diff --git "a/last-checkpoint/trainer_state.json" "b/last-checkpoint/trainer_state.json"
--- "a/last-checkpoint/trainer_state.json"
+++ "b/last-checkpoint/trainer_state.json"
@@ -1,8 +1,8 @@
 {
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 0.646752091838797,
-  "global_step": 8000,
+  "epoch": 0.9701281377581955,
+  "global_step": 12000,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
@@ -9727,11 +9727,4822 @@
       "eval_samples_per_second": 0.103,
       "eval_steps_per_second": 0.052,
       "step": 8000
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.923114237205918e-05,
+      "loss": 3.0919,
+      "step": 8005
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.922440509876843e-05,
+      "loss": 3.1249,
+      "step": 8010
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.921766782547767e-05,
+      "loss": 2.8888,
+      "step": 8015
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.921093055218692e-05,
+      "loss": 3.0429,
+      "step": 8020
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.920419327889617e-05,
+      "loss": 3.0775,
+      "step": 8025
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.919745600560542e-05,
+      "loss": 2.9262,
+      "step": 8030
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.919071873231466e-05,
+      "loss": 3.1079,
+      "step": 8035
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.91839814590239e-05,
+      "loss": 2.9723,
+      "step": 8040
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.917724418573315e-05,
+      "loss": 3.3847,
+      "step": 8045
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.91705069124424e-05,
+      "loss": 3.084,
+      "step": 8050
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.916376963915164e-05,
+      "loss": 2.9241,
+      "step": 8055
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.915703236586089e-05,
+      "loss": 2.8274,
+      "step": 8060
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.915029509257014e-05,
+      "loss": 3.0998,
+      "step": 8065
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.914355781927938e-05,
+      "loss": 3.2271,
+      "step": 8070
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.913682054598863e-05,
+      "loss": 3.0446,
+      "step": 8075
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.913008327269787e-05,
+      "loss": 3.2226,
+      "step": 8080
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.912334599940712e-05,
+      "loss": 2.9655,
+      "step": 8085
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9116608726116364e-05,
+      "loss": 3.2028,
+      "step": 8090
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.910987145282561e-05,
+      "loss": 3.1591,
+      "step": 8095
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.910313417953486e-05,
+      "loss": 2.9486,
+      "step": 8100
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.909639690624411e-05,
+      "loss": 3.0678,
+      "step": 8105
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.908965963295335e-05,
+      "loss": 2.7782,
+      "step": 8110
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.90829223596626e-05,
+      "loss": 3.3781,
+      "step": 8115
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9076185086371844e-05,
+      "loss": 3.2015,
+      "step": 8120
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9069447813081086e-05,
+      "loss": 3.2439,
+      "step": 8125
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9062710539790335e-05,
+      "loss": 2.8638,
+      "step": 8130
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9055973266499584e-05,
+      "loss": 2.9865,
+      "step": 8135
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.904923599320883e-05,
+      "loss": 3.0172,
+      "step": 8140
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9042498719918075e-05,
+      "loss": 2.7773,
+      "step": 8145
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9035761446627324e-05,
+      "loss": 3.3144,
+      "step": 8150
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.902902417333657e-05,
+      "loss": 3.3635,
+      "step": 8155
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.902228690004582e-05,
+      "loss": 2.9169,
+      "step": 8160
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9015549626755063e-05,
+      "loss": 2.8164,
+      "step": 8165
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9008812353464306e-05,
+      "loss": 2.8467,
+      "step": 8170
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9002075080173554e-05,
+      "loss": 3.0674,
+      "step": 8175
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.8995337806882797e-05,
+      "loss": 2.9448,
+      "step": 8180
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.8988600533592045e-05,
+      "loss": 3.3653,
+      "step": 8185
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.8981863260301294e-05,
+      "loss": 2.9465,
+      "step": 8190
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.897512598701054e-05,
+      "loss": 2.9197,
+      "step": 8195
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.8968388713719785e-05,
+      "loss": 3.0495,
+      "step": 8200
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.8961651440429034e-05,
+      "loss": 3.104,
+      "step": 8205
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.8954914167138276e-05,
+      "loss": 3.0136,
+      "step": 8210
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.8948176893847525e-05,
+      "loss": 2.9737,
+      "step": 8215
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.894143962055677e-05,
+      "loss": 3.0539,
+      "step": 8220
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.8934702347266016e-05,
+      "loss": 3.0518,
+      "step": 8225
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8927965073975265e-05,
+      "loss": 3.2109,
+      "step": 8230
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.892122780068451e-05,
+      "loss": 3.0303,
+      "step": 8235
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8914490527393756e-05,
+      "loss": 3.2467,
+      "step": 8240
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8907753254103005e-05,
+      "loss": 3.0201,
+      "step": 8245
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.890101598081225e-05,
+      "loss": 3.052,
+      "step": 8250
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.889427870752149e-05,
+      "loss": 3.2192,
+      "step": 8255
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.888754143423074e-05,
+      "loss": 2.8097,
+      "step": 8260
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.888080416093999e-05,
+      "loss": 2.9116,
+      "step": 8265
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.887406688764923e-05,
+      "loss": 3.1297,
+      "step": 8270
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.886732961435848e-05,
+      "loss": 3.2419,
+      "step": 8275
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8860592341067727e-05,
+      "loss": 2.9156,
+      "step": 8280
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8853855067776975e-05,
+      "loss": 3.1024,
+      "step": 8285
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.884711779448622e-05,
+      "loss": 2.9807,
+      "step": 8290
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8840380521195466e-05,
+      "loss": 3.2677,
+      "step": 8295
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.883364324790471e-05,
+      "loss": 3.0497,
+      "step": 8300
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.882690597461396e-05,
+      "loss": 3.1933,
+      "step": 8305
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.88201687013232e-05,
+      "loss": 2.9817,
+      "step": 8310
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.881343142803245e-05,
+      "loss": 2.9614,
+      "step": 8315
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.88066941547417e-05,
+      "loss": 3.2148,
+      "step": 8320
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.879995688145094e-05,
+      "loss": 3.4115,
+      "step": 8325
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.879321960816019e-05,
+      "loss": 3.109,
+      "step": 8330
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.878648233486944e-05,
+      "loss": 3.3976,
+      "step": 8335
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.877974506157868e-05,
+      "loss": 2.9871,
+      "step": 8340
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.877300778828792e-05,
+      "loss": 3.0941,
+      "step": 8345
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.876627051499717e-05,
+      "loss": 3.0284,
+      "step": 8350
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.875953324170642e-05,
+      "loss": 2.8833,
+      "step": 8355
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.875279596841567e-05,
+      "loss": 3.0523,
+      "step": 8360
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.874605869512491e-05,
+      "loss": 3.09,
+      "step": 8365
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.873932142183416e-05,
+      "loss": 3.1666,
+      "step": 8370
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.873258414854341e-05,
+      "loss": 3.1402,
+      "step": 8375
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.872584687525265e-05,
+      "loss": 3.0485,
+      "step": 8380
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.871910960196189e-05,
+      "loss": 3.1616,
+      "step": 8385
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.871237232867114e-05,
+      "loss": 2.9211,
+      "step": 8390
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.870563505538039e-05,
+      "loss": 2.9857,
+      "step": 8395
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.869889778208963e-05,
+      "loss": 3.1664,
+      "step": 8400
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.869216050879888e-05,
+      "loss": 3.0174,
+      "step": 8405
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.868542323550813e-05,
+      "loss": 3.0245,
+      "step": 8410
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.867868596221738e-05,
+      "loss": 3.1736,
+      "step": 8415
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.867194868892662e-05,
+      "loss": 3.1161,
+      "step": 8420
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.866521141563586e-05,
+      "loss": 3.0927,
+      "step": 8425
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.865847414234511e-05,
+      "loss": 3.0474,
+      "step": 8430
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8651736869054354e-05,
+      "loss": 2.7999,
+      "step": 8435
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.86449995957636e-05,
+      "loss": 2.9191,
+      "step": 8440
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.863826232247285e-05,
+      "loss": 2.9211,
+      "step": 8445
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.86315250491821e-05,
+      "loss": 3.1388,
+      "step": 8450
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.862478777589134e-05,
+      "loss": 3.2235,
+      "step": 8455
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.861805050260059e-05,
+      "loss": 3.0092,
+      "step": 8460
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.861131322930984e-05,
+      "loss": 3.0099,
+      "step": 8465
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.860457595601908e-05,
+      "loss": 3.1131,
+      "step": 8470
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8597838682728324e-05,
+      "loss": 3.0014,
+      "step": 8475
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.859110140943757e-05,
+      "loss": 3.2398,
+      "step": 8480
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.858436413614682e-05,
+      "loss": 3.0938,
+      "step": 8485
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8577626862856064e-05,
+      "loss": 2.9129,
+      "step": 8490
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.857088958956531e-05,
+      "loss": 3.3385,
+      "step": 8495
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.856415231627456e-05,
+      "loss": 2.9213,
+      "step": 8500
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.855741504298381e-05,
+      "loss": 3.254,
+      "step": 8505
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.855067776969305e-05,
+      "loss": 3.041,
+      "step": 8510
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8543940496402295e-05,
+      "loss": 3.0149,
+      "step": 8515
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8537203223111544e-05,
+      "loss": 2.9648,
+      "step": 8520
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8530465949820786e-05,
+      "loss": 2.8249,
+      "step": 8525
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8523728676530035e-05,
+      "loss": 3.4272,
+      "step": 8530
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8516991403239284e-05,
+      "loss": 3.2086,
+      "step": 8535
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.851025412994853e-05,
+      "loss": 3.3207,
+      "step": 8540
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8503516856657775e-05,
+      "loss": 3.0051,
+      "step": 8545
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.849677958336702e-05,
+      "loss": 3.0208,
+      "step": 8550
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8490042310076265e-05,
+      "loss": 3.1673,
+      "step": 8555
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8483305036785514e-05,
+      "loss": 3.1774,
+      "step": 8560
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8476567763494756e-05,
+      "loss": 2.9722,
+      "step": 8565
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8469830490204005e-05,
+      "loss": 2.7325,
+      "step": 8570
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8463093216913254e-05,
+      "loss": 3.0779,
+      "step": 8575
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8456355943622496e-05,
+      "loss": 3.1809,
+      "step": 8580
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8449618670331745e-05,
+      "loss": 2.9728,
+      "step": 8585
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8442881397040994e-05,
+      "loss": 2.8935,
+      "step": 8590
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.843614412375024e-05,
+      "loss": 3.0019,
+      "step": 8595
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8429406850459485e-05,
+      "loss": 3.1546,
+      "step": 8600
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.842266957716873e-05,
+      "loss": 2.9821,
+      "step": 8605
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8415932303877976e-05,
+      "loss": 2.9484,
+      "step": 8610
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8409195030587225e-05,
+      "loss": 3.1934,
+      "step": 8615
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.840245775729647e-05,
+      "loss": 3.2961,
+      "step": 8620
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8395720484005716e-05,
+      "loss": 3.0485,
+      "step": 8625
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8388983210714965e-05,
+      "loss": 3.0174,
+      "step": 8630
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.838224593742421e-05,
+      "loss": 3.1566,
+      "step": 8635
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8375508664133456e-05,
+      "loss": 2.893,
+      "step": 8640
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.83687713908427e-05,
+      "loss": 3.1131,
+      "step": 8645
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.836203411755195e-05,
+      "loss": 2.8845,
+      "step": 8650
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.835529684426119e-05,
+      "loss": 2.8261,
+      "step": 8655
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.834855957097044e-05,
+      "loss": 3.0335,
+      "step": 8660
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8341822297679686e-05,
+      "loss": 3.0395,
+      "step": 8665
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8335085024388935e-05,
+      "loss": 3.2093,
+      "step": 8670
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.832834775109818e-05,
+      "loss": 3.0218,
+      "step": 8675
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8321610477807426e-05,
+      "loss": 3.0037,
+      "step": 8680
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.831487320451667e-05,
+      "loss": 3.0408,
+      "step": 8685
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.830813593122591e-05,
+      "loss": 2.9611,
+      "step": 8690
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.830139865793516e-05,
+      "loss": 3.2521,
+      "step": 8695
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.829466138464441e-05,
+      "loss": 2.9996,
+      "step": 8700
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.828792411135366e-05,
+      "loss": 3.1156,
+      "step": 8705
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.82811868380629e-05,
+      "loss": 2.7829,
+      "step": 8710
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.827444956477215e-05,
+      "loss": 2.9055,
+      "step": 8715
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.82677122914814e-05,
+      "loss": 2.969,
+      "step": 8720
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.826097501819064e-05,
+      "loss": 2.7114,
+      "step": 8725
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.825423774489988e-05,
+      "loss": 2.9136,
+      "step": 8730
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.824750047160913e-05,
+      "loss": 2.8986,
+      "step": 8735
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.824076319831838e-05,
+      "loss": 3.0897,
+      "step": 8740
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.823402592502762e-05,
+      "loss": 3.1301,
+      "step": 8745
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.822728865173687e-05,
+      "loss": 2.9386,
+      "step": 8750
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.822055137844612e-05,
+      "loss": 3.1028,
+      "step": 8755
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.821381410515537e-05,
+      "loss": 2.7394,
+      "step": 8760
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.820707683186461e-05,
+      "loss": 3.0343,
+      "step": 8765
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.820033955857386e-05,
+      "loss": 3.0186,
+      "step": 8770
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.81936022852831e-05,
+      "loss": 3.1434,
+      "step": 8775
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.818686501199235e-05,
+      "loss": 3.1984,
+      "step": 8780
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.818012773870159e-05,
+      "loss": 2.9959,
+      "step": 8785
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.817339046541084e-05,
+      "loss": 2.9381,
+      "step": 8790
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.816665319212009e-05,
+      "loss": 3.2456,
+      "step": 8795
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.815991591882933e-05,
+      "loss": 3.0803,
+      "step": 8800
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.815317864553858e-05,
+      "loss": 2.755,
+      "step": 8805
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.814644137224783e-05,
+      "loss": 3.3534,
+      "step": 8810
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.813970409895707e-05,
+      "loss": 3.0963,
+      "step": 8815
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8132966825666313e-05,
+      "loss": 2.9195,
+      "step": 8820
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.812622955237556e-05,
+      "loss": 2.8593,
+      "step": 8825
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.811949227908481e-05,
+      "loss": 2.8676,
+      "step": 8830
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.811275500579405e-05,
+      "loss": 2.7496,
+      "step": 8835
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.81060177325033e-05,
+      "loss": 3.2668,
+      "step": 8840
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.809928045921255e-05,
+      "loss": 3.1165,
+      "step": 8845
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.80925431859218e-05,
+      "loss": 3.0659,
+      "step": 8850
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.808580591263104e-05,
+      "loss": 2.8842,
+      "step": 8855
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8079068639340284e-05,
+      "loss": 3.003,
+      "step": 8860
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.807233136604953e-05,
+      "loss": 3.0949,
+      "step": 8865
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.806559409275878e-05,
+      "loss": 2.9894,
+      "step": 8870
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8058856819468024e-05,
+      "loss": 2.8607,
+      "step": 8875
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.805211954617727e-05,
+      "loss": 3.3431,
+      "step": 8880
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.804538227288652e-05,
+      "loss": 3.0648,
+      "step": 8885
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8038644999595764e-05,
+      "loss": 3.2921,
+      "step": 8890
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.803190772630501e-05,
+      "loss": 2.9446,
+      "step": 8895
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.802517045301426e-05,
+      "loss": 3.1472,
+      "step": 8900
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8018433179723504e-05,
+      "loss": 3.2051,
+      "step": 8905
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8011695906432746e-05,
+      "loss": 3.4834,
+      "step": 8910
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8004958633141995e-05,
+      "loss": 3.0165,
+      "step": 8915
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7998221359851243e-05,
+      "loss": 3.028,
+      "step": 8920
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.799148408656049e-05,
+      "loss": 3.0388,
+      "step": 8925
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7984746813269734e-05,
+      "loss": 2.9662,
+      "step": 8930
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.797800953997898e-05,
+      "loss": 3.0036,
+      "step": 8935
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.797127226668823e-05,
+      "loss": 2.9179,
+      "step": 8940
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7964534993397474e-05,
+      "loss": 2.934,
+      "step": 8945
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7957797720106716e-05,
+      "loss": 3.2011,
+      "step": 8950
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7951060446815965e-05,
+      "loss": 3.1329,
+      "step": 8955
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7944323173525214e-05,
+      "loss": 3.2528,
+      "step": 8960
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7937585900234456e-05,
+      "loss": 3.3252,
+      "step": 8965
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7930848626943705e-05,
+      "loss": 3.2545,
+      "step": 8970
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7924111353652954e-05,
+      "loss": 3.2199,
+      "step": 8975
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.79173740803622e-05,
+      "loss": 2.9862,
+      "step": 8980
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7910636807071445e-05,
+      "loss": 3.0251,
+      "step": 8985
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.790389953378069e-05,
+      "loss": 2.9251,
+      "step": 8990
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7897162260489936e-05,
+      "loss": 3.0674,
+      "step": 8995
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.789042498719918e-05,
+      "loss": 3.0134,
+      "step": 9000
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.788368771390843e-05,
+      "loss": 3.4247,
+      "step": 9005
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.787829789527583e-05,
+      "loss": 3.4266,
+      "step": 9010
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.787156062198507e-05,
+      "loss": 3.334,
+      "step": 9015
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.786482334869432e-05,
+      "loss": 2.6583,
+      "step": 9020
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.785808607540356e-05,
+      "loss": 2.9451,
+      "step": 9025
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.785134880211281e-05,
+      "loss": 2.9675,
+      "step": 9030
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.784461152882205e-05,
+      "loss": 3.0311,
+      "step": 9035
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.78378742555313e-05,
+      "loss": 3.0225,
+      "step": 9040
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.783113698224055e-05,
+      "loss": 3.4978,
+      "step": 9045
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.78243997089498e-05,
+      "loss": 2.8633,
+      "step": 9050
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.781766243565904e-05,
+      "loss": 3.1776,
+      "step": 9055
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.781092516236829e-05,
+      "loss": 2.9358,
+      "step": 9060
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.780418788907754e-05,
+      "loss": 2.9577,
+      "step": 9065
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.779745061578678e-05,
+      "loss": 2.8513,
+      "step": 9070
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.779071334249602e-05,
+      "loss": 2.9536,
+      "step": 9075
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.778397606920527e-05,
+      "loss": 3.3119,
+      "step": 9080
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.777723879591452e-05,
+      "loss": 3.1788,
+      "step": 9085
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.777050152262376e-05,
+      "loss": 3.2256,
+      "step": 9090
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.776376424933301e-05,
+      "loss": 3.4169,
+      "step": 9095
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.775702697604226e-05,
+      "loss": 3.2875,
+      "step": 9100
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.775028970275151e-05,
+      "loss": 2.7868,
+      "step": 9105
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.774355242946075e-05,
+      "loss": 3.1224,
+      "step": 9110
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7736815156169994e-05,
+      "loss": 2.9381,
+      "step": 9115
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.773007788287924e-05,
+      "loss": 2.9636,
+      "step": 9120
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7723340609588485e-05,
+      "loss": 3.1651,
+      "step": 9125
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7716603336297734e-05,
+      "loss": 3.1566,
+      "step": 9130
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.770986606300698e-05,
+      "loss": 3.0716,
+      "step": 9135
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.770312878971623e-05,
+      "loss": 2.9066,
+      "step": 9140
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7696391516425474e-05,
+      "loss": 3.0514,
+      "step": 9145
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.768965424313472e-05,
+      "loss": 2.9148,
+      "step": 9150
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7682916969843965e-05,
+      "loss": 2.9026,
+      "step": 9155
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7676179696553214e-05,
+      "loss": 3.373,
+      "step": 9160
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7669442423262456e-05,
+      "loss": 3.0228,
+      "step": 9165
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7662705149971705e-05,
+      "loss": 2.9815,
+      "step": 9170
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7655967876680953e-05,
+      "loss": 2.9908,
+      "step": 9175
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7649230603390196e-05,
+      "loss": 3.0224,
+      "step": 9180
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7642493330099444e-05,
+      "loss": 3.5886,
+      "step": 9185
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.763575605680869e-05,
+      "loss": 2.9071,
+      "step": 9190
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7629018783517935e-05,
+      "loss": 2.9337,
+      "step": 9195
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7622281510227184e-05,
+      "loss": 3.4955,
+      "step": 9200
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7615544236936426e-05,
+      "loss": 2.9738,
+      "step": 9205
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7608806963645675e-05,
+      "loss": 2.9454,
+      "step": 9210
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7602069690354924e-05,
+      "loss": 3.086,
+      "step": 9215
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7595332417064166e-05,
+      "loss": 3.4329,
+      "step": 9220
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7588595143773415e-05,
+      "loss": 3.1704,
+      "step": 9225
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7581857870482664e-05,
+      "loss": 3.0783,
+      "step": 9230
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7575120597191906e-05,
+      "loss": 2.9674,
+      "step": 9235
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7568383323901155e-05,
+      "loss": 3.1008,
+      "step": 9240
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.75616460506104e-05,
+      "loss": 3.07,
+      "step": 9245
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7554908777319646e-05,
+      "loss": 3.3072,
+      "step": 9250
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.754817150402889e-05,
+      "loss": 3.1095,
+      "step": 9255
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.754143423073814e-05,
+      "loss": 3.0179,
+      "step": 9260
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7534696957447386e-05,
+      "loss": 2.901,
+      "step": 9265
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.752795968415663e-05,
+      "loss": 2.9,
+      "step": 9270
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.752122241086588e-05,
+      "loss": 3.0123,
+      "step": 9275
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7514485137575126e-05,
+      "loss": 3.2678,
+      "step": 9280
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.750774786428437e-05,
+      "loss": 2.8679,
+      "step": 9285
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.750101059099361e-05,
+      "loss": 2.9353,
+      "step": 9290
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.749427331770286e-05,
+      "loss": 2.8873,
+      "step": 9295
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.748753604441211e-05,
+      "loss": 3.2057,
+      "step": 9300
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7480798771121356e-05,
+      "loss": 3.1522,
+      "step": 9305
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.74740614978306e-05,
+      "loss": 3.1599,
+      "step": 9310
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.746732422453985e-05,
+      "loss": 2.9306,
+      "step": 9315
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7460586951249096e-05,
+      "loss": 3.013,
+      "step": 9320
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.745384967795834e-05,
+      "loss": 3.0665,
+      "step": 9325
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.744711240466758e-05,
+      "loss": 3.2568,
+      "step": 9330
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.744037513137683e-05,
+      "loss": 2.9778,
+      "step": 9335
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.743363785808608e-05,
+      "loss": 3.2069,
+      "step": 9340
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.742690058479532e-05,
+      "loss": 3.0228,
+      "step": 9345
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.742016331150457e-05,
+      "loss": 2.8045,
+      "step": 9350
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.741342603821382e-05,
+      "loss": 3.0413,
+      "step": 9355
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.740668876492307e-05,
+      "loss": 2.8799,
+      "step": 9360
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.739995149163231e-05,
+      "loss": 3.2395,
+      "step": 9365
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.739321421834156e-05,
+      "loss": 3.055,
+      "step": 9370
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.73864769450508e-05,
+      "loss": 3.046,
+      "step": 9375
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.737973967176004e-05,
+      "loss": 2.7768,
+      "step": 9380
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.737300239846929e-05,
+      "loss": 3.4218,
+      "step": 9385
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.736626512517854e-05,
+      "loss": 2.961,
+      "step": 9390
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.735952785188779e-05,
+      "loss": 3.1829,
+      "step": 9395
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.735279057859703e-05,
+      "loss": 3.1464,
+      "step": 9400
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.734605330530628e-05,
+      "loss": 3.0201,
+      "step": 9405
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.733931603201553e-05,
+      "loss": 2.8433,
+      "step": 9410
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.733257875872477e-05,
+      "loss": 2.9373,
+      "step": 9415
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.732584148543401e-05,
+      "loss": 3.1014,
+      "step": 9420
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.731910421214326e-05,
+      "loss": 2.9788,
+      "step": 9425
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.731236693885251e-05,
+      "loss": 3.1346,
+      "step": 9430
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.730562966556175e-05,
+      "loss": 3.0174,
+      "step": 9435
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7298892392271e-05,
+      "loss": 3.1674,
+      "step": 9440
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.729215511898025e-05,
+      "loss": 2.9165,
+      "step": 9445
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.72854178456895e-05,
+      "loss": 2.9635,
+      "step": 9450
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.727868057239874e-05,
+      "loss": 3.3654,
+      "step": 9455
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.727194329910798e-05,
+      "loss": 2.9462,
+      "step": 9460
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.726520602581723e-05,
+      "loss": 3.0271,
+      "step": 9465
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.725846875252648e-05,
+      "loss": 2.9914,
+      "step": 9470
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.725173147923572e-05,
+      "loss": 3.0899,
+      "step": 9475
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.724499420594497e-05,
+      "loss": 3.2328,
+      "step": 9480
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.723825693265422e-05,
+      "loss": 2.9155,
+      "step": 9485
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.723151965936346e-05,
+      "loss": 3.0138,
+      "step": 9490
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.722478238607271e-05,
+      "loss": 3.0283,
+      "step": 9495
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.721804511278196e-05,
+      "loss": 3.021,
+      "step": 9500
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.72113078394912e-05,
+      "loss": 3.3555,
+      "step": 9505
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7204570566200445e-05,
+      "loss": 2.9653,
+      "step": 9510
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7197833292909694e-05,
+      "loss": 2.8215,
+      "step": 9515
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.719109601961894e-05,
+      "loss": 2.9035,
+      "step": 9520
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.718435874632819e-05,
+      "loss": 3.1791,
+      "step": 9525
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7177621473037434e-05,
+      "loss": 3.0717,
+      "step": 9530
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.717088419974668e-05,
+      "loss": 3.1412,
+      "step": 9535
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.716414692645593e-05,
+      "loss": 2.8527,
+      "step": 9540
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7157409653165173e-05,
+      "loss": 3.1082,
+      "step": 9545
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7150672379874416e-05,
+      "loss": 3.0959,
+      "step": 9550
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7143935106583664e-05,
+      "loss": 2.8961,
+      "step": 9555
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.713719783329291e-05,
+      "loss": 3.0879,
+      "step": 9560
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7130460560002155e-05,
+      "loss": 3.0639,
+      "step": 9565
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7123723286711404e-05,
+      "loss": 2.9382,
+      "step": 9570
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.711698601342065e-05,
+      "loss": 2.9074,
+      "step": 9575
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7110248740129895e-05,
+      "loss": 3.1172,
+      "step": 9580
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7103511466839144e-05,
+      "loss": 2.6656,
+      "step": 9585
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.7096774193548386e-05,
+      "loss": 2.9794,
+      "step": 9590
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.7090036920257635e-05,
+      "loss": 3.0121,
+      "step": 9595
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.708329964696688e-05,
+      "loss": 2.9782,
+      "step": 9600
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.7076562373676126e-05,
+      "loss": 3.3433,
+      "step": 9605
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.7069825100385375e-05,
+      "loss": 2.7182,
+      "step": 9610
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.7063087827094624e-05,
+      "loss": 2.9858,
+      "step": 9615
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.7056350553803866e-05,
+      "loss": 3.4233,
+      "step": 9620
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.7049613280513115e-05,
+      "loss": 3.2665,
+      "step": 9625
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.704287600722236e-05,
+      "loss": 2.9364,
+      "step": 9630
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.70361387339316e-05,
+      "loss": 3.0708,
+      "step": 9635
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.702940146064085e-05,
+      "loss": 3.078,
+      "step": 9640
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.70226641873501e-05,
+      "loss": 2.9575,
+      "step": 9645
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.7015926914059346e-05,
+      "loss": 3.0181,
+      "step": 9650
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.700918964076859e-05,
+      "loss": 2.7718,
+      "step": 9655
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.7002452367477837e-05,
+      "loss": 3.2755,
+      "step": 9660
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.6995715094187085e-05,
+      "loss": 3.126,
+      "step": 9665
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.6988977820896334e-05,
+      "loss": 3.04,
+      "step": 9670
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.6982240547605576e-05,
+      "loss": 3.1031,
+      "step": 9675
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.697550327431482e-05,
+      "loss": 3.0017,
+      "step": 9680
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.697011345568222e-05,
+      "loss": 3.3596,
+      "step": 9685
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.696337618239146e-05,
+      "loss": 3.1731,
+      "step": 9690
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.695663890910071e-05,
+      "loss": 2.9105,
+      "step": 9695
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.694990163580996e-05,
+      "loss": 3.0292,
+      "step": 9700
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.69431643625192e-05,
+      "loss": 3.1361,
+      "step": 9705
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.693642708922845e-05,
+      "loss": 3.1294,
+      "step": 9710
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.692968981593769e-05,
+      "loss": 3.1243,
+      "step": 9715
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.692295254264694e-05,
+      "loss": 3.0418,
+      "step": 9720
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6916215269356184e-05,
+      "loss": 3.1548,
+      "step": 9725
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.690947799606543e-05,
+      "loss": 2.8778,
+      "step": 9730
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.690274072277468e-05,
+      "loss": 3.1158,
+      "step": 9735
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.689600344948393e-05,
+      "loss": 2.9114,
+      "step": 9740
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.688926617619317e-05,
+      "loss": 3.0476,
+      "step": 9745
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.688252890290242e-05,
+      "loss": 3.1116,
+      "step": 9750
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6875791629611664e-05,
+      "loss": 3.1288,
+      "step": 9755
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6869054356320906e-05,
+      "loss": 3.2127,
+      "step": 9760
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6862317083030155e-05,
+      "loss": 3.1348,
+      "step": 9765
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6855579809739404e-05,
+      "loss": 3.189,
+      "step": 9770
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.684884253644865e-05,
+      "loss": 2.9306,
+      "step": 9775
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6842105263157895e-05,
+      "loss": 3.0042,
+      "step": 9780
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6835367989867144e-05,
+      "loss": 2.8858,
+      "step": 9785
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.682863071657639e-05,
+      "loss": 3.2717,
+      "step": 9790
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6821893443285635e-05,
+      "loss": 3.0997,
+      "step": 9795
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.681515616999488e-05,
+      "loss": 3.1203,
+      "step": 9800
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6808418896704126e-05,
+      "loss": 3.0215,
+      "step": 9805
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6801681623413374e-05,
+      "loss": 3.1261,
+      "step": 9810
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6794944350122617e-05,
+      "loss": 3.4258,
+      "step": 9815
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6788207076831865e-05,
+      "loss": 2.8897,
+      "step": 9820
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6781469803541114e-05,
+      "loss": 2.9942,
+      "step": 9825
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.677473253025036e-05,
+      "loss": 2.9707,
+      "step": 9830
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6767995256959605e-05,
+      "loss": 3.0694,
+      "step": 9835
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6761257983668854e-05,
+      "loss": 2.8014,
+      "step": 9840
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6754520710378096e-05,
+      "loss": 3.0555,
+      "step": 9845
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6747783437087345e-05,
+      "loss": 2.8841,
+      "step": 9850
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.674104616379659e-05,
+      "loss": 3.1467,
+      "step": 9855
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6734308890505836e-05,
+      "loss": 2.8715,
+      "step": 9860
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6727571617215085e-05,
+      "loss": 3.08,
+      "step": 9865
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.672083434392433e-05,
+      "loss": 3.231,
+      "step": 9870
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6714097070633576e-05,
+      "loss": 3.0236,
+      "step": 9875
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6707359797342825e-05,
+      "loss": 3.2113,
+      "step": 9880
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.670062252405207e-05,
+      "loss": 3.0357,
+      "step": 9885
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.669388525076131e-05,
+      "loss": 3.2755,
+      "step": 9890
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.668714797747056e-05,
+      "loss": 2.7913,
+      "step": 9895
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.668041070417981e-05,
+      "loss": 2.9929,
+      "step": 9900
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6673673430889056e-05,
+      "loss": 3.2487,
+      "step": 9905
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.66669361575983e-05,
+      "loss": 3.0298,
+      "step": 9910
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6660198884307547e-05,
+      "loss": 2.9926,
+      "step": 9915
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6653461611016795e-05,
+      "loss": 3.0167,
+      "step": 9920
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.664672433772604e-05,
+      "loss": 2.9078,
+      "step": 9925
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.663998706443528e-05,
+      "loss": 2.8785,
+      "step": 9930
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.663324979114453e-05,
+      "loss": 3.0155,
+      "step": 9935
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.662651251785378e-05,
+      "loss": 2.9471,
+      "step": 9940
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.661977524456302e-05,
+      "loss": 2.8238,
+      "step": 9945
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.661303797127227e-05,
+      "loss": 2.7744,
+      "step": 9950
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.660630069798152e-05,
+      "loss": 3.2174,
+      "step": 9955
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.659956342469076e-05,
+      "loss": 3.0096,
+      "step": 9960
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.659282615140001e-05,
+      "loss": 2.9241,
+      "step": 9965
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.658608887810926e-05,
+      "loss": 2.8246,
+      "step": 9970
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.65793516048185e-05,
+      "loss": 3.0677,
+      "step": 9975
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.657261433152774e-05,
+      "loss": 3.3918,
+      "step": 9980
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.656587705823699e-05,
+      "loss": 3.0951,
+      "step": 9985
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.655913978494624e-05,
+      "loss": 3.2887,
+      "step": 9990
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6553749966313634e-05,
+      "loss": 3.2198,
+      "step": 9995
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.654701269302288e-05,
+      "loss": 3.1275,
+      "step": 10000
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.654027541973213e-05,
+      "loss": 2.8183,
+      "step": 10005
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6533538146441374e-05,
+      "loss": 3.2344,
+      "step": 10010
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6526800873150616e-05,
+      "loss": 3.0728,
+      "step": 10015
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6520063599859865e-05,
+      "loss": 3.1439,
+      "step": 10020
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6513326326569114e-05,
+      "loss": 2.9468,
+      "step": 10025
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.650658905327836e-05,
+      "loss": 3.0711,
+      "step": 10030
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6499851779987605e-05,
+      "loss": 2.9164,
+      "step": 10035
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6493114506696854e-05,
+      "loss": 3.1862,
+      "step": 10040
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.64863772334061e-05,
+      "loss": 2.7662,
+      "step": 10045
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6479639960115344e-05,
+      "loss": 2.7516,
+      "step": 10050
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6472902686824587e-05,
+      "loss": 3.0381,
+      "step": 10055
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6466165413533835e-05,
+      "loss": 2.9114,
+      "step": 10060
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6459428140243084e-05,
+      "loss": 3.1425,
+      "step": 10065
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6452690866952326e-05,
+      "loss": 3.1069,
+      "step": 10070
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6445953593661575e-05,
+      "loss": 3.1094,
+      "step": 10075
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6439216320370824e-05,
+      "loss": 3.1806,
+      "step": 10080
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6432479047080066e-05,
+      "loss": 3.2412,
+      "step": 10085
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6425741773789315e-05,
+      "loss": 2.6815,
+      "step": 10090
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.641900450049856e-05,
+      "loss": 2.9824,
+      "step": 10095
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6412267227207806e-05,
+      "loss": 3.0131,
+      "step": 10100
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.640552995391705e-05,
+      "loss": 3.2256,
+      "step": 10105
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.63987926806263e-05,
+      "loss": 3.0638,
+      "step": 10110
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6392055407335546e-05,
+      "loss": 2.9276,
+      "step": 10115
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6385318134044795e-05,
+      "loss": 3.1751,
+      "step": 10120
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.637858086075404e-05,
+      "loss": 3.1735,
+      "step": 10125
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6371843587463286e-05,
+      "loss": 2.7972,
+      "step": 10130
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.636510631417253e-05,
+      "loss": 2.9994,
+      "step": 10135
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.635836904088178e-05,
+      "loss": 2.9632,
+      "step": 10140
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.635163176759102e-05,
+      "loss": 2.9621,
+      "step": 10145
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.634489449430027e-05,
+      "loss": 2.826,
+      "step": 10150
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.633815722100952e-05,
+      "loss": 3.0373,
+      "step": 10155
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.633141994771876e-05,
+      "loss": 2.8925,
+      "step": 10160
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.632468267442801e-05,
+      "loss": 3.0813,
+      "step": 10165
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6317945401137256e-05,
+      "loss": 2.8686,
+      "step": 10170
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6311208127846505e-05,
+      "loss": 3.0871,
+      "step": 10175
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.630447085455575e-05,
+      "loss": 2.9655,
+      "step": 10180
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.629773358126499e-05,
+      "loss": 2.9345,
+      "step": 10185
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.629099630797424e-05,
+      "loss": 3.0776,
+      "step": 10190
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.628425903468348e-05,
+      "loss": 3.0574,
+      "step": 10195
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.627752176139273e-05,
+      "loss": 2.9316,
+      "step": 10200
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.627078448810198e-05,
+      "loss": 2.8688,
+      "step": 10205
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.626404721481123e-05,
+      "loss": 2.9406,
+      "step": 10210
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.625730994152047e-05,
+      "loss": 3.0147,
+      "step": 10215
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.625057266822972e-05,
+      "loss": 3.0162,
+      "step": 10220
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.624383539493896e-05,
+      "loss": 2.9309,
+      "step": 10225
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.623709812164821e-05,
+      "loss": 3.2362,
+      "step": 10230
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.623036084835745e-05,
+      "loss": 3.1789,
+      "step": 10235
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.62236235750667e-05,
+      "loss": 2.7308,
+      "step": 10240
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.621688630177595e-05,
+      "loss": 2.8587,
+      "step": 10245
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.621014902848519e-05,
+      "loss": 2.9517,
+      "step": 10250
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.620341175519444e-05,
+      "loss": 2.9824,
+      "step": 10255
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.619667448190369e-05,
+      "loss": 3.0932,
+      "step": 10260
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.618993720861293e-05,
+      "loss": 3.3439,
+      "step": 10265
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.618319993532217e-05,
+      "loss": 3.1795,
+      "step": 10270
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.617646266203142e-05,
+      "loss": 3.0621,
+      "step": 10275
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.616972538874067e-05,
+      "loss": 2.8596,
+      "step": 10280
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.616298811544992e-05,
+      "loss": 3.0604,
+      "step": 10285
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.615625084215916e-05,
+      "loss": 3.0342,
+      "step": 10290
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.614951356886841e-05,
+      "loss": 2.9337,
+      "step": 10295
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.614277629557766e-05,
+      "loss": 3.0446,
+      "step": 10300
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.61360390222869e-05,
+      "loss": 2.8938,
+      "step": 10305
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.612930174899615e-05,
+      "loss": 2.8491,
+      "step": 10310
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.612256447570539e-05,
+      "loss": 2.9921,
+      "step": 10315
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.611582720241464e-05,
+      "loss": 3.0225,
+      "step": 10320
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6109089929123883e-05,
+      "loss": 2.9918,
+      "step": 10325
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.610235265583313e-05,
+      "loss": 3.1685,
+      "step": 10330
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.609561538254238e-05,
+      "loss": 3.1366,
+      "step": 10335
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.608887810925163e-05,
+      "loss": 3.1979,
+      "step": 10340
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.608214083596087e-05,
+      "loss": 3.2472,
+      "step": 10345
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.607540356267012e-05,
+      "loss": 3.0492,
+      "step": 10350
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.606866628937936e-05,
+      "loss": 3.0889,
+      "step": 10355
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6061929016088605e-05,
+      "loss": 3.1712,
+      "step": 10360
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6055191742797854e-05,
+      "loss": 3.1994,
+      "step": 10365
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.60484544695071e-05,
+      "loss": 2.9661,
+      "step": 10370
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.604171719621635e-05,
+      "loss": 3.0717,
+      "step": 10375
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6034979922925594e-05,
+      "loss": 2.963,
+      "step": 10380
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.602824264963484e-05,
+      "loss": 2.9531,
+      "step": 10385
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.602150537634409e-05,
+      "loss": 2.7738,
+      "step": 10390
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6014768103053334e-05,
+      "loss": 3.2369,
+      "step": 10395
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6008030829762576e-05,
+      "loss": 3.0199,
+      "step": 10400
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6001293556471825e-05,
+      "loss": 2.9024,
+      "step": 10405
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.5994556283181074e-05,
+      "loss": 3.4314,
+      "step": 10410
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.5987819009890316e-05,
+      "loss": 2.9359,
+      "step": 10415
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.5981081736599565e-05,
+      "loss": 3.0674,
+      "step": 10420
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.5974344463308813e-05,
+      "loss": 2.8982,
+      "step": 10425
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.596760719001806e-05,
+      "loss": 3.1686,
+      "step": 10430
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.5960869916727304e-05,
+      "loss": 2.855,
+      "step": 10435
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.595413264343655e-05,
+      "loss": 2.9604,
+      "step": 10440
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.5947395370145795e-05,
+      "loss": 2.9602,
+      "step": 10445
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.594065809685504e-05,
+      "loss": 3.1929,
+      "step": 10450
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5933920823564286e-05,
+      "loss": 3.1893,
+      "step": 10455
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5927183550273535e-05,
+      "loss": 3.0879,
+      "step": 10460
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5920446276982784e-05,
+      "loss": 2.8213,
+      "step": 10465
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5913709003692026e-05,
+      "loss": 3.1896,
+      "step": 10470
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5906971730401275e-05,
+      "loss": 2.8873,
+      "step": 10475
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5900234457110524e-05,
+      "loss": 2.7271,
+      "step": 10480
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5893497183819766e-05,
+      "loss": 2.9885,
+      "step": 10485
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.588675991052901e-05,
+      "loss": 2.9931,
+      "step": 10490
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.588002263723826e-05,
+      "loss": 3.0486,
+      "step": 10495
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5873285363947506e-05,
+      "loss": 2.7903,
+      "step": 10500
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.586654809065675e-05,
+      "loss": 2.9754,
+      "step": 10505
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5859810817366e-05,
+      "loss": 3.2095,
+      "step": 10510
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5853073544075246e-05,
+      "loss": 3.1133,
+      "step": 10515
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5846336270784495e-05,
+      "loss": 2.7682,
+      "step": 10520
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.583959899749374e-05,
+      "loss": 2.941,
+      "step": 10525
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.583286172420298e-05,
+      "loss": 3.0385,
+      "step": 10530
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.582612445091223e-05,
+      "loss": 2.8513,
+      "step": 10535
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5819387177621477e-05,
+      "loss": 3.1885,
+      "step": 10540
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.581264990433072e-05,
+      "loss": 3.2157,
+      "step": 10545
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.580591263103997e-05,
+      "loss": 2.9265,
+      "step": 10550
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5799175357749216e-05,
+      "loss": 2.6739,
+      "step": 10555
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.579243808445846e-05,
+      "loss": 2.8451,
+      "step": 10560
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.578570081116771e-05,
+      "loss": 2.9205,
+      "step": 10565
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.577896353787695e-05,
+      "loss": 2.6902,
+      "step": 10570
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.57722262645862e-05,
+      "loss": 2.9319,
+      "step": 10575
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.576548899129544e-05,
+      "loss": 3.5273,
+      "step": 10580
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.575875171800469e-05,
+      "loss": 3.2621,
+      "step": 10585
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.575201444471394e-05,
+      "loss": 2.8811,
+      "step": 10590
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.574527717142319e-05,
+      "loss": 3.0016,
+      "step": 10595
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.573853989813243e-05,
+      "loss": 3.0808,
+      "step": 10600
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.573180262484168e-05,
+      "loss": 3.0308,
+      "step": 10605
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.572506535155093e-05,
+      "loss": 2.9164,
+      "step": 10610
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.571832807826017e-05,
+      "loss": 2.8661,
+      "step": 10615
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.571159080496941e-05,
+      "loss": 3.1764,
+      "step": 10620
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.570485353167866e-05,
+      "loss": 2.9688,
+      "step": 10625
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.569811625838791e-05,
+      "loss": 2.8292,
+      "step": 10630
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.569137898509715e-05,
+      "loss": 3.1102,
+      "step": 10635
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.56846417118064e-05,
+      "loss": 2.9934,
+      "step": 10640
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.567790443851565e-05,
+      "loss": 3.2563,
+      "step": 10645
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.56711671652249e-05,
+      "loss": 2.8752,
+      "step": 10650
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.566442989193414e-05,
+      "loss": 2.9797,
+      "step": 10655
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.565769261864338e-05,
+      "loss": 2.8909,
+      "step": 10660
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.565095534535263e-05,
+      "loss": 3.0765,
+      "step": 10665
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.564421807206187e-05,
+      "loss": 2.8784,
+      "step": 10670
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.563748079877112e-05,
+      "loss": 3.0186,
+      "step": 10675
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.563074352548037e-05,
+      "loss": 3.0764,
+      "step": 10680
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.562400625218962e-05,
+      "loss": 3.015,
+      "step": 10685
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.561726897889886e-05,
+      "loss": 2.9773,
+      "step": 10690
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.561053170560811e-05,
+      "loss": 2.9843,
+      "step": 10695
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.560379443231735e-05,
+      "loss": 2.8827,
+      "step": 10700
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5597057159026595e-05,
+      "loss": 3.1021,
+      "step": 10705
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.559031988573584e-05,
+      "loss": 3.0825,
+      "step": 10710
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.558358261244509e-05,
+      "loss": 2.9813,
+      "step": 10715
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.557684533915434e-05,
+      "loss": 2.8546,
+      "step": 10720
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.557010806586358e-05,
+      "loss": 3.0049,
+      "step": 10725
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.556337079257283e-05,
+      "loss": 2.6026,
+      "step": 10730
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.555663351928208e-05,
+      "loss": 3.0555,
+      "step": 10735
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.554989624599133e-05,
+      "loss": 2.9656,
+      "step": 10740
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.554315897270057e-05,
+      "loss": 2.9598,
+      "step": 10745
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5536421699409814e-05,
+      "loss": 2.9622,
+      "step": 10750
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.552968442611906e-05,
+      "loss": 3.2513,
+      "step": 10755
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5522947152828305e-05,
+      "loss": 3.0568,
+      "step": 10760
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5516209879537554e-05,
+      "loss": 3.1684,
+      "step": 10765
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.55094726062468e-05,
+      "loss": 3.0154,
+      "step": 10770
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.550273533295605e-05,
+      "loss": 2.8607,
+      "step": 10775
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5495998059665294e-05,
+      "loss": 2.953,
+      "step": 10780
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.548926078637454e-05,
+      "loss": 3.1728,
+      "step": 10785
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5482523513083785e-05,
+      "loss": 2.8006,
+      "step": 10790
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5475786239793034e-05,
+      "loss": 3.0427,
+      "step": 10795
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5469048966502276e-05,
+      "loss": 3.0487,
+      "step": 10800
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5462311693211525e-05,
+      "loss": 3.1628,
+      "step": 10805
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5455574419920773e-05,
+      "loss": 2.8268,
+      "step": 10810
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5448837146630015e-05,
+      "loss": 2.9314,
+      "step": 10815
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5442099873339264e-05,
+      "loss": 2.9674,
+      "step": 10820
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.543536260004851e-05,
+      "loss": 3.027,
+      "step": 10825
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5428625326757755e-05,
+      "loss": 3.0957,
+      "step": 10830
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5421888053467e-05,
+      "loss": 2.9217,
+      "step": 10835
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5415150780176246e-05,
+      "loss": 2.8224,
+      "step": 10840
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5408413506885495e-05,
+      "loss": 2.9139,
+      "step": 10845
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5401676233594744e-05,
+      "loss": 2.8247,
+      "step": 10850
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5394938960303986e-05,
+      "loss": 2.9044,
+      "step": 10855
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5388201687013235e-05,
+      "loss": 2.8095,
+      "step": 10860
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5381464413722484e-05,
+      "loss": 3.1797,
+      "step": 10865
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5374727140431726e-05,
+      "loss": 2.7943,
+      "step": 10870
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5367989867140975e-05,
+      "loss": 2.8645,
+      "step": 10875
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.536125259385022e-05,
+      "loss": 2.8684,
+      "step": 10880
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5354515320559466e-05,
+      "loss": 3.0959,
+      "step": 10885
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.534777804726871e-05,
+      "loss": 3.2173,
+      "step": 10890
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.534104077397796e-05,
+      "loss": 2.8572,
+      "step": 10895
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5334303500687206e-05,
+      "loss": 2.8573,
+      "step": 10900
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5327566227396455e-05,
+      "loss": 3.1226,
+      "step": 10905
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.53208289541057e-05,
+      "loss": 2.8455,
+      "step": 10910
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5314091680814946e-05,
+      "loss": 3.0792,
+      "step": 10915
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.530735440752419e-05,
+      "loss": 3.1342,
+      "step": 10920
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.530061713423343e-05,
+      "loss": 2.9695,
+      "step": 10925
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.529387986094268e-05,
+      "loss": 2.9094,
+      "step": 10930
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.528714258765193e-05,
+      "loss": 2.8054,
+      "step": 10935
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5280405314361176e-05,
+      "loss": 3.2511,
+      "step": 10940
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.527366804107042e-05,
+      "loss": 3.067,
+      "step": 10945
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.526693076777967e-05,
+      "loss": 3.0682,
+      "step": 10950
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5260193494488916e-05,
+      "loss": 3.2405,
+      "step": 10955
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.525345622119816e-05,
+      "loss": 2.8744,
+      "step": 10960
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.52467189479074e-05,
+      "loss": 2.8158,
+      "step": 10965
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.523998167461665e-05,
+      "loss": 3.1325,
+      "step": 10970
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.52332444013259e-05,
+      "loss": 2.8495,
+      "step": 10975
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.522650712803514e-05,
+      "loss": 2.9472,
+      "step": 10980
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.521976985474439e-05,
+      "loss": 3.1621,
+      "step": 10985
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.521303258145364e-05,
+      "loss": 2.998,
+      "step": 10990
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.520629530816289e-05,
+      "loss": 2.9869,
+      "step": 10995
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.519955803487213e-05,
+      "loss": 3.0895,
+      "step": 11000
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.519282076158137e-05,
+      "loss": 2.8368,
+      "step": 11005
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.518608348829062e-05,
+      "loss": 3.1186,
+      "step": 11010
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.517934621499986e-05,
+      "loss": 2.9144,
+      "step": 11015
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.517260894170911e-05,
+      "loss": 3.1524,
+      "step": 11020
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.516587166841836e-05,
+      "loss": 2.7921,
+      "step": 11025
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.515913439512761e-05,
+      "loss": 2.9172,
+      "step": 11030
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.515239712183685e-05,
+      "loss": 3.1107,
+      "step": 11035
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.51456598485461e-05,
+      "loss": 3.0204,
+      "step": 11040
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.513892257525535e-05,
+      "loss": 2.9076,
+      "step": 11045
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.513218530196459e-05,
+      "loss": 2.746,
+      "step": 11050
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.512544802867383e-05,
+      "loss": 3.1643,
+      "step": 11055
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.511871075538308e-05,
+      "loss": 2.9974,
+      "step": 11060
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.511197348209233e-05,
+      "loss": 2.9475,
+      "step": 11065
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.510523620880157e-05,
+      "loss": 2.8503,
+      "step": 11070
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.509849893551082e-05,
+      "loss": 2.8255,
+      "step": 11075
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.509176166222007e-05,
+      "loss": 3.0081,
+      "step": 11080
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.508502438892932e-05,
+      "loss": 3.0882,
+      "step": 11085
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.507828711563856e-05,
+      "loss": 3.3281,
+      "step": 11090
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.50715498423478e-05,
+      "loss": 2.7693,
+      "step": 11095
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.506481256905705e-05,
+      "loss": 2.9684,
+      "step": 11100
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.50580752957663e-05,
+      "loss": 2.8099,
+      "step": 11105
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.505133802247554e-05,
+      "loss": 2.9284,
+      "step": 11110
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.504460074918479e-05,
+      "loss": 3.0644,
+      "step": 11115
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.503786347589404e-05,
+      "loss": 3.0546,
+      "step": 11120
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.503112620260328e-05,
+      "loss": 3.0909,
+      "step": 11125
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.502438892931253e-05,
+      "loss": 3.0999,
+      "step": 11130
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5017651656021774e-05,
+      "loss": 2.9832,
+      "step": 11135
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.501091438273102e-05,
+      "loss": 3.2219,
+      "step": 11140
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5004177109440265e-05,
+      "loss": 2.7767,
+      "step": 11145
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.4997439836149514e-05,
+      "loss": 3.1374,
+      "step": 11150
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.499070256285876e-05,
+      "loss": 2.6789,
+      "step": 11155
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.498396528956801e-05,
+      "loss": 2.7915,
+      "step": 11160
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.4977228016277254e-05,
+      "loss": 3.1062,
+      "step": 11165
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.49704907429865e-05,
+      "loss": 2.7951,
+      "step": 11170
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.496375346969575e-05,
+      "loss": 3.0343,
+      "step": 11175
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.4957016196404993e-05,
+      "loss": 2.9418,
+      "step": 11180
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.4950278923114236e-05,
+      "loss": 2.9457,
+      "step": 11185
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.4943541649823484e-05,
+      "loss": 2.9444,
+      "step": 11190
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.493680437653273e-05,
+      "loss": 2.9274,
+      "step": 11195
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4930067103241975e-05,
+      "loss": 2.9349,
+      "step": 11200
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4923329829951224e-05,
+      "loss": 2.8807,
+      "step": 11205
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.491659255666047e-05,
+      "loss": 3.0897,
+      "step": 11210
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.490985528336972e-05,
+      "loss": 3.0386,
+      "step": 11215
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4903118010078964e-05,
+      "loss": 3.0223,
+      "step": 11220
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4896380736788206e-05,
+      "loss": 2.9861,
+      "step": 11225
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4889643463497455e-05,
+      "loss": 3.4111,
+      "step": 11230
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.48829061902067e-05,
+      "loss": 2.8458,
+      "step": 11235
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4876168916915946e-05,
+      "loss": 3.1452,
+      "step": 11240
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4869431643625195e-05,
+      "loss": 3.0766,
+      "step": 11245
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4862694370334444e-05,
+      "loss": 3.0981,
+      "step": 11250
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4855957097043686e-05,
+      "loss": 2.9809,
+      "step": 11255
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4849219823752935e-05,
+      "loss": 2.9177,
+      "step": 11260
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.484248255046218e-05,
+      "loss": 2.9045,
+      "step": 11265
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.483574527717142e-05,
+      "loss": 3.1389,
+      "step": 11270
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.482900800388067e-05,
+      "loss": 3.273,
+      "step": 11275
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.482227073058992e-05,
+      "loss": 3.0738,
+      "step": 11280
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4815533457299166e-05,
+      "loss": 3.0059,
+      "step": 11285
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.480879618400841e-05,
+      "loss": 3.1052,
+      "step": 11290
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4802058910717657e-05,
+      "loss": 2.9186,
+      "step": 11295
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4795321637426905e-05,
+      "loss": 3.4463,
+      "step": 11300
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.478858436413615e-05,
+      "loss": 2.9143,
+      "step": 11305
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4781847090845396e-05,
+      "loss": 3.0278,
+      "step": 11310
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.477510981755464e-05,
+      "loss": 2.7317,
+      "step": 11315
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.476837254426389e-05,
+      "loss": 3.2409,
+      "step": 11320
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.476163527097313e-05,
+      "loss": 2.8277,
+      "step": 11325
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.475489799768238e-05,
+      "loss": 2.7135,
+      "step": 11330
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.474816072439163e-05,
+      "loss": 2.9345,
+      "step": 11335
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4741423451100876e-05,
+      "loss": 3.0272,
+      "step": 11340
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.473468617781012e-05,
+      "loss": 3.1851,
+      "step": 11345
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.472794890451937e-05,
+      "loss": 3.0905,
+      "step": 11350
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.472121163122861e-05,
+      "loss": 3.0061,
+      "step": 11355
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.471447435793786e-05,
+      "loss": 3.1029,
+      "step": 11360
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.47077370846471e-05,
+      "loss": 3.0591,
+      "step": 11365
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.470099981135635e-05,
+      "loss": 3.0617,
+      "step": 11370
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.46942625380656e-05,
+      "loss": 3.0491,
+      "step": 11375
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.468752526477484e-05,
+      "loss": 3.0714,
+      "step": 11380
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.468078799148409e-05,
+      "loss": 2.9701,
+      "step": 11385
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.467405071819334e-05,
+      "loss": 3.2235,
+      "step": 11390
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.466731344490258e-05,
+      "loss": 2.8136,
+      "step": 11395
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.466057617161182e-05,
+      "loss": 2.9353,
+      "step": 11400
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.465383889832107e-05,
+      "loss": 3.0163,
+      "step": 11405
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.464710162503032e-05,
+      "loss": 2.9738,
+      "step": 11410
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.464036435173957e-05,
+      "loss": 3.0965,
+      "step": 11415
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.463362707844881e-05,
+      "loss": 2.955,
+      "step": 11420
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.462688980515806e-05,
+      "loss": 2.8734,
+      "step": 11425
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.462015253186731e-05,
+      "loss": 3.0841,
+      "step": 11430
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.461341525857655e-05,
+      "loss": 3.2559,
+      "step": 11435
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.460667798528579e-05,
+      "loss": 2.9961,
+      "step": 11440
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.459994071199504e-05,
+      "loss": 3.0601,
+      "step": 11445
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.459320343870429e-05,
+      "loss": 3.4935,
+      "step": 11450
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.458646616541353e-05,
+      "loss": 2.9135,
+      "step": 11455
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.457972889212278e-05,
+      "loss": 2.9466,
+      "step": 11460
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.457299161883203e-05,
+      "loss": 3.0844,
+      "step": 11465
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.456625434554128e-05,
+      "loss": 3.1374,
+      "step": 11470
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.455951707225052e-05,
+      "loss": 2.9251,
+      "step": 11475
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.455277979895977e-05,
+      "loss": 3.0553,
+      "step": 11480
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.454604252566901e-05,
+      "loss": 3.1865,
+      "step": 11485
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4539305252378254e-05,
+      "loss": 2.8396,
+      "step": 11490
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4533915433745656e-05,
+      "loss": 3.1356,
+      "step": 11495
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4527178160454905e-05,
+      "loss": 2.9385,
+      "step": 11500
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.452044088716415e-05,
+      "loss": 2.909,
+      "step": 11505
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4513703613873396e-05,
+      "loss": 3.1132,
+      "step": 11510
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4506966340582645e-05,
+      "loss": 2.9435,
+      "step": 11515
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.450157652195003e-05,
+      "loss": 3.6755,
+      "step": 11520
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.449483924865928e-05,
+      "loss": 3.2154,
+      "step": 11525
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.448810197536853e-05,
+      "loss": 3.0608,
+      "step": 11530
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.448136470207778e-05,
+      "loss": 3.012,
+      "step": 11535
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.447462742878702e-05,
+      "loss": 3.1076,
+      "step": 11540
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.446789015549627e-05,
+      "loss": 2.9761,
+      "step": 11545
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.446115288220552e-05,
+      "loss": 2.9443,
+      "step": 11550
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.445441560891476e-05,
+      "loss": 2.9833,
+      "step": 11555
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4447678335624004e-05,
+      "loss": 2.9272,
+      "step": 11560
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.444094106233325e-05,
+      "loss": 2.7781,
+      "step": 11565
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.44342037890425e-05,
+      "loss": 2.8401,
+      "step": 11570
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4427466515751743e-05,
+      "loss": 3.3159,
+      "step": 11575
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.442072924246099e-05,
+      "loss": 3.2537,
+      "step": 11580
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.441399196917024e-05,
+      "loss": 2.7323,
+      "step": 11585
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.440725469587949e-05,
+      "loss": 2.8754,
+      "step": 11590
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.440051742258873e-05,
+      "loss": 2.7398,
+      "step": 11595
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4393780149297974e-05,
+      "loss": 3.2692,
+      "step": 11600
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.438704287600722e-05,
+      "loss": 2.9441,
+      "step": 11605
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.438030560271647e-05,
+      "loss": 3.0601,
+      "step": 11610
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4373568329425714e-05,
+      "loss": 2.9512,
+      "step": 11615
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.436683105613496e-05,
+      "loss": 3.04,
+      "step": 11620
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.436009378284421e-05,
+      "loss": 2.9893,
+      "step": 11625
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4353356509553454e-05,
+      "loss": 2.9824,
+      "step": 11630
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.43466192362627e-05,
+      "loss": 2.9835,
+      "step": 11635
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4339881962971945e-05,
+      "loss": 2.8617,
+      "step": 11640
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4333144689681194e-05,
+      "loss": 2.8856,
+      "step": 11645
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4326407416390436e-05,
+      "loss": 3.131,
+      "step": 11650
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4319670143099685e-05,
+      "loss": 2.846,
+      "step": 11655
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4312932869808934e-05,
+      "loss": 3.2464,
+      "step": 11660
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.430619559651818e-05,
+      "loss": 3.1764,
+      "step": 11665
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4299458323227425e-05,
+      "loss": 3.0205,
+      "step": 11670
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4292721049936674e-05,
+      "loss": 3.0125,
+      "step": 11675
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.428598377664592e-05,
+      "loss": 3.0583,
+      "step": 11680
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4279246503355164e-05,
+      "loss": 3.2121,
+      "step": 11685
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4272509230064407e-05,
+      "loss": 3.0483,
+      "step": 11690
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4265771956773655e-05,
+      "loss": 2.9595,
+      "step": 11695
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4259034683482904e-05,
+      "loss": 2.8907,
+      "step": 11700
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4252297410192146e-05,
+      "loss": 3.2868,
+      "step": 11705
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4245560136901395e-05,
+      "loss": 2.9464,
+      "step": 11710
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4238822863610644e-05,
+      "loss": 3.4274,
+      "step": 11715
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.423208559031989e-05,
+      "loss": 3.1743,
+      "step": 11720
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4225348317029135e-05,
+      "loss": 3.2586,
+      "step": 11725
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.421861104373838e-05,
+      "loss": 2.7091,
+      "step": 11730
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4211873770447626e-05,
+      "loss": 2.7987,
+      "step": 11735
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.420513649715687e-05,
+      "loss": 2.9936,
+      "step": 11740
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.419839922386612e-05,
+      "loss": 3.0006,
+      "step": 11745
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4191661950575366e-05,
+      "loss": 3.2487,
+      "step": 11750
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4184924677284615e-05,
+      "loss": 2.9673,
+      "step": 11755
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.417818740399386e-05,
+      "loss": 2.9303,
+      "step": 11760
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4171450130703106e-05,
+      "loss": 2.7594,
+      "step": 11765
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.416471285741235e-05,
+      "loss": 2.9743,
+      "step": 11770
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.41579755841216e-05,
+      "loss": 2.7895,
+      "step": 11775
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.415123831083084e-05,
+      "loss": 3.0708,
+      "step": 11780
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.414450103754009e-05,
+      "loss": 2.9679,
+      "step": 11785
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.413776376424934e-05,
+      "loss": 2.998,
+      "step": 11790
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.413102649095858e-05,
+      "loss": 2.9087,
+      "step": 11795
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.412428921766783e-05,
+      "loss": 2.9266,
+      "step": 11800
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4117551944377076e-05,
+      "loss": 3.1942,
+      "step": 11805
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.411081467108632e-05,
+      "loss": 3.1146,
+      "step": 11810
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.410407739779557e-05,
+      "loss": 3.1195,
+      "step": 11815
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.409734012450481e-05,
+      "loss": 2.8045,
+      "step": 11820
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.409060285121406e-05,
+      "loss": 3.0467,
+      "step": 11825
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.40838655779233e-05,
+      "loss": 2.8286,
+      "step": 11830
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.407712830463255e-05,
+      "loss": 3.0586,
+      "step": 11835
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.40703910313418e-05,
+      "loss": 2.9959,
+      "step": 11840
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.406365375805105e-05,
+      "loss": 3.0457,
+      "step": 11845
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.405691648476029e-05,
+      "loss": 2.9982,
+      "step": 11850
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.405017921146954e-05,
+      "loss": 2.8829,
+      "step": 11855
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.404344193817878e-05,
+      "loss": 3.3715,
+      "step": 11860
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.403670466488803e-05,
+      "loss": 3.0459,
+      "step": 11865
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.402996739159727e-05,
+      "loss": 2.8927,
+      "step": 11870
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.402323011830652e-05,
+      "loss": 2.9431,
+      "step": 11875
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.401649284501577e-05,
+      "loss": 2.9148,
+      "step": 11880
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.400975557172501e-05,
+      "loss": 2.9739,
+      "step": 11885
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.400301829843426e-05,
+      "loss": 2.8028,
+      "step": 11890
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.399628102514351e-05,
+      "loss": 2.9279,
+      "step": 11895
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.398954375185275e-05,
+      "loss": 3.0049,
+      "step": 11900
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.398280647856199e-05,
+      "loss": 3.0606,
+      "step": 11905
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.397606920527124e-05,
+      "loss": 2.9626,
+      "step": 11910
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.396933193198049e-05,
+      "loss": 3.2141,
+      "step": 11915
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.396259465868974e-05,
+      "loss": 2.9912,
+      "step": 11920
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.395585738539898e-05,
+      "loss": 2.9204,
+      "step": 11925
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.394912011210823e-05,
+      "loss": 2.7776,
+      "step": 11930
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.394238283881748e-05,
+      "loss": 3.2191,
+      "step": 11935
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.393564556552672e-05,
+      "loss": 2.8625,
+      "step": 11940
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3928908292235964e-05,
+      "loss": 3.0099,
+      "step": 11945
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.392217101894521e-05,
+      "loss": 3.1315,
+      "step": 11950
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.391543374565446e-05,
+      "loss": 3.0517,
+      "step": 11955
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3908696472363703e-05,
+      "loss": 3.1481,
+      "step": 11960
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.390195919907295e-05,
+      "loss": 2.9481,
+      "step": 11965
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.38952219257822e-05,
+      "loss": 3.0222,
+      "step": 11970
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.388848465249145e-05,
+      "loss": 3.0445,
+      "step": 11975
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.388174737920069e-05,
+      "loss": 3.1314,
+      "step": 11980
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.387501010590994e-05,
+      "loss": 2.888,
+      "step": 11985
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.386827283261918e-05,
+      "loss": 3.1614,
+      "step": 11990
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3861535559328425e-05,
+      "loss": 3.1802,
+      "step": 11995
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3854798286037674e-05,
+      "loss": 3.0269,
+      "step": 12000
+    },
+    {
+      "epoch": 0.97,
+      "eval_loss": 2.9837048053741455,
+      "eval_rouge2_fmeasure": 0.0042,
+      "eval_rouge2_precision": 0.01,
+      "eval_rouge2_recall": 0.0033,
+      "eval_runtime": 2865.3591,
+      "eval_samples_per_second": 0.096,
+      "eval_steps_per_second": 0.048,
+      "step": 12000
     }
   ],
   "max_steps": 37107,
   "num_train_epochs": 3,
-  "total_flos": 3.12180234780672e+17,
+  "total_flos": 4.68270352171008e+17,
   "trial_name": null,
   "trial_params": null
 }