 0.0181,
+      "step": 10156
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.7019319852600396e-06,
+      "loss": 0.0216,
+      "step": 10158
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.6982070450236014e-06,
+      "loss": 0.0158,
+      "step": 10160
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.694482618901575e-06,
+      "loss": 0.0095,
+      "step": 10162
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.690758707473883e-06,
+      "loss": 0.0123,
+      "step": 10164
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.687035311320363e-06,
+      "loss": 0.0069,
+      "step": 10166
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.6833124310207785e-06,
+      "loss": 0.0223,
+      "step": 10168
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.679590067154807e-06,
+      "loss": 0.008,
+      "step": 10170
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.675868220302057e-06,
+      "loss": 0.029,
+      "step": 10172
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.672146891042041e-06,
+      "loss": 0.006,
+      "step": 10174
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.668426079954201e-06,
+      "loss": 0.0089,
+      "step": 10176
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.664705787617894e-06,
+      "loss": 0.0099,
+      "step": 10178
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.660986014612405e-06,
+      "loss": 0.0066,
+      "step": 10180
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.65726676151692e-06,
+      "loss": 0.0062,
+      "step": 10182
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.653548028910565e-06,
+      "loss": 0.0174,
+      "step": 10184
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.649829817372369e-06,
+      "loss": 0.0059,
+      "step": 10186
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.64611212748129e-06,
+      "loss": 0.0071,
+      "step": 10188
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.642394959816198e-06,
+      "loss": 0.0125,
+      "step": 10190
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.638678314955882e-06,
+      "loss": 0.0108,
+      "step": 10192
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.634962193479057e-06,
+      "loss": 0.0113,
+      "step": 10194
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.631246595964354e-06,
+      "loss": 0.0139,
+      "step": 10196
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 6.627531522990311e-06,
+      "loss": 0.0217,
+      "step": 10198
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.623816975135398e-06,
+      "loss": 0.0177,
+      "step": 10200
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.620102952978e-06,
+      "loss": 0.0096,
+      "step": 10202
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.616389457096417e-06,
+      "loss": 0.01,
+      "step": 10204
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.61267648806887e-06,
+      "loss": 0.0123,
+      "step": 10206
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.608964046473491e-06,
+      "loss": 0.0145,
+      "step": 10208
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.605252132888345e-06,
+      "loss": 0.0123,
+      "step": 10210
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.6015407478914e-06,
+      "loss": 0.0123,
+      "step": 10212
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.597829892060544e-06,
+      "loss": 0.0151,
+      "step": 10214
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.594119565973589e-06,
+      "loss": 0.0147,
+      "step": 10216
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.5904097702082635e-06,
+      "loss": 0.0127,
+      "step": 10218
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.586700505342212e-06,
+      "loss": 0.0095,
+      "step": 10220
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.582991771952986e-06,
+      "loss": 0.0146,
+      "step": 10222
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.5792835706180725e-06,
+      "loss": 0.0133,
+      "step": 10224
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.5755759019148616e-06,
+      "loss": 0.0155,
+      "step": 10226
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.571868766420672e-06,
+      "loss": 0.0127,
+      "step": 10228
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.568162164712727e-06,
+      "loss": 0.0148,
+      "step": 10230
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.564456097368176e-06,
+      "loss": 0.0095,
+      "step": 10232
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.560750564964079e-06,
+      "loss": 0.0073,
+      "step": 10234
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.557045568077422e-06,
+      "loss": 0.0126,
+      "step": 10236
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 6.553341107285092e-06,
+      "loss": 0.0202,
+      "step": 10238
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 6.549637183163911e-06,
+      "loss": 0.0122,
+      "step": 10240
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 6.5459337962906e-06,
+      "loss": 0.0146,
+      "step": 10242
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 6.542230947241815e-06,
+      "loss": 0.0085,
+      "step": 10244
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 6.538528636594108e-06,
+      "loss": 0.0157,
+      "step": 10246
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 6.53482686492396e-06,
+      "loss": 0.0123,
+      "step": 10248
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 6.531125632807767e-06,
+      "loss": 0.012,
+      "step": 10250
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 6.52742494082184e-06,
+      "loss": 0.0161,
+      "step": 10252
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 6.523724789542404e-06,
+      "loss": 0.0084,
+      "step": 10254
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 6.520025179545597e-06,
+      "loss": 0.0177,
+      "step": 10256
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 6.516326111407478e-06,
+      "loss": 0.015,
+      "step": 10258
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 6.512627585704028e-06,
+      "loss": 0.008,
+      "step": 10260
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 6.508929603011127e-06,
+      "loss": 0.0166,
+      "step": 10262
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 6.50523216390458e-06,
+      "loss": 0.0187,
+      "step": 10264
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 6.501535268960109e-06,
+      "loss": 0.0083,
+      "step": 10266
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 6.497838918753352e-06,
+      "loss": 0.0102,
+      "step": 10268
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 6.49414311385985e-06,
+      "loss": 0.0068,
+      "step": 10270
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 6.4904478548550746e-06,
+      "loss": 0.009,
+      "step": 10272
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 6.486753142314406e-06,
+      "loss": 0.015,
+      "step": 10274
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 6.483058976813139e-06,
+      "loss": 0.0126,
+      "step": 10276
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 6.479365358926477e-06,
+      "loss": 0.0074,
+      "step": 10278
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 6.475672289229555e-06,
+      "loss": 0.0134,
+      "step": 10280
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 6.471979768297402e-06,
+      "loss": 0.0106,
+      "step": 10282
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 6.468287796704982e-06,
+      "loss": 0.0104,
+      "step": 10284
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 6.4645963750271525e-06,
+      "loss": 0.0093,
+      "step": 10286
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 6.4609055038387045e-06,
+      "loss": 0.0043,
+      "step": 10288
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 6.4572151837143295e-06,
+      "loss": 0.0154,
+      "step": 10290
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 6.453525415228645e-06,
+      "loss": 0.0152,
+      "step": 10292
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 6.449836198956168e-06,
+      "loss": 0.0079,
+      "step": 10294
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 6.4461475354713435e-06,
+      "loss": 0.0161,
+      "step": 10296
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 6.442459425348522e-06,
+      "loss": 0.0104,
+      "step": 10298
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 6.4387718691619735e-06,
+      "loss": 0.0116,
+      "step": 10300
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 6.435084867485875e-06,
+      "loss": 0.0076,
+      "step": 10302
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 6.431398420894322e-06,
+      "loss": 0.0146,
+      "step": 10304
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 6.42771252996132e-06,
+      "loss": 0.0126,
+      "step": 10306
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 6.424027195260798e-06,
+      "loss": 0.0124,
+      "step": 10308
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 6.420342417366584e-06,
+      "loss": 0.0017,
+      "step": 10310
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 6.416658196852426e-06,
+      "loss": 0.0137,
+      "step": 10312
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 6.412974534291988e-06,
+      "loss": 0.0175,
+      "step": 10314
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 6.409291430258847e-06,
+      "loss": 0.0082,
+      "step": 10316
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 6.405608885326486e-06,
+      "loss": 0.0068,
+      "step": 10318
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 6.401926900068304e-06,
+      "loss": 0.0082,
+      "step": 10320
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 6.3982454750576185e-06,
+      "loss": 0.0104,
+      "step": 10322
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 6.394564610867656e-06,
+      "loss": 0.0115,
+      "step": 10324
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 6.39088430807155e-06,
+      "loss": 0.0149,
+      "step": 10326
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 6.3872045672423545e-06,
+      "loss": 0.0098,
+      "step": 10328
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 6.383525388953036e-06,
+      "loss": 0.0144,
+      "step": 10330
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 6.379846773776469e-06,
+      "loss": 0.0182,
+      "step": 10332
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 6.376168722285438e-06,
+      "loss": 0.0096,
+      "step": 10334
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 6.3724912350526465e-06,
+      "loss": 0.0189,
+      "step": 10336
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 6.368814312650712e-06,
+      "loss": 0.0121,
+      "step": 10338
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 6.365137955652156e-06,
+      "loss": 0.0111,
+      "step": 10340
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 6.36146216462941e-06,
+      "loss": 0.0099,
+      "step": 10342
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 6.357786940154832e-06,
+      "loss": 0.0058,
+      "step": 10344
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 6.354112282800675e-06,
+      "loss": 0.0102,
+      "step": 10346
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 6.350438193139118e-06,
+      "loss": 0.0128,
+      "step": 10348
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 6.346764671742238e-06,
+      "loss": 0.016,
+      "step": 10350
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 6.3430917191820354e-06,
+      "loss": 0.0173,
+      "step": 10352
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 6.339419336030412e-06,
+      "loss": 0.009,
+      "step": 10354
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 6.335747522859195e-06,
+      "loss": 0.01,
+      "step": 10356
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 6.332076280240103e-06,
+      "loss": 0.0157,
+      "step": 10358
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 6.328405608744783e-06,
+      "loss": 0.0083,
+      "step": 10360
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 6.324735508944783e-06,
+      "loss": 0.0108,
+      "step": 10362
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 6.321065981411574e-06,
+      "loss": 0.0125,
+      "step": 10364
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 6.317397026716519e-06,
+      "loss": 0.0059,
+      "step": 10366
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 6.313728645430906e-06,
+      "loss": 0.0078,
+      "step": 10368
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 6.310060838125929e-06,
+      "loss": 0.0077,
+      "step": 10370
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 6.3063936053726994e-06,
+      "loss": 0.0148,
+      "step": 10372
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 6.302726947742228e-06,
+      "loss": 0.0064,
+      "step": 10374
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 6.29906086580544e-06,
+      "loss": 0.0113,
+      "step": 10376
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 6.295395360133177e-06,
+      "loss": 0.0119,
+      "step": 10378
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 6.2917304312961865e-06,
+      "loss": 0.0064,
+      "step": 10380
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 6.288066079865121e-06,
+      "loss": 0.0118,
+      "step": 10382
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 6.2844023064105484e-06,
+      "loss": 0.0148,
+      "step": 10384
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 6.280739111502951e-06,
+      "loss": 0.0096,
+      "step": 10386
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 6.277076495712718e-06,
+      "loss": 0.0071,
+      "step": 10388
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 6.2734144596101364e-06,
+      "loss": 0.0129,
+      "step": 10390
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 6.26975300376542e-06,
+      "loss": 0.0076,
+      "step": 10392
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 6.266092128748687e-06,
+      "loss": 0.0062,
+      "step": 10394
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 6.262431835129963e-06,
+      "loss": 0.0171,
+      "step": 10396
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 6.2587721234791774e-06,
+      "loss": 0.0168,
+      "step": 10398
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 6.2551129943661825e-06,
+      "loss": 0.0126,
+      "step": 10400
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 6.251454448360729e-06,
+      "loss": 0.0073,
+      "step": 10402
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 6.2477964860324844e-06,
+      "loss": 0.0189,
+      "step": 10404
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 6.244139107951017e-06,
+      "loss": 0.0055,
+      "step": 10406
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 6.240482314685811e-06,
+      "loss": 0.0117,
+      "step": 10408
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 6.236826106806253e-06,
+      "loss": 0.0074,
+      "step": 10410
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 6.233170484881652e-06,
+      "loss": 0.0185,
+      "step": 10412
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 6.229515449481205e-06,
+      "loss": 0.0093,
+      "step": 10414
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 6.225861001174036e-06,
+      "loss": 0.0148,
+      "step": 10416
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 6.222207140529167e-06,
+      "loss": 0.0112,
+      "step": 10418
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 6.218553868115538e-06,
+      "loss": 0.0103,
+      "step": 10420
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 6.214901184501984e-06,
+      "loss": 0.0092,
+      "step": 10422
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 6.211249090257259e-06,
+      "loss": 0.0095,
+      "step": 10424
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 6.20759758595002e-06,
+      "loss": 0.0123,
+      "step": 10426
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 6.2039466721488405e-06,
+      "loss": 0.0098,
+      "step": 10428
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 6.20029634942219e-06,
+      "loss": 0.0106,
+      "step": 10430
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 6.19664661833845e-06,
+      "loss": 0.0074,
+      "step": 10432
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 6.192997479465914e-06,
+      "loss": 0.015,
+      "step": 10434
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 6.189348933372787e-06,
+      "loss": 0.0095,
+      "step": 10436
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 6.185700980627167e-06,
+      "loss": 0.0063,
+      "step": 10438
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 6.182053621797068e-06,
+      "loss": 0.0142,
+      "step": 10440
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 6.178406857450417e-06,
+      "loss": 0.0157,
+      "step": 10442
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 6.174760688155044e-06,
+      "loss": 0.0135,
+      "step": 10444
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 6.171115114478677e-06,
+      "loss": 0.0107,
+      "step": 10446
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 6.167470136988964e-06,
+      "loss": 0.003,
+      "step": 10448
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 6.163825756253461e-06,
+      "loss": 0.0075,
+      "step": 10450
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 6.1601819728396216e-06,
+      "loss": 0.0155,
+      "step": 10452
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 6.156538787314808e-06,
+      "loss": 0.0164,
+      "step": 10454
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 6.152896200246297e-06,
+      "loss": 0.01,
+      "step": 10456
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 6.149254212201261e-06,
+      "loss": 0.0061,
+      "step": 10458
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 6.145612823746795e-06,
+      "loss": 0.0114,
+      "step": 10460
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 6.141972035449881e-06,
+      "loss": 0.0159,
+      "step": 10462
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 6.1383318478774255e-06,
+      "loss": 0.0092,
+      "step": 10464
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 6.134692261596227e-06,
+      "loss": 0.0128,
+      "step": 10466
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 6.131053277173003e-06,
+      "loss": 0.0094,
+      "step": 10468
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 6.127414895174366e-06,
+      "loss": 0.0189,
+      "step": 10470
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 6.123777116166844e-06,
+      "loss": 0.0069,
+      "step": 10472
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 6.120139940716862e-06,
+      "loss": 0.0061,
+      "step": 10474
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 6.116503369390764e-06,
+      "loss": 0.017,
+      "step": 10476
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 6.112867402754785e-06,
+      "loss": 0.0162,
+      "step": 10478
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 6.1092320413750725e-06,
+      "loss": 0.006,
+      "step": 10480
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 6.105597285817682e-06,
+      "loss": 0.0069,
+      "step": 10482
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 6.1019631366485785e-06,
+      "loss": 0.0195,
+      "step": 10484
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 6.09832959443362e-06,
+      "loss": 0.0135,
+      "step": 10486
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 6.094696659738575e-06,
+      "loss": 0.0094,
+      "step": 10488
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 6.091064333129123e-06,
+      "loss": 0.0122,
+      "step": 10490
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 6.087432615170849e-06,
+      "loss": 0.0074,
+      "step": 10492
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 6.0838015064292325e-06,
+      "loss": 0.0127,
+      "step": 10494
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 6.080171007469664e-06,
+      "loss": 0.0098,
+      "step": 10496
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 6.076541118857448e-06,
+      "loss": 0.0142,
+      "step": 10498
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 6.07291184115778e-06,
+      "loss": 0.0136,
+      "step": 10500
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 6.069283174935766e-06,
+      "loss": 0.0149,
+      "step": 10502
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 6.065655120756417e-06,
+      "loss": 0.0119,
+      "step": 10504
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 6.062027679184653e-06,
+      "loss": 0.0124,
+      "step": 10506
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 6.058400850785293e-06,
+      "loss": 0.0088,
+      "step": 10508
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 6.054774636123058e-06,
+      "loss": 0.0107,
+      "step": 10510
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 6.051149035762578e-06,
+      "loss": 0.013,
+      "step": 10512
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 6.047524050268392e-06,
+      "loss": 0.0104,
+      "step": 10514
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 6.043899680204937e-06,
+      "loss": 0.0077,
+      "step": 10516
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 6.040275926136547e-06,
+      "loss": 0.0138,
+      "step": 10518
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 6.036652788627477e-06,
+      "loss": 0.0157,
+      "step": 10520
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 6.033030268241871e-06,
+      "loss": 0.0064,
+      "step": 10522
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 6.029408365543792e-06,
+      "loss": 0.0137,
+      "step": 10524
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 6.025787081097188e-06,
+      "loss": 0.0159,
+      "step": 10526
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 6.022166415465925e-06,
+      "loss": 0.0268,
+      "step": 10528
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 6.0185463692137666e-06,
+      "loss": 0.0108,
+      "step": 10530
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 6.014926942904388e-06,
+      "loss": 0.0054,
+      "step": 10532
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 6.011308137101355e-06,
+      "loss": 0.0074,
+      "step": 10534
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 6.007689952368144e-06,
+      "loss": 0.0118,
+      "step": 10536
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 6.004072389268134e-06,
+      "loss": 0.0085,
+      "step": 10538
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 6.0004554483646135e-06,
+      "loss": 0.0085,
+      "step": 10540
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.996839130220761e-06,
+      "loss": 0.0138,
+      "step": 10542
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.993223435399663e-06,
+      "loss": 0.0068,
+      "step": 10544
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.989608364464317e-06,
+      "loss": 0.0113,
+      "step": 10546
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.9859939179776164e-06,
+      "loss": 0.0078,
+      "step": 10548
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.982380096502355e-06,
+      "loss": 0.0091,
+      "step": 10550
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.978766900601232e-06,
+      "loss": 0.0098,
+      "step": 10552
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.975154330836854e-06,
+      "loss": 0.0102,
+      "step": 10554
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.971542387771725e-06,
+      "loss": 0.0094,
+      "step": 10556
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.967931071968246e-06,
+      "loss": 0.0062,
+      "step": 10558
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.964320383988731e-06,
+      "loss": 0.017,
+      "step": 10560
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.960710324395394e-06,
+      "loss": 0.0115,
+      "step": 10562
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.957100893750349e-06,
+      "loss": 0.0073,
+      "step": 10564
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.9534920926156044e-06,
+      "loss": 0.0121,
+      "step": 10566
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.9498839215530846e-06,
+      "loss": 0.0122,
+      "step": 10568
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.9462763811246095e-06,
+      "loss": 0.0111,
+      "step": 10570
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.9426694718919e-06,
+      "loss": 0.0117,
+      "step": 10572
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.9390631944165764e-06,
+      "loss": 0.0084,
+      "step": 10574
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.935457549260167e-06,
+      "loss": 0.0136,
+      "step": 10576
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.931852536984096e-06,
+      "loss": 0.009,
+      "step": 10578
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.928248158149697e-06,
+      "loss": 0.0111,
+      "step": 10580
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.9246444133181905e-06,
+      "loss": 0.0069,
+      "step": 10582
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.921041303050713e-06,
+      "loss": 0.0072,
+      "step": 10584
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.917438827908293e-06,
+      "loss": 0.0122,
+      "step": 10586
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.913836988451871e-06,
+      "loss": 0.0061,
+      "step": 10588
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.9102357852422695e-06,
+      "loss": 0.0112,
+      "step": 10590
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.906635218840233e-06,
+      "loss": 0.0127,
+      "step": 10592
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.903035289806389e-06,
+      "loss": 0.0091,
+      "step": 10594
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.899435998701285e-06,
+      "loss": 0.0106,
+      "step": 10596
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.895837346085349e-06,
+      "loss": 0.0122,
+      "step": 10598
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.892239332518919e-06,
+      "loss": 0.0081,
+      "step": 10600
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.888641958562236e-06,
+      "loss": 0.0098,
+      "step": 10602
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.885045224775441e-06,
+      "loss": 0.0131,
+      "step": 10604
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.881449131718568e-06,
+      "loss": 0.0168,
+      "step": 10606
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.877853679951557e-06,
+      "loss": 0.0173,
+      "step": 10608
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.8742588700342505e-06,
+      "loss": 0.0098,
+      "step": 10610
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.870664702526387e-06,
+      "loss": 0.0067,
+      "step": 10612
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.867071177987604e-06,
+      "loss": 0.0151,
+      "step": 10614
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.8634782969774395e-06,
+      "loss": 0.0109,
+      "step": 10616
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.859886060055338e-06,
+      "loss": 0.0069,
+      "step": 10618
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.8562944677806346e-06,
+      "loss": 0.0134,
+      "step": 10620
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.8527035207125656e-06,
+      "loss": 0.0088,
+      "step": 10622
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.849113219410272e-06,
+      "loss": 0.0118,
+      "step": 10624
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.845523564432791e-06,
+      "loss": 0.0104,
+      "step": 10626
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.841934556339062e-06,
+      "loss": 0.0151,
+      "step": 10628
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.838346195687915e-06,
+      "loss": 0.0098,
+      "step": 10630
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.834758483038087e-06,
+      "loss": 0.0123,
+      "step": 10632
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.8311714189482115e-06,
+      "loss": 0.0137,
+      "step": 10634
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.82758500397683e-06,
+      "loss": 0.0087,
+      "step": 10636
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.823999238682363e-06,
+      "loss": 0.0121,
+      "step": 10638
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.8204141236231485e-06,
+      "loss": 0.0104,
+      "step": 10640
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.816829659357417e-06,
+      "loss": 0.0177,
+      "step": 10642
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.813245846443295e-06,
+      "loss": 0.0169,
+      "step": 10644
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.809662685438806e-06,
+      "loss": 0.0071,
+      "step": 10646
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.806080176901879e-06,
+      "loss": 0.0133,
+      "step": 10648
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.8024983213903374e-06,
+      "loss": 0.0064,
+      "step": 10650
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.798917119461908e-06,
+      "loss": 0.0059,
+      "step": 10652
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.795336571674203e-06,
+      "loss": 0.0094,
+      "step": 10654
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.791756678584746e-06,
+      "loss": 0.0103,
+      "step": 10656
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.788177440750958e-06,
+      "loss": 0.0109,
+      "step": 10658
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.784598858730146e-06,
+      "loss": 0.0067,
+      "step": 10660
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.781020933079524e-06,
+      "loss": 0.0089,
+      "step": 10662
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.777443664356203e-06,
+      "loss": 0.0142,
+      "step": 10664
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.773867053117192e-06,
+      "loss": 0.0147,
+      "step": 10666
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.7702910999194e-06,
+      "loss": 0.0105,
+      "step": 10668
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.766715805319623e-06,
+      "loss": 0.0058,
+      "step": 10670
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.76314116987457e-06,
+      "loss": 0.0097,
+      "step": 10672
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.759567194140834e-06,
+      "loss": 0.0142,
+      "step": 10674
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.755993878674908e-06,
+      "loss": 0.0119,
+      "step": 10676
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.752421224033187e-06,
+      "loss": 0.011,
+      "step": 10678
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.74884923077196e-06,
+      "loss": 0.0136,
+      "step": 10680
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.745277899447421e-06,
+      "loss": 0.0132,
+      "step": 10682
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.741707230615643e-06,
+      "loss": 0.0099,
+      "step": 10684
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.738137224832614e-06,
+      "loss": 0.0094,
+      "step": 10686
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.734567882654204e-06,
+      "loss": 0.0151,
+      "step": 10688
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.730999204636195e-06,
+      "loss": 0.0065,
+      "step": 10690
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.727431191334249e-06,
+      "loss": 0.01,
+      "step": 10692
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.723863843303938e-06,
+      "loss": 0.0113,
+      "step": 10694
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.720297161100725e-06,
+      "loss": 0.0031,
+      "step": 10696
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.7167311452799745e-06,
+      "loss": 0.0104,
+      "step": 10698
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.713165796396931e-06,
+      "loss": 0.0108,
+      "step": 10700
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.709601115006759e-06,
+      "loss": 0.0087,
+      "step": 10702
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.706037101664495e-06,
+      "loss": 0.0117,
+      "step": 10704
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.702473756925093e-06,
+      "loss": 0.0142,
+      "step": 10706
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.698911081343386e-06,
+      "loss": 0.0122,
+      "step": 10708
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.69534907547411e-06,
+      "loss": 0.0132,
+      "step": 10710
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.691787739871901e-06,
+      "loss": 0.0085,
+      "step": 10712
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.688227075091288e-06,
+      "loss": 0.0137,
+      "step": 10714
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.68466708168669e-06,
+      "loss": 0.0102,
+      "step": 10716
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.681107760212422e-06,
+      "loss": 0.0135,
+      "step": 10718
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.6775491112227e-06,
+      "loss": 0.0211,
+      "step": 10720
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.673991135271637e-06,
+      "loss": 0.0136,
+      "step": 10722
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.670433832913231e-06,
+      "loss": 0.0135,
+      "step": 10724
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.666877204701383e-06,
+      "loss": 0.0175,
+      "step": 10726
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.663321251189893e-06,
+      "loss": 0.0111,
+      "step": 10728
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.659765972932445e-06,
+      "loss": 0.0077,
+      "step": 10730
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.656211370482618e-06,
+      "loss": 0.0137,
+      "step": 10732
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.652657444393898e-06,
+      "loss": 0.0064,
+      "step": 10734
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.649104195219654e-06,
+      "loss": 0.0127,
+      "step": 10736
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.645551623513163e-06,
+      "loss": 0.0078,
+      "step": 10738
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.641999729827576e-06,
+      "loss": 0.0062,
+      "step": 10740
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.638448514715959e-06,
+      "loss": 0.0103,
+      "step": 10742
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.634897978731257e-06,
+      "loss": 0.0067,
+      "step": 10744
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.6313481224263215e-06,
+      "loss": 0.0105,
+      "step": 10746
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.627798946353884e-06,
+      "loss": 0.0062,
+      "step": 10748
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.624250451066584e-06,
+      "loss": 0.0146,
+      "step": 10750
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.6207026371169485e-06,
+      "loss": 0.0105,
+      "step": 10752
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.6171555050574035e-06,
+      "loss": 0.0085,
+      "step": 10754
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.613609055440256e-06,
+      "loss": 0.015,
+      "step": 10756
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.610063288817723e-06,
+      "loss": 0.0162,
+      "step": 10758
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.606518205741902e-06,
+      "loss": 0.0036,
+      "step": 10760
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.602973806764794e-06,
+      "loss": 0.006,
+      "step": 10762
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.599430092438285e-06,
+      "loss": 0.0117,
+      "step": 10764
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.595887063314158e-06,
+      "loss": 0.0151,
+      "step": 10766
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.5923447199440935e-06,
+      "loss": 0.008,
+      "step": 10768
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.588803062879663e-06,
+      "loss": 0.014,
+      "step": 10770
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.585262092672328e-06,
+      "loss": 0.0097,
+      "step": 10772
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.581721809873437e-06,
+      "loss": 0.0054,
+      "step": 10774
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.578182215034247e-06,
+      "loss": 0.0086,
+      "step": 10776
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.5746433087059e-06,
+      "loss": 0.0097,
+      "step": 10778
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.571105091439427e-06,
+      "loss": 0.0081,
+      "step": 10780
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.567567563785758e-06,
+      "loss": 0.0059,
+      "step": 10782
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.564030726295715e-06,
+      "loss": 0.0051,
+      "step": 10784
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.560494579520008e-06,
+      "loss": 0.0142,
+      "step": 10786
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.55695912400924e-06,
+      "loss": 0.015,
+      "step": 10788
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.553424360313909e-06,
+      "loss": 0.0122,
+      "step": 10790
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.549890288984408e-06,
+      "loss": 0.0113,
+      "step": 10792
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.54635691057102e-06,
+      "loss": 0.0076,
+      "step": 10794
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.542824225623914e-06,
+      "loss": 0.0106,
+      "step": 10796
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.539292234693158e-06,
+      "loss": 0.0074,
+      "step": 10798
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.535760938328714e-06,
+      "loss": 0.0049,
+      "step": 10800
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.532230337080429e-06,
+      "loss": 0.0107,
+      "step": 10802
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.52870043149804e-06,
+      "loss": 0.0073,
+      "step": 10804
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.5251712221311834e-06,
+      "loss": 0.0131,
+      "step": 10806
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.521642709529387e-06,
+      "loss": 0.0067,
+      "step": 10808
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.518114894242067e-06,
+      "loss": 0.0128,
+      "step": 10810
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.514587776818526e-06,
+      "loss": 0.0078,
+      "step": 10812
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.511061357807971e-06,
+      "loss": 0.0125,
+      "step": 10814
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.507535637759483e-06,
+      "loss": 0.0063,
+      "step": 10816
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.504010617222053e-06,
+      "loss": 0.0068,
+      "step": 10818
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.500486296744546e-06,
+      "loss": 0.0066,
+      "step": 10820
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.496962676875728e-06,
+      "loss": 0.0121,
+      "step": 10822
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.493439758164254e-06,
+      "loss": 0.0057,
+      "step": 10824
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.489917541158674e-06,
+      "loss": 0.0117,
+      "step": 10826
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.4863960264074215e-06,
+      "loss": 0.0059,
+      "step": 10828
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.482875214458816e-06,
+      "loss": 0.0098,
+      "step": 10830
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.479355105861081e-06,
+      "loss": 0.0161,
+      "step": 10832
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.475835701162326e-06,
+      "loss": 0.0062,
+      "step": 10834
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.472317000910545e-06,
+      "loss": 0.0122,
+      "step": 10836
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.468799005653629e-06,
+      "loss": 0.0156,
+      "step": 10838
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.465281715939359e-06,
+      "loss": 0.0084,
+      "step": 10840
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.461765132315402e-06,
+      "loss": 0.0037,
+      "step": 10842
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.458249255329311e-06,
+      "loss": 0.0048,
+      "step": 10844
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.454734085528541e-06,
+      "loss": 0.013,
+      "step": 10846
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.451219623460431e-06,
+      "loss": 0.0088,
+      "step": 10848
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.447705869672211e-06,
+      "loss": 0.0178,
+      "step": 10850
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.444192824710994e-06,
+      "loss": 0.0071,
+      "step": 10852
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.440680489123791e-06,
+      "loss": 0.0062,
+      "step": 10854
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.4371688634575024e-06,
+      "loss": 0.0117,
+      "step": 10856
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.433657948258912e-06,
+      "loss": 0.0089,
+      "step": 10858
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.430147744074693e-06,
+      "loss": 0.0055,
+      "step": 10860
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.426638251451414e-06,
+      "loss": 0.0092,
+      "step": 10862
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.423129470935531e-06,
+      "loss": 0.0051,
+      "step": 10864
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.41962140307339e-06,
+      "loss": 0.0111,
+      "step": 10866
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.4161140484112165e-06,
+      "loss": 0.0163,
+      "step": 10868
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.412607407495143e-06,
+      "loss": 0.0126,
+      "step": 10870
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.409101480871168e-06,
+      "loss": 0.0079,
+      "step": 10872
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.4055962690852025e-06,
+      "loss": 0.0057,
+      "step": 10874
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.402091772683027e-06,
+      "loss": 0.0079,
+      "step": 10876
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.39858799221032e-06,
+      "loss": 0.0072,
+      "step": 10878
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.395084928212648e-06,
+      "loss": 0.0107,
+      "step": 10880
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.391582581235468e-06,
+      "loss": 0.0132,
+      "step": 10882
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.388080951824121e-06,
+      "loss": 0.0103,
+      "step": 10884
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.38458004052383e-06,
+      "loss": 0.0076,
+      "step": 10886
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.38107984787972e-06,
+      "loss": 0.0104,
+      "step": 10888
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.377580374436801e-06,
+      "loss": 0.0071,
+      "step": 10890
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.374081620739959e-06,
+      "loss": 0.0075,
+      "step": 10892
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.3705835873339814e-06,
+      "loss": 0.0058,
+      "step": 10894
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.367086274763544e-06,
+      "loss": 0.0094,
+      "step": 10896
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.3635896835731945e-06,
+      "loss": 0.0072,
+      "step": 10898
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.36009381430739e-06,
+      "loss": 0.0105,
+      "step": 10900
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.356598667510453e-06,
+      "loss": 0.0052,
+      "step": 10902
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.35310424372661e-06,
+      "loss": 0.0088,
+      "step": 10904
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.349610543499973e-06,
+      "loss": 0.0075,
+      "step": 10906
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.346117567374531e-06,
+      "loss": 0.0109,
+      "step": 10908
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.34262531589417e-06,
+      "loss": 0.0062,
+      "step": 10910
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.339133789602666e-06,
+      "loss": 0.008,
+      "step": 10912
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.33564298904367e-06,
+      "loss": 0.0048,
+      "step": 10914
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.3321529147607224e-06,
+      "loss": 0.0058,
+      "step": 10916
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.328663567297261e-06,
+      "loss": 0.0063,
+      "step": 10918
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.325174947196601e-06,
+      "loss": 0.0078,
+      "step": 10920
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.321687055001953e-06,
+      "loss": 0.0065,
+      "step": 10922
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.318199891256399e-06,
+      "loss": 0.0201,
+      "step": 10924
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.314713456502928e-06,
+      "loss": 0.0052,
+      "step": 10926
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.3112277512843935e-06,
+      "loss": 0.0088,
+      "step": 10928
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.307742776143555e-06,
+      "loss": 0.0102,
+      "step": 10930
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.304258531623043e-06,
+      "loss": 0.0042,
+      "step": 10932
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.300775018265385e-06,
+      "loss": 0.0073,
+      "step": 10934
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.297292236612989e-06,
+      "loss": 0.0134,
+      "step": 10936
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.293810187208155e-06,
+      "loss": 0.011,
+      "step": 10938
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.290328870593062e-06,
+      "loss": 0.0084,
+      "step": 10940
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.286848287309774e-06,
+      "loss": 0.0066,
+      "step": 10942
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.283368437900247e-06,
+      "loss": 0.013,
+      "step": 10944
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.2798893229063245e-06,
+      "loss": 0.0103,
+      "step": 10946
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.276410942869723e-06,
+      "loss": 0.0053,
+      "step": 10948
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.272933298332059e-06,
+      "loss": 0.0124,
+      "step": 10950
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.269456389834825e-06,
+      "loss": 0.0061,
+      "step": 10952
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.265980217919408e-06,
+      "loss": 0.0062,
+      "step": 10954
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.262504783127071e-06,
+      "loss": 0.0097,
+      "step": 10956
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.259030085998962e-06,
+      "loss": 0.0066,
+      "step": 10958
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.25555612707612e-06,
+      "loss": 0.0064,
+      "step": 10960
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.2520829068994724e-06,
+      "loss": 0.0094,
+      "step": 10962
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.248610426009818e-06,
+      "loss": 0.0116,
+      "step": 10964
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.245138684947853e-06,
+      "loss": 0.006,
+      "step": 10966
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.241667684254157e-06,
+      "loss": 0.0085,
+      "step": 10968
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.238197424469187e-06,
+      "loss": 0.0092,
+      "step": 10970
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.234727906133287e-06,
+      "loss": 0.0129,
+      "step": 10972
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.23125912978669e-06,
+      "loss": 0.0088,
+      "step": 10974
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.227791095969512e-06,
+      "loss": 0.0081,
+      "step": 10976
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.224323805221755e-06,
+      "loss": 0.0089,
+      "step": 10978
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.220857258083296e-06,
+      "loss": 0.0062,
+      "step": 10980
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.21739145509391e-06,
+      "loss": 0.0169,
+      "step": 10982
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.213926396793241e-06,
+      "loss": 0.0072,
+      "step": 10984
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.210462083720833e-06,
+      "loss": 0.0093,
+      "step": 10986
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.206998516416099e-06,
+      "loss": 0.0104,
+      "step": 10988
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.203535695418348e-06,
+      "loss": 0.0099,
+      "step": 10990
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.200073621266765e-06,
+      "loss": 0.0167,
+      "step": 10992
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.196612294500426e-06,
+      "loss": 0.0058,
+      "step": 10994
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.1931517156582835e-06,
+      "loss": 0.0083,
+      "step": 10996
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.189691885279171e-06,
+      "loss": 0.0147,
+      "step": 10998
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.186232803901814e-06,
+      "loss": 0.0099,
+      "step": 11000
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.182774472064822e-06,
+      "loss": 0.0081,
+      "step": 11002
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.179316890306678e-06,
+      "loss": 0.0085,
+      "step": 11004
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.175860059165756e-06,
+      "loss": 0.0063,
+      "step": 11006
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.17240397918031e-06,
+      "loss": 0.0041,
+      "step": 11008
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.168948650888486e-06,
+      "loss": 0.0143,
+      "step": 11010
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.165494074828296e-06,
+      "loss": 0.007,
+      "step": 11012
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.1620402515376435e-06,
+      "loss": 0.0099,
+      "step": 11014
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.158587181554318e-06,
+      "loss": 0.0082,
+      "step": 11016
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.155134865415992e-06,
+      "loss": 0.0136,
+      "step": 11018
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.151683303660211e-06,
+      "loss": 0.0093,
+      "step": 11020
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.148232496824412e-06,
+      "loss": 0.0081,
+      "step": 11022
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.144782445445918e-06,
+      "loss": 0.0104,
+      "step": 11024
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.141333150061924e-06,
+      "loss": 0.0093,
+      "step": 11026
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.137884611209506e-06,
+      "loss": 0.0069,
+      "step": 11028
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.134436829425633e-06,
+      "loss": 0.0088,
+      "step": 11030
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.130989805247152e-06,
+      "loss": 0.0109,
+      "step": 11032
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.127543539210793e-06,
+      "loss": 0.0052,
+      "step": 11034
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.1240980318531595e-06,
+      "loss": 0.0092,
+      "step": 11036
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.12065328371075e-06,
+      "loss": 0.0189,
+      "step": 11038
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.117209295319931e-06,
+      "loss": 0.0108,
+      "step": 11040
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.113766067216967e-06,
+      "loss": 0.0086,
+      "step": 11042
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.110323599937985e-06,
+      "loss": 0.0156,
+      "step": 11044
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.1068818940190065e-06,
+      "loss": 0.0117,
+      "step": 11046
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.103440949995936e-06,
+      "loss": 0.014,
+      "step": 11048
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.100000768404554e-06,
+      "loss": 0.0044,
+      "step": 11050
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.096561349780518e-06,
+      "loss": 0.0054,
+      "step": 11052
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.093122694659377e-06,
+      "loss": 0.0062,
+      "step": 11054
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.089684803576551e-06,
+      "loss": 0.0118,
+      "step": 11056
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.0862476770673516e-06,
+      "loss": 0.0041,
+      "step": 11058
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.08281131566696e-06,
+      "loss": 0.0075,
+      "step": 11060
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.079375719910446e-06,
+      "loss": 0.0083,
+      "step": 11062
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.075940890332758e-06,
+      "loss": 0.0038,
+      "step": 11064
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.072506827468731e-06,
+      "loss": 0.0097,
+      "step": 11066
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.069073531853068e-06,
+      "loss": 0.0089,
+      "step": 11068
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.065641004020359e-06,
+      "loss": 0.011,
+      "step": 11070
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.062209244505076e-06,
+      "loss": 0.0109,
+      "step": 11072
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.058778253841577e-06,
+      "loss": 0.0087,
+      "step": 11074
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.055348032564081e-06,
+      "loss": 0.0095,
+      "step": 11076
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.051918581206708e-06,
+      "loss": 0.0103,
+      "step": 11078
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.0484899003034515e-06,
+      "loss": 0.0099,
+      "step": 11080
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.0450619903881805e-06,
+      "loss": 0.0054,
+      "step": 11082
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.041634851994642e-06,
+      "loss": 0.0048,
+      "step": 11084
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.038208485656471e-06,
+      "loss": 0.0101,
+      "step": 11086
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.034782891907182e-06,
+      "loss": 0.0079,
+      "step": 11088
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.031358071280165e-06,
+      "loss": 0.0067,
+      "step": 11090
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.02793402430869e-06,
+      "loss": 0.0061,
+      "step": 11092
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.024510751525908e-06,
+      "loss": 0.0131,
+      "step": 11094
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.0210882534648455e-06,
+      "loss": 0.0082,
+      "step": 11096
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.017666530658416e-06,
+      "loss": 0.0057,
+      "step": 11098
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.014245583639403e-06,
+      "loss": 0.0064,
+      "step": 11100
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.010825412940477e-06,
+      "loss": 0.0124,
+      "step": 11102
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.007406019094185e-06,
+      "loss": 0.0118,
+      "step": 11104
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.003987402632955e-06,
+      "loss": 0.0091,
+      "step": 11106
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.000569564089086e-06,
+      "loss": 0.0134,
+      "step": 11108
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.997152503994769e-06,
+      "loss": 0.0163,
+      "step": 11110
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.993736222882058e-06,
+      "loss": 0.0102,
+      "step": 11112
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.990320721282902e-06,
+      "loss": 0.0057,
+      "step": 11114
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.986905999729114e-06,
+      "loss": 0.0087,
+      "step": 11116
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.9834920587523935e-06,
+      "loss": 0.004,
+      "step": 11118
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.9800788988843205e-06,
+      "loss": 0.0056,
+      "step": 11120
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.976666520656352e-06,
+      "loss": 0.0083,
+      "step": 11122
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.973254924599818e-06,
+      "loss": 0.0104,
+      "step": 11124
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.9698441112459266e-06,
+      "loss": 0.0136,
+      "step": 11126
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.966434081125769e-06,
+      "loss": 0.0049,
+      "step": 11128
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.96302483477032e-06,
+      "loss": 0.0073,
+      "step": 11130
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.959616372710416e-06,
+      "loss": 0.0115,
+      "step": 11132
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.956208695476786e-06,
+      "loss": 0.0112,
+      "step": 11134
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.952801803600033e-06,
+      "loss": 0.0104,
+      "step": 11136
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.949395697610634e-06,
+      "loss": 0.0085,
+      "step": 11138
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.945990378038941e-06,
+      "loss": 0.013,
+      "step": 11140
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.942585845415192e-06,
+      "loss": 0.0053,
+      "step": 11142
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.9391821002694996e-06,
+      "loss": 0.0064,
+      "step": 11144
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.935779143131856e-06,
+      "loss": 0.0063,
+      "step": 11146
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.932376974532121e-06,
+      "loss": 0.0125,
+      "step": 11148
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.928975595000044e-06,
+      "loss": 0.0064,
+      "step": 11150
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.925575005065241e-06,
+      "loss": 0.0174,
+      "step": 11152
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.922175205257215e-06,
+      "loss": 0.0052,
+      "step": 11154
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.9187761961053335e-06,
+      "loss": 0.0023,
+      "step": 11156
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.915377978138853e-06,
+      "loss": 0.0089,
+      "step": 11158
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.911980551886902e-06,
+      "loss": 0.0051,
+      "step": 11160
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.908583917878489e-06,
+      "loss": 0.01,
+      "step": 11162
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.9051880766424885e-06,
+      "loss": 0.0129,
+      "step": 11164
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.901793028707666e-06,
+      "loss": 0.0093,
+      "step": 11166
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.898398774602651e-06,
+      "loss": 0.0065,
+      "step": 11168
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.895005314855959e-06,
+      "loss": 0.0135,
+      "step": 11170
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.891612649995973e-06,
+      "loss": 0.0138,
+      "step": 11172
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.888220780550958e-06,
+      "loss": 0.0057,
+      "step": 11174
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.884829707049057e-06,
+      "loss": 0.0112,
+      "step": 11176
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.881439430018288e-06,
+      "loss": 0.0129,
+      "step": 11178
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.878049949986539e-06,
+      "loss": 0.0106,
+      "step": 11180
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.874661267481574e-06,
+      "loss": 0.007,
+      "step": 11182
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.871273383031043e-06,
+      "loss": 0.005,
+      "step": 11184
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.867886297162467e-06,
+      "loss": 0.0087,
+      "step": 11186
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.864500010403234e-06,
+      "loss": 0.0104,
+      "step": 11188
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.861114523280619e-06,
+      "loss": 0.0109,
+      "step": 11190
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.857729836321772e-06,
+      "loss": 0.0042,
+      "step": 11192
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.854345950053711e-06,
+      "loss": 0.0053,
+      "step": 11194
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.850962865003327e-06,
+      "loss": 0.012,
+      "step": 11196
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.8475805816974e-06,
+      "loss": 0.0092,
+      "step": 11198
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.844199100662578e-06,
+      "loss": 0.0084,
+      "step": 11200
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.840818422425376e-06,
+      "loss": 0.0074,
+      "step": 11202
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.837438547512197e-06,
+      "loss": 0.0122,
+      "step": 11204
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.834059476449312e-06,
+      "loss": 0.0114,
+      "step": 11206
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.830681209762873e-06,
+      "loss": 0.0086,
+      "step": 11208
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.827303747978898e-06,
+      "loss": 0.0051,
+      "step": 11210
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.823927091623278e-06,
+      "loss": 0.0142,
+      "step": 11212
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.820551241221791e-06,
+      "loss": 0.0071,
+      "step": 11214
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.8171761973000845e-06,
+      "loss": 0.0087,
+      "step": 11216
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.813801960383672e-06,
+      "loss": 0.0064,
+      "step": 11218
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.81042853099795e-06,
+      "loss": 0.0104,
+      "step": 11220
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.807055909668193e-06,
+      "loss": 0.0082,
+      "step": 11222
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.80368409691954e-06,
+      "loss": 0.0087,
+      "step": 11224
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.800313093277002e-06,
+      "loss": 0.0078,
+      "step": 11226
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.7969428992654775e-06,
+      "loss": 0.0045,
+      "step": 11228
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.793573515409729e-06,
+      "loss": 0.0042,
+      "step": 11230
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.7902049422344e-06,
+      "loss": 0.0071,
+      "step": 11232
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.786837180263994e-06,
+      "loss": 0.0095,
+      "step": 11234
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.783470230022908e-06,
+      "loss": 0.0131,
+      "step": 11236
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.780104092035392e-06,
+      "loss": 0.0091,
+      "step": 11238
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.776738766825585e-06,
+      "loss": 0.0087,
+      "step": 11240
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.773374254917491e-06,
+      "loss": 0.012,
+      "step": 11242
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.7700105568349905e-06,
+      "loss": 0.0114,
+      "step": 11244
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.766647673101839e-06,
+      "loss": 0.0095,
+      "step": 11246
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.763285604241665e-06,
+      "loss": 0.0039,
+      "step": 11248
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.759924350777968e-06,
+      "loss": 0.0097,
+      "step": 11250
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.756563913234113e-06,
+      "loss": 0.0119,
+      "step": 11252
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.753204292133352e-06,
+      "loss": 0.0145,
+      "step": 11254
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.749845487998806e-06,
+      "loss": 0.0067,
+      "step": 11256
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.7464875013534614e-06,
+      "loss": 0.0075,
+      "step": 11258
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.743130332720184e-06,
+      "loss": 0.0043,
+      "step": 11260
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.739773982621712e-06,
+      "loss": 0.0077,
+      "step": 11262
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.736418451580656e-06,
+      "loss": 0.0097,
+      "step": 11264
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.733063740119497e-06,
+      "loss": 0.0044,
+      "step": 11266
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.729709848760584e-06,
+      "loss": 0.0041,
+      "step": 11268
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.726356778026148e-06,
+      "loss": 0.0147,
+      "step": 11270
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.723004528438291e-06,
+      "loss": 0.0047,
+      "step": 11272
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.719653100518976e-06,
+      "loss": 0.0129,
+      "step": 11274
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.7163024947900505e-06,
+      "loss": 0.0111,
+      "step": 11276
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.712952711773233e-06,
+      "loss": 0.0086,
+      "step": 11278
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.7096037519901065e-06,
+      "loss": 0.0077,
+      "step": 11280
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.706255615962127e-06,
+      "loss": 0.0088,
+      "step": 11282
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.702908304210625e-06,
+      "loss": 0.0122,
+      "step": 11284
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.6995618172568066e-06,
+      "loss": 0.011,
+      "step": 11286
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.696216155621748e-06,
+      "loss": 0.0049,
+      "step": 11288
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.692871319826385e-06,
+      "loss": 0.0045,
+      "step": 11290
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.689527310391544e-06,
+      "loss": 0.0065,
+      "step": 11292
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.686184127837905e-06,
+      "loss": 0.0143,
+      "step": 11294
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.682841772686033e-06,
+      "loss": 0.0078,
+      "step": 11296
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.679500245456352e-06,
+      "loss": 0.0071,
+      "step": 11298
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.676159546669167e-06,
+      "loss": 0.0039,
+      "step": 11300
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.672819676844649e-06,
+      "loss": 0.0079,
+      "step": 11302
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.669480636502847e-06,
+      "loss": 0.0052,
+      "step": 11304
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.666142426163667e-06,
+      "loss": 0.0088,
+      "step": 11306
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.662805046346901e-06,
+      "loss": 0.0042,
+      "step": 11308
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.6594684975721974e-06,
+      "loss": 0.0102,
+      "step": 11310
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.656132780359089e-06,
+      "loss": 0.0039,
+      "step": 11312
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.652797895226966e-06,
+      "loss": 0.0132,
+      "step": 11314
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.649463842695099e-06,
+      "loss": 0.0032,
+      "step": 11316
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.646130623282625e-06,
+      "loss": 0.0061,
+      "step": 11318
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.642798237508555e-06,
+      "loss": 0.0055,
+      "step": 11320
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.639466685891766e-06,
+      "loss": 0.0071,
+      "step": 11322
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.636135968951e-06,
+      "loss": 0.0126,
+      "step": 11324
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.632806087204878e-06,
+      "loss": 0.0097,
+      "step": 11326
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.629477041171895e-06,
+      "loss": 0.0104,
+      "step": 11328
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.6261488313703985e-06,
+      "loss": 0.0064,
+      "step": 11330
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.6228214583186205e-06,
+      "loss": 0.0154,
+      "step": 11332
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.619494922534663e-06,
+      "loss": 0.0102,
+      "step": 11334
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.616169224536489e-06,
+      "loss": 0.013,
+      "step": 11336
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.612844364841931e-06,
+      "loss": 0.0042,
+      "step": 11338
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.6095203439686985e-06,
+      "loss": 0.0059,
+      "step": 11340
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.606197162434368e-06,
+      "loss": 0.0044,
+      "step": 11342
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.602874820756388e-06,
+      "loss": 0.007,
+      "step": 11344
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.599553319452065e-06,
+      "loss": 0.0096,
+      "step": 11346
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.596232659038588e-06,
+      "loss": 0.0086,
+      "step": 11348
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.5929128400330035e-06,
+      "loss": 0.0071,
+      "step": 11350
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.5895938629522396e-06,
+      "loss": 0.0156,
+      "step": 11352
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.58627572831308e-06,
+      "loss": 0.0077,
+      "step": 11354
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.582958436632185e-06,
+      "loss": 0.0052,
+      "step": 11356
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.579641988426084e-06,
+      "loss": 0.008,
+      "step": 11358
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.576326384211177e-06,
+      "loss": 0.006,
+      "step": 11360
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.573011624503721e-06,
+      "loss": 0.0058,
+      "step": 11362
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.569697709819857e-06,
+      "loss": 0.0174,
+      "step": 11364
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.566384640675579e-06,
+      "loss": 0.0012,
+      "step": 11366
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.563072417586767e-06,
+      "loss": 0.0041,
+      "step": 11368
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.559761041069149e-06,
+      "loss": 0.0123,
+      "step": 11370
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.556450511638336e-06,
+      "loss": 0.0089,
+      "step": 11372
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.553140829809804e-06,
+      "loss": 0.0056,
+      "step": 11374
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.5498319960988975e-06,
+      "loss": 0.0022,
+      "step": 11376
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.546524011020826e-06,
+      "loss": 0.0062,
+      "step": 11378
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.543216875090663e-06,
+      "loss": 0.0078,
+      "step": 11380
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.539910588823359e-06,
+      "loss": 0.0044,
+      "step": 11382
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.53660515273373e-06,
+      "loss": 0.0097,
+      "step": 11384
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.533300567336454e-06,
+      "loss": 0.0136,
+      "step": 11386
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.52999683314608e-06,
+      "loss": 0.0097,
+      "step": 11388
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.5266939506770305e-06,
+      "loss": 0.0061,
+      "step": 11390
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.523391920443584e-06,
+      "loss": 0.0036,
+      "step": 11392
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.5200907429598906e-06,
+      "loss": 0.0095,
+      "step": 11394
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.516790418739972e-06,
+      "loss": 0.0091,
+      "step": 11396
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.513490948297713e-06,
+      "loss": 0.0052,
+      "step": 11398
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.51019233214687e-06,
+      "loss": 0.0093,
+      "step": 11400
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.5068945708010556e-06,
+      "loss": 0.0086,
+      "step": 11402
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.503597664773761e-06,
+      "loss": 0.0096,
+      "step": 11404
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.500301614578343e-06,
+      "loss": 0.0039,
+      "step": 11406
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.4970064207280175e-06,
+      "loss": 0.0068,
+      "step": 11408
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.493712083735868e-06,
+      "loss": 0.0082,
+      "step": 11410
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.490418604114852e-06,
+      "loss": 0.0083,
+      "step": 11412
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.487125982377789e-06,
+      "loss": 0.0079,
+      "step": 11414
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.483834219037369e-06,
+      "loss": 0.0095,
+      "step": 11416
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.480543314606138e-06,
+      "loss": 0.0043,
+      "step": 11418
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.477253269596522e-06,
+      "loss": 0.0054,
+      "step": 11420
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.473964084520799e-06,
+      "loss": 0.0028,
+      "step": 11422
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.470675759891126e-06,
+      "loss": 0.006,
+      "step": 11424
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.467388296219516e-06,
+      "loss": 0.0034,
+      "step": 11426
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.4641016940178535e-06,
+      "loss": 0.0079,
+      "step": 11428
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.460815953797889e-06,
+      "loss": 0.0072,
+      "step": 11430
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.4575310760712395e-06,
+      "loss": 0.0098,
+      "step": 11432
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.4542470613493824e-06,
+      "loss": 0.0097,
+      "step": 11434
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.450963910143662e-06,
+      "loss": 0.0127,
+      "step": 11436
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.447681622965292e-06,
+      "loss": 0.0068,
+      "step": 11438
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.444400200325353e-06,
+      "loss": 0.0026,
+      "step": 11440
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.441119642734781e-06,
+      "loss": 0.0093,
+      "step": 11442
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.437839950704388e-06,
+      "loss": 0.0104,
+      "step": 11444
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.434561124744849e-06,
+      "loss": 0.0123,
+      "step": 11446
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.431283165366701e-06,
+      "loss": 0.0111,
+      "step": 11448
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.428006073080342e-06,
+      "loss": 0.0109,
+      "step": 11450
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.424729848396045e-06,
+      "loss": 0.0086,
+      "step": 11452
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.421454491823942e-06,
+      "loss": 0.0134,
+      "step": 11454
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.418180003874036e-06,
+      "loss": 0.003,
+      "step": 11456
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.4149063850561825e-06,
+      "loss": 0.0056,
+      "step": 11458
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.411633635880112e-06,
+      "loss": 0.0057,
+      "step": 11460
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.408361756855419e-06,
+      "loss": 0.0064,
+      "step": 11462
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.405090748491558e-06,
+      "loss": 0.0122,
+      "step": 11464
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.4018206112978475e-06,
+      "loss": 0.0118,
+      "step": 11466
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.398551345783474e-06,
+      "loss": 0.0109,
+      "step": 11468
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.395282952457489e-06,
+      "loss": 0.0077,
+      "step": 11470
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.392015431828809e-06,
+      "loss": 0.0095,
+      "step": 11472
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.388748784406205e-06,
+      "loss": 0.004,
+      "step": 11474
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.385483010698326e-06,
+      "loss": 0.006,
+      "step": 11476
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.382218111213671e-06,
+      "loss": 0.0066,
+      "step": 11478
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.378954086460616e-06,
+      "loss": 0.0054,
+      "step": 11480
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.37569093694739e-06,
+      "loss": 0.0054,
+      "step": 11482
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.372428663182091e-06,
+      "loss": 0.0091,
+      "step": 11484
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.369167265672681e-06,
+      "loss": 0.0098,
+      "step": 11486
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.36590674492699e-06,
+      "loss": 0.0049,
+      "step": 11488
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.362647101452699e-06,
+      "loss": 0.0075,
+      "step": 11490
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.359388335757358e-06,
+      "loss": 0.0094,
+      "step": 11492
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.3561304483483855e-06,
+      "loss": 0.0071,
+      "step": 11494
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.352873439733063e-06,
+      "loss": 0.0049,
+      "step": 11496
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.349617310418523e-06,
+      "loss": 0.0124,
+      "step": 11498
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.346362060911774e-06,
+      "loss": 0.0144,
+      "step": 11500
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.343107691719688e-06,
+      "loss": 0.0186,
+      "step": 11502
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.339854203348987e-06,
+      "loss": 0.0121,
+      "step": 11504
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.3366015963062714e-06,
+      "loss": 0.0165,
+      "step": 11506
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.333349871097988e-06,
+      "loss": 0.0066,
+      "step": 11508
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.330099028230462e-06,
+      "loss": 0.0093,
+      "step": 11510
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.326849068209877e-06,
+      "loss": 0.0078,
+      "step": 11512
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.323599991542269e-06,
+      "loss": 0.0138,
+      "step": 11514
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.320351798733547e-06,
+      "loss": 0.0073,
+      "step": 11516
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.317104490289484e-06,
+      "loss": 0.0069,
+      "step": 11518
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.313858066715707e-06,
+      "loss": 0.007,
+      "step": 11520
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.310612528517706e-06,
+      "loss": 0.012,
+      "step": 11522
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.307367876200839e-06,
+      "loss": 0.0055,
+      "step": 11524
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.3041241102703225e-06,
+      "loss": 0.0059,
+      "step": 11526
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.30088123123124e-06,
+      "loss": 0.0056,
+      "step": 11528
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.297639239588526e-06,
+      "loss": 0.0045,
+      "step": 11530
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.2943981358469885e-06,
+      "loss": 0.0111,
+      "step": 11532
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.291157920511289e-06,
+      "loss": 0.0085,
+      "step": 11534
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.287918594085957e-06,
+      "loss": 0.007,
+      "step": 11536
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.284680157075374e-06,
+      "loss": 0.0066,
+      "step": 11538
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.281442609983793e-06,
+      "loss": 0.0082,
+      "step": 11540
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.278205953315327e-06,
+      "loss": 0.0046,
+      "step": 11542
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.2749701875739505e-06,
+      "loss": 0.0043,
+      "step": 11544
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.271735313263493e-06,
+      "loss": 0.0065,
+      "step": 11546
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.268501330887644e-06,
+      "loss": 0.0097,
+      "step": 11548
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.2652682409499666e-06,
+      "loss": 0.006,
+      "step": 11550
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.262036043953878e-06,
+      "loss": 0.0099,
+      "step": 11552
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.25880474040265e-06,
+      "loss": 0.0189,
+      "step": 11554
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.255574330799426e-06,
+      "loss": 0.0045,
+      "step": 11556
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.252344815647202e-06,
+      "loss": 0.0038,
+      "step": 11558
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.249116195448845e-06,
+      "loss": 0.0036,
+      "step": 11560
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.245888470707074e-06,
+      "loss": 0.0075,
+      "step": 11562
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.242661641924461e-06,
+      "loss": 0.0086,
+      "step": 11564
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.239435709603455e-06,
+      "loss": 0.0082,
+      "step": 11566
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.2362106742463635e-06,
+      "loss": 0.0134,
+      "step": 11568
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.232986536355339e-06,
+      "loss": 0.0052,
+      "step": 11570
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.229763296432409e-06,
+      "loss": 0.004,
+      "step": 11572
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.226540954979461e-06,
+      "loss": 0.0085,
+      "step": 11574
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.223319512498233e-06,
+      "loss": 0.0068,
+      "step": 11576
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.220098969490326e-06,
+      "loss": 0.0092,
+      "step": 11578
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.216879326457206e-06,
+      "loss": 0.0054,
+      "step": 11580
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.213660583900198e-06,
+      "loss": 0.0015,
+      "step": 11582
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.210442742320485e-06,
+      "loss": 0.0137,
+      "step": 11584
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.207225802219105e-06,
+      "loss": 0.0078,
+      "step": 11586
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.204009764096966e-06,
+      "loss": 0.0043,
+      "step": 11588
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.200794628454823e-06,
+      "loss": 0.0084,
+      "step": 11590
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.197580395793305e-06,
+      "loss": 0.0045,
+      "step": 11592
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.194367066612884e-06,
+      "loss": 0.0051,
+      "step": 11594
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.191154641413905e-06,
+      "loss": 0.0075,
+      "step": 11596
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.187943120696567e-06,
+      "loss": 0.0052,
+      "step": 11598
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.184732504960931e-06,
+      "loss": 0.0052,
+      "step": 11600
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.18152279470691e-06,
+      "loss": 0.0042,
+      "step": 11602
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.178313990434281e-06,
+      "loss": 0.0059,
+      "step": 11604
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.1751060926426775e-06,
+      "loss": 0.0061,
+      "step": 11606
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.1718991018316015e-06,
+      "loss": 0.0072,
+      "step": 11608
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.168693018500396e-06,
+      "loss": 0.0083,
+      "step": 11610
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.1654878431482784e-06,
+      "loss": 0.009,
+      "step": 11612
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.162283576274317e-06,
+      "loss": 0.0079,
+      "step": 11614
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.159080218377447e-06,
+      "loss": 0.0076,
+      "step": 11616
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.15587776995645e-06,
+      "loss": 0.0058,
+      "step": 11618
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.152676231509968e-06,
+      "loss": 0.0074,
+      "step": 11620
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.149475603536509e-06,
+      "loss": 0.0151,
+      "step": 11622
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.14627588653444e-06,
+      "loss": 0.0109,
+      "step": 11624
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.143077081001973e-06,
+      "loss": 0.0078,
+      "step": 11626
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.13987918743719e-06,
+      "loss": 0.0077,
+      "step": 11628
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.136682206338031e-06,
+      "loss": 0.0066,
+      "step": 11630
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.133486138202288e-06,
+      "loss": 0.0096,
+      "step": 11632
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.1302909835276084e-06,
+      "loss": 0.007,
+      "step": 11634
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.127096742811506e-06,
+      "loss": 0.0073,
+      "step": 11636
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.12390341655135e-06,
+      "loss": 0.0064,
+      "step": 11638
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.1207110052443675e-06,
+      "loss": 0.0103,
+      "step": 11640
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.117519509387634e-06,
+      "loss": 0.0043,
+      "step": 11642
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.114328929478098e-06,
+      "loss": 0.0103,
+      "step": 11644
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.111139266012551e-06,
+      "loss": 0.0079,
+      "step": 11646
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.107950519487653e-06,
+      "loss": 0.0045,
+      "step": 11648
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.1047626903999106e-06,
+      "loss": 0.0066,
+      "step": 11650
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.101575779245696e-06,
+      "loss": 0.0126,
+      "step": 11652
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.098389786521234e-06,
+      "loss": 0.0107,
+      "step": 11654
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.095204712722614e-06,
+      "loss": 0.0064,
+      "step": 11656
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.092020558345771e-06,
+      "loss": 0.0102,
+      "step": 11658
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.088837323886504e-06,
+      "loss": 0.008,
+      "step": 11660
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.0856550098404645e-06,
+      "loss": 0.006,
+      "step": 11662
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.082473616703167e-06,
+      "loss": 0.0055,
+      "step": 11664
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.0792931449699745e-06,
+      "loss": 0.0074,
+      "step": 11666
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.076113595136113e-06,
+      "loss": 0.0069,
+      "step": 11668
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.0729349676966625e-06,
+      "loss": 0.0075,
+      "step": 11670
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.069757263146562e-06,
+      "loss": 0.0077,
+      "step": 11672
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.066580481980603e-06,
+      "loss": 0.0055,
+      "step": 11674
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.06340462469343e-06,
+      "loss": 0.0113,
+      "step": 11676
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.060229691779552e-06,
+      "loss": 0.0084,
+      "step": 11678
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.057055683733334e-06,
+      "loss": 0.007,
+      "step": 11680
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.053882601048987e-06,
+      "loss": 0.0068,
+      "step": 11682
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.050710444220585e-06,
+      "loss": 0.0025,
+      "step": 11684
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.047539213742064e-06,
+      "loss": 0.012,
+      "step": 11686
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.044368910107204e-06,
+      "loss": 0.0068,
+      "step": 11688
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.041199533809641e-06,
+      "loss": 0.0075,
+      "step": 11690
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.038031085342875e-06,
+      "loss": 0.009,
+      "step": 11692
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.034863565200259e-06,
+      "loss": 0.0105,
+      "step": 11694
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.031696973875003e-06,
+      "loss": 0.0142,
+      "step": 11696
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.028531311860161e-06,
+      "loss": 0.0038,
+      "step": 11698
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.02536657964866e-06,
+      "loss": 0.0039,
+      "step": 11700
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.022202777733264e-06,
+      "loss": 0.0039,
+      "step": 11702
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.01903990660661e-06,
+      "loss": 0.0032,
+      "step": 11704
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.015877966761173e-06,
+      "loss": 0.0074,
+      "step": 11706
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.0127169586892955e-06,
+      "loss": 0.0067,
+      "step": 11708
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.00955688288317e-06,
+      "loss": 0.0041,
+      "step": 11710
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.006397739834848e-06,
+      "loss": 0.0051,
+      "step": 11712
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.003239530036226e-06,
+      "loss": 0.007,
+      "step": 11714
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.0000822539790675e-06,
+      "loss": 0.0054,
+      "step": 11716
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.9969259121549805e-06,
+      "loss": 0.0051,
+      "step": 11718
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.99377050505543e-06,
+      "loss": 0.0128,
+      "step": 11720
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.990616033171738e-06,
+      "loss": 0.0083,
+      "step": 11722
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.987462496995082e-06,
+      "loss": 0.0076,
+      "step": 11724
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.984309897016495e-06,
+      "loss": 0.0052,
+      "step": 11726
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.981158233726854e-06,
+      "loss": 0.0057,
+      "step": 11728
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.978007507616903e-06,
+      "loss": 0.0045,
+      "step": 11730
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.974857719177227e-06,
+      "loss": 0.0074,
+      "step": 11732
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.971708868898279e-06,
+      "loss": 0.0034,
+      "step": 11734
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.9685609572703544e-06,
+      "loss": 0.004,
+      "step": 11736
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.965413984783609e-06,
+      "loss": 0.0062,
+      "step": 11738
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.9622679519280504e-06,
+      "loss": 0.0044,
+      "step": 11740
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.959122859193543e-06,
+      "loss": 0.0083,
+      "step": 11742
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.9559787070698e-06,
+      "loss": 0.006,
+      "step": 11744
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.952835496046383e-06,
+      "loss": 0.0018,
+      "step": 11746
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.949693226612722e-06,
+      "loss": 0.0035,
+      "step": 11748
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.946551899258093e-06,
+      "loss": 0.0053,
+      "step": 11750
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.943411514471619e-06,
+      "loss": 0.009,
+      "step": 11752
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.9402720727422835e-06,
+      "loss": 0.006,
+      "step": 11754
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 3.937133574558925e-06,
+      "loss": 0.0121,
+      "step": 11756
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.933996020410231e-06,
+      "loss": 0.0074,
+      "step": 11758
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.930859410784742e-06,
+      "loss": 0.0041,
+      "step": 11760
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.927723746170848e-06,
+      "loss": 0.0058,
+      "step": 11762
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.9245890270568e-06,
+      "loss": 0.0102,
+      "step": 11764
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.921455253930699e-06,
+      "loss": 0.0054,
+      "step": 11766
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.918322427280493e-06,
+      "loss": 0.0042,
+      "step": 11768
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.9151905475939886e-06,
+      "loss": 0.0079,
+      "step": 11770
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.912059615358849e-06,
+      "loss": 0.0049,
+      "step": 11772
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.908929631062579e-06,
+      "loss": 0.005,
+      "step": 11774
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.905800595192538e-06,
+      "loss": 0.0103,
+      "step": 11776
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.902672508235945e-06,
+      "loss": 0.0124,
+      "step": 11778
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.899545370679867e-06,
+      "loss": 0.0067,
+      "step": 11780
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.896419183011226e-06,
+      "loss": 0.0107,
+      "step": 11782
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.893293945716786e-06,
+      "loss": 0.0078,
+      "step": 11784
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.89016965928318e-06,
+      "loss": 0.007,
+      "step": 11786
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.887046324196873e-06,
+      "loss": 0.0039,
+      "step": 11788
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.8839239409442e-06,
+      "loss": 0.0046,
+      "step": 11790
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.8808025100113335e-06,
+      "loss": 0.0074,
+      "step": 11792
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.877682031884308e-06,
+      "loss": 0.0121,
+      "step": 11794
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.874562507049005e-06,
+      "loss": 0.0084,
+      "step": 11796
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.871443935991161e-06,
+      "loss": 0.0074,
+      "step": 11798
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.868326319196362e-06,
+      "loss": 0.0057,
+      "step": 11800
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.865209657150036e-06,
+      "loss": 0.0042,
+      "step": 11802
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.8620939503374775e-06,
+      "loss": 0.0046,
+      "step": 11804
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.85897919924383e-06,
+      "loss": 0.0045,
+      "step": 11806
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.855865404354074e-06,
+      "loss": 0.0045,
+      "step": 11808
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.852752566153059e-06,
+      "loss": 0.0062,
+      "step": 11810
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.849640685125473e-06,
+      "loss": 0.0094,
+      "step": 11812
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.846529761755867e-06,
+      "loss": 0.0069,
+      "step": 11814
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.8434197965286304e-06,
+      "loss": 0.0027,
+      "step": 11816
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.840310789928005e-06,
+      "loss": 0.0061,
+      "step": 11818
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.83720274243809e-06,
+      "loss": 0.0053,
+      "step": 11820
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.834095654542836e-06,
+      "loss": 0.0128,
+      "step": 11822
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.830989526726036e-06,
+      "loss": 0.0097,
+      "step": 11824
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.8278843594713365e-06,
+      "loss": 0.0048,
+      "step": 11826
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.824780153262242e-06,
+      "loss": 0.0069,
+      "step": 11828
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.821676908582098e-06,
+      "loss": 0.0068,
+      "step": 11830
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.8185746259141e-06,
+      "loss": 0.0068,
+      "step": 11832
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.815473305741299e-06,
+      "loss": 0.0024,
+      "step": 11834
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.8123729485465953e-06,
+      "loss": 0.0063,
+      "step": 11836
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.8092735548127413e-06,
+      "loss": 0.0139,
+      "step": 11838
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.80617512502233e-06,
+      "loss": 0.0045,
+      "step": 11840
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.8030776596578177e-06,
+      "loss": 0.0048,
+      "step": 11842
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.7999811592014956e-06,
+      "loss": 0.0071,
+      "step": 11844
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.79688562413552e-06,
+      "loss": 0.0088,
+      "step": 11846
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.793791054941882e-06,
+      "loss": 0.0034,
+      "step": 11848
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.7906974521024343e-06,
+      "loss": 0.0032,
+      "step": 11850
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.7876048160988734e-06,
+      "loss": 0.0063,
+      "step": 11852
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.7845131474127493e-06,
+      "loss": 0.0059,
+      "step": 11854
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.7814224465254525e-06,
+      "loss": 0.0053,
+      "step": 11856
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.7783327139182357e-06,
+      "loss": 0.0126,
+      "step": 11858
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.775243950072187e-06,
+      "loss": 0.0067,
+      "step": 11860
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.772156155468257e-06,
+      "loss": 0.0061,
+      "step": 11862
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.7690693305872327e-06,
+      "loss": 0.0047,
+      "step": 11864
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.7659834759097603e-06,
+      "loss": 0.0024,
+      "step": 11866
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.7628985919163284e-06,
+      "loss": 0.0093,
+      "step": 11868
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.7598146790872825e-06,
+      "loss": 0.0043,
+      "step": 11870
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.7567317379028077e-06,
+      "loss": 0.0055,
+      "step": 11872
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.7536497688429384e-06,
+      "loss": 0.0022,
+      "step": 11874
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.750568772387564e-06,
+      "loss": 0.0065,
+      "step": 11876
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.7474887490164213e-06,
+      "loss": 0.0083,
+      "step": 11878
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.744409699209088e-06,
+      "loss": 0.0071,
+      "step": 11880
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.741331623444999e-06,
+      "loss": 0.0079,
+      "step": 11882
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.7382545222034385e-06,
+      "loss": 0.0039,
+      "step": 11884
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.735178395963529e-06,
+      "loss": 0.0046,
+      "step": 11886
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.732103245204245e-06,
+      "loss": 0.0032,
+      "step": 11888
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.729029070404414e-06,
+      "loss": 0.0051,
+      "step": 11890
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.725955872042709e-06,
+      "loss": 0.0062,
+      "step": 11892
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.722883650597654e-06,
+      "loss": 0.0053,
+      "step": 11894
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.719812406547609e-06,
+      "loss": 0.005,
+      "step": 11896
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.716742140370799e-06,
+      "loss": 0.0156,
+      "step": 11898
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.7136728525452803e-06,
+      "loss": 0.007,
+      "step": 11900
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.710604543548971e-06,
+      "loss": 0.0022,
+      "step": 11902
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.707537213859623e-06,
+      "loss": 0.0054,
+      "step": 11904
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.7044708639548477e-06,
+      "loss": 0.0023,
+      "step": 11906
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.701405494312099e-06,
+      "loss": 0.0041,
+      "step": 11908
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.6983411054086804e-06,
+      "loss": 0.0046,
+      "step": 11910
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.6952776977217355e-06,
+      "loss": 0.0056,
+      "step": 11912
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.6922152717282667e-06,
+      "loss": 0.0034,
+      "step": 11914
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.68915382790511e-06,
+      "loss": 0.0014,
+      "step": 11916
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.686093366728962e-06,
+      "loss": 0.0099,
+      "step": 11918
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.683033888676354e-06,
+      "loss": 0.0078,
+      "step": 11920
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.679975394223673e-06,
+      "loss": 0.0072,
+      "step": 11922
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.6769178838471508e-06,
+      "loss": 0.006,
+      "step": 11924
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.6738613580228664e-06,
+      "loss": 0.0055,
+      "step": 11926
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.6708058172267434e-06,
+      "loss": 0.008,
+      "step": 11928
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.667751261934549e-06,
+      "loss": 0.009,
+      "step": 11930
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.6646976926219025e-06,
+      "loss": 0.0031,
+      "step": 11932
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.6616451097642734e-06,
+      "loss": 0.0076,
+      "step": 11934
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.6585935138369644e-06,
+      "loss": 0.0048,
+      "step": 11936
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.655542905315135e-06,
+      "loss": 0.0109,
+      "step": 11938
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.6524932846737926e-06,
+      "loss": 0.0058,
+      "step": 11940
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.6494446523877835e-06,
+      "loss": 0.0073,
+      "step": 11942
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.646397008931799e-06,
+      "loss": 0.0032,
+      "step": 11944
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.643350354780384e-06,
+      "loss": 0.0053,
+      "step": 11946
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.6403046904079255e-06,
+      "loss": 0.0047,
+      "step": 11948
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.6372600162886605e-06,
+      "loss": 0.0059,
+      "step": 11950
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.6342163328966617e-06,
+      "loss": 0.0057,
+      "step": 11952
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.6311736407058607e-06,
+      "loss": 0.0032,
+      "step": 11954
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.6281319401900194e-06,
+      "loss": 0.0028,
+      "step": 11956
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.625091231822763e-06,
+      "loss": 0.0053,
+      "step": 11958
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.6220515160775436e-06,
+      "loss": 0.0049,
+      "step": 11960
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.6190127934276743e-06,
+      "loss": 0.0043,
+      "step": 11962
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.6159750643463064e-06,
+      "loss": 0.0045,
+      "step": 11964
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.61293832930644e-06,
+      "loss": 0.0095,
+      "step": 11966
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.6099025887809123e-06,
+      "loss": 0.0041,
+      "step": 11968
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.6068678432424175e-06,
+      "loss": 0.0018,
+      "step": 11970
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.603834093163483e-06,
+      "loss": 0.0045,
+      "step": 11972
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.600801339016492e-06,
+      "loss": 0.0076,
+      "step": 11974
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.5977695812736613e-06,
+      "loss": 0.0091,
+      "step": 11976
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.594738820407063e-06,
+      "loss": 0.0041,
+      "step": 11978
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.5917090568886092e-06,
+      "loss": 0.0048,
+      "step": 11980
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.588680291190061e-06,
+      "loss": 0.0072,
+      "step": 11982
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.5856525237830164e-06,
+      "loss": 0.0093,
+      "step": 11984
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.582625755138918e-06,
+      "loss": 0.0062,
+      "step": 11986
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.579599985729062e-06,
+      "loss": 0.0066,
+      "step": 11988
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.5765752160245848e-06,
+      "loss": 0.0067,
+      "step": 11990
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.5735514464964615e-06,
+      "loss": 0.0056,
+      "step": 11992
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.570528677615519e-06,
+      "loss": 0.0019,
+      "step": 11994
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.5675069098524282e-06,
+      "loss": 0.0092,
+      "step": 11996
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.564486143677699e-06,
+      "loss": 0.0089,
+      "step": 11998
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.5614663795616843e-06,
+      "loss": 0.0064,
+      "step": 12000
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.5584476179745876e-06,
+      "loss": 0.0041,
+      "step": 12002
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.5554298593864523e-06,
+      "loss": 0.0055,
+      "step": 12004
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.5524131042671705e-06,
+      "loss": 0.0069,
+      "step": 12006
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.549397353086468e-06,
+      "loss": 0.0064,
+      "step": 12008
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.546382606313923e-06,
+      "loss": 0.0086,
+      "step": 12010
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.5433688644189577e-06,
+      "loss": 0.0112,
+      "step": 12012
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.540356127870833e-06,
+      "loss": 0.007,
+      "step": 12014
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.53734439713865e-06,
+      "loss": 0.0087,
+      "step": 12016
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.5343336726913614e-06,
+      "loss": 0.0026,
+      "step": 12018
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.5313239549977606e-06,
+      "loss": 0.0118,
+      "step": 12020
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.5283152445264877e-06,
+      "loss": 0.0107,
+      "step": 12022
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.5253075417460146e-06,
+      "loss": 0.0105,
+      "step": 12024
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.52230084712467e-06,
+      "loss": 0.007,
+      "step": 12026
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.519295161130614e-06,
+      "loss": 0.0048,
+      "step": 12028
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.516290484231859e-06,
+      "loss": 0.0048,
+      "step": 12030
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.513286816896251e-06,
+      "loss": 0.0074,
+      "step": 12032
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.510284159591488e-06,
+      "loss": 0.0034,
+      "step": 12034
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.507282512785105e-06,
+      "loss": 0.0056,
+      "step": 12036
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.5042818769444866e-06,
+      "loss": 0.0075,
+      "step": 12038
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.5012822525368506e-06,
+      "loss": 0.0042,
+      "step": 12040
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.4982836400292573e-06,
+      "loss": 0.0121,
+      "step": 12042
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.495286039888618e-06,
+      "loss": 0.0046,
+      "step": 12044
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.492289452581684e-06,
+      "loss": 0.0029,
+      "step": 12046
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.489293878575042e-06,
+      "loss": 0.0044,
+      "step": 12048
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.486299318335128e-06,
+      "loss": 0.0053,
+      "step": 12050
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.4833057723282214e-06,
+      "loss": 0.0048,
+      "step": 12052
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.4803132410204364e-06,
+      "loss": 0.0069,
+      "step": 12054
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.4773217248777313e-06,
+      "loss": 0.0027,
+      "step": 12056
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.47433122436591e-06,
+      "loss": 0.0055,
+      "step": 12058
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.4713417399506154e-06,
+      "loss": 0.0033,
+      "step": 12060
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.468353272097339e-06,
+      "loss": 0.0087,
+      "step": 12062
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.4653658212714003e-06,
+      "loss": 0.0089,
+      "step": 12064
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.4623793879379695e-06,
+      "loss": 0.0032,
+      "step": 12066
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.459393972562064e-06,
+      "loss": 0.0053,
+      "step": 12068
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.4564095756085302e-06,
+      "loss": 0.0092,
+      "step": 12070
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.4534261975420578e-06,
+      "loss": 0.0104,
+      "step": 12072
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.4504438388271866e-06,
+      "loss": 0.0027,
+      "step": 12074
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.4474624999282923e-06,
+      "loss": 0.0097,
+      "step": 12076
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.444482181309594e-06,
+      "loss": 0.0064,
+      "step": 12078
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.441502883435146e-06,
+      "loss": 0.0073,
+      "step": 12080
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.4385246067688517e-06,
+      "loss": 0.0048,
+      "step": 12082
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.4355473517744464e-06,
+      "loss": 0.0047,
+      "step": 12084
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.4325711189155177e-06,
+      "loss": 0.0017,
+      "step": 12086
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.429595908655482e-06,
+      "loss": 0.0051,
+      "step": 12088
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.4266217214576035e-06,
+      "loss": 0.0096,
+      "step": 12090
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.4236485577849886e-06,
+      "loss": 0.0037,
+      "step": 12092
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.4206764181005834e-06,
+      "loss": 0.0064,
+      "step": 12094
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.4177053028671693e-06,
+      "loss": 0.0048,
+      "step": 12096
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.4147352125473687e-06,
+      "loss": 0.0041,
+      "step": 12098
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.4117661476036514e-06,
+      "loss": 0.0068,
+      "step": 12100
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.4087981084983258e-06,
+      "loss": 0.0042,
+      "step": 12102
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.4058310956935314e-06,
+      "loss": 0.0079,
+      "step": 12104
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.40286510965126e-06,
+      "loss": 0.0055,
+      "step": 12106
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.3999001508333396e-06,
+      "loss": 0.0041,
+      "step": 12108
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.39693621970143e-06,
+      "loss": 0.0081,
+      "step": 12110
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.3939733167170475e-06,
+      "loss": 0.0016,
+      "step": 12112
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.3910114423415296e-06,
+      "loss": 0.0027,
+      "step": 12114
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.3880505970360667e-06,
+      "loss": 0.0033,
+      "step": 12116
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.3850907812616873e-06,
+      "loss": 0.0088,
+      "step": 12118
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.3821319954792533e-06,
+      "loss": 0.0082,
+      "step": 12120
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.379174240149472e-06,
+      "loss": 0.0142,
+      "step": 12122
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.3762175157328915e-06,
+      "loss": 0.0031,
+      "step": 12124
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.3732618226898937e-06,
+      "loss": 0.0034,
+      "step": 12126
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.3703071614806994e-06,
+      "loss": 0.0055,
+      "step": 12128
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.367353532565375e-06,
+      "loss": 0.0066,
+      "step": 12130
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.364400936403822e-06,
+      "loss": 0.0012,
+      "step": 12132
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.3614493734557884e-06,
+      "loss": 0.0087,
+      "step": 12134
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.358498844180845e-06,
+      "loss": 0.0184,
+      "step": 12136
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.3555493490384215e-06,
+      "loss": 0.0033,
+      "step": 12138
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.3526008884877683e-06,
+      "loss": 0.0092,
+      "step": 12140
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.3496534629879905e-06,
+      "loss": 0.0051,
+      "step": 12142
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.3467070729980177e-06,
+      "loss": 0.0097,
+      "step": 12144
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.34376171897663e-06,
+      "loss": 0.0036,
+      "step": 12146
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.3408174013824402e-06,
+      "loss": 0.0054,
+      "step": 12148
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.337874120673904e-06,
+      "loss": 0.0057,
+      "step": 12150
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.334931877309311e-06,
+      "loss": 0.0063,
+      "step": 12152
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.3319906717467864e-06,
+      "loss": 0.0076,
+      "step": 12154
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.3290505044443023e-06,
+      "loss": 0.0033,
+      "step": 12156
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.3261113758596686e-06,
+      "loss": 0.0041,
+      "step": 12158
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.323173286450523e-06,
+      "loss": 0.009,
+      "step": 12160
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.3202362366743523e-06,
+      "loss": 0.001,
+      "step": 12162
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.317300226988477e-06,
+      "loss": 0.0087,
+      "step": 12164
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.3143652578500607e-06,
+      "loss": 0.0063,
+      "step": 12166
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.311431329716096e-06,
+      "loss": 0.0048,
+      "step": 12168
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.3084984430434152e-06,
+      "loss": 0.004,
+      "step": 12170
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.305566598288694e-06,
+      "loss": 0.0045,
+      "step": 12172
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.3026357959084465e-06,
+      "loss": 0.0094,
+      "step": 12174
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.299706036359015e-06,
+      "loss": 0.0046,
+      "step": 12176
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.296777320096589e-06,
+      "loss": 0.0077,
+      "step": 12178
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.293849647577194e-06,
+      "loss": 0.008,
+      "step": 12180
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.29092301925669e-06,
+      "loss": 0.0025,
+      "step": 12182
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.2879974355907684e-06,
+      "loss": 0.003,
+      "step": 12184
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.2850728970349722e-06,
+      "loss": 0.0138,
+      "step": 12186
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.282149404044672e-06,
+      "loss": 0.0045,
+      "step": 12188
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.279226957075081e-06,
+      "loss": 0.0034,
+      "step": 12190
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.276305556581243e-06,
+      "loss": 0.006,
+      "step": 12192
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.2733852030180444e-06,
+      "loss": 0.0038,
+      "step": 12194
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.270465896840205e-06,
+      "loss": 0.004,
+      "step": 12196
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.2675476385022853e-06,
+      "loss": 0.0033,
+      "step": 12198
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.264630428458676e-06,
+      "loss": 0.0039,
+      "step": 12200
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.2617142671636127e-06,
+      "loss": 0.0042,
+      "step": 12202
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.2587991550711627e-06,
+      "loss": 0.0049,
+      "step": 12204
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.255885092635236e-06,
+      "loss": 0.0054,
+      "step": 12206
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.252972080309569e-06,
+      "loss": 0.0091,
+      "step": 12208
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.250060118547739e-06,
+      "loss": 0.0052,
+      "step": 12210
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.247149207803163e-06,
+      "loss": 0.0076,
+      "step": 12212
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.2442393485290947e-06,
+      "loss": 0.0073,
+      "step": 12214
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.2413305411786155e-06,
+      "loss": 0.0051,
+      "step": 12216
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.2384227862046525e-06,
+      "loss": 0.0035,
+      "step": 12218
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.2355160840599654e-06,
+      "loss": 0.0072,
+      "step": 12220
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.232610435197152e-06,
+      "loss": 0.0072,
+      "step": 12222
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.2297058400686433e-06,
+      "loss": 0.0031,
+      "step": 12224
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.2268022991267e-06,
+      "loss": 0.005,
+      "step": 12226
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.2238998128234324e-06,
+      "loss": 0.0043,
+      "step": 12228
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.2209983816107816e-06,
+      "loss": 0.0055,
+      "step": 12230
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.2180980059405166e-06,
+      "loss": 0.0052,
+      "step": 12232
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.2151986862642504e-06,
+      "loss": 0.005,
+      "step": 12234
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.212300423033432e-06,
+      "loss": 0.0046,
+      "step": 12236
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.2094032166993425e-06,
+      "loss": 0.0086,
+      "step": 12238
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.2065070677130938e-06,
+      "loss": 0.0026,
+      "step": 12240
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.203611976525641e-06,
+      "loss": 0.0073,
+      "step": 12242
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.200717943587778e-06,
+      "loss": 0.0089,
+      "step": 12244
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.1978249693501185e-06,
+      "loss": 0.0077,
+      "step": 12246
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.194933054263126e-06,
+      "loss": 0.0051,
+      "step": 12248
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.1920421987770956e-06,
+      "loss": 0.008,
+      "step": 12250
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.1891524033421527e-06,
+      "loss": 0.004,
+      "step": 12252
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.1862636684082582e-06,
+      "loss": 0.0069,
+      "step": 12254
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.183375994425213e-06,
+      "loss": 0.0092,
+      "step": 12256
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.1804893818426498e-06,
+      "loss": 0.009,
+      "step": 12258
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.177603831110039e-06,
+      "loss": 0.006,
+      "step": 12260
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.1747193426766763e-06,
+      "loss": 0.0082,
+      "step": 12262
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.1718359169917033e-06,
+      "loss": 0.0062,
+      "step": 12264
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.168953554504094e-06,
+      "loss": 0.0022,
+      "step": 12266
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.16607225566265e-06,
+      "loss": 0.0059,
+      "step": 12268
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.1631920209160106e-06,
+      "loss": 0.0032,
+      "step": 12270
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.1603128507126515e-06,
+      "loss": 0.005,
+      "step": 12272
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.1574347455008813e-06,
+      "loss": 0.004,
+      "step": 12274
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.154557705728849e-06,
+      "loss": 0.0037,
+      "step": 12276
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.1516817318445225e-06,
+      "loss": 0.0105,
+      "step": 12278
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.148806824295719e-06,
+      "loss": 0.005,
+      "step": 12280
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.14593298353008e-06,
+      "loss": 0.0061,
+      "step": 12282
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.1430602099950892e-06,
+      "loss": 0.0064,
+      "step": 12284
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.140188504138053e-06,
+      "loss": 0.0035,
+      "step": 12286
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.137317866406121e-06,
+      "loss": 0.0061,
+      "step": 12288
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.134448297246274e-06,
+      "loss": 0.0032,
+      "step": 12290
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.1315797971053295e-06,
+      "loss": 0.0025,
+      "step": 12292
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.12871236642993e-06,
+      "loss": 0.0024,
+      "step": 12294
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.1258460056665553e-06,
+      "loss": 0.0048,
+      "step": 12296
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.1229807152615198e-06,
+      "loss": 0.0053,
+      "step": 12298
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.1201164956609777e-06,
+      "loss": 0.0025,
+      "step": 12300
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.1172533473109022e-06,
+      "loss": 0.0006,
+      "step": 12302
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.11439127065711e-06,
+      "loss": 0.0017,
+      "step": 12304
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.1115302661452527e-06,
+      "loss": 0.0041,
+      "step": 12306
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.1086703342208026e-06,
+      "loss": 0.0037,
+      "step": 12308
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.10581147532908e-06,
+      "loss": 0.0053,
+      "step": 12310
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.1029536899152257e-06,
+      "loss": 0.0054,
+      "step": 12312
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.1000969784242217e-06,
+      "loss": 0.0023,
+      "step": 12314
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.097241341300882e-06,
+      "loss": 0.0053,
+      "step": 12316
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.0943867789898453e-06,
+      "loss": 0.004,
+      "step": 12318
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.0915332919355912e-06,
+      "loss": 0.0024,
+      "step": 12320
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.088680880582434e-06,
+      "loss": 0.0017,
+      "step": 12322
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.0858295453745123e-06,
+      "loss": 0.0019,
+      "step": 12324
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.0829792867557973e-06,
+      "loss": 0.004,
+      "step": 12326
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.0801301051700992e-06,
+      "loss": 0.0037,
+      "step": 12328
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.077282001061057e-06,
+      "loss": 0.002,
+      "step": 12330
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.074434974872147e-06,
+      "loss": 0.0034,
+      "step": 12332
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.0715890270466652e-06,
+      "loss": 0.0009,
+      "step": 12334
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.0687441580277546e-06,
+      "loss": 0.0026,
+      "step": 12336
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.0659003682583766e-06,
+      "loss": 0.0038,
+      "step": 12338
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.0630576581813365e-06,
+      "loss": 0.004,
+      "step": 12340
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.06021602823926e-06,
+      "loss": 0.0034,
+      "step": 12342
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.057375478874615e-06,
+      "loss": 0.0027,
+      "step": 12344
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.054536010529694e-06,
+      "loss": 0.0017,
+      "step": 12346
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.05169762364663e-06,
+      "loss": 0.0055,
+      "step": 12348
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.0488603186673772e-06,
+      "loss": 0.0036,
+      "step": 12350
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.046024096033723e-06,
+      "loss": 0.0037,
+      "step": 12352
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.0431889561872905e-06,
+      "loss": 0.0031,
+      "step": 12354
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.0403548995695387e-06,
+      "loss": 0.0063,
+      "step": 12356
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.0375219266217428e-06,
+      "loss": 0.003,
+      "step": 12358
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.0346900377850218e-06,
+      "loss": 0.0024,
+      "step": 12360
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.031859233500325e-06,
+      "loss": 0.0021,
+      "step": 12362
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.02902951420843e-06,
+      "loss": 0.0017,
+      "step": 12364
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.0262008803499455e-06,
+      "loss": 0.0026,
+      "step": 12366
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.023373332365306e-06,
+      "loss": 0.0063,
+      "step": 12368
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.020546870694787e-06,
+      "loss": 0.0027,
+      "step": 12370
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.017721495778493e-06,
+      "loss": 0.0034,
+      "step": 12372
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.01489720805635e-06,
+      "loss": 0.0023,
+      "step": 12374
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.0120740079681233e-06,
+      "loss": 0.004,
+      "step": 12376
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.0092518959534125e-06,
+      "loss": 0.0037,
+      "step": 12378
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.0064308724516357e-06,
+      "loss": 0.0056,
+      "step": 12380
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.0036109379020474e-06,
+      "loss": 0.0012,
+      "step": 12382
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.000792092743735e-06,
+      "loss": 0.0037,
+      "step": 12384
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2.9979743374156136e-06,
+      "loss": 0.0066,
+      "step": 12386
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2.9951576723564335e-06,
+      "loss": 0.002,
+      "step": 12388
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2.9923420980047647e-06,
+      "loss": 0.0041,
+      "step": 12390
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2.9895276147990192e-06,
+      "loss": 0.0025,
+      "step": 12392
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2.986714223177427e-06,
+      "loss": 0.0019,
+      "step": 12394
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2.983901923578062e-06,
+      "loss": 0.0038,
+      "step": 12396
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2.981090716438814e-06,
+      "loss": 0.0045,
+      "step": 12398
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2.978280602197412e-06,
+      "loss": 0.0033,
+      "step": 12400
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2.9754715812914135e-06,
+      "loss": 0.0026,
+      "step": 12402
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2.9726636541582076e-06,
+      "loss": 0.0022,
+      "step": 12404
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2.9698568212350056e-06,
+      "loss": 0.0018,
+      "step": 12406
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2.9670510829588508e-06,
+      "loss": 0.0045,
+      "step": 12408
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2.9642464397666202e-06,
+      "loss": 0.0013,
+      "step": 12410
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2.9614428920950234e-06,
+      "loss": 0.0051,
+      "step": 12412
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 2.9586404403805847e-06,
+      "loss": 0.0036,
+      "step": 12414
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2.9558390850596743e-06,
+      "loss": 0.0057,
+      "step": 12416
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2.9530388265684806e-06,
+      "loss": 0.0029,
+      "step": 12418
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2.950239665343032e-06,
+      "loss": 0.0027,
+      "step": 12420
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2.947441601819173e-06,
+      "loss": 0.002,
+      "step": 12422
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2.944644636432583e-06,
+      "loss": 0.0051,
+      "step": 12424
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2.941848769618774e-06,
+      "loss": 0.0067,
+      "step": 12426
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2.939054001813084e-06,
+      "loss": 0.0042,
+      "step": 12428
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2.9362603334506755e-06,
+      "loss": 0.0019,
+      "step": 12430
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2.933467764966548e-06,
+      "loss": 0.0042,
+      "step": 12432
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2.930676296795527e-06,
+      "loss": 0.0038,
+      "step": 12434
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2.927885929372264e-06,
+      "loss": 0.003,
+      "step": 12436
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2.925096663131237e-06,
+      "loss": 0.0052,
+      "step": 12438
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2.9223084985067583e-06,
+      "loss": 0.0061,
+      "step": 12440
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2.919521435932967e-06,
+      "loss": 0.0026,
+      "step": 12442
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2.9167354758438337e-06,
+      "loss": 0.0026,
+      "step": 12444
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2.9139506186731457e-06,
+      "loss": 0.0017,
+      "step": 12446
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2.9111668648545354e-06,
+      "loss": 0.0028,
+      "step": 12448
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2.908384214821446e-06,
+      "loss": 0.0021,
+      "step": 12450
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2.9056026690071646e-06,
+      "loss": 0.0045,
+      "step": 12452
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 2.9028222278447936e-06,
+      "loss": 0.0021,
+      "step": 12454
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2.9000428917672717e-06,
+      "loss": 0.0039,
+      "step": 12456
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2.897264661207362e-06,
+      "loss": 0.0037,
+      "step": 12458
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2.894487536597659e-06,
+      "loss": 0.0041,
+      "step": 12460
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2.891711518370578e-06,
+      "loss": 0.0013,
+      "step": 12462
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2.8889366069583703e-06,
+      "loss": 0.0029,
+      "step": 12464
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2.8861628027931065e-06,
+      "loss": 0.0063,
+      "step": 12466
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2.883390106306694e-06,
+      "loss": 0.0018,
+      "step": 12468
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2.8806185179308567e-06,
+      "loss": 0.0007,
+      "step": 12470
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2.877848038097156e-06,
+      "loss": 0.0053,
+      "step": 12472
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2.8750786672369757e-06,
+      "loss": 0.0016,
+      "step": 12474
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2.872310405781533e-06,
+      "loss": 0.0059,
+      "step": 12476
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2.8695432541618627e-06,
+      "loss": 0.0007,
+      "step": 12478
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2.866777212808829e-06,
+      "loss": 0.009,
+      "step": 12480
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2.8640122821531278e-06,
+      "loss": 0.0036,
+      "step": 12482
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2.8612484626252836e-06,
+      "loss": 0.0019,
+      "step": 12484
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2.8584857546556387e-06,
+      "loss": 0.0013,
+      "step": 12486
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2.855724158674371e-06,
+      "loss": 0.0007,
+      "step": 12488
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2.852963675111484e-06,
+      "loss": 0.0045,
+      "step": 12490
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2.8502043043968042e-06,
+      "loss": 0.0081,
+      "step": 12492
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2.847446046959982e-06,
+      "loss": 0.0025,
+      "step": 12494
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 2.844688903230505e-06,
+      "loss": 0.0019,
+      "step": 12496
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2.841932873637678e-06,
+      "loss": 0.0026,
+      "step": 12498
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2.8391779586106427e-06,
+      "loss": 0.0037,
+      "step": 12500
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2.8364241585783514e-06,
+      "loss": 0.0047,
+      "step": 12502
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2.8336714739695993e-06,
+      "loss": 0.0047,
+      "step": 12504
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2.8309199052129944e-06,
+      "loss": 0.0006,
+      "step": 12506
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2.828169452736983e-06,
+      "loss": 0.0023,
+      "step": 12508
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2.8254201169698246e-06,
+      "loss": 0.0106,
+      "step": 12510
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2.822671898339615e-06,
+      "loss": 0.0034,
+      "step": 12512
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2.8199247972742747e-06,
+      "loss": 0.0014,
+      "step": 12514
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2.817178814201549e-06,
+      "loss": 0.004,
+      "step": 12516
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2.8144339495490035e-06,
+      "loss": 0.0031,
+      "step": 12518
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2.8116902037440408e-06,
+      "loss": 0.0031,
+      "step": 12520
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2.8089475772138775e-06,
+      "loss": 0.004,
+      "step": 12522
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2.8062060703855666e-06,
+      "loss": 0.0053,
+      "step": 12524
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2.803465683685975e-06,
+      "loss": 0.0024,
+      "step": 12526
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2.800726417541807e-06,
+      "loss": 0.0038,
+      "step": 12528
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2.797988272379586e-06,
+      "loss": 0.0035,
+      "step": 12530
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2.7952512486256646e-06,
+      "loss": 0.003,
+      "step": 12532
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2.792515346706216e-06,
+      "loss": 0.0062,
+      "step": 12534
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 2.789780567047239e-06,
+      "loss": 0.0017,
+      "step": 12536
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2.7870469100745612e-06,
+      "loss": 0.0041,
+      "step": 12538
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2.784314376213837e-06,
+      "loss": 0.0019,
+      "step": 12540
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2.781582965890538e-06,
+      "loss": 0.0024,
+      "step": 12542
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2.7788526795299675e-06,
+      "loss": 0.0006,
+      "step": 12544
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2.7761235175572554e-06,
+      "loss": 0.0016,
+      "step": 12546
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2.7733954803973505e-06,
+      "loss": 0.001,
+      "step": 12548
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2.7706685684750245e-06,
+      "loss": 0.004,
+      "step": 12550
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2.767942782214884e-06,
+      "loss": 0.003,
+      "step": 12552
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2.7652181220413523e-06,
+      "loss": 0.0029,
+      "step": 12554
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2.762494588378685e-06,
+      "loss": 0.0024,
+      "step": 12556
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2.7597721816509482e-06,
+      "loss": 0.0023,
+      "step": 12558
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2.757050902282051e-06,
+      "loss": 0.0028,
+      "step": 12560
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2.7543307506957084e-06,
+      "loss": 0.0008,
+      "step": 12562
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2.7516117273154765e-06,
+      "loss": 0.0022,
+      "step": 12564
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2.7488938325647217e-06,
+      "loss": 0.0012,
+      "step": 12566
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2.746177066866643e-06,
+      "loss": 0.0027,
+      "step": 12568
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2.743461430644264e-06,
+      "loss": 0.0058,
+      "step": 12570
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2.740746924320431e-06,
+      "loss": 0.0035,
+      "step": 12572
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2.7380335483178068e-06,
+      "loss": 0.0027,
+      "step": 12574
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2.735321303058893e-06,
+      "loss": 0.0013,
+      "step": 12576
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 2.7326101889659983e-06,
+      "loss": 0.0055,
+      "step": 12578
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2.7299002064612734e-06,
+      "loss": 0.0036,
+      "step": 12580
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2.7271913559666742e-06,
+      "loss": 0.0033,
+      "step": 12582
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2.724483637903993e-06,
+      "loss": 0.0017,
+      "step": 12584
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2.721777052694844e-06,
+      "loss": 0.0032,
+      "step": 12586
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2.7190716007606634e-06,
+      "loss": 0.0045,
+      "step": 12588
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2.7163672825227104e-06,
+      "loss": 0.0021,
+      "step": 12590
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2.7136640984020636e-06,
+      "loss": 0.0073,
+      "step": 12592
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2.710962048819633e-06,
+      "loss": 0.0043,
+      "step": 12594
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2.7082611341961506e-06,
+      "loss": 0.0044,
+      "step": 12596
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2.7055613549521643e-06,
+      "loss": 0.002,
+      "step": 12598
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2.702862711508053e-06,
+      "loss": 0.0052,
+      "step": 12600
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2.70016520428402e-06,
+      "loss": 0.0032,
+      "step": 12602
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2.6974688337000832e-06,
+      "loss": 0.0029,
+      "step": 12604
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2.694773600176085e-06,
+      "loss": 0.0054,
+      "step": 12606
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2.6920795041316994e-06,
+      "loss": 0.0024,
+      "step": 12608
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2.6893865459864154e-06,
+      "loss": 0.001,
+      "step": 12610
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2.686694726159551e-06,
+      "loss": 0.0049,
+      "step": 12612
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2.6840040450702366e-06,
+      "loss": 0.0027,
+      "step": 12614
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2.6813145031374356e-06,
+      "loss": 0.0036,
+      "step": 12616
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 2.678626100779933e-06,
+      "loss": 0.0044,
+      "step": 12618
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2.67593883841633e-06,
+      "loss": 0.0016,
+      "step": 12620
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2.6732527164650523e-06,
+      "loss": 0.0041,
+      "step": 12622
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2.6705677353443514e-06,
+      "loss": 0.0021,
+      "step": 12624
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2.6678838954722997e-06,
+      "loss": 0.0043,
+      "step": 12626
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2.6652011972667945e-06,
+      "loss": 0.0044,
+      "step": 12628
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2.6625196411455456e-06,
+      "loss": 0.0021,
+      "step": 12630
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2.6598392275261e-06,
+      "loss": 0.0029,
+      "step": 12632
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2.657159956825811e-06,
+      "loss": 0.0051,
+      "step": 12634
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2.654481829461868e-06,
+      "loss": 0.0023,
+      "step": 12636
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2.6518048458512692e-06,
+      "loss": 0.0072,
+      "step": 12638
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2.6491290064108454e-06,
+      "loss": 0.0047,
+      "step": 12640
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2.646454311557245e-06,
+      "loss": 0.0029,
+      "step": 12642
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2.643780761706941e-06,
+      "loss": 0.0058,
+      "step": 12644
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2.641108357276223e-06,
+      "loss": 0.0009,
+      "step": 12646
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2.6384370986812027e-06,
+      "loss": 0.0008,
+      "step": 12648
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2.6357669863378155e-06,
+      "loss": 0.0003,
+      "step": 12650
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2.633098020661824e-06,
+      "loss": 0.005,
+      "step": 12652
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2.6304302020687997e-06,
+      "loss": 0.0035,
+      "step": 12654
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2.627763530974147e-06,
+      "loss": 0.0042,
+      "step": 12656
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2.6250980077930864e-06,
+      "loss": 0.0033,
+      "step": 12658
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 2.6224336329406607e-06,
+      "loss": 0.0013,
+      "step": 12660
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2.6197704068317274e-06,
+      "loss": 0.0012,
+      "step": 12662
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2.6171083298809773e-06,
+      "loss": 0.0041,
+      "step": 12664
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2.6144474025029145e-06,
+      "loss": 0.0078,
+      "step": 12666
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2.611787625111868e-06,
+      "loss": 0.0024,
+      "step": 12668
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2.6091289981219813e-06,
+      "loss": 0.0074,
+      "step": 12670
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2.606471521947225e-06,
+      "loss": 0.0033,
+      "step": 12672
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2.6038151970013903e-06,
+      "loss": 0.0067,
+      "step": 12674
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2.601160023698086e-06,
+      "loss": 0.0036,
+      "step": 12676
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2.5985060024507403e-06,
+      "loss": 0.0034,
+      "step": 12678
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2.5958531336726057e-06,
+      "loss": 0.0026,
+      "step": 12680
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2.593201417776755e-06,
+      "loss": 0.0011,
+      "step": 12682
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2.590550855176085e-06,
+      "loss": 0.0017,
+      "step": 12684
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2.5879014462832997e-06,
+      "loss": 0.0023,
+      "step": 12686
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2.5852531915109415e-06,
+      "loss": 0.0032,
+      "step": 12688
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2.5826060912713544e-06,
+      "loss": 0.0011,
+      "step": 12690
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2.579960145976722e-06,
+      "loss": 0.0019,
+      "step": 12692
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2.5773153560390297e-06,
+      "loss": 0.0027,
+      "step": 12694
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2.5746717218700946e-06,
+      "loss": 0.0036,
+      "step": 12696
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2.572029243881552e-06,
+      "loss": 0.0065,
+      "step": 12698
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 2.5693879224848585e-06,
+      "loss": 0.0035,
+      "step": 12700
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2.5667477580912836e-06,
+      "loss": 0.0034,
+      "step": 12702
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2.56410875111192e-06,
+      "loss": 0.0022,
+      "step": 12704
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2.561470901957683e-06,
+      "loss": 0.0019,
+      "step": 12706
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2.5588342110393105e-06,
+      "loss": 0.0012,
+      "step": 12708
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2.5561986787673477e-06,
+      "loss": 0.0047,
+      "step": 12710
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2.553564305552171e-06,
+      "loss": 0.0055,
+      "step": 12712
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2.5509310918039755e-06,
+      "loss": 0.0014,
+      "step": 12714
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2.5482990379327654e-06,
+      "loss": 0.0059,
+      "step": 12716
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2.545668144348379e-06,
+      "loss": 0.0036,
+      "step": 12718
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2.54303841146046e-06,
+      "loss": 0.0026,
+      "step": 12720
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2.5404098396784815e-06,
+      "loss": 0.0042,
+      "step": 12722
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2.537782429411736e-06,
+      "loss": 0.0037,
+      "step": 12724
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2.5351561810693225e-06,
+      "loss": 0.0022,
+      "step": 12726
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2.5325310950601734e-06,
+      "loss": 0.0012,
+      "step": 12728
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2.5299071717930367e-06,
+      "loss": 0.0019,
+      "step": 12730
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2.527284411676475e-06,
+      "loss": 0.0025,
+      "step": 12732
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2.524662815118868e-06,
+      "loss": 0.0027,
+      "step": 12734
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2.5220423825284223e-06,
+      "loss": 0.0028,
+      "step": 12736
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2.519423114313159e-06,
+      "loss": 0.004,
+      "step": 12738
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2.5168050108809206e-06,
+      "loss": 0.0011,
+      "step": 12740
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 2.5141880726393607e-06,
+      "loss": 0.0071,
+      "step": 12742
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2.5115722999959636e-06,
+      "loss": 0.0039,
+      "step": 12744
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2.508957693358016e-06,
+      "loss": 0.003,
+      "step": 12746
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2.506344253132641e-06,
+      "loss": 0.0011,
+      "step": 12748
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2.5037319797267635e-06,
+      "loss": 0.0075,
+      "step": 12750
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2.5011208735471383e-06,
+      "loss": 0.0071,
+      "step": 12752
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2.4985109350003344e-06,
+      "loss": 0.0055,
+      "step": 12754
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2.4959021644927427e-06,
+      "loss": 0.0034,
+      "step": 12756
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2.493294562430565e-06,
+      "loss": 0.004,
+      "step": 12758
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2.4906881292198213e-06,
+      "loss": 0.0034,
+      "step": 12760
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2.4880828652663613e-06,
+      "loss": 0.0025,
+      "step": 12762
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2.4854787709758366e-06,
+      "loss": 0.0035,
+      "step": 12764
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2.4828758467537274e-06,
+      "loss": 0.0022,
+      "step": 12766
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2.4802740930053305e-06,
+      "loss": 0.0021,
+      "step": 12768
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2.4776735101357606e-06,
+      "loss": 0.0029,
+      "step": 12770
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2.4750740985499434e-06,
+      "loss": 0.0038,
+      "step": 12772
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2.472475858652632e-06,
+      "loss": 0.0028,
+      "step": 12774
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2.4698787908483867e-06,
+      "loss": 0.0033,
+      "step": 12776
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2.467282895541597e-06,
+      "loss": 0.0068,
+      "step": 12778
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2.464688173136458e-06,
+      "loss": 0.0032,
+      "step": 12780
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 2.4620946240369904e-06,
+      "loss": 0.0028,
+      "step": 12782
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2.459502248647029e-06,
+      "loss": 0.0026,
+      "step": 12784
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2.4569110473702306e-06,
+      "loss": 0.0051,
+      "step": 12786
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2.454321020610061e-06,
+      "loss": 0.0024,
+      "step": 12788
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2.4517321687698047e-06,
+      "loss": 0.0022,
+      "step": 12790
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2.4491444922525687e-06,
+      "loss": 0.0026,
+      "step": 12792
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2.446557991461277e-06,
+      "loss": 0.0054,
+      "step": 12794
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2.443972666798662e-06,
+      "loss": 0.001,
+      "step": 12796
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2.4413885186672804e-06,
+      "loss": 0.0067,
+      "step": 12798
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2.4388055474695084e-06,
+      "loss": 0.0021,
+      "step": 12800
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2.4362237536075295e-06,
+      "loss": 0.0064,
+      "step": 12802
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2.433643137483347e-06,
+      "loss": 0.0017,
+      "step": 12804
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2.4310636994987856e-06,
+      "loss": 0.0065,
+      "step": 12806
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2.4284854400554834e-06,
+      "loss": 0.0023,
+      "step": 12808
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2.425908359554897e-06,
+      "loss": 0.0014,
+      "step": 12810
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2.423332458398293e-06,
+      "loss": 0.0024,
+      "step": 12812
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2.420757736986762e-06,
+      "loss": 0.0014,
+      "step": 12814
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2.4181841957212095e-06,
+      "loss": 0.0004,
+      "step": 12816
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2.4156118350023527e-06,
+      "loss": 0.0057,
+      "step": 12818
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2.413040655230726e-06,
+      "loss": 0.0036,
+      "step": 12820
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2.410470656806685e-06,
+      "loss": 0.0045,
+      "step": 12822
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 2.4079018401303967e-06,
+      "loss": 0.0046,
+      "step": 12824
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2.405334205601848e-06,
+      "loss": 0.0018,
+      "step": 12826
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2.402767753620835e-06,
+      "loss": 0.0008,
+      "step": 12828
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2.4002024845869786e-06,
+      "loss": 0.0019,
+      "step": 12830
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2.3976383988997064e-06,
+      "loss": 0.003,
+      "step": 12832
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2.39507549695827e-06,
+      "loss": 0.0013,
+      "step": 12834
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2.392513779161729e-06,
+      "loss": 0.004,
+      "step": 12836
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2.3899532459089634e-06,
+      "loss": 0.003,
+      "step": 12838
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2.3873938975986698e-06,
+      "loss": 0.0016,
+      "step": 12840
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2.3848357346293593e-06,
+      "loss": 0.0038,
+      "step": 12842
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2.3822787573993557e-06,
+      "loss": 0.0034,
+      "step": 12844
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2.3797229663067965e-06,
+      "loss": 0.0011,
+      "step": 12846
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2.377168361749642e-06,
+      "loss": 0.0044,
+      "step": 12848
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2.374614944125665e-06,
+      "loss": 0.003,
+      "step": 12850
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2.372062713832447e-06,
+      "loss": 0.0029,
+      "step": 12852
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2.3695116712673927e-06,
+      "loss": 0.0018,
+      "step": 12854
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2.366961816827721e-06,
+      "loss": 0.0037,
+      "step": 12856
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2.364413150910463e-06,
+      "loss": 0.0024,
+      "step": 12858
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2.36186567391246e-06,
+      "loss": 0.0019,
+      "step": 12860
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2.3593193862303775e-06,
+      "loss": 0.0063,
+      "step": 12862
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2.356774288260694e-06,
+      "loss": 0.002,
+      "step": 12864
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 2.3542303803997004e-06,
+      "loss": 0.0039,
+      "step": 12866
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2.3516876630434982e-06,
+      "loss": 0.0034,
+      "step": 12868
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2.349146136588011e-06,
+      "loss": 0.0034,
+      "step": 12870
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2.346605801428976e-06,
+      "loss": 0.0062,
+      "step": 12872
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2.344066657961942e-06,
+      "loss": 0.0018,
+      "step": 12874
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2.341528706582268e-06,
+      "loss": 0.0019,
+      "step": 12876
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2.3389919476851354e-06,
+      "loss": 0.0044,
+      "step": 12878
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2.3364563816655384e-06,
+      "loss": 0.004,
+      "step": 12880
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2.333922008918286e-06,
+      "loss": 0.0028,
+      "step": 12882
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2.3313888298379937e-06,
+      "loss": 0.003,
+      "step": 12884
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2.328856844819103e-06,
+      "loss": 0.0028,
+      "step": 12886
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2.3263260542558564e-06,
+      "loss": 0.0022,
+      "step": 12888
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2.3237964585423244e-06,
+      "loss": 0.0051,
+      "step": 12890
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2.321268058072379e-06,
+      "loss": 0.0009,
+      "step": 12892
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2.3187408532397126e-06,
+      "loss": 0.0035,
+      "step": 12894
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2.3162148444378318e-06,
+      "loss": 0.0022,
+      "step": 12896
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2.3136900320600574e-06,
+      "loss": 0.0025,
+      "step": 12898
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2.3111664164995196e-06,
+      "loss": 0.0041,
+      "step": 12900
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2.3086439981491616e-06,
+      "loss": 0.0019,
+      "step": 12902
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2.306122777401746e-06,
+      "loss": 0.0026,
+      "step": 12904
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 2.3036027546498495e-06,
+      "loss": 0.0035,
+      "step": 12906
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2.3010839302858535e-06,
+      "loss": 0.0062,
+      "step": 12908
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2.29856630470196e-06,
+      "loss": 0.0046,
+      "step": 12910
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2.296049878290185e-06,
+      "loss": 0.0034,
+      "step": 12912
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2.2935346514423517e-06,
+      "loss": 0.0027,
+      "step": 12914
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2.2910206245501043e-06,
+      "loss": 0.0014,
+      "step": 12916
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2.2885077980048907e-06,
+      "loss": 0.0145,
+      "step": 12918
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2.2859961721979807e-06,
+      "loss": 0.0044,
+      "step": 12920
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2.2834857475204555e-06,
+      "loss": 0.0092,
+      "step": 12922
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2.2809765243632008e-06,
+      "loss": 0.0017,
+      "step": 12924
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2.2784685031169275e-06,
+      "loss": 0.0012,
+      "step": 12926
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2.275961684172154e-06,
+      "loss": 0.0012,
+      "step": 12928
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2.27345606791921e-06,
+      "loss": 0.0034,
+      "step": 12930
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2.2709516547482347e-06,
+      "loss": 0.003,
+      "step": 12932
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2.2684484450491894e-06,
+      "loss": 0.0038,
+      "step": 12934
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2.26594643921184e-06,
+      "loss": 0.004,
+      "step": 12936
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2.263445637625774e-06,
+      "loss": 0.0022,
+      "step": 12938
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2.2609460406803775e-06,
+      "loss": 0.0088,
+      "step": 12940
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2.258447648764863e-06,
+      "loss": 0.0027,
+      "step": 12942
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2.2559504622682436e-06,
+      "loss": 0.0018,
+      "step": 12944
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2.253454481579357e-06,
+      "loss": 0.0015,
+      "step": 12946
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 2.2509597070868393e-06,
+      "loss": 0.0015,
+      "step": 12948
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2.2484661391791494e-06,
+      "loss": 0.0017,
+      "step": 12950
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2.2459737782445546e-06,
+      "loss": 0.004,
+      "step": 12952
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2.2434826246711384e-06,
+      "loss": 0.0011,
+      "step": 12954
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2.2409926788467883e-06,
+      "loss": 0.0048,
+      "step": 12956
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2.238503941159206e-06,
+      "loss": 0.0041,
+      "step": 12958
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2.2360164119959093e-06,
+      "loss": 0.0009,
+      "step": 12960
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2.2335300917442293e-06,
+      "loss": 0.0022,
+      "step": 12962
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2.231044980791298e-06,
+      "loss": 0.0043,
+      "step": 12964
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2.2285610795240695e-06,
+      "loss": 0.0025,
+      "step": 12966
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2.226078388329307e-06,
+      "loss": 0.0052,
+      "step": 12968
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2.223596907593586e-06,
+      "loss": 0.0025,
+      "step": 12970
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2.221116637703291e-06,
+      "loss": 0.0071,
+      "step": 12972
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2.2186375790446148e-06,
+      "loss": 0.0079,
+      "step": 12974
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2.216159732003568e-06,
+      "loss": 0.0048,
+      "step": 12976
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2.213683096965975e-06,
+      "loss": 0.0024,
+      "step": 12978
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2.2112076743174593e-06,
+      "loss": 0.0014,
+      "step": 12980
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2.208733464443468e-06,
+      "loss": 0.004,
+      "step": 12982
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2.2062604677292554e-06,
+      "loss": 0.0035,
+      "step": 12984
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2.2037886845598845e-06,
+      "loss": 0.0028,
+      "step": 12986
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 2.201318115320227e-06,
+      "loss": 0.0032,
+      "step": 12988
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2.1988487603949726e-06,
+      "loss": 0.0038,
+      "step": 12990
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2.19638062016862e-06,
+      "loss": 0.0045,
+      "step": 12992
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2.193913695025478e-06,
+      "loss": 0.0031,
+      "step": 12994
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2.1914479853496618e-06,
+      "loss": 0.0019,
+      "step": 12996
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2.1889834915251063e-06,
+      "loss": 0.0085,
+      "step": 12998
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2.1865202139355467e-06,
+      "loss": 0.0017,
+      "step": 13000
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2.18405815296454e-06,
+      "loss": 0.0049,
+      "step": 13002
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2.181597308995441e-06,
+      "loss": 0.0026,
+      "step": 13004
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2.1791376824114265e-06,
+      "loss": 0.0048,
+      "step": 13006
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2.176679273595477e-06,
+      "loss": 0.0019,
+      "step": 13008
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2.1742220829303904e-06,
+      "loss": 0.0043,
+      "step": 13010
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2.1717661107987663e-06,
+      "loss": 0.001,
+      "step": 13012
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2.169311357583016e-06,
+      "loss": 0.0035,
+      "step": 13014
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2.1668578236653647e-06,
+      "loss": 0.0027,
+      "step": 13016
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2.1644055094278515e-06,
+      "loss": 0.0045,
+      "step": 13018
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2.1619544152523143e-06,
+      "loss": 0.0032,
+      "step": 13020
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2.1595045415204087e-06,
+      "loss": 0.0011,
+      "step": 13022
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2.1570558886135997e-06,
+      "loss": 0.0027,
+      "step": 13024
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2.154608456913163e-06,
+      "loss": 0.0017,
+      "step": 13026
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2.1521622468001814e-06,
+      "loss": 0.0009,
+      "step": 13028
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 2.1497172586555447e-06,
+      "loss": 0.0038,
+      "step": 13030
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2.1472734928599593e-06,
+      "loss": 0.0014,
+      "step": 13032
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2.14483094979394e-06,
+      "loss": 0.0025,
+      "step": 13034
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2.1423896298378066e-06,
+      "loss": 0.0036,
+      "step": 13036
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2.1399495333716912e-06,
+      "loss": 0.0082,
+      "step": 13038
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2.137510660775539e-06,
+      "loss": 0.0008,
+      "step": 13040
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2.1350730124290985e-06,
+      "loss": 0.0031,
+      "step": 13042
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2.132636588711928e-06,
+      "loss": 0.0055,
+      "step": 13044
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2.1302013900034e-06,
+      "loss": 0.0034,
+      "step": 13046
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2.1277674166826935e-06,
+      "loss": 0.0052,
+      "step": 13048
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2.125334669128798e-06,
+      "loss": 0.0048,
+      "step": 13050
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2.1229031477205075e-06,
+      "loss": 0.0017,
+      "step": 13052
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2.120472852836433e-06,
+      "loss": 0.0025,
+      "step": 13054
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2.118043784854984e-06,
+      "loss": 0.0039,
+      "step": 13056
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2.115615944154391e-06,
+      "loss": 0.0025,
+      "step": 13058
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2.1131893311126815e-06,
+      "loss": 0.0016,
+      "step": 13060
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2.1107639461077024e-06,
+      "loss": 0.0043,
+      "step": 13062
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2.108339789517102e-06,
+      "loss": 0.0066,
+      "step": 13064
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2.1059168617183433e-06,
+      "loss": 0.0024,
+      "step": 13066
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2.1034951630886913e-06,
+      "loss": 0.0016,
+      "step": 13068
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 2.101074694005226e-06,
+      "loss": 0.0032,
+      "step": 13070
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2.098655454844829e-06,
+      "loss": 0.0022,
+      "step": 13072
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2.0962374459841982e-06,
+      "loss": 0.001,
+      "step": 13074
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2.0938206677998328e-06,
+      "loss": 0.0033,
+      "step": 13076
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2.091405120668045e-06,
+      "loss": 0.001,
+      "step": 13078
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2.0889908049649543e-06,
+      "loss": 0.0053,
+      "step": 13080
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2.0865777210664904e-06,
+      "loss": 0.0048,
+      "step": 13082
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2.084165869348387e-06,
+      "loss": 0.0019,
+      "step": 13084
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2.0817552501861836e-06,
+      "loss": 0.003,
+      "step": 13086
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2.0793458639552365e-06,
+      "loss": 0.004,
+      "step": 13088
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2.0769377110307076e-06,
+      "loss": 0.0057,
+      "step": 13090
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2.074530791787559e-06,
+      "loss": 0.0038,
+      "step": 13092
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2.07212510660057e-06,
+      "loss": 0.0034,
+      "step": 13094
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2.069720655844325e-06,
+      "loss": 0.0028,
+      "step": 13096
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2.067317439893215e-06,
+      "loss": 0.0033,
+      "step": 13098
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2.064915459121435e-06,
+      "loss": 0.0045,
+      "step": 13100
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2.0625147139029956e-06,
+      "loss": 0.0027,
+      "step": 13102
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2.06011520461171e-06,
+      "loss": 0.0017,
+      "step": 13104
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2.0577169316212033e-06,
+      "loss": 0.0024,
+      "step": 13106
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2.055319895304899e-06,
+      "loss": 0.0007,
+      "step": 13108
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2.0529240960360396e-06,
+      "loss": 0.0081,
+      "step": 13110
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 2.050529534187665e-06,
+      "loss": 0.0029,
+      "step": 13112
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2.048136210132632e-06,
+      "loss": 0.0007,
+      "step": 13114
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2.045744124243593e-06,
+      "loss": 0.0005,
+      "step": 13116
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2.0433532768930176e-06,
+      "loss": 0.0028,
+      "step": 13118
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2.0409636684531774e-06,
+      "loss": 0.0018,
+      "step": 13120
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2.0385752992961584e-06,
+      "loss": 0.0033,
+      "step": 13122
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2.03618816979384e-06,
+      "loss": 0.0014,
+      "step": 13124
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2.0338022803179225e-06,
+      "loss": 0.0038,
+      "step": 13126
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2.0314176312399024e-06,
+      "loss": 0.0038,
+      "step": 13128
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2.0290342229310923e-06,
+      "loss": 0.0029,
+      "step": 13130
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2.0266520557626022e-06,
+      "loss": 0.0027,
+      "step": 13132
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2.0242711301053573e-06,
+      "loss": 0.0038,
+      "step": 13134
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2.0218914463300843e-06,
+      "loss": 0.0047,
+      "step": 13136
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2.019513004807322e-06,
+      "loss": 0.003,
+      "step": 13138
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2.0171358059074085e-06,
+      "loss": 0.0044,
+      "step": 13140
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2.0147598500004904e-06,
+      "loss": 0.0014,
+      "step": 13142
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2.012385137456523e-06,
+      "loss": 0.0049,
+      "step": 13144
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2.0100116686452708e-06,
+      "loss": 0.0033,
+      "step": 13146
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2.0076394439362968e-06,
+      "loss": 0.0015,
+      "step": 13148
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2.005268463698975e-06,
+      "loss": 0.0034,
+      "step": 13150
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 2.0028987283024902e-06,
+      "loss": 0.0003,
+      "step": 13152
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 2.0005302381158232e-06,
+      "loss": 0.0015,
+      "step": 13154
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.9981629935077663e-06,
+      "loss": 0.0004,
+      "step": 13156
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.995796994846917e-06,
+      "loss": 0.0046,
+      "step": 13158
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.9934322425016816e-06,
+      "loss": 0.0025,
+      "step": 13160
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.9910687368402715e-06,
+      "loss": 0.0041,
+      "step": 13162
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.9887064782306984e-06,
+      "loss": 0.0049,
+      "step": 13164
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.986345467040789e-06,
+      "loss": 0.001,
+      "step": 13166
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.983985703638166e-06,
+      "loss": 0.0052,
+      "step": 13168
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.981627188390266e-06,
+      "loss": 0.0017,
+      "step": 13170
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.979269921664324e-06,
+      "loss": 0.0027,
+      "step": 13172
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.9769139038273875e-06,
+      "loss": 0.0038,
+      "step": 13174
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.9745591352463055e-06,
+      "loss": 0.0023,
+      "step": 13176
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.9722056162877367e-06,
+      "loss": 0.0024,
+      "step": 13178
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.969853347318137e-06,
+      "loss": 0.0015,
+      "step": 13180
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.9675023287037766e-06,
+      "loss": 0.0046,
+      "step": 13182
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.9651525608107237e-06,
+      "loss": 0.0028,
+      "step": 13184
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.962804044004859e-06,
+      "loss": 0.003,
+      "step": 13186
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.960456778651859e-06,
+      "loss": 0.0027,
+      "step": 13188
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.958110765117215e-06,
+      "loss": 0.0009,
+      "step": 13190
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.9557660037662175e-06,
+      "loss": 0.0037,
+      "step": 13192
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.953422494963968e-06,
+      "loss": 0.0019,
+      "step": 13194
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.951080239075366e-06,
+      "loss": 0.0022,
+      "step": 13196
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.9487392364651135e-06,
+      "loss": 0.0007,
+      "step": 13198
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.946399487497729e-06,
+      "loss": 0.0047,
+      "step": 13200
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.9440609925375285e-06,
+      "loss": 0.0057,
+      "step": 13202
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.941723751948631e-06,
+      "loss": 0.0044,
+      "step": 13204
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.9393877660949646e-06,
+      "loss": 0.0037,
+      "step": 13206
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.9370530353402604e-06,
+      "loss": 0.0029,
+      "step": 13208
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.9347195600480552e-06,
+      "loss": 0.0023,
+      "step": 13210
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.9323873405816827e-06,
+      "loss": 0.0037,
+      "step": 13212
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.9300563773042924e-06,
+      "loss": 0.0056,
+      "step": 13214
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.927726670578831e-06,
+      "loss": 0.0034,
+      "step": 13216
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.9253982207680566e-06,
+      "loss": 0.003,
+      "step": 13218
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.92307102823452e-06,
+      "loss": 0.0011,
+      "step": 13220
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.9207450933405857e-06,
+      "loss": 0.0019,
+      "step": 13222
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.9184204164484223e-06,
+      "loss": 0.0046,
+      "step": 13224
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.9160969979199974e-06,
+      "loss": 0.0014,
+      "step": 13226
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.9137748381170818e-06,
+      "loss": 0.0004,
+      "step": 13228
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.9114539374012564e-06,
+      "loss": 0.0013,
+      "step": 13230
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.909134296133903e-06,
+      "loss": 0.0043,
+      "step": 13232
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.906815914676211e-06,
+      "loss": 0.0046,
+      "step": 13234
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.9044987933891635e-06,
+      "loss": 0.0021,
+      "step": 13236
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.9021829326335606e-06,
+      "loss": 0.0011,
+      "step": 13238
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.8998683327699929e-06,
+      "loss": 0.002,
+      "step": 13240
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.8975549941588679e-06,
+      "loss": 0.0022,
+      "step": 13242
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.8952429171603849e-06,
+      "loss": 0.0015,
+      "step": 13244
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.8929321021345526e-06,
+      "loss": 0.0016,
+      "step": 13246
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.8906225494411844e-06,
+      "loss": 0.0064,
+      "step": 13248
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.8883142594398974e-06,
+      "loss": 0.0023,
+      "step": 13250
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.8860072324901079e-06,
+      "loss": 0.0027,
+      "step": 13252
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.8837014689510337e-06,
+      "loss": 0.0054,
+      "step": 13254
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.8813969691817047e-06,
+      "loss": 0.0017,
+      "step": 13256
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.8790937335409487e-06,
+      "loss": 0.0049,
+      "step": 13258
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.8767917623873943e-06,
+      "loss": 0.0014,
+      "step": 13260
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.8744910560794783e-06,
+      "loss": 0.005,
+      "step": 13262
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.8721916149754392e-06,
+      "loss": 0.0017,
+      "step": 13264
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.8698934394333146e-06,
+      "loss": 0.0019,
+      "step": 13266
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.8675965298109522e-06,
+      "loss": 0.004,
+      "step": 13268
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.8653008864659926e-06,
+      "loss": 0.0031,
+      "step": 13270
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.863006509755888e-06,
+      "loss": 0.004,
+      "step": 13272
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.8607134000378945e-06,
+      "loss": 0.0057,
+      "step": 13274
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.858421557669059e-06,
+      "loss": 0.0022,
+      "step": 13276
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.8561309830062434e-06,
+      "loss": 0.0038,
+      "step": 13278
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.8538416764061097e-06,
+      "loss": 0.0013,
+      "step": 13280
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.8515536382251176e-06,
+      "loss": 0.0009,
+      "step": 13282
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.8492668688195293e-06,
+      "loss": 0.0024,
+      "step": 13284
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.8469813685454164e-06,
+      "loss": 0.0017,
+      "step": 13286
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.8446971377586497e-06,
+      "loss": 0.0015,
+      "step": 13288
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.8424141768148962e-06,
+      "loss": 0.0021,
+      "step": 13290
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.8401324860696347e-06,
+      "loss": 0.0043,
+      "step": 13292
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.8378520658781429e-06,
+      "loss": 0.0035,
+      "step": 13294
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.8355729165954973e-06,
+      "loss": 0.0052,
+      "step": 13296
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.833295038576577e-06,
+      "loss": 0.0015,
+      "step": 13298
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.831018432176066e-06,
+      "loss": 0.0028,
+      "step": 13300
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.8287430977484522e-06,
+      "loss": 0.0016,
+      "step": 13302
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.826469035648022e-06,
+      "loss": 0.0084,
+      "step": 13304
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.8241962462288609e-06,
+      "loss": 0.0057,
+      "step": 13306
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.821924729844865e-06,
+      "loss": 0.0027,
+      "step": 13308
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.8196544868497201e-06,
+      "loss": 0.0024,
+      "step": 13310
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.8173855175969268e-06,
+      "loss": 0.0045,
+      "step": 13312
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.8151178224397758e-06,
+      "loss": 0.0042,
+      "step": 13314
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.812851401731367e-06,
+      "loss": 0.0047,
+      "step": 13316
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.8105862558245979e-06,
+      "loss": 0.0032,
+      "step": 13318
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.8083223850721743e-06,
+      "loss": 0.0015,
+      "step": 13320
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.8060597898265907e-06,
+      "loss": 0.0009,
+      "step": 13322
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.8037984704401579e-06,
+      "loss": 0.0023,
+      "step": 13324
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.8015384272649726e-06,
+      "loss": 0.0023,
+      "step": 13326
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.7992796606529483e-06,
+      "loss": 0.0031,
+      "step": 13328
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.7970221709557868e-06,
+      "loss": 0.0016,
+      "step": 13330
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.7947659585249977e-06,
+      "loss": 0.001,
+      "step": 13332
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.7925110237118914e-06,
+      "loss": 0.0024,
+      "step": 13334
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.7902573668675815e-06,
+      "loss": 0.0025,
+      "step": 13336
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.7880049883429774e-06,
+      "loss": 0.0026,
+      "step": 13338
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.785753888488787e-06,
+      "loss": 0.0028,
+      "step": 13340
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.783504067655528e-06,
+      "loss": 0.0034,
+      "step": 13342
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.7812555261935172e-06,
+      "loss": 0.003,
+      "step": 13344
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.779008264452864e-06,
+      "loss": 0.0033,
+      "step": 13346
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.7767622827834873e-06,
+      "loss": 0.0065,
+      "step": 13348
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.7745175815351057e-06,
+      "loss": 0.0006,
+      "step": 13350
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.7722741610572348e-06,
+      "loss": 0.0037,
+      "step": 13352
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.7700320216991874e-06,
+      "loss": 0.0033,
+      "step": 13354
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.7677911638100876e-06,
+      "loss": 0.0062,
+      "step": 13356
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.765551587738852e-06,
+      "loss": 0.0038,
+      "step": 13358
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.7633132938342023e-06,
+      "loss": 0.003,
+      "step": 13360
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.7610762824446537e-06,
+      "loss": 0.0017,
+      "step": 13362
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.7588405539185304e-06,
+      "loss": 0.0036,
+      "step": 13364
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.7566061086039476e-06,
+      "loss": 0.0031,
+      "step": 13366
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.7543729468488301e-06,
+      "loss": 0.0061,
+      "step": 13368
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.7521410690008944e-06,
+      "loss": 0.0027,
+      "step": 13370
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.7499104754076623e-06,
+      "loss": 0.0007,
+      "step": 13372
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.747681166416455e-06,
+      "loss": 0.0032,
+      "step": 13374
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.7454531423743948e-06,
+      "loss": 0.0012,
+      "step": 13376
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.7432264036283986e-06,
+      "loss": 0.0028,
+      "step": 13378
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.741000950525189e-06,
+      "loss": 0.001,
+      "step": 13380
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.7387767834112833e-06,
+      "loss": 0.0023,
+      "step": 13382
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.7365539026330058e-06,
+      "loss": 0.0067,
+      "step": 13384
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.7343323085364717e-06,
+      "loss": 0.0049,
+      "step": 13386
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.7321120014676006e-06,
+      "loss": 0.0043,
+      "step": 13388
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.7298929817721122e-06,
+      "loss": 0.002,
+      "step": 13390
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.7276752497955286e-06,
+      "loss": 0.0031,
+      "step": 13392
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.7254588058831633e-06,
+      "loss": 0.0047,
+      "step": 13394
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.7232436503801332e-06,
+      "loss": 0.0022,
+      "step": 13396
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.721029783631355e-06,
+      "loss": 0.0018,
+      "step": 13398
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.718817205981549e-06,
+      "loss": 0.004,
+      "step": 13400
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.7166059177752249e-06,
+      "loss": 0.0045,
+      "step": 13402
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.7143959193566995e-06,
+      "loss": 0.0024,
+      "step": 13404
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.7121872110700888e-06,
+      "loss": 0.002,
+      "step": 13406
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.7099797932593043e-06,
+      "loss": 0.0031,
+      "step": 13408
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.7077736662680533e-06,
+      "loss": 0.002,
+      "step": 13410
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.705568830439851e-06,
+      "loss": 0.0057,
+      "step": 13412
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.703365286118006e-06,
+      "loss": 0.0038,
+      "step": 13414
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.7011630336456296e-06,
+      "loss": 0.0031,
+      "step": 13416
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.6989620733656253e-06,
+      "loss": 0.0039,
+      "step": 13418
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.696762405620701e-06,
+      "loss": 0.0052,
+      "step": 13420
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.694564030753365e-06,
+      "loss": 0.0029,
+      "step": 13422
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.692366949105918e-06,
+      "loss": 0.0063,
+      "step": 13424
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.6901711610204597e-06,
+      "loss": 0.0019,
+      "step": 13426
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.6879766668388943e-06,
+      "loss": 0.0046,
+      "step": 13428
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.6857834669029216e-06,
+      "loss": 0.0026,
+      "step": 13430
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.683591561554041e-06,
+      "loss": 0.0025,
+      "step": 13432
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.681400951133544e-06,
+      "loss": 0.0028,
+      "step": 13434
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.6792116359825317e-06,
+      "loss": 0.002,
+      "step": 13436
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.6770236164418906e-06,
+      "loss": 0.0028,
+      "step": 13438
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.6748368928523174e-06,
+      "loss": 0.0017,
+      "step": 13440
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.672651465554298e-06,
+      "loss": 0.0086,
+      "step": 13442
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.6704673348881217e-06,
+      "loss": 0.0029,
+      "step": 13444
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.6682845011938742e-06,
+      "loss": 0.0024,
+      "step": 13446
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.6661029648114425e-06,
+      "loss": 0.0078,
+      "step": 13448
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.6639227260805047e-06,
+      "loss": 0.0035,
+      "step": 13450
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.6617437853405395e-06,
+      "loss": 0.009,
+      "step": 13452
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.6595661429308273e-06,
+      "loss": 0.0015,
+      "step": 13454
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.657389799190444e-06,
+      "loss": 0.0021,
+      "step": 13456
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.655214754458261e-06,
+      "loss": 0.0019,
+      "step": 13458
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.6530410090729498e-06,
+      "loss": 0.001,
+      "step": 13460
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.6508685633729826e-06,
+      "loss": 0.0029,
+      "step": 13462
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.648697417696622e-06,
+      "loss": 0.0033,
+      "step": 13464
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.646527572381932e-06,
+      "loss": 0.0022,
+      "step": 13466
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.6443590277667743e-06,
+      "loss": 0.0009,
+      "step": 13468
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.642191784188808e-06,
+      "loss": 0.0063,
+      "step": 13470
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.640025841985493e-06,
+      "loss": 0.0035,
+      "step": 13472
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.6378612014940788e-06,
+      "loss": 0.0015,
+      "step": 13474
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.6356978630516163e-06,
+      "loss": 0.002,
+      "step": 13476
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.6335358269949586e-06,
+      "loss": 0.0026,
+      "step": 13478
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.6313750936607487e-06,
+      "loss": 0.0035,
+      "step": 13480
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.629215663385425e-06,
+      "loss": 0.0034,
+      "step": 13482
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.627057536505232e-06,
+      "loss": 0.0031,
+      "step": 13484
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.6249007133562056e-06,
+      "loss": 0.0027,
+      "step": 13486
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.6227451942741812e-06,
+      "loss": 0.0017,
+      "step": 13488
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.620590979594786e-06,
+      "loss": 0.0009,
+      "step": 13490
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.6184380696534518e-06,
+      "loss": 0.0023,
+      "step": 13492
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.6162864647853993e-06,
+      "loss": 0.003,
+      "step": 13494
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.6141361653256538e-06,
+      "loss": 0.0048,
+      "step": 13496
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.6119871716090285e-06,
+      "loss": 0.0023,
+      "step": 13498
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.6098394839701403e-06,
+      "loss": 0.0007,
+      "step": 13500
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.6076931027434017e-06,
+      "loss": 0.0035,
+      "step": 13502
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.605548028263022e-06,
+      "loss": 0.006,
+      "step": 13504
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.6034042608630041e-06,
+      "loss": 0.0029,
+      "step": 13506
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.6012618008771464e-06,
+      "loss": 0.0013,
+      "step": 13508
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.5991206486390487e-06,
+      "loss": 0.0018,
+      "step": 13510
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.5969808044821068e-06,
+      "loss": 0.0033,
+      "step": 13512
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.594842268739506e-06,
+      "loss": 0.0023,
+      "step": 13514
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.5927050417442347e-06,
+      "loss": 0.0042,
+      "step": 13516
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.5905691238290788e-06,
+      "loss": 0.0019,
+      "step": 13518
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.5884345153266124e-06,
+      "loss": 0.0035,
+      "step": 13520
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.586301216569215e-06,
+      "loss": 0.0004,
+      "step": 13522
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.584169227889052e-06,
+      "loss": 0.0015,
+      "step": 13524
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.5820385496180923e-06,
+      "loss": 0.001,
+      "step": 13526
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.579909182088103e-06,
+      "loss": 0.002,
+      "step": 13528
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.5777811256306374e-06,
+      "loss": 0.0024,
+      "step": 13530
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.575654380577052e-06,
+      "loss": 0.0031,
+      "step": 13532
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.5735289472584991e-06,
+      "loss": 0.0017,
+      "step": 13534
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.5714048260059244e-06,
+      "loss": 0.0048,
+      "step": 13536
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.5692820171500655e-06,
+      "loss": 0.0022,
+      "step": 13538
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.5671605210214647e-06,
+      "loss": 0.0041,
+      "step": 13540
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.565040337950453e-06,
+      "loss": 0.0036,
+      "step": 13542
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.5629214682671623e-06,
+      "loss": 0.003,
+      "step": 13544
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.5608039123015117e-06,
+      "loss": 0.0065,
+      "step": 13546
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.5586876703832254e-06,
+      "loss": 0.0017,
+      "step": 13548
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.5565727428418153e-06,
+      "loss": 0.0058,
+      "step": 13550
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.5544591300065947e-06,
+      "loss": 0.0019,
+      "step": 13552
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.5523468322066659e-06,
+      "loss": 0.0014,
+      "step": 13554
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.5502358497709314e-06,
+      "loss": 0.001,
+      "step": 13556
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.548126183028088e-06,
+      "loss": 0.0019,
+      "step": 13558
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.5460178323066289e-06,
+      "loss": 0.0019,
+      "step": 13560
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.5439107979348366e-06,
+      "loss": 0.003,
+      "step": 13562
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.5418050802407924e-06,
+      "loss": 0.0033,
+      "step": 13564
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.539700679552374e-06,
+      "loss": 0.0066,
+      "step": 13566
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.5375975961972556e-06,
+      "loss": 0.002,
+      "step": 13568
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.5354958305028967e-06,
+      "loss": 0.0021,
+      "step": 13570
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.5333953827965631e-06,
+      "loss": 0.0009,
+      "step": 13572
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.53129625340531e-06,
+      "loss": 0.0019,
+      "step": 13574
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.529198442655989e-06,
+      "loss": 0.0045,
+      "step": 13576
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.5271019508752438e-06,
+      "loss": 0.0059,
+      "step": 13578
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.5250067783895128e-06,
+      "loss": 0.0022,
+      "step": 13580
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.5229129255250309e-06,
+      "loss": 0.0029,
+      "step": 13582
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.5208203926078302e-06,
+      "loss": 0.0018,
+      "step": 13584
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.5187291799637293e-06,
+      "loss": 0.0016,
+      "step": 13586
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.516639287918349e-06,
+      "loss": 0.0023,
+      "step": 13588
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.5145507167971019e-06,
+      "loss": 0.0016,
+      "step": 13590
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.5124634669251948e-06,
+      "loss": 0.0029,
+      "step": 13592
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.5103775386276243e-06,
+      "loss": 0.0035,
+      "step": 13594
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.5082929322291883e-06,
+      "loss": 0.007,
+      "step": 13596
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.5062096480544752e-06,
+      "loss": 0.0042,
+      "step": 13598
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.504127686427872e-06,
+      "loss": 0.0012,
+      "step": 13600
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.5020470476735505e-06,
+      "loss": 0.0028,
+      "step": 13602
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.499967732115487e-06,
+      "loss": 0.0026,
+      "step": 13604
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.4978897400774416e-06,
+      "loss": 0.0026,
+      "step": 13606
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.49581307188298e-06,
+      "loss": 0.0045,
+      "step": 13608
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.4937377278554477e-06,
+      "loss": 0.002,
+      "step": 13610
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.4916637083179964e-06,
+      "loss": 0.0037,
+      "step": 13612
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.4895910135935666e-06,
+      "loss": 0.0037,
+      "step": 13614
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.4875196440048934e-06,
+      "loss": 0.0022,
+      "step": 13616
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.4854495998745044e-06,
+      "loss": 0.0038,
+      "step": 13618
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.4833808815247175e-06,
+      "loss": 0.0003,
+      "step": 13620
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.4813134892776504e-06,
+      "loss": 0.0049,
+      "step": 13622
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.4792474234552156e-06,
+      "loss": 0.0026,
+      "step": 13624
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.4771826843791104e-06,
+      "loss": 0.0022,
+      "step": 13626
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.4751192723708318e-06,
+      "loss": 0.0038,
+      "step": 13628
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.4730571877516686e-06,
+      "loss": 0.0022,
+      "step": 13630
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.470996430842707e-06,
+      "loss": 0.0023,
+      "step": 13632
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.4689370019648198e-06,
+      "loss": 0.0022,
+      "step": 13634
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.466878901438672e-06,
+      "loss": 0.0081,
+      "step": 13636
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.4648221295847298e-06,
+      "loss": 0.0027,
+      "step": 13638
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.4627666867232493e-06,
+      "loss": 0.0039,
+      "step": 13640
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.4607125731742756e-06,
+      "loss": 0.004,
+      "step": 13642
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.4586597892576503e-06,
+      "loss": 0.0007,
+      "step": 13644
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.4566083352930105e-06,
+      "loss": 0.0021,
+      "step": 13646
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.4545582115997825e-06,
+      "loss": 0.0021,
+      "step": 13648
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.4525094184971812e-06,
+      "loss": 0.0016,
+      "step": 13650
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.4504619563042244e-06,
+      "loss": 0.003,
+      "step": 13652
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.4484158253397152e-06,
+      "loss": 0.0026,
+      "step": 13654
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.4463710259222563e-06,
+      "loss": 0.0022,
+      "step": 13656
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.444327558370232e-06,
+      "loss": 0.0023,
+      "step": 13658
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.4422854230018324e-06,
+      "loss": 0.0014,
+      "step": 13660
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.4402446201350273e-06,
+      "loss": 0.0023,
+      "step": 13662
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.4382051500875916e-06,
+      "loss": 0.0005,
+      "step": 13664
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.43616701317708e-06,
+      "loss": 0.0012,
+      "step": 13666
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.4341302097208487e-06,
+      "loss": 0.0032,
+      "step": 13668
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.432094740036044e-06,
+      "loss": 0.0024,
+      "step": 13670
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.430060604439607e-06,
+      "loss": 0.0031,
+      "step": 13672
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.4280278032482631e-06,
+      "loss": 0.0025,
+      "step": 13674
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.425996336778539e-06,
+      "loss": 0.0042,
+      "step": 13676
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.4239662053467452e-06,
+      "loss": 0.0035,
+      "step": 13678
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.4219374092689941e-06,
+      "loss": 0.0018,
+      "step": 13680
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.4199099488611789e-06,
+      "loss": 0.0023,
+      "step": 13682
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.4178838244389937e-06,
+      "loss": 0.0024,
+      "step": 13684
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.4158590363179215e-06,
+      "loss": 0.0008,
+      "step": 13686
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.4138355848132402e-06,
+      "loss": 0.0013,
+      "step": 13688
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.4118134702400133e-06,
+      "loss": 0.0041,
+      "step": 13690
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.409792692913098e-06,
+      "loss": 0.0016,
+      "step": 13692
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.4077732531471455e-06,
+      "loss": 0.0044,
+      "step": 13694
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.4057551512566025e-06,
+      "loss": 0.0028,
+      "step": 13696
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.4037383875556977e-06,
+      "loss": 0.0016,
+      "step": 13698
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.4017229623584583e-06,
+      "loss": 0.0017,
+      "step": 13700
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3997088759787036e-06,
+      "loss": 0.0043,
+      "step": 13702
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3976961287300407e-06,
+      "loss": 0.0031,
+      "step": 13704
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3956847209258672e-06,
+      "loss": 0.001,
+      "step": 13706
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3936746528793765e-06,
+      "loss": 0.0018,
+      "step": 13708
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3916659249035514e-06,
+      "loss": 0.0016,
+      "step": 13710
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3896585373111703e-06,
+      "loss": 0.0017,
+      "step": 13712
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3876524904147915e-06,
+      "loss": 0.0041,
+      "step": 13714
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3856477845267791e-06,
+      "loss": 0.0034,
+      "step": 13716
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3836444199592746e-06,
+      "loss": 0.0021,
+      "step": 13718
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3816423970242232e-06,
+      "loss": 0.0039,
+      "step": 13720
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3796417160333485e-06,
+      "loss": 0.0036,
+      "step": 13722
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.3776423772981772e-06,
+      "loss": 0.0026,
+      "step": 13724
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.375644381130019e-06,
+      "loss": 0.0033,
+      "step": 13726
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.3736477278399806e-06,
+      "loss": 0.0031,
+      "step": 13728
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.3716524177389511e-06,
+      "loss": 0.0072,
+      "step": 13730
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.36965845113762e-06,
+      "loss": 0.0002,
+      "step": 13732
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.3676658283464595e-06,
+      "loss": 0.003,
+      "step": 13734
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.3656745496757407e-06,
+      "loss": 0.0004,
+      "step": 13736
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.3636846154355155e-06,
+      "loss": 0.0021,
+      "step": 13738
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.3616960259356348e-06,
+      "loss": 0.0004,
+      "step": 13740
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.3597087814857368e-06,
+      "loss": 0.0023,
+      "step": 13742
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.3577228823952526e-06,
+      "loss": 0.0011,
+      "step": 13744
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.3557383289734004e-06,
+      "loss": 0.0063,
+      "step": 13746
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.353755121529189e-06,
+      "loss": 0.0044,
+      "step": 13748
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.3517732603714183e-06,
+      "loss": 0.0016,
+      "step": 13750
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.3497927458086836e-06,
+      "loss": 0.0019,
+      "step": 13752
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.3478135781493617e-06,
+      "loss": 0.0035,
+      "step": 13754
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.3458357577016257e-06,
+      "loss": 0.002,
+      "step": 13756
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.34385928477344e-06,
+      "loss": 0.0046,
+      "step": 13758
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.341884159672554e-06,
+      "loss": 0.0054,
+      "step": 13760
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.3399103827065075e-06,
+      "loss": 0.0049,
+      "step": 13762
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.3379379541826353e-06,
+      "loss": 0.0014,
+      "step": 13764
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.3359668744080601e-06,
+      "loss": 0.0022,
+      "step": 13766
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.3339971436896949e-06,
+      "loss": 0.0058,
+      "step": 13768
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.3320287623342376e-06,
+      "loss": 0.004,
+      "step": 13770
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.3300617306481833e-06,
+      "loss": 0.0026,
+      "step": 13772
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.328096048937816e-06,
+      "loss": 0.0038,
+      "step": 13774
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.3261317175092047e-06,
+      "loss": 0.0023,
+      "step": 13776
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.32416873666821e-06,
+      "loss": 0.0042,
+      "step": 13778
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.3222071067204822e-06,
+      "loss": 0.0028,
+      "step": 13780
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.3202468279714653e-06,
+      "loss": 0.0023,
+      "step": 13782
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.3182879007263906e-06,
+      "loss": 0.0017,
+      "step": 13784
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.3163303252902727e-06,
+      "loss": 0.0023,
+      "step": 13786
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.3143741019679268e-06,
+      "loss": 0.0073,
+      "step": 13788
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.312419231063946e-06,
+      "loss": 0.0066,
+      "step": 13790
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.3104657128827247e-06,
+      "loss": 0.0048,
+      "step": 13792
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.308513547728436e-06,
+      "loss": 0.0046,
+      "step": 13794
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.3065627359050491e-06,
+      "loss": 0.0013,
+      "step": 13796
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.304613277716319e-06,
+      "loss": 0.0047,
+      "step": 13798
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.3026651734657947e-06,
+      "loss": 0.0014,
+      "step": 13800
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.3007184234568082e-06,
+      "loss": 0.0037,
+      "step": 13802
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.2987730279924815e-06,
+      "loss": 0.002,
+      "step": 13804
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.2968289873757311e-06,
+      "loss": 0.001,
+      "step": 13806
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.294886301909256e-06,
+      "loss": 0.0045,
+      "step": 13808
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.292944971895549e-06,
+      "loss": 0.0028,
+      "step": 13810
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.2910049976368888e-06,
+      "loss": 0.0025,
+      "step": 13812
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.2890663794353476e-06,
+      "loss": 0.0028,
+      "step": 13814
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.2871291175927814e-06,
+      "loss": 0.0024,
+      "step": 13816
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.2851932124108323e-06,
+      "loss": 0.0019,
+      "step": 13818
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.2832586641909405e-06,
+      "loss": 0.0041,
+      "step": 13820
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.2813254732343294e-06,
+      "loss": 0.0023,
+      "step": 13822
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.2793936398420093e-06,
+      "loss": 0.0036,
+      "step": 13824
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.2774631643147827e-06,
+      "loss": 0.0027,
+      "step": 13826
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.2755340469532407e-06,
+      "loss": 0.0007,
+      "step": 13828
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.2736062880577616e-06,
+      "loss": 0.0019,
+      "step": 13830
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.2716798879285108e-06,
+      "loss": 0.004,
+      "step": 13832
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.269754846865443e-06,
+      "loss": 0.0036,
+      "step": 13834
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.2678311651683017e-06,
+      "loss": 0.0014,
+      "step": 13836
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.2659088431366218e-06,
+      "loss": 0.0033,
+      "step": 13838
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.2639878810697204e-06,
+      "loss": 0.0016,
+      "step": 13840
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.262068279266706e-06,
+      "loss": 0.004,
+      "step": 13842
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.2601500380264798e-06,
+      "loss": 0.0019,
+      "step": 13844
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.2582331576477225e-06,
+      "loss": 0.0027,
+      "step": 13846
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.2563176384289055e-06,
+      "loss": 0.0023,
+      "step": 13848
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.2544034806682926e-06,
+      "loss": 0.0018,
+      "step": 13850
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.252490684663933e-06,
+      "loss": 0.0033,
+      "step": 13852
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.2505792507136638e-06,
+      "loss": 0.0009,
+      "step": 13854
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.2486691791151072e-06,
+      "loss": 0.0027,
+      "step": 13856
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.24676047016568e-06,
+      "loss": 0.0037,
+      "step": 13858
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.2448531241625784e-06,
+      "loss": 0.0015,
+      "step": 13860
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.2429471414027949e-06,
+      "loss": 0.0018,
+      "step": 13862
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.2410425221831024e-06,
+      "loss": 0.0018,
+      "step": 13864
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.2391392668000646e-06,
+      "loss": 0.003,
+      "step": 13866
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.2372373755500356e-06,
+      "loss": 0.003,
+      "step": 13868
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.235336848729154e-06,
+      "loss": 0.0019,
+      "step": 13870
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.2334376866333441e-06,
+      "loss": 0.0044,
+      "step": 13872
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.231539889558323e-06,
+      "loss": 0.0019,
+      "step": 13874
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.2296434577995875e-06,
+      "loss": 0.0029,
+      "step": 13876
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.2277483916524324e-06,
+      "loss": 0.0028,
+      "step": 13878
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.2258546914119295e-06,
+      "loss": 0.0029,
+      "step": 13880
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.223962357372943e-06,
+      "loss": 0.0021,
+      "step": 13882
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.222071389830125e-06,
+      "loss": 0.0031,
+      "step": 13884
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.2201817890779155e-06,
+      "loss": 0.0017,
+      "step": 13886
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.2182935554105379e-06,
+      "loss": 0.002,
+      "step": 13888
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.2164066891220017e-06,
+      "loss": 0.0016,
+      "step": 13890
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.2145211905061094e-06,
+      "loss": 0.0038,
+      "step": 13892
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.2126370598564497e-06,
+      "loss": 0.0013,
+      "step": 13894
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.210754297466391e-06,
+      "loss": 0.0026,
+      "step": 13896
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.208872903629097e-06,
+      "loss": 0.0044,
+      "step": 13898
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.2069928786375173e-06,
+      "loss": 0.0034,
+      "step": 13900
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.2051142227843826e-06,
+      "loss": 0.002,
+      "step": 13902
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.203236936362212e-06,
+      "loss": 0.0052,
+      "step": 13904
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.2013610196633174e-06,
+      "loss": 0.0017,
+      "step": 13906
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.1994864729797918e-06,
+      "loss": 0.0027,
+      "step": 13908
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.1976132966035192e-06,
+      "loss": 0.0031,
+      "step": 13910
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.1957414908261623e-06,
+      "loss": 0.0024,
+      "step": 13912
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.1938710559391809e-06,
+      "loss": 0.001,
+      "step": 13914
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.1920019922338112e-06,
+      "loss": 0.0016,
+      "step": 13916
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.1901343000010856e-06,
+      "loss": 0.0018,
+      "step": 13918
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.1882679795318131e-06,
+      "loss": 0.0018,
+      "step": 13920
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.1864030311165965e-06,
+      "loss": 0.0076,
+      "step": 13922
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.1845394550458223e-06,
+      "loss": 0.0018,
+      "step": 13924
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.1826772516096652e-06,
+      "loss": 0.0018,
+      "step": 13926
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.1808164210980821e-06,
+      "loss": 0.0015,
+      "step": 13928
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.1789569638008202e-06,
+      "loss": 0.0032,
+      "step": 13930
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.177098880007409e-06,
+      "loss": 0.0016,
+      "step": 13932
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.1752421700071704e-06,
+      "loss": 0.0026,
+      "step": 13934
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.173386834089203e-06,
+      "loss": 0.0033,
+      "step": 13936
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.1715328725424002e-06,
+      "loss": 0.0044,
+      "step": 13938
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.1696802856554368e-06,
+      "loss": 0.0043,
+      "step": 13940
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.1678290737167785e-06,
+      "loss": 0.0014,
+      "step": 13942
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.1659792370146694e-06,
+      "loss": 0.0029,
+      "step": 13944
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.1641307758371412e-06,
+      "loss": 0.0011,
+      "step": 13946
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.1622836904720159e-06,
+      "loss": 0.0029,
+      "step": 13948
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.160437981206901e-06,
+      "loss": 0.0022,
+      "step": 13950
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.1585936483291837e-06,
+      "loss": 0.0029,
+      "step": 13952
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.156750692126043e-06,
+      "loss": 0.0019,
+      "step": 13954
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.1549091128844425e-06,
+      "loss": 0.002,
+      "step": 13956
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.1530689108911285e-06,
+      "loss": 0.0017,
+      "step": 13958
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.151230086432632e-06,
+      "loss": 0.0025,
+      "step": 13960
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.1493926397952748e-06,
+      "loss": 0.0032,
+      "step": 13962
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.1475565712651615e-06,
+      "loss": 0.001,
+      "step": 13964
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.1457218811281834e-06,
+      "loss": 0.0013,
+      "step": 13966
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.1438885696700108e-06,
+      "loss": 0.0009,
+      "step": 13968
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.142056637176111e-06,
+      "loss": 0.0036,
+      "step": 13970
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.1402260839317237e-06,
+      "loss": 0.0062,
+      "step": 13972
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.138396910221885e-06,
+      "loss": 0.0024,
+      "step": 13974
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.1365691163314074e-06,
+      "loss": 0.0051,
+      "step": 13976
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.1347427025448931e-06,
+      "loss": 0.0015,
+      "step": 13978
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.1329176691467303e-06,
+      "loss": 0.0047,
+      "step": 13980
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.1310940164210916e-06,
+      "loss": 0.0014,
+      "step": 13982
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.1292717446519296e-06,
+      "loss": 0.0037,
+      "step": 13984
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.12745085412299e-06,
+      "loss": 0.0051,
+      "step": 13986
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.1256313451177968e-06,
+      "loss": 0.0022,
+      "step": 13988
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.1238132179196636e-06,
+      "loss": 0.0016,
+      "step": 13990
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.1219964728116839e-06,
+      "loss": 0.0024,
+      "step": 13992
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.12018111007674e-06,
+      "loss": 0.0027,
+      "step": 13994
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.1183671299974984e-06,
+      "loss": 0.0006,
+      "step": 13996
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.116554532856412e-06,
+      "loss": 0.0037,
+      "step": 13998
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.1147433189357125e-06,
+      "loss": 0.0027,
+      "step": 14000
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.1129334885174193e-06,
+      "loss": 0.0036,
+      "step": 14002
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.1111250418833376e-06,
+      "loss": 0.002,
+      "step": 14004
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.1093179793150598e-06,
+      "loss": 0.0013,
+      "step": 14006
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.1075123010939538e-06,
+      "loss": 0.0023,
+      "step": 14008
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.1057080075011805e-06,
+      "loss": 0.0025,
+      "step": 14010
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.1039050988176847e-06,
+      "loss": 0.0043,
+      "step": 14012
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.1021035753241893e-06,
+      "loss": 0.0058,
+      "step": 14014
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.1003034373012055e-06,
+      "loss": 0.0018,
+      "step": 14016
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.0985046850290281e-06,
+      "loss": 0.0012,
+      "step": 14018
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.0967073187877386e-06,
+      "loss": 0.003,
+      "step": 14020
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.094911338857202e-06,
+      "loss": 0.0023,
+      "step": 14022
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.0931167455170622e-06,
+      "loss": 0.0026,
+      "step": 14024
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.0913235390467524e-06,
+      "loss": 0.0032,
+      "step": 14026
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.0895317197254919e-06,
+      "loss": 0.0038,
+      "step": 14028
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.087741287832277e-06,
+      "loss": 0.0041,
+      "step": 14030
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.0859522436458903e-06,
+      "loss": 0.0017,
+      "step": 14032
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.0841645874449025e-06,
+      "loss": 0.0007,
+      "step": 14034
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.0823783195076654e-06,
+      "loss": 0.0019,
+      "step": 14036
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.0805934401123153e-06,
+      "loss": 0.0018,
+      "step": 14038
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.0788099495367677e-06,
+      "loss": 0.0029,
+      "step": 14040
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.0770278480587303e-06,
+      "loss": 0.0006,
+      "step": 14042
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.075247135955686e-06,
+      "loss": 0.0038,
+      "step": 14044
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.0734678135049093e-06,
+      "loss": 0.0039,
+      "step": 14046
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.0716898809834497e-06,
+      "loss": 0.0027,
+      "step": 14048
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.069913338668148e-06,
+      "loss": 0.0008,
+      "step": 14050
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.0681381868356245e-06,
+      "loss": 0.0014,
+      "step": 14052
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.066364425762285e-06,
+      "loss": 0.0014,
+      "step": 14054
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.0645920557243184e-06,
+      "loss": 0.0031,
+      "step": 14056
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.0628210769976922e-06,
+      "loss": 0.0029,
+      "step": 14058
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.0610514898581636e-06,
+      "loss": 0.0011,
+      "step": 14060
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.0592832945812725e-06,
+      "loss": 0.0033,
+      "step": 14062
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.0575164914423374e-06,
+      "loss": 0.0018,
+      "step": 14064
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.055751080716465e-06,
+      "loss": 0.0019,
+      "step": 14066
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.0539870626785452e-06,
+      "loss": 0.0044,
+      "step": 14068
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.0522244376032464e-06,
+      "loss": 0.0024,
+      "step": 14070
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.050463205765021e-06,
+      "loss": 0.0029,
+      "step": 14072
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.0487033674381086e-06,
+      "loss": 0.0025,
+      "step": 14074
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.0469449228965289e-06,
+      "loss": 0.0038,
+      "step": 14076
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.0451878724140884e-06,
+      "loss": 0.0011,
+      "step": 14078
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.0434322162643673e-06,
+      "loss": 0.0055,
+      "step": 14080
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.041677954720739e-06,
+      "loss": 0.0072,
+      "step": 14082
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.039925088056356e-06,
+      "loss": 0.0028,
+      "step": 14084
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.0381736165441514e-06,
+      "loss": 0.0027,
+      "step": 14086
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.0364235404568402e-06,
+      "loss": 0.0046,
+      "step": 14088
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.0346748600669254e-06,
+      "loss": 0.0031,
+      "step": 14090
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.0329275756466905e-06,
+      "loss": 0.003,
+      "step": 14092
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.0311816874682012e-06,
+      "loss": 0.0007,
+      "step": 14094
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.0294371958033023e-06,
+      "loss": 0.0025,
+      "step": 14096
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.0276941009236296e-06,
+      "loss": 0.0044,
+      "step": 14098
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.0259524031005907e-06,
+      "loss": 0.003,
+      "step": 14100
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.0242121026053865e-06,
+      "loss": 0.0026,
+      "step": 14102
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.022473199708991e-06,
+      "loss": 0.0022,
+      "step": 14104
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.0207356946821655e-06,
+      "loss": 0.0024,
+      "step": 14106
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.018999587795454e-06,
+      "loss": 0.0017,
+      "step": 14108
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.0172648793191831e-06,
+      "loss": 0.0054,
+      "step": 14110
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.015531569523459e-06,
+      "loss": 0.003,
+      "step": 14112
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.0137996586781684e-06,
+      "loss": 0.0035,
+      "step": 14114
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.0120691470529843e-06,
+      "loss": 0.0007,
+      "step": 14116
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.0103400349173643e-06,
+      "loss": 0.0027,
+      "step": 14118
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.0086123225405398e-06,
+      "loss": 0.003,
+      "step": 14120
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.0068860101915312e-06,
+      "loss": 0.0021,
+      "step": 14122
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.0051610981391392e-06,
+      "loss": 0.0027,
+      "step": 14124
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.003437586651943e-06,
+      "loss": 0.0015,
+      "step": 14126
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.0017154759983094e-06,
+      "loss": 0.0037,
+      "step": 14128
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 9.999947664463815e-07,
+      "loss": 0.0015,
+      "step": 14130
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 9.982754582640885e-07,
+      "loss": 0.0038,
+      "step": 14132
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 9.965575517191418e-07,
+      "loss": 0.0008,
+      "step": 14134
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 9.94841047079028e-07,
+      "loss": 0.0028,
+      "step": 14136
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 9.93125944611023e-07,
+      "loss": 0.0042,
+      "step": 14138
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 9.914122445821828e-07,
+      "loss": 0.0031,
+      "step": 14140
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 9.896999472593415e-07,
+      "loss": 0.003,
+      "step": 14142
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 9.879890529091152e-07,
+      "loss": 0.0021,
+      "step": 14144
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 9.862795617979049e-07,
+      "loss": 0.0065,
+      "step": 14146
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 9.84571474191891e-07,
+      "loss": 0.0049,
+      "step": 14148
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 9.828647903570387e-07,
+      "loss": 0.0054,
+      "step": 14150
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 9.811595105590876e-07,
+      "loss": 0.0009,
+      "step": 14152
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 9.794556350635675e-07,
+      "loss": 0.0033,
+      "step": 14154
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 9.777531641357784e-07,
+      "loss": 0.0029,
+      "step": 14156
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 9.76052098040815e-07,
+      "loss": 0.0011,
+      "step": 14158
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 9.74352437043541e-07,
+      "loss": 0.0027,
+      "step": 14160
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 9.726541814086076e-07,
+      "loss": 0.0015,
+      "step": 14162
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 9.709573314004473e-07,
+      "loss": 0.0015,
+      "step": 14164
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 9.692618872832748e-07,
+      "loss": 0.0023,
+      "step": 14166
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 9.675678493210806e-07,
+      "loss": 0.0024,
+      "step": 14168
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 9.658752177776386e-07,
+      "loss": 0.0018,
+      "step": 14170
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 9.641839929165063e-07,
+      "loss": 0.0026,
+      "step": 14172
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 9.62494175001022e-07,
+      "loss": 0.001,
+      "step": 14174
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 9.608057642942992e-07,
+      "loss": 0.0014,
+      "step": 14176
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 9.59118761059238e-07,
+      "loss": 0.0037,
+      "step": 14178
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 9.57433165558519e-07,
+      "loss": 0.007,
+      "step": 14180
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 9.557489780546025e-07,
+      "loss": 0.0017,
+      "step": 14182
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 9.540661988097289e-07,
+      "loss": 0.0026,
+      "step": 14184
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 9.523848280859172e-07,
+      "loss": 0.0017,
+      "step": 14186
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 9.507048661449714e-07,
+      "loss": 0.0029,
+      "step": 14188
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 9.490263132484778e-07,
+      "loss": 0.0018,
+      "step": 14190
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 9.473491696577941e-07,
+      "loss": 0.0034,
+      "step": 14192
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 9.456734356340668e-07,
+      "loss": 0.0012,
+      "step": 14194
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 9.43999111438223e-07,
+      "loss": 0.0031,
+      "step": 14196
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 9.423261973309661e-07,
+      "loss": 0.0031,
+      "step": 14198
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 9.40654693572779e-07,
+      "loss": 0.0013,
+      "step": 14200
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 9.3898460042393e-07,
+      "loss": 0.0038,
+      "step": 14202
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 9.373159181444647e-07,
+      "loss": 0.0035,
+      "step": 14204
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 9.356486469942128e-07,
+      "loss": 0.0025,
+      "step": 14206
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 9.339827872327756e-07,
+      "loss": 0.0022,
+      "step": 14208
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 9.323183391195457e-07,
+      "loss": 0.0024,
+      "step": 14210
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 9.306553029136855e-07,
+      "loss": 0.0037,
+      "step": 14212
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 9.289936788741472e-07,
+      "loss": 0.0031,
+      "step": 14214
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 9.273334672596535e-07,
+      "loss": 0.004,
+      "step": 14216
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 9.256746683287143e-07,
+      "loss": 0.0014,
+      "step": 14218
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 9.240172823396176e-07,
+      "loss": 0.0027,
+      "step": 14220
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 9.223613095504336e-07,
+      "loss": 0.0007,
+      "step": 14222
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 9.207067502190037e-07,
+      "loss": 0.0015,
+      "step": 14224
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 9.190536046029618e-07,
+      "loss": 0.0018,
+      "step": 14226
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 9.174018729597112e-07,
+      "loss": 0.0019,
+      "step": 14228
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 9.157515555464414e-07,
+      "loss": 0.0045,
+      "step": 14230
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 9.14102652620118e-07,
+      "loss": 0.0008,
+      "step": 14232
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 9.124551644374868e-07,
+      "loss": 0.0031,
+      "step": 14234
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 9.10809091255076e-07,
+      "loss": 0.0036,
+      "step": 14236
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 9.091644333291938e-07,
+      "loss": 0.002,
+      "step": 14238
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 9.075211909159242e-07,
+      "loss": 0.0047,
+      "step": 14240
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 9.058793642711294e-07,
+      "loss": 0.0032,
+      "step": 14242
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 9.042389536504581e-07,
+      "loss": 0.0038,
+      "step": 14244
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 9.025999593093349e-07,
+      "loss": 0.0024,
+      "step": 14246
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 9.009623815029611e-07,
+      "loss": 0.0021,
+      "step": 14248
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 8.993262204863218e-07,
+      "loss": 0.0045,
+      "step": 14250
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 8.976914765141809e-07,
+      "loss": 0.0021,
+      "step": 14252
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 8.960581498410803e-07,
+      "loss": 0.0035,
+      "step": 14254
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 8.944262407213378e-07,
+      "loss": 0.0018,
+      "step": 14256
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 8.927957494090567e-07,
+      "loss": 0.0007,
+      "step": 14258
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 8.911666761581173e-07,
+      "loss": 0.0065,
+      "step": 14260
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 8.895390212221811e-07,
+      "loss": 0.0014,
+      "step": 14262
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 8.879127848546809e-07,
+      "loss": 0.0007,
+      "step": 14264
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 8.862879673088398e-07,
+      "loss": 0.0065,
+      "step": 14266
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 8.846645688376488e-07,
+      "loss": 0.0061,
+      "step": 14268
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 8.830425896938888e-07,
+      "loss": 0.0015,
+      "step": 14270
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 8.81422030130109e-07,
+      "loss": 0.0022,
+      "step": 14272
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 8.798028903986467e-07,
+      "loss": 0.0023,
+      "step": 14274
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 8.781851707516131e-07,
+      "loss": 0.0019,
+      "step": 14276
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 8.765688714409016e-07,
+      "loss": 0.0041,
+      "step": 14278
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 8.749539927181782e-07,
+      "loss": 0.0055,
+      "step": 14280
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 8.733405348348967e-07,
+      "loss": 0.0025,
+      "step": 14282
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 8.717284980422791e-07,
+      "loss": 0.0039,
+      "step": 14284
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 8.701178825913382e-07,
+      "loss": 0.0015,
+      "step": 14286
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 8.685086887328542e-07,
+      "loss": 0.005,
+      "step": 14288
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 8.669009167173925e-07,
+      "loss": 0.0014,
+      "step": 14290
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 8.65294566795295e-07,
+      "loss": 0.0042,
+      "step": 14292
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 8.63689639216686e-07,
+      "loss": 0.0014,
+      "step": 14294
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 8.620861342314624e-07,
+      "loss": 0.0025,
+      "step": 14296
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 8.604840520892998e-07,
+      "loss": 0.0031,
+      "step": 14298
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 8.588833930396578e-07,
+      "loss": 0.0035,
+      "step": 14300
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 8.572841573317714e-07,
+      "loss": 0.0045,
+      "step": 14302
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 8.556863452146513e-07,
+      "loss": 0.003,
+      "step": 14304
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 8.540899569370909e-07,
+      "loss": 0.004,
+      "step": 14306
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 8.524949927476611e-07,
+      "loss": 0.002,
+      "step": 14308
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 8.50901452894709e-07,
+      "loss": 0.0029,
+      "step": 14310
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 8.493093376263584e-07,
+      "loss": 0.0022,
+      "step": 14312
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 8.477186471905164e-07,
+      "loss": 0.0008,
+      "step": 14314
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 8.46129381834866e-07,
+      "loss": 0.0049,
+      "step": 14316
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 8.445415418068681e-07,
+      "loss": 0.0005,
+      "step": 14318
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 8.429551273537595e-07,
+      "loss": 0.0034,
+      "step": 14320
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 8.413701387225604e-07,
+      "loss": 0.0063,
+      "step": 14322
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 8.39786576160061e-07,
+      "loss": 0.0025,
+      "step": 14324
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 8.382044399128386e-07,
+      "loss": 0.003,
+      "step": 14326
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 8.366237302272407e-07,
+      "loss": 0.0029,
+      "step": 14328
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 8.350444473493968e-07,
+      "loss": 0.0027,
+      "step": 14330
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 8.33466591525216e-07,
+      "loss": 0.0032,
+      "step": 14332
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 8.318901630003773e-07,
+      "loss": 0.0048,
+      "step": 14334
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 8.303151620203464e-07,
+      "loss": 0.0021,
+      "step": 14336
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 8.287415888303641e-07,
+      "loss": 0.0051,
+      "step": 14338
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 8.271694436754451e-07,
+      "loss": 0.0035,
+      "step": 14340
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 8.25598726800384e-07,
+      "loss": 0.0014,
+      "step": 14342
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 8.240294384497538e-07,
+      "loss": 0.0024,
+      "step": 14344
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 8.224615788679058e-07,
+      "loss": 0.0036,
+      "step": 14346
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 8.208951482989691e-07,
+      "loss": 0.003,
+      "step": 14348
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 8.193301469868464e-07,
+      "loss": 0.0035,
+      "step": 14350
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 8.177665751752217e-07,
+      "loss": 0.0032,
+      "step": 14352
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 8.162044331075536e-07,
+      "loss": 0.0016,
+      "step": 14354
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 8.146437210270819e-07,
+      "loss": 0.0009,
+      "step": 14356
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 8.130844391768189e-07,
+      "loss": 0.0027,
+      "step": 14358
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 8.11526587799557e-07,
+      "loss": 0.0026,
+      "step": 14360
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 8.099701671378668e-07,
+      "loss": 0.0018,
+      "step": 14362
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 8.084151774340965e-07,
+      "loss": 0.0008,
+      "step": 14364
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 8.068616189303679e-07,
+      "loss": 0.0014,
+      "step": 14366
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 8.053094918685799e-07,
+      "loss": 0.003,
+      "step": 14368
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 8.037587964904136e-07,
+      "loss": 0.0008,
+      "step": 14370
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 8.022095330373236e-07,
+      "loss": 0.0023,
+      "step": 14372
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 8.006617017505402e-07,
+      "loss": 0.0011,
+      "step": 14374
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 7.991153028710741e-07,
+      "loss": 0.0031,
+      "step": 14376
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 7.975703366397114e-07,
+      "loss": 0.0016,
+      "step": 14378
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 7.960268032970175e-07,
+      "loss": 0.0003,
+      "step": 14380
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 7.944847030833292e-07,
+      "loss": 0.0029,
+      "step": 14382
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 7.929440362387619e-07,
+      "loss": 0.0015,
+      "step": 14384
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 7.914048030032117e-07,
+      "loss": 0.001,
+      "step": 14386
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 7.898670036163503e-07,
+      "loss": 0.0038,
+      "step": 14388
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 7.883306383176215e-07,
+      "loss": 0.0021,
+      "step": 14390
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 7.867957073462507e-07,
+      "loss": 0.0061,
+      "step": 14392
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 7.8526221094124e-07,
+      "loss": 0.0035,
+      "step": 14394
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 7.83730149341364e-07,
+      "loss": 0.0008,
+      "step": 14396
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 7.821995227851775e-07,
+      "loss": 0.0023,
+      "step": 14398
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 7.806703315110098e-07,
+      "loss": 0.0026,
+      "step": 14400
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 7.791425757569682e-07,
+      "loss": 0.0031,
+      "step": 14402
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 7.776162557609379e-07,
+      "loss": 0.002,
+      "step": 14404
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 7.760913717605756e-07,
+      "loss": 0.001,
+      "step": 14406
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 7.745679239933202e-07,
+      "loss": 0.0014,
+      "step": 14408
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 7.730459126963808e-07,
+      "loss": 0.0051,
+      "step": 14410
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 7.7152533810675e-07,
+      "loss": 0.005,
+      "step": 14412
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 7.700062004611897e-07,
+      "loss": 0.0026,
+      "step": 14414
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 7.684884999962428e-07,
+      "loss": 0.0006,
+      "step": 14416
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 7.669722369482258e-07,
+      "loss": 0.0015,
+      "step": 14418
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 7.654574115532353e-07,
+      "loss": 0.004,
+      "step": 14420
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 7.639440240471385e-07,
+      "loss": 0.0015,
+      "step": 14422
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 7.624320746655811e-07,
+      "loss": 0.0028,
+      "step": 14424
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 7.60921563643986e-07,
+      "loss": 0.0024,
+      "step": 14426
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 7.594124912175527e-07,
+      "loss": 0.0034,
+      "step": 14428
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 7.579048576212534e-07,
+      "loss": 0.001,
+      "step": 14430
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 7.563986630898379e-07,
+      "loss": 0.0015,
+      "step": 14432
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 7.548939078578332e-07,
+      "loss": 0.0014,
+      "step": 14434
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 7.53390592159543e-07,
+      "loss": 0.0023,
+      "step": 14436
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 7.518887162290433e-07,
+      "loss": 0.0035,
+      "step": 14438
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 7.50388280300186e-07,
+      "loss": 0.0033,
+      "step": 14440
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 7.48889284606602e-07,
+      "loss": 0.0026,
+      "step": 14442
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 7.473917293816979e-07,
+      "loss": 0.007,
+      "step": 14444
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 7.458956148586516e-07,
+      "loss": 0.0022,
+      "step": 14446
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 7.444009412704211e-07,
+      "loss": 0.003,
+      "step": 14448
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 7.429077088497393e-07,
+      "loss": 0.0048,
+      "step": 14450
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 7.414159178291136e-07,
+      "loss": 0.0005,
+      "step": 14452
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 7.399255684408246e-07,
+      "loss": 0.008,
+      "step": 14454
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 7.384366609169336e-07,
+      "loss": 0.0022,
+      "step": 14456
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 7.369491954892749e-07,
+      "loss": 0.0035,
+      "step": 14458
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 7.35463172389459e-07,
+      "loss": 0.0021,
+      "step": 14460
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 7.339785918488673e-07,
+      "loss": 0.0017,
+      "step": 14462
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 7.32495454098665e-07,
+      "loss": 0.0013,
+      "step": 14464
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 7.310137593697853e-07,
+      "loss": 0.0038,
+      "step": 14466
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 7.2953350789294e-07,
+      "loss": 0.0014,
+      "step": 14468
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 7.280546998986149e-07,
+      "loss": 0.0025,
+      "step": 14470
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 7.265773356170724e-07,
+      "loss": 0.0037,
+      "step": 14472
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 7.251014152783487e-07,
+      "loss": 0.001,
+      "step": 14474
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 7.236269391122586e-07,
+      "loss": 0.0025,
+      "step": 14476
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 7.221539073483863e-07,
+      "loss": 0.0036,
+      "step": 14478
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 7.206823202160951e-07,
+      "loss": 0.0021,
+      "step": 14480
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 7.192121779445227e-07,
+      "loss": 0.0011,
+      "step": 14482
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 7.177434807625816e-07,
+      "loss": 0.0038,
+      "step": 14484
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 7.162762288989567e-07,
+      "loss": 0.0018,
+      "step": 14486
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 7.148104225821128e-07,
+      "loss": 0.0015,
+      "step": 14488
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 7.133460620402877e-07,
+      "loss": 0.0024,
+      "step": 14490
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 7.118831475014931e-07,
+      "loss": 0.0032,
+      "step": 14492
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 7.104216791935148e-07,
+      "loss": 0.0019,
+      "step": 14494
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 7.089616573439151e-07,
+      "loss": 0.0016,
+      "step": 14496
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 7.075030821800299e-07,
+      "loss": 0.0014,
+      "step": 14498
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 7.060459539289733e-07,
+      "loss": 0.003,
+      "step": 14500
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 7.045902728176268e-07,
+      "loss": 0.0021,
+      "step": 14502
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 7.03136039072655e-07,
+      "loss": 0.0036,
+      "step": 14504
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 7.01683252920492e-07,
+      "loss": 0.0045,
+      "step": 14506
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 7.002319145873482e-07,
+      "loss": 0.0043,
+      "step": 14508
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.98782024299206e-07,
+      "loss": 0.0058,
+      "step": 14510
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.97333582281825e-07,
+      "loss": 0.0037,
+      "step": 14512
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.958865887607402e-07,
+      "loss": 0.0019,
+      "step": 14514
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.944410439612603e-07,
+      "loss": 0.0009,
+      "step": 14516
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.929969481084642e-07,
+      "loss": 0.0019,
+      "step": 14518
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.91554301427213e-07,
+      "loss": 0.0003,
+      "step": 14520
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.901131041421327e-07,
+      "loss": 0.0005,
+      "step": 14522
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.886733564776349e-07,
+      "loss": 0.004,
+      "step": 14524
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.872350586578935e-07,
+      "loss": 0.0017,
+      "step": 14526
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.857982109068639e-07,
+      "loss": 0.0039,
+      "step": 14528
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.843628134482771e-07,
+      "loss": 0.0013,
+      "step": 14530
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.829288665056344e-07,
+      "loss": 0.0058,
+      "step": 14532
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.814963703022104e-07,
+      "loss": 0.0069,
+      "step": 14534
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.800653250610578e-07,
+      "loss": 0.0012,
+      "step": 14536
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.786357310049984e-07,
+      "loss": 0.0019,
+      "step": 14538
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.772075883566353e-07,
+      "loss": 0.0035,
+      "step": 14540
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.757808973383373e-07,
+      "loss": 0.0036,
+      "step": 14542
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.743556581722532e-07,
+      "loss": 0.003,
+      "step": 14544
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.729318710803024e-07,
+      "loss": 0.0032,
+      "step": 14546
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 6.715095362841817e-07,
+      "loss": 0.0044,
+      "step": 14548
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.700886540053575e-07,
+      "loss": 0.002,
+      "step": 14550
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.686692244650716e-07,
+      "loss": 0.0019,
+      "step": 14552
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.672512478843407e-07,
+      "loss": 0.003,
+      "step": 14554
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.658347244839558e-07,
+      "loss": 0.0023,
+      "step": 14556
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.644196544844784e-07,
+      "loss": 0.0022,
+      "step": 14558
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.630060381062464e-07,
+      "loss": 0.0019,
+      "step": 14560
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.61593875569373e-07,
+      "loss": 0.0044,
+      "step": 14562
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.601831670937409e-07,
+      "loss": 0.005,
+      "step": 14564
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.587739128990056e-07,
+      "loss": 0.0041,
+      "step": 14566
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.573661132046016e-07,
+      "loss": 0.0021,
+      "step": 14568
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.559597682297337e-07,
+      "loss": 0.0018,
+      "step": 14570
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.545548781933819e-07,
+      "loss": 0.0034,
+      "step": 14572
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.53151443314296e-07,
+      "loss": 0.0037,
+      "step": 14574
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.517494638110033e-07,
+      "loss": 0.0028,
+      "step": 14576
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.503489399018004e-07,
+      "loss": 0.0035,
+      "step": 14578
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.489498718047626e-07,
+      "loss": 0.0038,
+      "step": 14580
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.475522597377326e-07,
+      "loss": 0.002,
+      "step": 14582
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.461561039183306e-07,
+      "loss": 0.0037,
+      "step": 14584
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.447614045639494e-07,
+      "loss": 0.0023,
+      "step": 14586
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 6.433681618917542e-07,
+      "loss": 0.003,
+      "step": 14588
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.419763761186826e-07,
+      "loss": 0.001,
+      "step": 14590
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.405860474614478e-07,
+      "loss": 0.0041,
+      "step": 14592
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.391971761365323e-07,
+      "loss": 0.0035,
+      "step": 14594
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.378097623601964e-07,
+      "loss": 0.0013,
+      "step": 14596
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.364238063484684e-07,
+      "loss": 0.0037,
+      "step": 14598
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.350393083171535e-07,
+      "loss": 0.0043,
+      "step": 14600
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.336562684818292e-07,
+      "loss": 0.0039,
+      "step": 14602
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.322746870578477e-07,
+      "loss": 0.007,
+      "step": 14604
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.308945642603281e-07,
+      "loss": 0.0037,
+      "step": 14606
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.295159003041651e-07,
+      "loss": 0.0021,
+      "step": 14608
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.281386954040303e-07,
+      "loss": 0.0013,
+      "step": 14610
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.267629497743643e-07,
+      "loss": 0.0011,
+      "step": 14612
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.253886636293805e-07,
+      "loss": 0.0014,
+      "step": 14614
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.240158371830662e-07,
+      "loss": 0.0032,
+      "step": 14616
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.226444706491819e-07,
+      "loss": 0.0019,
+      "step": 14618
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.212745642412587e-07,
+      "loss": 0.0017,
+      "step": 14620
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.199061181726007e-07,
+      "loss": 0.0027,
+      "step": 14622
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.185391326562862e-07,
+      "loss": 0.0028,
+      "step": 14624
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.171736079051661e-07,
+      "loss": 0.0054,
+      "step": 14626
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.158095441318634e-07,
+      "loss": 0.0014,
+      "step": 14628
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 6.144469415487709e-07,
+      "loss": 0.0021,
+      "step": 14630
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 6.130858003680574e-07,
+      "loss": 0.0024,
+      "step": 14632
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 6.117261208016645e-07,
+      "loss": 0.0023,
+      "step": 14634
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 6.103679030613042e-07,
+      "loss": 0.0007,
+      "step": 14636
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 6.090111473584581e-07,
+      "loss": 0.0028,
+      "step": 14638
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 6.076558539043875e-07,
+      "loss": 0.002,
+      "step": 14640
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 6.063020229101191e-07,
+      "loss": 0.0008,
+      "step": 14642
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 6.049496545864586e-07,
+      "loss": 0.004,
+      "step": 14644
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 6.035987491439754e-07,
+      "loss": 0.0012,
+      "step": 14646
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 6.022493067930191e-07,
+      "loss": 0.0014,
+      "step": 14648
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 6.009013277437059e-07,
+      "loss": 0.0026,
+      "step": 14650
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.995548122059292e-07,
+      "loss": 0.0056,
+      "step": 14652
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.982097603893488e-07,
+      "loss": 0.0016,
+      "step": 14654
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.968661725034008e-07,
+      "loss": 0.0033,
+      "step": 14656
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.955240487572922e-07,
+      "loss": 0.0039,
+      "step": 14658
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.941833893600036e-07,
+      "loss": 0.0022,
+      "step": 14660
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.928441945202846e-07,
+      "loss": 0.0012,
+      "step": 14662
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.915064644466562e-07,
+      "loss": 0.0006,
+      "step": 14664
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.90170199347414e-07,
+      "loss": 0.0008,
+      "step": 14666
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.888353994306273e-07,
+      "loss": 0.0003,
+      "step": 14668
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.875020649041318e-07,
+      "loss": 0.0054,
+      "step": 14670
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.861701959755384e-07,
+      "loss": 0.0012,
+      "step": 14672
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.848397928522309e-07,
+      "loss": 0.0017,
+      "step": 14674
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.835108557413627e-07,
+      "loss": 0.0016,
+      "step": 14676
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.82183384849857e-07,
+      "loss": 0.0048,
+      "step": 14678
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.808573803844131e-07,
+      "loss": 0.0038,
+      "step": 14680
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.795328425515001e-07,
+      "loss": 0.0023,
+      "step": 14682
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.782097715573609e-07,
+      "loss": 0.0016,
+      "step": 14684
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.768881676080029e-07,
+      "loss": 0.0017,
+      "step": 14686
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.755680309092127e-07,
+      "loss": 0.0025,
+      "step": 14688
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.742493616665468e-07,
+      "loss": 0.0022,
+      "step": 14690
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.729321600853311e-07,
+      "loss": 0.0026,
+      "step": 14692
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.716164263706614e-07,
+      "loss": 0.0028,
+      "step": 14694
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.703021607274095e-07,
+      "loss": 0.0052,
+      "step": 14696
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.689893633602173e-07,
+      "loss": 0.0017,
+      "step": 14698
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.676780344734989e-07,
+      "loss": 0.0023,
+      "step": 14700
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.663681742714344e-07,
+      "loss": 0.0036,
+      "step": 14702
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.650597829579818e-07,
+      "loss": 0.0007,
+      "step": 14704
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.637528607368658e-07,
+      "loss": 0.0029,
+      "step": 14706
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.62447407811586e-07,
+      "loss": 0.0027,
+      "step": 14708
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.611434243854097e-07,
+      "loss": 0.0046,
+      "step": 14710
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.598409106613778e-07,
+      "loss": 0.0028,
+      "step": 14712
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.585398668423014e-07,
+      "loss": 0.0013,
+      "step": 14714
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.572402931307641e-07,
+      "loss": 0.0014,
+      "step": 14716
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.559421897291195e-07,
+      "loss": 0.0041,
+      "step": 14718
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.546455568394904e-07,
+      "loss": 0.0028,
+      "step": 14720
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.53350394663772e-07,
+      "loss": 0.0047,
+      "step": 14722
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.520567034036351e-07,
+      "loss": 0.0031,
+      "step": 14724
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.50764483260513e-07,
+      "loss": 0.0019,
+      "step": 14726
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.49473734435615e-07,
+      "loss": 0.0032,
+      "step": 14728
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.481844571299222e-07,
+      "loss": 0.0024,
+      "step": 14730
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.468966515441854e-07,
+      "loss": 0.0008,
+      "step": 14732
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.456103178789252e-07,
+      "loss": 0.0016,
+      "step": 14734
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.443254563344302e-07,
+      "loss": 0.0039,
+      "step": 14736
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.430420671107672e-07,
+      "loss": 0.0009,
+      "step": 14738
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.417601504077686e-07,
+      "loss": 0.0018,
+      "step": 14740
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.404797064250378e-07,
+      "loss": 0.0034,
+      "step": 14742
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.3920073536195e-07,
+      "loss": 0.0044,
+      "step": 14744
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.379232374176524e-07,
+      "loss": 0.0013,
+      "step": 14746
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.366472127910605e-07,
+      "loss": 0.0047,
+      "step": 14748
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.353726616808596e-07,
+      "loss": 0.0012,
+      "step": 14750
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.340995842855068e-07,
+      "loss": 0.0016,
+      "step": 14752
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.328279808032322e-07,
+      "loss": 0.0004,
+      "step": 14754
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.31557851432033e-07,
+      "loss": 0.0044,
+      "step": 14756
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.302891963696788e-07,
+      "loss": 0.0037,
+      "step": 14758
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.290220158137083e-07,
+      "loss": 0.0042,
+      "step": 14760
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.277563099614302e-07,
+      "loss": 0.0047,
+      "step": 14762
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.26492079009927e-07,
+      "loss": 0.0037,
+      "step": 14764
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.252293231560468e-07,
+      "loss": 0.0014,
+      "step": 14766
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.2396804259641e-07,
+      "loss": 0.0009,
+      "step": 14768
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.227082375274095e-07,
+      "loss": 0.0034,
+      "step": 14770
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.214499081452084e-07,
+      "loss": 0.0015,
+      "step": 14772
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.201930546457345e-07,
+      "loss": 0.0015,
+      "step": 14774
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.189376772246901e-07,
+      "loss": 0.005,
+      "step": 14776
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.176837760775466e-07,
+      "loss": 0.002,
+      "step": 14778
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.1643135139955e-07,
+      "loss": 0.0024,
+      "step": 14780
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.151804033857077e-07,
+      "loss": 0.002,
+      "step": 14782
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.139309322308029e-07,
+      "loss": 0.0014,
+      "step": 14784
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.126829381293896e-07,
+      "loss": 0.0009,
+      "step": 14786
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.114364212757894e-07,
+      "loss": 0.0017,
+      "step": 14788
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.101913818640958e-07,
+      "loss": 0.0018,
+      "step": 14790
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.089478200881659e-07,
+      "loss": 0.0016,
+      "step": 14792
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.077057361416371e-07,
+      "loss": 0.0064,
+      "step": 14794
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.064651302179091e-07,
+      "loss": 0.0027,
+      "step": 14796
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.052260025101541e-07,
+      "loss": 0.0016,
+      "step": 14798
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.039883532113132e-07,
+      "loss": 0.0019,
+      "step": 14800
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.027521825140991e-07,
+      "loss": 0.0047,
+      "step": 14802
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.015174906109932e-07,
+      "loss": 0.0015,
+      "step": 14804
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.00284277694244e-07,
+      "loss": 0.0022,
+      "step": 14806
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 4.990525439558735e-07,
+      "loss": 0.0046,
+      "step": 14808
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 4.978222895876727e-07,
+      "loss": 0.003,
+      "step": 14810
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 4.965935147812028e-07,
+      "loss": 0.0026,
+      "step": 14812
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 4.953662197277898e-07,
+      "loss": 0.0046,
+      "step": 14814
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 4.941404046185372e-07,
+      "loss": 0.0032,
+      "step": 14816
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 4.929160696443103e-07,
+      "loss": 0.0037,
+      "step": 14818
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 4.916932149957488e-07,
+      "loss": 0.003,
+      "step": 14820
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 4.904718408632602e-07,
+      "loss": 0.0062,
+      "step": 14822
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 4.892519474370217e-07,
+      "loss": 0.002,
+      "step": 14824
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 4.880335349069809e-07,
+      "loss": 0.0017,
+      "step": 14826
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 4.868166034628541e-07,
+      "loss": 0.0033,
+      "step": 14828
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 4.856011532941252e-07,
+      "loss": 0.0009,
+      "step": 14830
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 4.843871845900505e-07,
+      "loss": 0.0022,
+      "step": 14832
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 4.831746975396534e-07,
+      "loss": 0.0019,
+      "step": 14834
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 4.819636923317284e-07,
+      "loss": 0.0029,
+      "step": 14836
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 4.807541691548368e-07,
+      "loss": 0.0045,
+      "step": 14838
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 4.795461281973113e-07,
+      "loss": 0.0037,
+      "step": 14840
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 4.783395696472526e-07,
+      "loss": 0.0047,
+      "step": 14842
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 4.771344936925337e-07,
+      "loss": 0.0018,
+      "step": 14844
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 4.7593090052079237e-07,
+      "loss": 0.0023,
+      "step": 14846
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 4.747287903194353e-07,
+      "loss": 0.0046,
+      "step": 14848
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 4.735281632756439e-07,
+      "loss": 0.003,
+      "step": 14850
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 4.723290195763608e-07,
+      "loss": 0.0017,
+      "step": 14852
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 4.7113135940830447e-07,
+      "loss": 0.0025,
+      "step": 14854
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 4.6993518295796015e-07,
+      "loss": 0.0022,
+      "step": 14856
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 4.6874049041158107e-07,
+      "loss": 0.0011,
+      "step": 14858
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 4.675472819551907e-07,
+      "loss": 0.0024,
+      "step": 14860
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 4.663555577745782e-07,
+      "loss": 0.0028,
+      "step": 14862
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 4.6516531805530615e-07,
+      "loss": 0.0019,
+      "step": 14864
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 4.639765629827042e-07,
+      "loss": 0.0001,
+      "step": 14866
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 4.627892927418698e-07,
+      "loss": 0.002,
+      "step": 14868
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 4.6160350751766945e-07,
+      "loss": 0.0009,
+      "step": 14870
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 4.604192074947411e-07,
+      "loss": 0.0038,
+      "step": 14872
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 4.592363928574883e-07,
+      "loss": 0.003,
+      "step": 14874
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 4.580550637900827e-07,
+      "loss": 0.0014,
+      "step": 14876
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 4.5687522047646813e-07,
+      "loss": 0.0021,
+      "step": 14878
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 4.5569686310035444e-07,
+      "loss": 0.0025,
+      "step": 14880
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 4.5451999184522145e-07,
+      "loss": 0.0026,
+      "step": 14882
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 4.533446068943159e-07,
+      "loss": 0.0007,
+      "step": 14884
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 4.5217070843065593e-07,
+      "loss": 0.0011,
+      "step": 14886
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 4.509982966370252e-07,
+      "loss": 0.0023,
+      "step": 14888
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 4.498273716959789e-07,
+      "loss": 0.0034,
+      "step": 14890
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 4.486579337898356e-07,
+      "loss": 0.0016,
+      "step": 14892
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 4.474899831006885e-07,
+      "loss": 0.0015,
+      "step": 14894
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 4.4632351981039543e-07,
+      "loss": 0.002,
+      "step": 14896
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 4.451585441005857e-07,
+      "loss": 0.0034,
+      "step": 14898
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 4.439950561526507e-07,
+      "loss": 0.0018,
+      "step": 14900
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 4.4283305614775894e-07,
+      "loss": 0.0034,
+      "step": 14902
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 4.41672544266839e-07,
+      "loss": 0.0057,
+      "step": 14904
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 4.4051352069059526e-07,
+      "loss": 0.0016,
+      "step": 14906
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 4.393559855994922e-07,
+      "loss": 0.0027,
+      "step": 14908
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 4.381999391737701e-07,
+      "loss": 0.0024,
+      "step": 14910
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 4.370453815934328e-07,
+      "loss": 0.0026,
+      "step": 14912
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 4.358923130382553e-07,
+      "loss": 0.0022,
+      "step": 14914
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 4.3474073368777736e-07,
+      "loss": 0.0045,
+      "step": 14916
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 4.3359064372130886e-07,
+      "loss": 0.0027,
+      "step": 14918
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.324420433179288e-07,
+      "loss": 0.0021,
+      "step": 14920
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.312949326564819e-07,
+      "loss": 0.0023,
+      "step": 14922
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.3014931191558307e-07,
+      "loss": 0.0034,
+      "step": 14924
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.290051812736129e-07,
+      "loss": 0.0016,
+      "step": 14926
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.278625409087234e-07,
+      "loss": 0.004,
+      "step": 14928
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.2672139099882995e-07,
+      "loss": 0.004,
+      "step": 14930
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.255817317216204e-07,
+      "loss": 0.0015,
+      "step": 14932
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.244435632545463e-07,
+      "loss": 0.0019,
+      "step": 14934
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.2330688577483014e-07,
+      "loss": 0.0006,
+      "step": 14936
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.221716994594627e-07,
+      "loss": 0.0014,
+      "step": 14938
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.2103800448519914e-07,
+      "loss": 0.0019,
+      "step": 14940
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.1990580102856504e-07,
+      "loss": 0.0014,
+      "step": 14942
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.1877508926585486e-07,
+      "loss": 0.0012,
+      "step": 14944
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.176458693731278e-07,
+      "loss": 0.0028,
+      "step": 14946
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.1651814152620985e-07,
+      "loss": 0.001,
+      "step": 14948
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.1539190590069946e-07,
+      "loss": 0.0011,
+      "step": 14950
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.1426716267195853e-07,
+      "loss": 0.002,
+      "step": 14952
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.131439120151215e-07,
+      "loss": 0.0077,
+      "step": 14954
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.1202215410508284e-07,
+      "loss": 0.0021,
+      "step": 14956
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 4.1090188911651174e-07,
+      "loss": 0.0025,
+      "step": 14958
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 4.0978311722383977e-07,
+      "loss": 0.0058,
+      "step": 14960
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 4.0866583860127095e-07,
+      "loss": 0.003,
+      "step": 14962
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 4.0755005342277167e-07,
+      "loss": 0.001,
+      "step": 14964
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 4.0643576186207954e-07,
+      "loss": 0.0015,
+      "step": 14966
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 4.05322964092697e-07,
+      "loss": 0.0005,
+      "step": 14968
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 4.042116602878976e-07,
+      "loss": 0.0013,
+      "step": 14970
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 4.031018506207185e-07,
+      "loss": 0.0034,
+      "step": 14972
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 4.0199353526396477e-07,
+      "loss": 0.0017,
+      "step": 14974
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 4.0088671439020953e-07,
+      "loss": 0.0035,
+      "step": 14976
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.997813881717949e-07,
+      "loss": 0.004,
+      "step": 14978
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.986775567808265e-07,
+      "loss": 0.0045,
+      "step": 14980
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.9757522038918137e-07,
+      "loss": 0.0031,
+      "step": 14982
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.9647437916849995e-07,
+      "loss": 0.002,
+      "step": 14984
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.95375033290194e-07,
+      "loss": 0.0018,
+      "step": 14986
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.942771829254388e-07,
+      "loss": 0.0041,
+      "step": 14988
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.9318082824517656e-07,
+      "loss": 0.0027,
+      "step": 14990
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.9208596942011956e-07,
+      "loss": 0.0009,
+      "step": 14992
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.909926066207459e-07,
+      "loss": 0.0049,
+      "step": 14994
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.899007400172994e-07,
+      "loss": 0.0017,
+      "step": 14996
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.888103697797929e-07,
+      "loss": 0.0012,
+      "step": 14998
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.8772149607800624e-07,
+      "loss": 0.0025,
+      "step": 15000
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.8663411908148375e-07,
+      "loss": 0.0029,
+      "step": 15002
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.8554823895953885e-07,
+      "loss": 0.0037,
+      "step": 15004
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.844638558812508e-07,
+      "loss": 0.0008,
+      "step": 15006
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.8338097001546783e-07,
+      "loss": 0.0028,
+      "step": 15008
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.822995815308028e-07,
+      "loss": 0.0008,
+      "step": 15010
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.812196905956356e-07,
+      "loss": 0.0014,
+      "step": 15012
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.80141297378116e-07,
+      "loss": 0.0027,
+      "step": 15014
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.7906440204615423e-07,
+      "loss": 0.0056,
+      "step": 15016
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.779890047674339e-07,
+      "loss": 0.0049,
+      "step": 15018
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.7691510570940115e-07,
+      "loss": 0.0026,
+      "step": 15020
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.758427050392699e-07,
+      "loss": 0.0036,
+      "step": 15022
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.7477180292402214e-07,
+      "loss": 0.0004,
+      "step": 15024
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.737023995304079e-07,
+      "loss": 0.0018,
+      "step": 15026
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.726344950249372e-07,
+      "loss": 0.0021,
+      "step": 15028
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.7156808957389266e-07,
+      "loss": 0.0034,
+      "step": 15030
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.7050318334332145e-07,
+      "loss": 0.0025,
+      "step": 15032
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.694397764990398e-07,
+      "loss": 0.0017,
+      "step": 15034
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.6837786920662534e-07,
+      "loss": 0.0035,
+      "step": 15036
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.6731746163142567e-07,
+      "loss": 0.0026,
+      "step": 15038
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.6625855393855437e-07,
+      "loss": 0.0004,
+      "step": 15040
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.6520114629289504e-07,
+      "loss": 0.0046,
+      "step": 15042
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.641452388590894e-07,
+      "loss": 0.0025,
+      "step": 15044
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.6309083180155247e-07,
+      "loss": 0.0025,
+      "step": 15046
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.620379252844619e-07,
+      "loss": 0.0019,
+      "step": 15048
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.6098651947176657e-07,
+      "loss": 0.0007,
+      "step": 15050
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.5993661452717433e-07,
+      "loss": 0.0011,
+      "step": 15052
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.5888821061416556e-07,
+      "loss": 0.0026,
+      "step": 15054
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.578413078959864e-07,
+      "loss": 0.0027,
+      "step": 15056
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.567959065356452e-07,
+      "loss": 0.0018,
+      "step": 15058
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.557520066959186e-07,
+      "loss": 0.0034,
+      "step": 15060
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.5470960853935086e-07,
+      "loss": 0.001,
+      "step": 15062
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.536687122282512e-07,
+      "loss": 0.0024,
+      "step": 15064
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.5262931792469646e-07,
+      "loss": 0.0022,
+      "step": 15066
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.515914257905262e-07,
+      "loss": 0.0023,
+      "step": 15068
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.5055503598734996e-07,
+      "loss": 0.0026,
+      "step": 15070
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.495201486765387e-07,
+      "loss": 0.004,
+      "step": 15072
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.484867640192358e-07,
+      "loss": 0.0018,
+      "step": 15074
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.474548821763446e-07,
+      "loss": 0.0038,
+      "step": 15076
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.464245033085367e-07,
+      "loss": 0.003,
+      "step": 15078
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.453956275762527e-07,
+      "loss": 0.0051,
+      "step": 15080
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.443682551396954e-07,
+      "loss": 0.0029,
+      "step": 15082
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.433423861588325e-07,
+      "loss": 0.0012,
+      "step": 15084
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.423180207934029e-07,
+      "loss": 0.0013,
+      "step": 15086
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4129515920290455e-07,
+      "loss": 0.0026,
+      "step": 15088
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4027380154660893e-07,
+      "loss": 0.002,
+      "step": 15090
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.3925394798354437e-07,
+      "loss": 0.0028,
+      "step": 15092
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.382355986725139e-07,
+      "loss": 0.0043,
+      "step": 15094
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.3721875377208056e-07,
+      "loss": 0.003,
+      "step": 15096
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.362034134405756e-07,
+      "loss": 0.0016,
+      "step": 15098
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.3518957783609476e-07,
+      "loss": 0.0019,
+      "step": 15100
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.3417724711649944e-07,
+      "loss": 0.0016,
+      "step": 15102
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.3316642143941814e-07,
+      "loss": 0.0032,
+      "step": 15104
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.3215710096224483e-07,
+      "loss": 0.0046,
+      "step": 15106
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.3114928584213614e-07,
+      "loss": 0.0008,
+      "step": 15108
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.3014297623601865e-07,
+      "loss": 0.0005,
+      "step": 15110
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.2913817230058265e-07,
+      "loss": 0.0021,
+      "step": 15112
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.2813487419228295e-07,
+      "loss": 0.0018,
+      "step": 15114
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.2713308206733907e-07,
+      "loss": 0.0031,
+      "step": 15116
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.2613279608173953e-07,
+      "loss": 0.0012,
+      "step": 15118
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.2513401639123643e-07,
+      "loss": 0.0028,
+      "step": 15120
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.241367431513487e-07,
+      "loss": 0.0021,
+      "step": 15122
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.2314097651735657e-07,
+      "loss": 0.0026,
+      "step": 15124
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.221467166443115e-07,
+      "loss": 0.0011,
+      "step": 15126
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.211539636870242e-07,
+      "loss": 0.0025,
+      "step": 15128
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.2016271780007766e-07,
+      "loss": 0.0038,
+      "step": 15130
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.1917297913781176e-07,
+      "loss": 0.0001,
+      "step": 15132
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.1818474785433985e-07,
+      "loss": 0.0046,
+      "step": 15134
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.171980241035355e-07,
+      "loss": 0.0042,
+      "step": 15136
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.162128080390414e-07,
+      "loss": 0.0014,
+      "step": 15138
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.152290998142604e-07,
+      "loss": 0.0024,
+      "step": 15140
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.1424689958236556e-07,
+      "loss": 0.0024,
+      "step": 15142
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.132662074962911e-07,
+      "loss": 0.0007,
+      "step": 15144
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.122870237087405e-07,
+      "loss": 0.0037,
+      "step": 15146
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.113093483721774e-07,
+      "loss": 0.0016,
+      "step": 15148
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.1033318163883553e-07,
+      "loss": 0.0036,
+      "step": 15150
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.0935852366070995e-07,
+      "loss": 0.0018,
+      "step": 15152
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.0838537458956487e-07,
+      "loss": 0.0032,
+      "step": 15154
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.074137345769257e-07,
+      "loss": 0.004,
+      "step": 15156
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.064436037740814e-07,
+      "loss": 0.0039,
+      "step": 15158
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.054749823320924e-07,
+      "loss": 0.0009,
+      "step": 15160
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.045078704017801e-07,
+      "loss": 0.0027,
+      "step": 15162
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.0354226813372967e-07,
+      "loss": 0.0024,
+      "step": 15164
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.025781756782931e-07,
+      "loss": 0.0009,
+      "step": 15166
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.0161559318558796e-07,
+      "loss": 0.0025,
+      "step": 15168
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.0065452080549564e-07,
+      "loss": 0.0012,
+      "step": 15170
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2.996949586876607e-07,
+      "loss": 0.0029,
+      "step": 15172
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2.98736906981496e-07,
+      "loss": 0.0028,
+      "step": 15174
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2.9778036583617664e-07,
+      "loss": 0.004,
+      "step": 15176
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2.9682533540064564e-07,
+      "loss": 0.0019,
+      "step": 15178
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2.958718158236051e-07,
+      "loss": 0.0023,
+      "step": 15180
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2.9491980725352753e-07,
+      "loss": 0.0034,
+      "step": 15182
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2.939693098386465e-07,
+      "loss": 0.0015,
+      "step": 15184
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2.9302032372696356e-07,
+      "loss": 0.0026,
+      "step": 15186
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2.920728490662417e-07,
+      "loss": 0.0061,
+      "step": 15188
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2.911268860040095e-07,
+      "loss": 0.0012,
+      "step": 15190
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2.901824346875626e-07,
+      "loss": 0.0055,
+      "step": 15192
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2.892394952639588e-07,
+      "loss": 0.0004,
+      "step": 15194
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2.8829806788001846e-07,
+      "loss": 0.0033,
+      "step": 15196
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2.8735815268233323e-07,
+      "loss": 0.0029,
+      "step": 15198
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2.864197498172516e-07,
+      "loss": 0.0042,
+      "step": 15200
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2.8548285943089226e-07,
+      "loss": 0.0037,
+      "step": 15202
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 2.845474816691329e-07,
+      "loss": 0.003,
+      "step": 15204
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2.836136166776227e-07,
+      "loss": 0.0044,
+      "step": 15206
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2.826812646017696e-07,
+      "loss": 0.0014,
+      "step": 15208
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2.8175042558675094e-07,
+      "loss": 0.004,
+      "step": 15210
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2.808210997775018e-07,
+      "loss": 0.0014,
+      "step": 15212
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2.7989328731872543e-07,
+      "loss": 0.002,
+      "step": 15214
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2.7896698835489065e-07,
+      "loss": 0.0053,
+      "step": 15216
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2.78042203030231e-07,
+      "loss": 0.0015,
+      "step": 15218
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2.7711893148873904e-07,
+      "loss": 0.0025,
+      "step": 15220
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2.7619717387417645e-07,
+      "loss": 0.0013,
+      "step": 15222
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2.7527693033007063e-07,
+      "loss": 0.0002,
+      "step": 15224
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2.74358200999707e-07,
+      "loss": 0.0009,
+      "step": 15226
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2.7344098602614e-07,
+      "loss": 0.0017,
+      "step": 15228
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2.725252855521865e-07,
+      "loss": 0.0033,
+      "step": 15230
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2.716110997204291e-07,
+      "loss": 0.005,
+      "step": 15232
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2.706984286732139e-07,
+      "loss": 0.003,
+      "step": 15234
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2.697872725526496e-07,
+      "loss": 0.003,
+      "step": 15236
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2.6887763150060917e-07,
+      "loss": 0.0028,
+      "step": 15238
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2.679695056587339e-07,
+      "loss": 0.0028,
+      "step": 15240
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2.67062895168424e-07,
+      "loss": 0.0025,
+      "step": 15242
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2.661578001708442e-07,
+      "loss": 0.0015,
+      "step": 15244
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 2.6525422080692644e-07,
+      "loss": 0.0043,
+      "step": 15246
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2.6435215721736575e-07,
+      "loss": 0.0012,
+      "step": 15248
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2.6345160954261874e-07,
+      "loss": 0.0024,
+      "step": 15250
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2.625525779229077e-07,
+      "loss": 0.0019,
+      "step": 15252
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2.616550624982206e-07,
+      "loss": 0.0028,
+      "step": 15254
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2.607590634083046e-07,
+      "loss": 0.0037,
+      "step": 15256
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2.5986458079267587e-07,
+      "loss": 0.0015,
+      "step": 15258
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2.5897161479061073e-07,
+      "loss": 0.0037,
+      "step": 15260
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2.5808016554115136e-07,
+      "loss": 0.0013,
+      "step": 15262
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2.571902331831033e-07,
+      "loss": 0.0022,
+      "step": 15264
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2.5630181785503583e-07,
+      "loss": 0.0026,
+      "step": 15266
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2.5541491969528264e-07,
+      "loss": 0.0045,
+      "step": 15268
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2.5452953884193996e-07,
+      "loss": 0.0026,
+      "step": 15270
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2.536456754328664e-07,
+      "loss": 0.0007,
+      "step": 15272
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2.527633296056908e-07,
+      "loss": 0.0029,
+      "step": 15274
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2.518825014977966e-07,
+      "loss": 0.0011,
+      "step": 15276
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2.5100319124633734e-07,
+      "loss": 0.0035,
+      "step": 15278
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2.501253989882302e-07,
+      "loss": 0.0026,
+      "step": 15280
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2.492491248601503e-07,
+      "loss": 0.0031,
+      "step": 15282
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2.4837436899854407e-07,
+      "loss": 0.0021,
+      "step": 15284
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 2.4750113153961477e-07,
+      "loss": 0.0016,
+      "step": 15286
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.466294126193325e-07,
+      "loss": 0.0021,
+      "step": 15288
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.4575921237343316e-07,
+      "loss": 0.0017,
+      "step": 15290
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.4489053093741055e-07,
+      "loss": 0.0028,
+      "step": 15292
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.440233684465254e-07,
+      "loss": 0.0027,
+      "step": 15294
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.4315772503580416e-07,
+      "loss": 0.0015,
+      "step": 15296
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.422936008400323e-07,
+      "loss": 0.0029,
+      "step": 15298
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.414309959937589e-07,
+      "loss": 0.0027,
+      "step": 15300
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.4056991063130084e-07,
+      "loss": 0.0048,
+      "step": 15302
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.397103448867344e-07,
+      "loss": 0.0012,
+      "step": 15304
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.3885229889390126e-07,
+      "loss": 0.0019,
+      "step": 15306
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.3799577278640463e-07,
+      "loss": 0.0007,
+      "step": 15308
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.3714076669761333e-07,
+      "loss": 0.0008,
+      "step": 15310
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.3628728076065754e-07,
+      "loss": 0.0012,
+      "step": 15312
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.354353151084321e-07,
+      "loss": 0.0033,
+      "step": 15314
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.345848698735942e-07,
+      "loss": 0.0013,
+      "step": 15316
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.3373594518856458e-07,
+      "loss": 0.0012,
+      "step": 15318
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.3288854118552639e-07,
+      "loss": 0.0019,
+      "step": 15320
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.3204265799643077e-07,
+      "loss": 0.0038,
+      "step": 15322
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.311982957529846e-07,
+      "loss": 0.0029,
+      "step": 15324
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.3035545458666154e-07,
+      "loss": 0.0051,
+      "step": 15326
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 2.2951413462869886e-07,
+      "loss": 0.0038,
+      "step": 15328
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2.286743360100985e-07,
+      "loss": 0.0027,
+      "step": 15330
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2.2783605886162018e-07,
+      "loss": 0.0033,
+      "step": 15332
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2.2699930331379182e-07,
+      "loss": 0.0017,
+      "step": 15334
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2.2616406949690362e-07,
+      "loss": 0.0017,
+      "step": 15336
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2.2533035754100708e-07,
+      "loss": 0.001,
+      "step": 15338
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2.2449816757591835e-07,
+      "loss": 0.0048,
+      "step": 15340
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2.236674997312127e-07,
+      "loss": 0.0025,
+      "step": 15342
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2.2283835413623444e-07,
+      "loss": 0.0007,
+      "step": 15344
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2.2201073092008696e-07,
+      "loss": 0.0021,
+      "step": 15346
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2.2118463021163715e-07,
+      "loss": 0.0062,
+      "step": 15348
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2.2036005213951662e-07,
+      "loss": 0.0038,
+      "step": 15350
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2.1953699683211704e-07,
+      "loss": 0.0028,
+      "step": 15352
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2.1871546441759484e-07,
+      "loss": 0.0022,
+      "step": 15354
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2.1789545502386877e-07,
+      "loss": 0.0025,
+      "step": 15356
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2.1707696877862005e-07,
+      "loss": 0.001,
+      "step": 15358
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2.162600058092945e-07,
+      "loss": 0.0004,
+      "step": 15360
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2.1544456624309927e-07,
+      "loss": 0.002,
+      "step": 15362
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2.146306502070039e-07,
+      "loss": 0.0005,
+      "step": 15364
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2.1381825782774145e-07,
+      "loss": 0.0017,
+      "step": 15366
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 2.130073892318074e-07,
+      "loss": 0.0014,
+      "step": 15368
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2.1219804454546188e-07,
+      "loss": 0.0021,
+      "step": 15370
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2.1139022389472297e-07,
+      "loss": 0.0017,
+      "step": 15372
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2.105839274053767e-07,
+      "loss": 0.0033,
+      "step": 15374
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2.0977915520297042e-07,
+      "loss": 0.001,
+      "step": 15376
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2.089759074128117e-07,
+      "loss": 0.0022,
+      "step": 15378
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2.081741841599727e-07,
+      "loss": 0.0046,
+      "step": 15380
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2.0737398556928689e-07,
+      "loss": 0.0031,
+      "step": 15382
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2.065753117653535e-07,
+      "loss": 0.0041,
+      "step": 15384
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2.0577816287253082e-07,
+      "loss": 0.0013,
+      "step": 15386
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2.049825390149396e-07,
+      "loss": 0.0039,
+      "step": 15388
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2.0418844031646735e-07,
+      "loss": 0.0016,
+      "step": 15390
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2.0339586690076074e-07,
+      "loss": 0.0018,
+      "step": 15392
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2.0260481889122775e-07,
+      "loss": 0.0023,
+      "step": 15394
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2.0181529641104315e-07,
+      "loss": 0.0026,
+      "step": 15396
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2.010272995831386e-07,
+      "loss": 0.0019,
+      "step": 15398
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 2.0024082853021487e-07,
+      "loss": 0.0012,
+      "step": 15400
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.9945588337472733e-07,
+      "loss": 0.0046,
+      "step": 15402
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.986724642389004e-07,
+      "loss": 0.0037,
+      "step": 15404
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.9789057124471876e-07,
+      "loss": 0.0014,
+      "step": 15406
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.9711020451392837e-07,
+      "loss": 0.0023,
+      "step": 15408
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.9633136416803867e-07,
+      "loss": 0.0034,
+      "step": 15410
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.9555405032832043e-07,
+      "loss": 0.0009,
+      "step": 15412
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.9477826311580793e-07,
+      "loss": 0.0004,
+      "step": 15414
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.9400400265129682e-07,
+      "loss": 0.0018,
+      "step": 15416
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.9323126905534618e-07,
+      "loss": 0.0031,
+      "step": 15418
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.924600624482742e-07,
+      "loss": 0.0018,
+      "step": 15420
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.9169038295016707e-07,
+      "loss": 0.0026,
+      "step": 15422
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.9092223068086891e-07,
+      "loss": 0.0022,
+      "step": 15424
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.9015560575998403e-07,
+      "loss": 0.001,
+      "step": 15426
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.8939050830688366e-07,
+      "loss": 0.0017,
+      "step": 15428
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.8862693844070023e-07,
+      "loss": 0.0011,
+      "step": 15430
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.8786489628032644e-07,
+      "loss": 0.0014,
+      "step": 15432
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.8710438194441848e-07,
+      "loss": 0.0022,
+      "step": 15434
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.8634539555139385e-07,
+      "loss": 0.0028,
+      "step": 15436
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.8558793721943248e-07,
+      "loss": 0.0011,
+      "step": 15438
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.8483200706647774e-07,
+      "loss": 0.0033,
+      "step": 15440
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.840776052102322e-07,
+      "loss": 0.0009,
+      "step": 15442
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.833247317681619e-07,
+      "loss": 0.0039,
+      "step": 15444
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.8257338685749526e-07,
+      "loss": 0.0014,
+      "step": 15446
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.8182357059522427e-07,
+      "loss": 0.0016,
+      "step": 15448
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.8107528309809774e-07,
+      "loss": 0.0042,
+      "step": 15450
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.8032852448263248e-07,
+      "loss": 0.0042,
+      "step": 15452
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.795832948651033e-07,
+      "loss": 0.0018,
+      "step": 15454
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.7883959436154952e-07,
+      "loss": 0.0017,
+      "step": 15456
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.7809742308776856e-07,
+      "loss": 0.0027,
+      "step": 15458
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.7735678115932353e-07,
+      "loss": 0.0015,
+      "step": 15460
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.7661766869153772e-07,
+      "loss": 0.0005,
+      "step": 15462
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.75880085799498e-07,
+      "loss": 0.0011,
+      "step": 15464
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.7514403259805135e-07,
+      "loss": 0.0027,
+      "step": 15466
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.7440950920180388e-07,
+      "loss": 0.0037,
+      "step": 15468
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.7367651572512967e-07,
+      "loss": 0.003,
+      "step": 15470
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.7294505228216075e-07,
+      "loss": 0.0019,
+      "step": 15472
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.7221511898679043e-07,
+      "loss": 0.0007,
+      "step": 15474
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.7148671595267564e-07,
+      "loss": 0.0031,
+      "step": 15476
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.707598432932356e-07,
+      "loss": 0.0016,
+      "step": 15478
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.7003450112164866e-07,
+      "loss": 0.0022,
+      "step": 15480
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.693106895508556e-07,
+      "loss": 0.0018,
+      "step": 15482
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.6858840869355963e-07,
+      "loss": 0.0034,
+      "step": 15484
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.6786765866222522e-07,
+      "loss": 0.0047,
+      "step": 15486
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.6714843956908145e-07,
+      "loss": 0.0012,
+      "step": 15488
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.664307515261121e-07,
+      "loss": 0.0024,
+      "step": 15490
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.6571459464506888e-07,
+      "loss": 0.0024,
+      "step": 15492
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.6499996903746374e-07,
+      "loss": 0.0003,
+      "step": 15494
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.6428687481456762e-07,
+      "loss": 0.0031,
+      "step": 15496
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.6357531208741507e-07,
+      "loss": 0.0025,
+      "step": 15498
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.62865280966803e-07,
+      "loss": 0.003,
+      "step": 15500
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.621567815632863e-07,
+      "loss": 0.0018,
+      "step": 15502
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.6144981398718674e-07,
+      "loss": 0.002,
+      "step": 15504
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.6074437834858293e-07,
+      "loss": 0.0009,
+      "step": 15506
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.6004047475731587e-07,
+      "loss": 0.0021,
+      "step": 15508
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.5933810332299015e-07,
+      "loss": 0.003,
+      "step": 15510
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.586372641549705e-07,
+      "loss": 0.0018,
+      "step": 15512
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.5793795736238182e-07,
+      "loss": 0.003,
+      "step": 15514
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.5724018305411148e-07,
+      "loss": 0.0012,
+      "step": 15516
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.5654394133880812e-07,
+      "loss": 0.0024,
+      "step": 15518
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.5584923232488502e-07,
+      "loss": 0.0027,
+      "step": 15520
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.551560561205101e-07,
+      "loss": 0.0062,
+      "step": 15522
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.544644128336159e-07,
+      "loss": 0.0016,
+      "step": 15524
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.5377430257189962e-07,
+      "loss": 0.0034,
+      "step": 15526
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.530857254428153e-07,
+      "loss": 0.0034,
+      "step": 15528
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.523986815535783e-07,
+      "loss": 0.0028,
+      "step": 15530
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.5171317101116746e-07,
+      "loss": 0.0035,
+      "step": 15532
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.5102919392232407e-07,
+      "loss": 0.0012,
+      "step": 15534
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.5034675039354517e-07,
+      "loss": 0.0018,
+      "step": 15536
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.4966584053109579e-07,
+      "loss": 0.002,
+      "step": 15538
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.4898646444099553e-07,
+      "loss": 0.0032,
+      "step": 15540
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.4830862222902975e-07,
+      "loss": 0.0014,
+      "step": 15542
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.4763231400074519e-07,
+      "loss": 0.0016,
+      "step": 15544
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.4695753986144646e-07,
+      "loss": 0.0011,
+      "step": 15546
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.4628429991620175e-07,
+      "loss": 0.0007,
+      "step": 15548
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.4561259426983942e-07,
+      "loss": 0.0013,
+      "step": 15550
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.4494242302695029e-07,
+      "loss": 0.0008,
+      "step": 15552
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.4427378629188306e-07,
+      "loss": 0.0032,
+      "step": 15554
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.4360668416875002e-07,
+      "loss": 0.0028,
+      "step": 15556
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.4294111676142474e-07,
+      "loss": 0.0045,
+      "step": 15558
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.4227708417354213e-07,
+      "loss": 0.0025,
+      "step": 15560
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.4161458650849502e-07,
+      "loss": 0.0014,
+      "step": 15562
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.409536238694409e-07,
+      "loss": 0.0007,
+      "step": 15564
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.402941963592963e-07,
+      "loss": 0.0012,
+      "step": 15566
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.3963630408073914e-07,
+      "loss": 0.0016,
+      "step": 15568
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.3897994713620855e-07,
+      "loss": 0.0009,
+      "step": 15570
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.383251256279028e-07,
+      "loss": 0.0021,
+      "step": 15572
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.3767183965778365e-07,
+      "loss": 0.0049,
+      "step": 15574
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.3702008932757416e-07,
+      "loss": 0.0031,
+      "step": 15576
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.363698747387554e-07,
+      "loss": 0.0005,
+      "step": 15578
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.3572119599257083e-07,
+      "loss": 0.0006,
+      "step": 15580
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.350740531900241e-07,
+      "loss": 0.0013,
+      "step": 15582
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.3442844643188124e-07,
+      "loss": 0.0029,
+      "step": 15584
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.3378437581866855e-07,
+      "loss": 0.0029,
+      "step": 15586
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.3314184145067023e-07,
+      "loss": 0.0015,
+      "step": 15588
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.3250084342793734e-07,
+      "loss": 0.0008,
+      "step": 15590
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.3186138185027563e-07,
+      "loss": 0.0021,
+      "step": 15592
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.3122345681725545e-07,
+      "loss": 0.0014,
+      "step": 15594
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.3058706842820624e-07,
+      "loss": 0.0017,
+      "step": 15596
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.2995221678221758e-07,
+      "loss": 0.0022,
+      "step": 15598
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.2931890197814267e-07,
+      "loss": 0.0011,
+      "step": 15600
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.286871241145915e-07,
+      "loss": 0.0034,
+      "step": 15602
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.280568832899376e-07,
+      "loss": 0.0027,
+      "step": 15604
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.274281796023158e-07,
+      "loss": 0.0021,
+      "step": 15606
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.268010131496178e-07,
+      "loss": 0.0011,
+      "step": 15608
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.261753840294977e-07,
+      "loss": 0.0011,
+      "step": 15610
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.2555129233937313e-07,
+      "loss": 0.0014,
+      "step": 15612
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.249287381764186e-07,
+      "loss": 0.0029,
+      "step": 15614
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.24307721637571e-07,
+      "loss": 0.0016,
+      "step": 15616
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.2368824281952629e-07,
+      "loss": 0.0015,
+      "step": 15618
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.2307030181874402e-07,
+      "loss": 0.0019,
+      "step": 15620
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.2245389873144052e-07,
+      "loss": 0.0005,
+      "step": 15622
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.2183903365359574e-07,
+      "loss": 0.0019,
+      "step": 15624
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.2122570668094746e-07,
+      "loss": 0.0022,
+      "step": 15626
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.2061391790899602e-07,
+      "loss": 0.001,
+      "step": 15628
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.2000366743300184e-07,
+      "loss": 0.0057,
+      "step": 15630
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.1939495534798672e-07,
+      "loss": 0.005,
+      "step": 15632
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.1878778174872929e-07,
+      "loss": 0.0065,
+      "step": 15634
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.181821467297728e-07,
+      "loss": 0.0006,
+      "step": 15636
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.1757805038541958e-07,
+      "loss": 0.0013,
+      "step": 15638
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.1697549280973108e-07,
+      "loss": 0.0018,
+      "step": 15640
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.1637447409653002e-07,
+      "loss": 0.0024,
+      "step": 15642
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.1577499433939932e-07,
+      "loss": 0.0136,
+      "step": 15644
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.151770536316843e-07,
+      "loss": 0.0039,
+      "step": 15646
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.1458065206648717e-07,
+      "loss": 0.0037,
+      "step": 15648
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.1398578973667362e-07,
+      "loss": 0.0018,
+      "step": 15650
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.1339246673486737e-07,
+      "loss": 0.0013,
+      "step": 15652
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.1280068315345338e-07,
+      "loss": 0.0047,
+      "step": 15654
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.1221043908457685e-07,
+      "loss": 0.0013,
+      "step": 15656
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.1162173462014313e-07,
+      "loss": 0.0044,
+      "step": 15658
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.1103456985181893e-07,
+      "loss": 0.0008,
+      "step": 15660
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.1044894487102997e-07,
+      "loss": 0.0013,
+      "step": 15662
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.098648597689611e-07,
+      "loss": 0.0038,
+      "step": 15664
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.0928231463656069e-07,
+      "loss": 0.0025,
+      "step": 15666
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.0870130956453396e-07,
+      "loss": 0.0016,
+      "step": 15668
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.0812184464334963e-07,
+      "loss": 0.0016,
+      "step": 15670
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.0754391996323332e-07,
+      "loss": 0.0009,
+      "step": 15672
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.0696753561417306e-07,
+      "loss": 0.0024,
+      "step": 15674
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.0639269168591703e-07,
+      "loss": 0.002,
+      "step": 15676
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.0581938826797034e-07,
+      "loss": 0.0021,
+      "step": 15678
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.0524762544960265e-07,
+      "loss": 0.0011,
+      "step": 15680
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.0467740331984166e-07,
+      "loss": 0.0019,
+      "step": 15682
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.0410872196747524e-07,
+      "loss": 0.0012,
+      "step": 15684
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.0354158148105032e-07,
+      "loss": 0.0014,
+      "step": 15686
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.0297598194887736e-07,
+      "loss": 0.0012,
+      "step": 15688
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.0241192345902152e-07,
+      "loss": 0.0019,
+      "step": 15690
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.018494060993136e-07,
+      "loss": 0.0013,
+      "step": 15692
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.0128842995734023e-07,
+      "loss": 0.0017,
+      "step": 15694
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.0072899512045154e-07,
+      "loss": 0.0009,
+      "step": 15696
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.001711016757534e-07,
+      "loss": 0.0012,
+      "step": 15698
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 9.96147497101152e-08,
+      "loss": 0.0034,
+      "step": 15700
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 9.905993931016544e-08,
+      "loss": 0.0014,
+      "step": 15702
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 9.85066705622928e-08,
+      "loss": 0.0021,
+      "step": 15704
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 9.795494355264502e-08,
+      "loss": 0.0026,
+      "step": 15706
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 9.740475836713003e-08,
+      "loss": 0.0015,
+      "step": 15708
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 9.68561150914149e-08,
+      "loss": 0.0042,
+      "step": 15710
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 9.630901381092905e-08,
+      "loss": 0.0011,
+      "step": 15712
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 9.57634546108599e-08,
+      "loss": 0.007,
+      "step": 15714
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 9.521943757615393e-08,
+      "loss": 0.0008,
+      "step": 15716
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 9.467696279152116e-08,
+      "loss": 0.0025,
+      "step": 15718
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 9.413603034142849e-08,
+      "loss": 0.0027,
+      "step": 15720
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 9.359664031010185e-08,
+      "loss": 0.0046,
+      "step": 15722
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 9.305879278152851e-08,
+      "loss": 0.0028,
+      "step": 15724
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 9.252248783945594e-08,
+      "loss": 0.0006,
+      "step": 15726
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 9.198772556739177e-08,
+      "loss": 0.0041,
+      "step": 15728
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 9.145450604860163e-08,
+      "loss": 0.0022,
+      "step": 15730
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 9.09228293661124e-08,
+      "loss": 0.0023,
+      "step": 15732
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 9.039269560271013e-08,
+      "loss": 0.0021,
+      "step": 15734
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 8.986410484093988e-08,
+      "loss": 0.0048,
+      "step": 15736
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 8.933705716310804e-08,
+      "loss": 0.0042,
+      "step": 15738
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 8.881155265128005e-08,
+      "loss": 0.0018,
+      "step": 15740
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 8.828759138727939e-08,
+      "loss": 0.0017,
+      "step": 15742
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 8.776517345269408e-08,
+      "loss": 0.0018,
+      "step": 15744
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 8.724429892886577e-08,
+      "loss": 0.002,
+      "step": 15746
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 8.672496789689955e-08,
+      "loss": 0.0029,
+      "step": 15748
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 8.620718043765853e-08,
+      "loss": 0.0038,
+      "step": 15750
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 8.56909366317682e-08,
+      "loss": 0.0009,
+      "step": 15752
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 8.517623655960872e-08,
+      "loss": 0.0007,
+      "step": 15754
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 8.466308030132597e-08,
+      "loss": 0.0024,
+      "step": 15756
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 8.415146793681939e-08,
+      "loss": 0.0026,
+      "step": 15758
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 8.364139954575301e-08,
+      "loss": 0.0042,
+      "step": 15760
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 8.313287520754776e-08,
+      "loss": 0.0029,
+      "step": 15762
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 8.262589500138473e-08,
+      "loss": 0.0017,
+      "step": 15764
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 8.212045900620414e-08,
+      "loss": 0.0027,
+      "step": 15766
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 8.161656730070744e-08,
+      "loss": 0.0023,
+      "step": 15768
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 8.1114219963353e-08,
+      "loss": 0.0019,
+      "step": 15770
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 8.061341707236048e-08,
+      "loss": 0.003,
+      "step": 15772
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 8.011415870570971e-08,
+      "loss": 0.0031,
+      "step": 15774
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 7.961644494113741e-08,
+      "loss": 0.0038,
+      "step": 15776
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 7.912027585614268e-08,
+      "loss": 0.0025,
+      "step": 15778
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 7.862565152798263e-08,
+      "loss": 0.0019,
+      "step": 15780
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 7.813257203367452e-08,
+      "loss": 0.0038,
+      "step": 15782
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 7.764103744999363e-08,
+      "loss": 0.0009,
+      "step": 15784
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 7.715104785347539e-08,
+      "loss": 0.0025,
+      "step": 15786
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 7.666260332041653e-08,
+      "loss": 0.0027,
+      "step": 15788
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 7.617570392686958e-08,
+      "loss": 0.0026,
+      "step": 15790
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 7.569034974865053e-08,
+      "loss": 0.0016,
+      "step": 15792
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 7.520654086133117e-08,
+      "loss": 0.0019,
+      "step": 15794
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 7.472427734024567e-08,
+      "loss": 0.0043,
+      "step": 15796
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 7.4243559260484e-08,
+      "loss": 0.0025,
+      "step": 15798
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 7.376438669690067e-08,
+      "loss": 0.0005,
+      "step": 15800
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 7.328675972410271e-08,
+      "loss": 0.0061,
+      "step": 15802
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 7.281067841646394e-08,
+      "loss": 0.0043,
+      "step": 15804
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 7.233614284811174e-08,
+      "loss": 0.0011,
+      "step": 15806
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 7.18631530929359e-08,
+      "loss": 0.0032,
+      "step": 15808
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 7.139170922458417e-08,
+      "loss": 0.0009,
+      "step": 15810
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 7.092181131646336e-08,
+      "loss": 0.0016,
+      "step": 15812
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 7.045345944174053e-08,
+      "loss": 0.0038,
+      "step": 15814
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 6.99866536733429e-08,
+      "loss": 0.001,
+      "step": 15816
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 6.952139408395342e-08,
+      "loss": 0.0006,
+      "step": 15818
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 6.905768074601859e-08,
+      "loss": 0.0015,
+      "step": 15820
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.859551373174067e-08,
+      "loss": 0.0017,
+      "step": 15822
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.81348931130843e-08,
+      "loss": 0.0032,
+      "step": 15824
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.767581896176989e-08,
+      "loss": 0.0047,
+      "step": 15826
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.721829134927915e-08,
+      "loss": 0.0038,
+      "step": 15828
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.676231034685398e-08,
+      "loss": 0.0034,
+      "step": 15830
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.630787602549204e-08,
+      "loss": 0.0009,
+      "step": 15832
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.58549884559545e-08,
+      "loss": 0.0031,
+      "step": 15834
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.540364770875717e-08,
+      "loss": 0.0015,
+      "step": 15836
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.495385385417829e-08,
+      "loss": 0.0017,
+      "step": 15838
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.450560696225405e-08,
+      "loss": 0.0032,
+      "step": 15840
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.405890710278084e-08,
+      "loss": 0.002,
+      "step": 15842
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.361375434531192e-08,
+      "loss": 0.0027,
+      "step": 15844
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.317014875916183e-08,
+      "loss": 0.0002,
+      "step": 15846
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.272809041340311e-08,
+      "loss": 0.0004,
+      "step": 15848
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.228757937686735e-08,
+      "loss": 0.0014,
+      "step": 15850
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.184861571814527e-08,
+      "loss": 0.0018,
+      "step": 15852
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.141119950558772e-08,
+      "loss": 0.0031,
+      "step": 15854
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.097533080730466e-08,
+      "loss": 0.0019,
+      "step": 15856
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.054100969116072e-08,
+      "loss": 0.0014,
+      "step": 15858
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 6.010823622478623e-08,
+      "loss": 0.0013,
+      "step": 15860
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 5.967701047556618e-08,
+      "loss": 0.0018,
+      "step": 15862
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 5.924733251064574e-08,
+      "loss": 0.0015,
+      "step": 15864
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 5.881920239692918e-08,
+      "loss": 0.002,
+      "step": 15866
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 5.839262020107872e-08,
+      "loss": 0.0035,
+      "step": 15868
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 5.796758598951791e-08,
+      "loss": 0.0026,
+      "step": 15870
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 5.7544099828428237e-08,
+      "loss": 0.001,
+      "step": 15872
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 5.7122161783748074e-08,
+      "loss": 0.0019,
+      "step": 15874
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 5.670177192117599e-08,
+      "loss": 0.0005,
+      "step": 15876
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 5.6282930306171824e-08,
+      "loss": 0.0025,
+      "step": 15878
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 5.58656370039512e-08,
+      "loss": 0.0008,
+      "step": 15880
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 5.544989207949103e-08,
+      "loss": 0.0004,
+      "step": 15882
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 5.5035695597523977e-08,
+      "loss": 0.0024,
+      "step": 15884
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 5.4623047622546225e-08,
+      "loss": 0.0036,
+      "step": 15886
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 5.421194821880749e-08,
+      "loss": 0.0021,
+      "step": 15888
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 5.380239745032101e-08,
+      "loss": 0.0018,
+      "step": 15890
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 5.339439538085578e-08,
+      "loss": 0.0026,
+      "step": 15892
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 5.2987942073943196e-08,
+      "loss": 0.0006,
+      "step": 15894
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 5.258303759286709e-08,
+      "loss": 0.0035,
+      "step": 15896
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 5.2179682000677023e-08,
+      "loss": 0.0029,
+      "step": 15898
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 5.1777875360178307e-08,
+      "loss": 0.0019,
+      "step": 15900
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 5.137761773393535e-08,
+      "loss": 0.0041,
+      "step": 15902
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 5.097890918427162e-08,
+      "loss": 0.0031,
+      "step": 15904
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 5.058174977326746e-08,
+      "loss": 0.0031,
+      "step": 15906
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 5.0186139562764526e-08,
+      "loss": 0.0021,
+      "step": 15908
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.979207861436242e-08,
+      "loss": 0.001,
+      "step": 15910
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.9399566989419834e-08,
+      "loss": 0.0023,
+      "step": 15912
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.900860474905234e-08,
+      "loss": 0.0012,
+      "step": 15914
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.861919195413789e-08,
+      "loss": 0.002,
+      "step": 15916
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.8231328665310216e-08,
+      "loss": 0.0019,
+      "step": 15918
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.784501494296212e-08,
+      "loss": 0.0015,
+      "step": 15920
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.746025084724548e-08,
+      "loss": 0.0007,
+      "step": 15922
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.707703643807127e-08,
+      "loss": 0.0021,
+      "step": 15924
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.669537177510952e-08,
+      "loss": 0.0008,
+      "step": 15926
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.6315256917787155e-08,
+      "loss": 0.0036,
+      "step": 15928
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.593669192529237e-08,
+      "loss": 0.002,
+      "step": 15930
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.5559676856570255e-08,
+      "loss": 0.0026,
+      "step": 15932
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.518421177032384e-08,
+      "loss": 0.0012,
+      "step": 15934
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.481029672501635e-08,
+      "loss": 0.0012,
+      "step": 15936
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.4437931778870127e-08,
+      "loss": 0.0013,
+      "step": 15938
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.4067116989864325e-08,
+      "loss": 0.0007,
+      "step": 15940
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.369785241573832e-08,
+      "loss": 0.0008,
+      "step": 15942
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.333013811398834e-08,
+      "loss": 0.0019,
+      "step": 15944
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 4.296397414187192e-08,
+      "loss": 0.003,
+      "step": 15946
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 4.259936055640235e-08,
+      "loss": 0.0012,
+      "step": 15948
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 4.22362974143542e-08,
+      "loss": 0.0029,
+      "step": 15950
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 4.1874784772256707e-08,
+      "loss": 0.006,
+      "step": 15952
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 4.15148226864015e-08,
+      "loss": 0.0028,
+      "step": 15954
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 4.11564112128382e-08,
+      "loss": 0.0016,
+      "step": 15956
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 4.0799550407373267e-08,
+      "loss": 0.0002,
+      "step": 15958
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 4.044424032557226e-08,
+      "loss": 0.0046,
+      "step": 15960
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 4.009048102276092e-08,
+      "loss": 0.0019,
+      "step": 15962
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.973827255402185e-08,
+      "loss": 0.0023,
+      "step": 15964
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.938761497419563e-08,
+      "loss": 0.0035,
+      "step": 15966
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.903850833788303e-08,
+      "loss": 0.0024,
+      "step": 15968
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.86909526994439e-08,
+      "loss": 0.004,
+      "step": 15970
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.834494811299272e-08,
+      "loss": 0.003,
+      "step": 15972
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.800049463240751e-08,
+      "loss": 0.0021,
+      "step": 15974
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.765759231132093e-08,
+      "loss": 0.0042,
+      "step": 15976
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.731624120312582e-08,
+      "loss": 0.0004,
+      "step": 15978
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.6976441360971896e-08,
+      "loss": 0.0025,
+      "step": 15980
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.663819283777126e-08,
+      "loss": 0.0024,
+      "step": 15982
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.630149568618957e-08,
+      "loss": 0.0023,
+      "step": 15984
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.596634995865489e-08,
+      "loss": 0.0027,
+      "step": 15986
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.563275570735103e-08,
+      "loss": 0.0012,
+      "step": 15988
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.5300712984219774e-08,
+      "loss": 0.0026,
+      "step": 15990
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.497022184096532e-08,
+      "loss": 0.0019,
+      "step": 15992
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.46412823290454e-08,
+      "loss": 0.0026,
+      "step": 15994
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.4313894499680146e-08,
+      "loss": 0.0021,
+      "step": 15996
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.398805840384545e-08,
+      "loss": 0.0016,
+      "step": 15998
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.366377409227739e-08,
+      "loss": 0.0016,
+      "step": 16000
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.334104161546781e-08,
+      "loss": 0.0006,
+      "step": 16002
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.301986102366983e-08,
+      "loss": 0.004,
+      "step": 16004
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.270023236689457e-08,
+      "loss": 0.0039,
+      "step": 16006
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.238215569490777e-08,
+      "loss": 0.0021,
+      "step": 16008
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.206563105723981e-08,
+      "loss": 0.003,
+      "step": 16010
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.175065850317349e-08,
+      "loss": 0.003,
+      "step": 16012
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.143723808175403e-08,
+      "loss": 0.0093,
+      "step": 16014
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.112536984178238e-08,
+      "loss": 0.0018,
+      "step": 16016
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.0815053831818596e-08,
+      "loss": 0.0023,
+      "step": 16018
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.050629010018291e-08,
+      "loss": 0.0016,
+      "step": 16020
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.01990786949502e-08,
+      "loss": 0.0007,
+      "step": 16022
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.989341966395665e-08,
+      "loss": 0.003,
+      "step": 16024
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.95893130547964e-08,
+      "loss": 0.0009,
+      "step": 16026
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.9286758914819358e-08,
+      "loss": 0.0016,
+      "step": 16028
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.8985757291137838e-08,
+      "loss": 0.0021,
+      "step": 16030
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.868630823061769e-08,
+      "loss": 0.0045,
+      "step": 16032
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.838841177988605e-08,
+      "loss": 0.0033,
+      "step": 16034
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.809206798532915e-08,
+      "loss": 0.0021,
+      "step": 16036
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.7797276893087865e-08,
+      "loss": 0.0009,
+      "step": 16038
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.7504038549065472e-08,
+      "loss": 0.0014,
+      "step": 16040
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.721235299891989e-08,
+      "loss": 0.0036,
+      "step": 16042
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.6922220288070345e-08,
+      "loss": 0.0018,
+      "step": 16044
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.6633640461691812e-08,
+      "loss": 0.004,
+      "step": 16046
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.634661356471724e-08,
+      "loss": 0.0051,
+      "step": 16048
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.606113964183976e-08,
+      "loss": 0.0027,
+      "step": 16050
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.57772187375116e-08,
+      "loss": 0.0017,
+      "step": 16052
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.5494850895938505e-08,
+      "loss": 0.0031,
+      "step": 16054
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.521403616108975e-08,
+      "loss": 0.0031,
+      "step": 16056
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.4934774576688137e-08,
+      "loss": 0.0015,
+      "step": 16058
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.465706618621888e-08,
+      "loss": 0.0028,
+      "step": 16060
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.4380911032921838e-08,
+      "loss": 0.0025,
+      "step": 16062
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.410630915979817e-08,
+      "loss": 0.0013,
+      "step": 16064
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.383326060960256e-08,
+      "loss": 0.0023,
+      "step": 16066
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.356176542485322e-08,
+      "loss": 0.0015,
+      "step": 16068
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.329182364782412e-08,
+      "loss": 0.0009,
+      "step": 16070
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3023435320546073e-08,
+      "loss": 0.0013,
+      "step": 16072
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.2756600484808987e-08,
+      "loss": 0.0056,
+      "step": 16074
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.2491319182162964e-08,
+      "loss": 0.0044,
+      "step": 16076
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.2227591453911623e-08,
+      "loss": 0.0015,
+      "step": 16078
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.196541734112212e-08,
+      "loss": 0.0028,
+      "step": 16080
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.170479688461513e-08,
+      "loss": 0.0013,
+      "step": 16082
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.144573012497153e-08,
+      "loss": 0.0026,
+      "step": 16084
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.118821710253127e-08,
+      "loss": 0.0015,
+      "step": 16086
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.093225785738895e-08,
+      "loss": 0.0035,
+      "step": 16088
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.0677852429400458e-08,
+      "loss": 0.0021,
+      "step": 16090
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.0425000858179665e-08,
+      "loss": 0.0027,
+      "step": 16092
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.0173703183096194e-08,
+      "loss": 0.0011,
+      "step": 16094
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.992395944327874e-08,
+      "loss": 0.0032,
+      "step": 16096
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.9675769677616196e-08,
+      "loss": 0.0005,
+      "step": 16098
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.9429133924752096e-08,
+      "loss": 0.002,
+      "step": 16100
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.9184052223089054e-08,
+      "loss": 0.0039,
+      "step": 16102
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.8940524610789878e-08,
+      "loss": 0.0036,
+      "step": 16104
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.8698551125772015e-08,
+      "loss": 0.0007,
+      "step": 16106
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.8458131805713096e-08,
+      "loss": 0.0015,
+      "step": 16108
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.8219266688048742e-08,
+      "loss": 0.0017,
+      "step": 16110
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.798195580997142e-08,
+      "loss": 0.0011,
+      "step": 16112
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.774619920843268e-08,
+      "loss": 0.0008,
+      "step": 16114
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.751199692014205e-08,
+      "loss": 0.0007,
+      "step": 16116
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.7279348981565914e-08,
+      "loss": 0.0023,
+      "step": 16118
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.704825542892974e-08,
+      "loss": 0.0021,
+      "step": 16120
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.6818716298215855e-08,
+      "loss": 0.0044,
+      "step": 16122
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.6590731625165666e-08,
+      "loss": 0.0032,
+      "step": 16124
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.6364301445278563e-08,
+      "loss": 0.0015,
+      "step": 16126
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.613942579381189e-08,
+      "loss": 0.0027,
+      "step": 16128
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.5916104705778757e-08,
+      "loss": 0.0008,
+      "step": 16130
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.5694338215952453e-08,
+      "loss": 0.0025,
+      "step": 16132
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.547412635886536e-08,
+      "loss": 0.0008,
+      "step": 16134
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.5255469168804492e-08,
+      "loss": 0.0013,
+      "step": 16136
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.503836667981595e-08,
+      "loss": 0.0026,
+      "step": 16138
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.4822818925707139e-08,
+      "loss": 0.0024,
+      "step": 16140
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.4608825940036764e-08,
+      "loss": 0.0012,
+      "step": 16142
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.4396387756128172e-08,
+      "loss": 0.0018,
+      "step": 16144
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.4185504407058237e-08,
+      "loss": 0.0018,
+      "step": 16146
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.3976175925662917e-08,
+      "loss": 0.0012,
+      "step": 16148
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.3768402344537246e-08,
+      "loss": 0.0007,
+      "step": 16150
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.3562183696032016e-08,
+      "loss": 0.0004,
+      "step": 16152
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.3357520012258207e-08,
+      "loss": 0.0048,
+      "step": 16154
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.315441132508255e-08,
+      "loss": 0.0031,
+      "step": 16156
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.2952857666130858e-08,
+      "loss": 0.0025,
+      "step": 16158
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.2752859066786915e-08,
+      "loss": 0.0039,
+      "step": 16160
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.2554415558191368e-08,
+      "loss": 0.0012,
+      "step": 16162
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.2357527171243943e-08,
+      "loss": 0.0025,
+      "step": 16164
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.216219393660234e-08,
+      "loss": 0.0018,
+      "step": 16166
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.1968415884680007e-08,
+      "loss": 0.0005,
+      "step": 16168
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.1776193045650585e-08,
+      "loss": 0.0023,
+      "step": 16170
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.1585525449443468e-08,
+      "loss": 0.001,
+      "step": 16172
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.1396413125749351e-08,
+      "loss": 0.0007,
+      "step": 16174
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.1208856104012455e-08,
+      "loss": 0.0005,
+      "step": 16176
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.10228544134372e-08,
+      "loss": 0.004,
+      "step": 16178
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.0838408082985974e-08,
+      "loss": 0.0009,
+      "step": 16180
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.0655517141378024e-08,
+      "loss": 0.0021,
+      "step": 16182
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.0474181617091684e-08,
+      "loss": 0.0018,
+      "step": 16184
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.0294401538361032e-08,
+      "loss": 0.0036,
+      "step": 16186
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.0116176933180343e-08,
+      "loss": 0.0013,
+      "step": 16188
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 9.939507829299644e-09,
+      "loss": 0.0021,
+      "step": 16190
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 9.764394254228037e-09,
+      "loss": 0.0047,
+      "step": 16192
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 9.590836235232604e-09,
+      "loss": 0.002,
+      "step": 16194
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 9.41883379933728e-09,
+      "loss": 0.0014,
+      "step": 16196
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 9.248386973323975e-09,
+      "loss": 0.0024,
+      "step": 16198
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 9.079495783731462e-09,
+      "loss": 0.0027,
+      "step": 16200
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 8.912160256859813e-09,
+      "loss": 0.0018,
+      "step": 16202
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 8.746380418762635e-09,
+      "loss": 0.003,
+      "step": 16204
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 8.58215629525372e-09,
+      "loss": 0.0026,
+      "step": 16206
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 8.419487911903724e-09,
+      "loss": 0.0027,
+      "step": 16208
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 8.258375294042386e-09,
+      "loss": 0.0023,
+      "step": 16210
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 8.098818466755198e-09,
+      "loss": 0.0013,
+      "step": 16212
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 7.940817454885619e-09,
+      "loss": 0.0027,
+      "step": 16214
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 7.7843722830373e-09,
+      "loss": 0.0028,
+      "step": 16216
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 7.629482975569646e-09,
+      "loss": 0.0014,
+      "step": 16218
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 7.476149556598922e-09,
+      "loss": 0.0035,
+      "step": 16220
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 7.324372050001583e-09,
+      "loss": 0.0036,
+      "step": 16222
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 7.174150479409836e-09,
+      "loss": 0.001,
+      "step": 16224
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 7.025484868213861e-09,
+      "loss": 0.0032,
+      "step": 16226
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 6.878375239562918e-09,
+      "loss": 0.0018,
+      "step": 16228
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 6.732821616363128e-09,
+      "loss": 0.0011,
+      "step": 16230
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 6.588824021278584e-09,
+      "loss": 0.0018,
+      "step": 16232
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 6.4463824767291294e-09,
+      "loss": 0.0022,
+      "step": 16234
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 6.3054970048959105e-09,
+      "loss": 0.0014,
+      "step": 16236
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 6.166167627715825e-09,
+      "loss": 0.0024,
+      "step": 16238
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 6.028394366881518e-09,
+      "loss": 0.0042,
+      "step": 16240
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.892177243846942e-09,
+      "loss": 0.0014,
+      "step": 16242
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.757516279821796e-09,
+      "loss": 0.0014,
+      "step": 16244
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.624411495774862e-09,
+      "loss": 0.0023,
+      "step": 16246
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.492862912429564e-09,
+      "loss": 0.0027,
+      "step": 16248
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.362870550269517e-09,
+      "loss": 0.0017,
+      "step": 16250
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.234434429537416e-09,
+      "loss": 0.0017,
+      "step": 16252
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 5.107554570229489e-09,
+      "loss": 0.0011,
+      "step": 16254
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 4.982230992103265e-09,
+      "loss": 0.0033,
+      "step": 16256
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 4.8584637146709135e-09,
+      "loss": 0.0032,
+      "step": 16258
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 4.736252757205906e-09,
+      "loss": 0.0047,
+      "step": 16260
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 4.615598138737465e-09,
+      "loss": 0.002,
+      "step": 16262
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 4.496499878050564e-09,
+      "loss": 0.0015,
+      "step": 16264
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 4.3789579936914795e-09,
+      "loss": 0.0022,
+      "step": 16266
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 4.262972503961127e-09,
+      "loss": 0.0025,
+      "step": 16268
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 4.148543426919505e-09,
+      "loss": 0.0004,
+      "step": 16270
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 4.035670780385692e-09,
+      "loss": 0.0024,
+      "step": 16272
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.924354581932299e-09,
+      "loss": 0.002,
+      "step": 16274
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.814594848894349e-09,
+      "loss": 0.001,
+      "step": 16276
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.7063915983603925e-09,
+      "loss": 0.0024,
+      "step": 16278
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.5997448471802866e-09,
+      "loss": 0.0017,
+      "step": 16280
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.4946546119585257e-09,
+      "loss": 0.0015,
+      "step": 16282
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.3911209090586874e-09,
+      "loss": 0.0024,
+      "step": 16284
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.2891437546023196e-09,
+      "loss": 0.0016,
+      "step": 16286
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.188723164467833e-09,
+      "loss": 0.002,
+      "step": 16288
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.089859154290498e-09,
+      "loss": 0.0012,
+      "step": 16290
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.9925517394657764e-09,
+      "loss": 0.0006,
+      "step": 16292
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.896800935143773e-09,
+      "loss": 0.0017,
+      "step": 16294
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.8026067562347824e-09,
+      "loss": 0.0025,
+      "step": 16296
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.709969217404851e-09,
+      "loss": 0.0016,
+      "step": 16298
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.6188883330779958e-09,
+      "loss": 0.0008,
+      "step": 16300
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.529364117437316e-09,
+      "loss": 0.0033,
+      "step": 16302
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.4413965844216625e-09,
+      "loss": 0.0017,
+      "step": 16304
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.354985747727856e-09,
+      "loss": 0.0008,
+      "step": 16306
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.2701316208117997e-09,
+      "loss": 0.0014,
+      "step": 16308
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1868342168851474e-09,
+      "loss": 0.0017,
+      "step": 16310
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.105093548918635e-09,
+      "loss": 0.0018,
+      "step": 16312
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.0249096296387495e-09,
+      "loss": 0.0051,
+      "step": 16314
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.946282471532168e-09,
+      "loss": 0.0016,
+      "step": 16316
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.869212086841321e-09,
+      "loss": 0.0014,
+      "step": 16318
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.7936984875654983e-09,
+      "loss": 0.002,
+      "step": 16320
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.7197416854641824e-09,
+      "loss": 0.0024,
+      "step": 16322
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.6473416920526064e-09,
+      "loss": 0.0044,
+      "step": 16324
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.5764985186028648e-09,
+      "loss": 0.0054,
+      "step": 16326
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.5072121761472436e-09,
+      "loss": 0.0033,
+      "step": 16328
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.43948267547378e-09,
+      "loss": 0.0007,
+      "step": 16330
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.3733100271284826e-09,
+      "loss": 0.0009,
+      "step": 16332
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.3086942414153315e-09,
+      "loss": 0.0027,
+      "step": 16334
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.2456353283940569e-09,
+      "loss": 0.0021,
+      "step": 16336
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.184133297884582e-09,
+      "loss": 0.0021,
+      "step": 16338
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.1241881594636905e-09,
+      "loss": 0.0022,
+      "step": 16340
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.0657999224639171e-09,
+      "loss": 0.0019,
+      "step": 16342
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.0089685959779882e-09,
+      "loss": 0.0027,
+      "step": 16344
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 9.536941888532713e-10,
+      "loss": 0.0055,
+      "step": 16346
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 8.999767096984358e-10,
+      "loss": 0.0026,
+      "step": 16348
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 8.478161668779017e-10,
+      "loss": 0.0034,
+      "step": 16350
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 7.972125685107301e-10,
+      "loss": 0.0007,
+      "step": 16352
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 7.481659224783943e-10,
+      "loss": 0.0008,
+      "step": 16354
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 7.006762364181185e-10,
+      "loss": 0.0029,
+      "step": 16356
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 6.547435177228778e-10,
+      "loss": 0.0014,
+      "step": 16358
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 6.103677735458391e-10,
+      "loss": 0.0014,
+      "step": 16360
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 5.675490107970305e-10,
+      "loss": 0.0013,
+      "step": 16362
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 5.262872361422311e-10,
+      "loss": 0.0029,
+      "step": 16364
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 4.865824560074118e-10,
+      "loss": 0.0017,
+      "step": 16366
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 4.484346765742942e-10,
+      "loss": 0.0025,
+      "step": 16368
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 4.1184390378257167e-10,
+      "loss": 0.0023,
+      "step": 16370
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.7681014333101897e-10,
+      "loss": 0.0054,
+      "step": 16372
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.433334006730515e-10,
+      "loss": 0.0036,
+      "step": 16374
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.1141368102227676e-10,
+      "loss": 0.0018,
+      "step": 16376
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.8105098934916307e-10,
+      "loss": 0.0035,
+      "step": 16378
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.522453303799299e-10,
+      "loss": 0.0036,
+      "step": 16380
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.2499670860098855e-10,
+      "loss": 0.0026,
+      "step": 16382
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.993051282545011e-10,
+      "loss": 0.0013,
+      "step": 16384
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.751705933417114e-10,
+      "loss": 0.0025,
+      "step": 16386
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.525931076196141e-10,
+      "loss": 0.001,
+      "step": 16388
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.3157267460428557e-10,
+      "loss": 0.0037,
+      "step": 16390
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.1210929756866329e-10,
+      "loss": 0.0015,
+      "step": 16392
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 9.420297954254587e-11,
+      "loss": 0.003,
+      "step": 16394
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 7.785372331592378e-11,
+      "loss": 0.0026,
+      "step": 16396
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 6.306153143231797e-11,
+      "loss": 0.0019,
+      "step": 16398
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 4.982640619766166e-11,
+      "loss": 0.0039,
+      "step": 16400
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.8148349669198113e-11,
+      "loss": 0.0028,
+      "step": 16402
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.802736366880332e-11,
+      "loss": 0.0032,
+      "step": 16404
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.9463449770773522e-11,
+      "loss": 0.002,
+      "step": 16406
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.2456609307376355e-11,
+      "loss": 0.0014,
+      "step": 16408
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 7.006843372181493e-12,
+      "loss": 0.0008,
+      "step": 16410
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.114152812289106e-12,
+      "loss": 0.0011,
+      "step": 16412
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 7.785382327707425e-13,
+      "loss": 0.0029,
+      "step": 16414
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 0.0,
+      "loss": 0.001,
+      "step": 16416
+    },
+    {
+      "epoch": 4.0,
+      "step": 16416,
+      "total_flos": 5.43675641046342e+18,
+      "train_loss": 0.10646877170433743,
+      "train_runtime": 71465.0126,
+      "train_samples_per_second": 7.35,
+      "train_steps_per_second": 0.23
+    }
+  ],
+  "max_steps": 16416,
+  "num_train_epochs": 4,
+  "total_flos": 5.43675641046342e+18,
+  "trial_name": null,
+  "trial_params": null
+}