diff --git "a/trainer_state.json" "b/trainer_state.json"
--- "a/trainer_state.json"
+++ "b/trainer_state.json"
@@ -1,8 +1,8 @@
 {
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 9.99997716868914,
-  "global_step": 218997,
+  "epoch": 13.99997716868914,
+  "global_step": 306596,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
@@ -131520,11 +131520,52619 @@
       "eval_steps_per_second": 1.307,
       "eval_wer": 0.9924314096499527,
       "step": 218997
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 3.335795999817205e-05,
+      "loss": 0.9012,
+      "step": 219000
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 3.335719834874404e-05,
+      "loss": 0.831,
+      "step": 219010
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 3.335643669931604e-05,
+      "loss": 0.8136,
+      "step": 219020
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 3.335567504988804e-05,
+      "loss": 0.7821,
+      "step": 219030
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 3.335491340046004e-05,
+      "loss": 0.8538,
+      "step": 219040
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 3.3354151751032035e-05,
+      "loss": 0.8824,
+      "step": 219050
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 3.335339010160404e-05,
+      "loss": 0.8587,
+      "step": 219060
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 3.335262845217603e-05,
+      "loss": 0.8558,
+      "step": 219070
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 3.3351866802748034e-05,
+      "loss": 0.8952,
+      "step": 219080
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 3.335110515332003e-05,
+      "loss": 0.8509,
+      "step": 219090
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 3.335034350389203e-05,
+      "loss": 0.8043,
+      "step": 219100
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 3.3349581854464026e-05,
+      "loss": 0.8634,
+      "step": 219110
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 3.334882020503603e-05,
+      "loss": 0.8458,
+      "step": 219120
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 3.334805855560803e-05,
+      "loss": 0.7927,
+      "step": 219130
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 3.3347296906180025e-05,
+      "loss": 0.8503,
+      "step": 219140
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 3.334653525675202e-05,
+      "loss": 0.8861,
+      "step": 219150
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 3.334577360732402e-05,
+      "loss": 0.795,
+      "step": 219160
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 3.3345011957896025e-05,
+      "loss": 0.8075,
+      "step": 219170
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 3.334425030846802e-05,
+      "loss": 0.9403,
+      "step": 219180
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 3.334348865904002e-05,
+      "loss": 0.8057,
+      "step": 219190
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 3.334272700961202e-05,
+      "loss": 0.88,
+      "step": 219200
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 3.334196536018402e-05,
+      "loss": 0.9273,
+      "step": 219210
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 3.334120371075601e-05,
+      "loss": 0.769,
+      "step": 219220
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 3.3340442061328016e-05,
+      "loss": 0.8413,
+      "step": 219230
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 3.3339680411900014e-05,
+      "loss": 0.7399,
+      "step": 219240
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 3.333891876247201e-05,
+      "loss": 0.8439,
+      "step": 219250
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 3.333815711304401e-05,
+      "loss": 0.7381,
+      "step": 219260
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 3.3337395463616006e-05,
+      "loss": 0.771,
+      "step": 219270
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 3.3336633814188004e-05,
+      "loss": 0.8658,
+      "step": 219280
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 3.333587216476001e-05,
+      "loss": 0.8658,
+      "step": 219290
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 3.3335110515332005e-05,
+      "loss": 0.8586,
+      "step": 219300
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 3.3334348865904e-05,
+      "loss": 0.8235,
+      "step": 219310
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 3.3333587216476e-05,
+      "loss": 0.8622,
+      "step": 219320
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 3.3332825567048e-05,
+      "loss": 0.8433,
+      "step": 219330
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 3.333206391762e-05,
+      "loss": 0.836,
+      "step": 219340
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 3.3331302268192e-05,
+      "loss": 0.8262,
+      "step": 219350
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 3.3330540618764e-05,
+      "loss": 0.8696,
+      "step": 219360
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 3.3329778969335994e-05,
+      "loss": 0.8863,
+      "step": 219370
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 3.3329017319908e-05,
+      "loss": 0.8375,
+      "step": 219380
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 3.332825567047999e-05,
+      "loss": 0.7529,
+      "step": 219390
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 3.3327494021051993e-05,
+      "loss": 0.8906,
+      "step": 219400
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 3.332673237162399e-05,
+      "loss": 0.8056,
+      "step": 219410
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 3.332597072219599e-05,
+      "loss": 0.7884,
+      "step": 219420
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 3.3325209072767986e-05,
+      "loss": 0.809,
+      "step": 219430
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 3.332444742333999e-05,
+      "loss": 0.7642,
+      "step": 219440
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 3.332368577391198e-05,
+      "loss": 0.8326,
+      "step": 219450
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 3.3322924124483985e-05,
+      "loss": 0.8249,
+      "step": 219460
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 3.332216247505598e-05,
+      "loss": 0.8799,
+      "step": 219470
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 3.332140082562798e-05,
+      "loss": 0.8816,
+      "step": 219480
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 3.332063917619998e-05,
+      "loss": 0.9649,
+      "step": 219490
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 3.331987752677198e-05,
+      "loss": 0.7322,
+      "step": 219500
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 3.331911587734398e-05,
+      "loss": 0.9043,
+      "step": 219510
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 3.331835422791598e-05,
+      "loss": 0.7784,
+      "step": 219520
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 3.3317592578487974e-05,
+      "loss": 0.8,
+      "step": 219530
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 3.331683092905997e-05,
+      "loss": 0.8272,
+      "step": 219540
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 3.3316069279631976e-05,
+      "loss": 0.8958,
+      "step": 219550
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 3.331530763020397e-05,
+      "loss": 0.8245,
+      "step": 219560
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 3.331454598077597e-05,
+      "loss": 0.8047,
+      "step": 219570
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 3.331378433134797e-05,
+      "loss": 0.8231,
+      "step": 219580
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 3.331302268191997e-05,
+      "loss": 0.7882,
+      "step": 219590
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 3.331226103249196e-05,
+      "loss": 0.9181,
+      "step": 219600
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 3.331149938306397e-05,
+      "loss": 0.8761,
+      "step": 219610
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 3.3310737733635965e-05,
+      "loss": 0.9225,
+      "step": 219620
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 3.330997608420796e-05,
+      "loss": 0.8974,
+      "step": 219630
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 3.330921443477996e-05,
+      "loss": 0.8208,
+      "step": 219640
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 3.3308452785351964e-05,
+      "loss": 0.8089,
+      "step": 219650
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 3.3307691135923955e-05,
+      "loss": 0.7505,
+      "step": 219660
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 3.330692948649596e-05,
+      "loss": 0.807,
+      "step": 219670
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 3.3306167837067956e-05,
+      "loss": 0.8053,
+      "step": 219680
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 3.3305406187639954e-05,
+      "loss": 0.9031,
+      "step": 219690
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 3.330464453821195e-05,
+      "loss": 0.8592,
+      "step": 219700
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 3.330388288878395e-05,
+      "loss": 0.7965,
+      "step": 219710
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 3.330312123935595e-05,
+      "loss": 0.7859,
+      "step": 219720
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 3.330235958992795e-05,
+      "loss": 0.7567,
+      "step": 219730
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 3.330159794049995e-05,
+      "loss": 0.9078,
+      "step": 219740
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 3.3300836291071945e-05,
+      "loss": 0.8551,
+      "step": 219750
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 3.330007464164395e-05,
+      "loss": 0.8601,
+      "step": 219760
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 3.329931299221594e-05,
+      "loss": 0.7908,
+      "step": 219770
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 3.3298551342787945e-05,
+      "loss": 0.9963,
+      "step": 219780
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 3.329778969335994e-05,
+      "loss": 0.8588,
+      "step": 219790
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 3.329702804393194e-05,
+      "loss": 0.8367,
+      "step": 219800
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 3.329626639450394e-05,
+      "loss": 0.8581,
+      "step": 219810
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 3.329550474507594e-05,
+      "loss": 0.8419,
+      "step": 219820
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 3.329474309564793e-05,
+      "loss": 0.8087,
+      "step": 219830
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 3.3293981446219936e-05,
+      "loss": 0.8231,
+      "step": 219840
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 3.3293219796791934e-05,
+      "loss": 0.82,
+      "step": 219850
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 3.329245814736393e-05,
+      "loss": 0.8424,
+      "step": 219860
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 3.329169649793593e-05,
+      "loss": 0.9256,
+      "step": 219870
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 3.329093484850793e-05,
+      "loss": 0.7529,
+      "step": 219880
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 3.329017319907993e-05,
+      "loss": 0.8053,
+      "step": 219890
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 3.328941154965193e-05,
+      "loss": 0.7805,
+      "step": 219900
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 3.3288649900223925e-05,
+      "loss": 0.7891,
+      "step": 219910
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 3.328788825079592e-05,
+      "loss": 0.9267,
+      "step": 219920
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 3.328712660136793e-05,
+      "loss": 0.8321,
+      "step": 219930
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 3.3286364951939924e-05,
+      "loss": 0.8146,
+      "step": 219940
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 3.328560330251192e-05,
+      "loss": 0.9695,
+      "step": 219950
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 3.328484165308392e-05,
+      "loss": 0.8715,
+      "step": 219960
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 3.3284080003655924e-05,
+      "loss": 0.8286,
+      "step": 219970
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 3.3283318354227914e-05,
+      "loss": 0.7885,
+      "step": 219980
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 3.328255670479992e-05,
+      "loss": 0.8541,
+      "step": 219990
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 3.3281795055371916e-05,
+      "loss": 0.7888,
+      "step": 220000
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 3.3281033405943913e-05,
+      "loss": 0.8295,
+      "step": 220010
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 3.328027175651591e-05,
+      "loss": 0.8415,
+      "step": 220020
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 3.3279510107087915e-05,
+      "loss": 0.8343,
+      "step": 220030
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 3.3278748457659906e-05,
+      "loss": 0.8375,
+      "step": 220040
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 3.327798680823191e-05,
+      "loss": 0.8179,
+      "step": 220050
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 3.327722515880391e-05,
+      "loss": 0.7844,
+      "step": 220060
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 3.3276463509375905e-05,
+      "loss": 0.8331,
+      "step": 220070
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 3.32757018599479e-05,
+      "loss": 0.8348,
+      "step": 220080
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 3.327494021051991e-05,
+      "loss": 0.9642,
+      "step": 220090
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 3.3274178561091904e-05,
+      "loss": 0.8441,
+      "step": 220100
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 3.32734169116639e-05,
+      "loss": 0.9004,
+      "step": 220110
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 3.32726552622359e-05,
+      "loss": 0.7862,
+      "step": 220120
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 3.32718936128079e-05,
+      "loss": 0.8176,
+      "step": 220130
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 3.32711319633799e-05,
+      "loss": 0.8935,
+      "step": 220140
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 3.32703703139519e-05,
+      "loss": 0.7984,
+      "step": 220150
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 3.3269608664523896e-05,
+      "loss": 0.8133,
+      "step": 220160
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 3.326884701509589e-05,
+      "loss": 0.8943,
+      "step": 220170
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 3.326808536566789e-05,
+      "loss": 0.9029,
+      "step": 220180
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 3.326732371623989e-05,
+      "loss": 0.8238,
+      "step": 220190
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 3.326656206681189e-05,
+      "loss": 0.7973,
+      "step": 220200
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 3.326580041738388e-05,
+      "loss": 0.8087,
+      "step": 220210
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 3.326503876795589e-05,
+      "loss": 0.9461,
+      "step": 220220
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 3.3264277118527885e-05,
+      "loss": 0.8329,
+      "step": 220230
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 3.326351546909988e-05,
+      "loss": 0.8782,
+      "step": 220240
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 3.326275381967188e-05,
+      "loss": 0.9228,
+      "step": 220250
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 3.3261992170243884e-05,
+      "loss": 0.7554,
+      "step": 220260
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 3.3261230520815875e-05,
+      "loss": 0.9912,
+      "step": 220270
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 3.326046887138788e-05,
+      "loss": 0.7711,
+      "step": 220280
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 3.3259707221959876e-05,
+      "loss": 0.7996,
+      "step": 220290
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 3.3258945572531874e-05,
+      "loss": 0.7475,
+      "step": 220300
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 3.325818392310388e-05,
+      "loss": 0.8617,
+      "step": 220310
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 3.3257422273675876e-05,
+      "loss": 0.8815,
+      "step": 220320
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 3.325666062424787e-05,
+      "loss": 0.8024,
+      "step": 220330
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 3.325589897481987e-05,
+      "loss": 0.8438,
+      "step": 220340
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 3.3255137325391875e-05,
+      "loss": 0.8465,
+      "step": 220350
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 3.3254375675963865e-05,
+      "loss": 1.0132,
+      "step": 220360
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 3.325361402653587e-05,
+      "loss": 0.7604,
+      "step": 220370
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 3.325285237710787e-05,
+      "loss": 0.9284,
+      "step": 220380
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 3.3252090727679865e-05,
+      "loss": 0.7838,
+      "step": 220390
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 3.325132907825186e-05,
+      "loss": 0.7946,
+      "step": 220400
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 3.3250567428823866e-05,
+      "loss": 0.9044,
+      "step": 220410
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 3.324980577939586e-05,
+      "loss": 0.8479,
+      "step": 220420
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 3.324904412996786e-05,
+      "loss": 0.812,
+      "step": 220430
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 3.324828248053986e-05,
+      "loss": 0.9145,
+      "step": 220440
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 3.3247520831111856e-05,
+      "loss": 0.8136,
+      "step": 220450
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 3.3246759181683854e-05,
+      "loss": 0.851,
+      "step": 220460
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 3.324599753225586e-05,
+      "loss": 0.9057,
+      "step": 220470
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 3.3245235882827855e-05,
+      "loss": 0.7176,
+      "step": 220480
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 3.324447423339985e-05,
+      "loss": 0.8533,
+      "step": 220490
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 3.324371258397185e-05,
+      "loss": 0.8468,
+      "step": 220500
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 3.324295093454385e-05,
+      "loss": 0.8306,
+      "step": 220510
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 3.324218928511585e-05,
+      "loss": 0.9325,
+      "step": 220520
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 3.324142763568785e-05,
+      "loss": 0.8079,
+      "step": 220530
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 3.324066598625985e-05,
+      "loss": 0.7752,
+      "step": 220540
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 3.3239904336831844e-05,
+      "loss": 0.7685,
+      "step": 220550
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 3.323914268740385e-05,
+      "loss": 0.8106,
+      "step": 220560
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 3.323838103797584e-05,
+      "loss": 0.8884,
+      "step": 220570
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 3.3237619388547844e-05,
+      "loss": 0.8795,
+      "step": 220580
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 3.323685773911984e-05,
+      "loss": 0.8781,
+      "step": 220590
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 3.323609608969184e-05,
+      "loss": 0.8848,
+      "step": 220600
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 3.3235334440263836e-05,
+      "loss": 0.7651,
+      "step": 220610
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 3.323457279083584e-05,
+      "loss": 0.7635,
+      "step": 220620
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 3.323381114140783e-05,
+      "loss": 0.8741,
+      "step": 220630
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 3.3233049491979835e-05,
+      "loss": 0.8967,
+      "step": 220640
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 3.3232287842551826e-05,
+      "loss": 0.925,
+      "step": 220650
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 3.323152619312383e-05,
+      "loss": 0.8083,
+      "step": 220660
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 3.323076454369583e-05,
+      "loss": 0.8381,
+      "step": 220670
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 3.3230002894267825e-05,
+      "loss": 0.8349,
+      "step": 220680
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 3.322924124483983e-05,
+      "loss": 0.8308,
+      "step": 220690
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 3.322847959541183e-05,
+      "loss": 0.7647,
+      "step": 220700
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 3.3227717945983824e-05,
+      "loss": 0.8228,
+      "step": 220710
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 3.322695629655582e-05,
+      "loss": 0.8018,
+      "step": 220720
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 3.3226194647127826e-05,
+      "loss": 0.8048,
+      "step": 220730
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 3.322543299769982e-05,
+      "loss": 0.8074,
+      "step": 220740
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 3.322467134827182e-05,
+      "loss": 0.7584,
+      "step": 220750
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 3.322390969884382e-05,
+      "loss": 0.8047,
+      "step": 220760
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 3.3223148049415816e-05,
+      "loss": 0.8824,
+      "step": 220770
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 3.322238639998781e-05,
+      "loss": 0.9157,
+      "step": 220780
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 3.322162475055982e-05,
+      "loss": 0.8877,
+      "step": 220790
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 3.322086310113181e-05,
+      "loss": 0.9225,
+      "step": 220800
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 3.322010145170381e-05,
+      "loss": 1.0095,
+      "step": 220810
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 3.321933980227581e-05,
+      "loss": 0.7551,
+      "step": 220820
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 3.321857815284781e-05,
+      "loss": 0.8215,
+      "step": 220830
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 3.3217816503419805e-05,
+      "loss": 0.8506,
+      "step": 220840
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 3.321705485399181e-05,
+      "loss": 0.9073,
+      "step": 220850
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 3.32162932045638e-05,
+      "loss": 0.7801,
+      "step": 220860
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 3.3215531555135804e-05,
+      "loss": 0.891,
+      "step": 220870
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 3.32147699057078e-05,
+      "loss": 0.8135,
+      "step": 220880
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 3.32140082562798e-05,
+      "loss": 0.8289,
+      "step": 220890
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 3.32132466068518e-05,
+      "loss": 0.8531,
+      "step": 220900
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 3.32124849574238e-05,
+      "loss": 0.7871,
+      "step": 220910
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 3.32117233079958e-05,
+      "loss": 0.8531,
+      "step": 220920
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 3.3210961658567796e-05,
+      "loss": 0.7822,
+      "step": 220930
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 3.32102000091398e-05,
+      "loss": 0.8241,
+      "step": 220940
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 3.320943835971179e-05,
+      "loss": 0.8308,
+      "step": 220950
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 3.3208676710283795e-05,
+      "loss": 0.858,
+      "step": 220960
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 3.320791506085579e-05,
+      "loss": 0.8198,
+      "step": 220970
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 3.320715341142779e-05,
+      "loss": 0.8157,
+      "step": 220980
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 3.320639176199979e-05,
+      "loss": 0.8164,
+      "step": 220990
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 3.320563011257179e-05,
+      "loss": 0.93,
+      "step": 221000
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 3.320486846314378e-05,
+      "loss": 0.8524,
+      "step": 221010
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 3.3204106813715786e-05,
+      "loss": 0.7541,
+      "step": 221020
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 3.3203345164287784e-05,
+      "loss": 0.8348,
+      "step": 221030
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 3.320258351485978e-05,
+      "loss": 0.7701,
+      "step": 221040
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 3.320182186543178e-05,
+      "loss": 0.7616,
+      "step": 221050
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 3.320106021600378e-05,
+      "loss": 0.8234,
+      "step": 221060
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 3.3200298566575774e-05,
+      "loss": 0.8255,
+      "step": 221070
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 3.319953691714778e-05,
+      "loss": 0.7753,
+      "step": 221080
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 3.3198775267719775e-05,
+      "loss": 0.7121,
+      "step": 221090
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 3.319801361829177e-05,
+      "loss": 0.7698,
+      "step": 221100
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 3.319725196886378e-05,
+      "loss": 0.7287,
+      "step": 221110
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 3.319649031943577e-05,
+      "loss": 0.8539,
+      "step": 221120
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 3.319572867000777e-05,
+      "loss": 0.8902,
+      "step": 221130
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 3.319496702057977e-05,
+      "loss": 0.9112,
+      "step": 221140
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 3.319420537115177e-05,
+      "loss": 0.8125,
+      "step": 221150
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 3.3193443721723764e-05,
+      "loss": 0.9308,
+      "step": 221160
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 3.319268207229577e-05,
+      "loss": 0.7592,
+      "step": 221170
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 3.319192042286776e-05,
+      "loss": 0.9476,
+      "step": 221180
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 3.3191158773439764e-05,
+      "loss": 0.8184,
+      "step": 221190
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 3.319039712401176e-05,
+      "loss": 0.9106,
+      "step": 221200
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 3.318963547458376e-05,
+      "loss": 0.7205,
+      "step": 221210
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 3.3188873825155756e-05,
+      "loss": 0.9031,
+      "step": 221220
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 3.318811217572776e-05,
+      "loss": 0.921,
+      "step": 221230
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 3.318735052629975e-05,
+      "loss": 0.77,
+      "step": 221240
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 3.3186588876871755e-05,
+      "loss": 0.8659,
+      "step": 221250
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 3.318582722744375e-05,
+      "loss": 0.9165,
+      "step": 221260
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 3.318506557801575e-05,
+      "loss": 0.8444,
+      "step": 221270
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 3.3184303928587754e-05,
+      "loss": 0.8391,
+      "step": 221280
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 3.318354227915975e-05,
+      "loss": 0.8814,
+      "step": 221290
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 3.318278062973175e-05,
+      "loss": 0.8446,
+      "step": 221300
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 3.318201898030375e-05,
+      "loss": 0.8966,
+      "step": 221310
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 3.318125733087575e-05,
+      "loss": 0.8367,
+      "step": 221320
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 3.318049568144774e-05,
+      "loss": 0.8516,
+      "step": 221330
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 3.3179734032019746e-05,
+      "loss": 0.8431,
+      "step": 221340
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 3.3178972382591743e-05,
+      "loss": 0.8762,
+      "step": 221350
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 3.317821073316374e-05,
+      "loss": 0.7672,
+      "step": 221360
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 3.317744908373574e-05,
+      "loss": 0.9653,
+      "step": 221370
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 3.317668743430774e-05,
+      "loss": 0.8049,
+      "step": 221380
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 3.317592578487973e-05,
+      "loss": 0.9021,
+      "step": 221390
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 3.317516413545174e-05,
+      "loss": 0.8893,
+      "step": 221400
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 3.3174402486023735e-05,
+      "loss": 0.8669,
+      "step": 221410
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 3.317364083659573e-05,
+      "loss": 0.8239,
+      "step": 221420
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 3.317287918716773e-05,
+      "loss": 0.7733,
+      "step": 221430
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 3.3172117537739734e-05,
+      "loss": 0.8393,
+      "step": 221440
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 3.3171355888311725e-05,
+      "loss": 0.9226,
+      "step": 221450
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 3.317059423888373e-05,
+      "loss": 0.9883,
+      "step": 221460
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 3.3169832589455727e-05,
+      "loss": 0.7959,
+      "step": 221470
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 3.3169070940027724e-05,
+      "loss": 0.8627,
+      "step": 221480
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 3.316830929059973e-05,
+      "loss": 0.7715,
+      "step": 221490
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 3.3167547641171726e-05,
+      "loss": 0.8765,
+      "step": 221500
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 3.316678599174372e-05,
+      "loss": 0.8854,
+      "step": 221510
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 3.316602434231572e-05,
+      "loss": 0.8065,
+      "step": 221520
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 3.3165262692887725e-05,
+      "loss": 0.8603,
+      "step": 221530
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 3.3164501043459716e-05,
+      "loss": 0.7951,
+      "step": 221540
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 3.316373939403172e-05,
+      "loss": 0.8461,
+      "step": 221550
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 3.316297774460371e-05,
+      "loss": 0.9074,
+      "step": 221560
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 3.3162216095175715e-05,
+      "loss": 0.8389,
+      "step": 221570
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 3.316145444574771e-05,
+      "loss": 0.7821,
+      "step": 221580
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 3.316069279631971e-05,
+      "loss": 0.8322,
+      "step": 221590
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 3.315993114689171e-05,
+      "loss": 0.8191,
+      "step": 221600
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 3.315916949746371e-05,
+      "loss": 0.8845,
+      "step": 221610
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 3.31584078480357e-05,
+      "loss": 0.8654,
+      "step": 221620
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 3.3157646198607706e-05,
+      "loss": 0.8611,
+      "step": 221630
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 3.3156884549179704e-05,
+      "loss": 0.7969,
+      "step": 221640
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 3.31561228997517e-05,
+      "loss": 0.8296,
+      "step": 221650
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 3.31553612503237e-05,
+      "loss": 0.8677,
+      "step": 221660
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 3.31545996008957e-05,
+      "loss": 0.8601,
+      "step": 221670
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 3.31538379514677e-05,
+      "loss": 0.9353,
+      "step": 221680
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 3.31530763020397e-05,
+      "loss": 0.8737,
+      "step": 221690
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 3.31523146526117e-05,
+      "loss": 0.861,
+      "step": 221700
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 3.315155300318369e-05,
+      "loss": 0.917,
+      "step": 221710
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 3.31507913537557e-05,
+      "loss": 0.8217,
+      "step": 221720
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 3.3150029704327695e-05,
+      "loss": 0.9146,
+      "step": 221730
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 3.314926805489969e-05,
+      "loss": 0.9137,
+      "step": 221740
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 3.314850640547169e-05,
+      "loss": 0.7845,
+      "step": 221750
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 3.3147744756043694e-05,
+      "loss": 0.7925,
+      "step": 221760
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 3.3146983106615684e-05,
+      "loss": 0.777,
+      "step": 221770
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 3.314622145718769e-05,
+      "loss": 0.9603,
+      "step": 221780
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 3.3145459807759686e-05,
+      "loss": 0.8545,
+      "step": 221790
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 3.3144698158331684e-05,
+      "loss": 0.8755,
+      "step": 221800
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 3.314393650890368e-05,
+      "loss": 0.8185,
+      "step": 221810
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 3.3143174859475685e-05,
+      "loss": 0.8636,
+      "step": 221820
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 3.3142413210047676e-05,
+      "loss": 0.8053,
+      "step": 221830
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 3.314165156061968e-05,
+      "loss": 0.9248,
+      "step": 221840
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 3.314088991119168e-05,
+      "loss": 0.7195,
+      "step": 221850
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 3.3140128261763675e-05,
+      "loss": 0.8166,
+      "step": 221860
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 3.313936661233567e-05,
+      "loss": 0.9295,
+      "step": 221870
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 3.313860496290768e-05,
+      "loss": 0.8094,
+      "step": 221880
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 3.3137843313479674e-05,
+      "loss": 0.8123,
+      "step": 221890
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 3.313708166405167e-05,
+      "loss": 0.8184,
+      "step": 221900
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 3.3136320014623676e-05,
+      "loss": 0.8145,
+      "step": 221910
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 3.313555836519567e-05,
+      "loss": 0.869,
+      "step": 221920
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 3.313479671576767e-05,
+      "loss": 0.7992,
+      "step": 221930
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 3.313403506633967e-05,
+      "loss": 0.8506,
+      "step": 221940
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 3.3133273416911666e-05,
+      "loss": 0.8404,
+      "step": 221950
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 3.3132511767483663e-05,
+      "loss": 0.7626,
+      "step": 221960
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 3.313175011805567e-05,
+      "loss": 0.8613,
+      "step": 221970
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 3.313098846862766e-05,
+      "loss": 0.8306,
+      "step": 221980
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 3.313022681919966e-05,
+      "loss": 0.8846,
+      "step": 221990
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 3.312946516977166e-05,
+      "loss": 0.8713,
+      "step": 222000
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 3.312870352034366e-05,
+      "loss": 0.8698,
+      "step": 222010
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 3.3127941870915655e-05,
+      "loss": 0.8591,
+      "step": 222020
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 3.312718022148765e-05,
+      "loss": 0.8588,
+      "step": 222030
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 3.312641857205965e-05,
+      "loss": 0.877,
+      "step": 222040
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 3.3125656922631654e-05,
+      "loss": 0.9098,
+      "step": 222050
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 3.312489527320365e-05,
+      "loss": 0.8594,
+      "step": 222060
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 3.312413362377565e-05,
+      "loss": 0.8038,
+      "step": 222070
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 3.312337197434765e-05,
+      "loss": 0.8276,
+      "step": 222080
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 3.3122610324919644e-05,
+      "loss": 0.8058,
+      "step": 222090
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 3.312184867549165e-05,
+      "loss": 0.9555,
+      "step": 222100
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 3.3121087026063646e-05,
+      "loss": 0.9215,
+      "step": 222110
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 3.312032537663564e-05,
+      "loss": 0.8492,
+      "step": 222120
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 3.311956372720764e-05,
+      "loss": 0.853,
+      "step": 222130
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 3.3118802077779645e-05,
+      "loss": 0.8228,
+      "step": 222140
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 3.3118040428351636e-05,
+      "loss": 0.8911,
+      "step": 222150
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 3.311727877892364e-05,
+      "loss": 0.8564,
+      "step": 222160
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 3.311651712949564e-05,
+      "loss": 0.781,
+      "step": 222170
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 3.3115755480067635e-05,
+      "loss": 0.988,
+      "step": 222180
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 3.311499383063963e-05,
+      "loss": 0.8189,
+      "step": 222190
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 3.3114232181211637e-05,
+      "loss": 0.8458,
+      "step": 222200
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 3.311347053178363e-05,
+      "loss": 0.8251,
+      "step": 222210
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 3.311270888235563e-05,
+      "loss": 0.7839,
+      "step": 222220
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 3.311194723292763e-05,
+      "loss": 0.8811,
+      "step": 222230
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 3.3111185583499626e-05,
+      "loss": 0.9644,
+      "step": 222240
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 3.3110423934071624e-05,
+      "loss": 0.8015,
+      "step": 222250
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 3.310966228464363e-05,
+      "loss": 0.8975,
+      "step": 222260
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 3.3108900635215626e-05,
+      "loss": 0.7909,
+      "step": 222270
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 3.310813898578762e-05,
+      "loss": 0.8876,
+      "step": 222280
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 3.310737733635963e-05,
+      "loss": 1.016,
+      "step": 222290
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 3.310661568693162e-05,
+      "loss": 0.9017,
+      "step": 222300
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 3.310585403750362e-05,
+      "loss": 0.8109,
+      "step": 222310
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 3.310509238807562e-05,
+      "loss": 0.9858,
+      "step": 222320
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 3.310433073864762e-05,
+      "loss": 0.8293,
+      "step": 222330
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 3.3103569089219615e-05,
+      "loss": 0.7991,
+      "step": 222340
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 3.310280743979162e-05,
+      "loss": 0.8111,
+      "step": 222350
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 3.310204579036361e-05,
+      "loss": 0.7876,
+      "step": 222360
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 3.3101284140935614e-05,
+      "loss": 0.8307,
+      "step": 222370
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 3.310052249150761e-05,
+      "loss": 0.876,
+      "step": 222380
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 3.309976084207961e-05,
+      "loss": 0.795,
+      "step": 222390
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 3.3098999192651606e-05,
+      "loss": 0.8641,
+      "step": 222400
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 3.309823754322361e-05,
+      "loss": 0.8977,
+      "step": 222410
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 3.30974758937956e-05,
+      "loss": 0.8694,
+      "step": 222420
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 3.3096714244367605e-05,
+      "loss": 0.8377,
+      "step": 222430
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 3.30959525949396e-05,
+      "loss": 0.8687,
+      "step": 222440
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 3.30951909455116e-05,
+      "loss": 0.8081,
+      "step": 222450
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 3.30944292960836e-05,
+      "loss": 0.7828,
+      "step": 222460
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 3.30936676466556e-05,
+      "loss": 0.7293,
+      "step": 222470
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 3.30929059972276e-05,
+      "loss": 0.8935,
+      "step": 222480
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 3.30921443477996e-05,
+      "loss": 0.8709,
+      "step": 222490
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 3.3091382698371594e-05,
+      "loss": 0.8234,
+      "step": 222500
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 3.309062104894359e-05,
+      "loss": 0.8538,
+      "step": 222510
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 3.3089859399515596e-05,
+      "loss": 0.7551,
+      "step": 222520
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 3.308909775008759e-05,
+      "loss": 0.7766,
+      "step": 222530
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 3.308833610065959e-05,
+      "loss": 0.8233,
+      "step": 222540
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 3.308757445123159e-05,
+      "loss": 0.8108,
+      "step": 222550
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 3.3086812801803586e-05,
+      "loss": 0.8365,
+      "step": 222560
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 3.3086051152375583e-05,
+      "loss": 0.7958,
+      "step": 222570
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 3.308528950294759e-05,
+      "loss": 0.8226,
+      "step": 222580
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 3.308452785351958e-05,
+      "loss": 0.8348,
+      "step": 222590
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 3.308376620409158e-05,
+      "loss": 0.7602,
+      "step": 222600
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 3.308300455466358e-05,
+      "loss": 0.8766,
+      "step": 222610
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 3.308224290523558e-05,
+      "loss": 0.9336,
+      "step": 222620
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 3.3081481255807575e-05,
+      "loss": 0.8116,
+      "step": 222630
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 3.308071960637958e-05,
+      "loss": 0.7903,
+      "step": 222640
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 3.307995795695158e-05,
+      "loss": 0.873,
+      "step": 222650
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 3.3079196307523574e-05,
+      "loss": 0.7282,
+      "step": 222660
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 3.307843465809558e-05,
+      "loss": 0.7497,
+      "step": 222670
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 3.307767300866757e-05,
+      "loss": 0.8171,
+      "step": 222680
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 3.307691135923957e-05,
+      "loss": 0.8169,
+      "step": 222690
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 3.307614970981157e-05,
+      "loss": 0.8005,
+      "step": 222700
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 3.307538806038357e-05,
+      "loss": 0.7434,
+      "step": 222710
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 3.3074626410955566e-05,
+      "loss": 0.7503,
+      "step": 222720
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 3.307386476152757e-05,
+      "loss": 0.8709,
+      "step": 222730
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 3.307310311209956e-05,
+      "loss": 0.8643,
+      "step": 222740
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 3.3072341462671565e-05,
+      "loss": 0.835,
+      "step": 222750
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 3.307157981324356e-05,
+      "loss": 0.7384,
+      "step": 222760
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 3.307081816381556e-05,
+      "loss": 0.9412,
+      "step": 222770
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 3.307005651438756e-05,
+      "loss": 0.8658,
+      "step": 222780
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 3.306929486495956e-05,
+      "loss": 0.7412,
+      "step": 222790
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 3.306853321553155e-05,
+      "loss": 0.8692,
+      "step": 222800
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 3.3067771566103557e-05,
+      "loss": 0.8602,
+      "step": 222810
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 3.3067009916675554e-05,
+      "loss": 0.7687,
+      "step": 222820
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 3.306624826724755e-05,
+      "loss": 0.8138,
+      "step": 222830
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 3.306548661781955e-05,
+      "loss": 0.7798,
+      "step": 222840
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 3.306472496839155e-05,
+      "loss": 0.8415,
+      "step": 222850
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 3.306396331896355e-05,
+      "loss": 0.7281,
+      "step": 222860
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 3.306320166953555e-05,
+      "loss": 0.8172,
+      "step": 222870
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 3.306244002010755e-05,
+      "loss": 0.7669,
+      "step": 222880
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 3.306167837067954e-05,
+      "loss": 0.7983,
+      "step": 222890
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 3.306091672125155e-05,
+      "loss": 0.9034,
+      "step": 222900
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 3.3060155071823545e-05,
+      "loss": 0.8266,
+      "step": 222910
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 3.305939342239554e-05,
+      "loss": 0.7645,
+      "step": 222920
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 3.305863177296754e-05,
+      "loss": 0.7842,
+      "step": 222930
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 3.305787012353954e-05,
+      "loss": 0.7561,
+      "step": 222940
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 3.3057108474111535e-05,
+      "loss": 0.7922,
+      "step": 222950
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 3.305634682468354e-05,
+      "loss": 0.8825,
+      "step": 222960
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 3.305558517525553e-05,
+      "loss": 0.7614,
+      "step": 222970
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 3.3054823525827534e-05,
+      "loss": 0.8357,
+      "step": 222980
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 3.305406187639953e-05,
+      "loss": 0.9255,
+      "step": 222990
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 3.305330022697153e-05,
+      "loss": 0.9559,
+      "step": 223000
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 3.3052538577543526e-05,
+      "loss": 0.9207,
+      "step": 223010
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 3.305177692811553e-05,
+      "loss": 0.834,
+      "step": 223020
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 3.305101527868753e-05,
+      "loss": 0.7987,
+      "step": 223030
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 3.3050253629259525e-05,
+      "loss": 0.8193,
+      "step": 223040
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 3.304949197983152e-05,
+      "loss": 0.8131,
+      "step": 223050
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 3.304873033040352e-05,
+      "loss": 0.9175,
+      "step": 223060
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 3.3047968680975525e-05,
+      "loss": 0.9631,
+      "step": 223070
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 3.304720703154752e-05,
+      "loss": 0.865,
+      "step": 223080
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 3.304644538211952e-05,
+      "loss": 0.8586,
+      "step": 223090
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 3.304568373269152e-05,
+      "loss": 0.8294,
+      "step": 223100
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 3.304492208326352e-05,
+      "loss": 0.8882,
+      "step": 223110
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 3.304416043383551e-05,
+      "loss": 0.8005,
+      "step": 223120
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 3.3043398784407516e-05,
+      "loss": 0.8985,
+      "step": 223130
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 3.3042637134979514e-05,
+      "loss": 0.7698,
+      "step": 223140
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 3.304187548555151e-05,
+      "loss": 0.8401,
+      "step": 223150
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 3.304111383612351e-05,
+      "loss": 0.8861,
+      "step": 223160
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 3.304035218669551e-05,
+      "loss": 0.8693,
+      "step": 223170
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 3.3039590537267503e-05,
+      "loss": 0.8327,
+      "step": 223180
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 3.303882888783951e-05,
+      "loss": 0.8844,
+      "step": 223190
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 3.3038067238411505e-05,
+      "loss": 0.8915,
+      "step": 223200
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 3.30373055889835e-05,
+      "loss": 0.8133,
+      "step": 223210
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 3.30365439395555e-05,
+      "loss": 0.8204,
+      "step": 223220
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 3.3035782290127504e-05,
+      "loss": 0.8305,
+      "step": 223230
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 3.30350206406995e-05,
+      "loss": 0.7702,
+      "step": 223240
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 3.30342589912715e-05,
+      "loss": 0.8695,
+      "step": 223250
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 3.30334973418435e-05,
+      "loss": 0.8665,
+      "step": 223260
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 3.3032735692415494e-05,
+      "loss": 0.8046,
+      "step": 223270
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 3.30319740429875e-05,
+      "loss": 0.7918,
+      "step": 223280
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 3.3031212393559496e-05,
+      "loss": 0.7888,
+      "step": 223290
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 3.3030450744131493e-05,
+      "loss": 0.805,
+      "step": 223300
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 3.302968909470349e-05,
+      "loss": 0.8248,
+      "step": 223310
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 3.3028927445275495e-05,
+      "loss": 0.7954,
+      "step": 223320
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 3.3028165795847486e-05,
+      "loss": 0.8271,
+      "step": 223330
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 3.302740414641949e-05,
+      "loss": 0.8415,
+      "step": 223340
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 3.302664249699149e-05,
+      "loss": 0.8352,
+      "step": 223350
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 3.3025880847563485e-05,
+      "loss": 0.719,
+      "step": 223360
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 3.302511919813548e-05,
+      "loss": 0.7778,
+      "step": 223370
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 3.302435754870749e-05,
+      "loss": 0.8127,
+      "step": 223380
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 3.302359589927948e-05,
+      "loss": 0.8225,
+      "step": 223390
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 3.302283424985148e-05,
+      "loss": 0.7606,
+      "step": 223400
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 3.302207260042348e-05,
+      "loss": 0.8326,
+      "step": 223410
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 3.3021310950995477e-05,
+      "loss": 0.8044,
+      "step": 223420
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 3.3020549301567474e-05,
+      "loss": 0.8153,
+      "step": 223430
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 3.301978765213947e-05,
+      "loss": 0.8416,
+      "step": 223440
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 3.3019026002711476e-05,
+      "loss": 0.8403,
+      "step": 223450
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 3.301826435328347e-05,
+      "loss": 0.8646,
+      "step": 223460
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 3.301750270385547e-05,
+      "loss": 0.9108,
+      "step": 223470
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 3.301674105442747e-05,
+      "loss": 0.8905,
+      "step": 223480
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 3.301597940499947e-05,
+      "loss": 0.9005,
+      "step": 223490
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 3.301521775557146e-05,
+      "loss": 0.837,
+      "step": 223500
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 3.301445610614347e-05,
+      "loss": 0.8634,
+      "step": 223510
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 3.3013694456715465e-05,
+      "loss": 0.8158,
+      "step": 223520
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 3.301293280728746e-05,
+      "loss": 0.8697,
+      "step": 223530
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 3.301217115785946e-05,
+      "loss": 0.8902,
+      "step": 223540
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 3.3011409508431464e-05,
+      "loss": 0.7944,
+      "step": 223550
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 3.3010647859003455e-05,
+      "loss": 0.773,
+      "step": 223560
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 3.300988620957546e-05,
+      "loss": 0.8751,
+      "step": 223570
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 3.3009124560147456e-05,
+      "loss": 0.8955,
+      "step": 223580
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 3.3008362910719454e-05,
+      "loss": 0.8334,
+      "step": 223590
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 3.300760126129145e-05,
+      "loss": 0.8493,
+      "step": 223600
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 3.3006839611863456e-05,
+      "loss": 0.9051,
+      "step": 223610
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 3.300607796243545e-05,
+      "loss": 0.8405,
+      "step": 223620
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 3.300531631300745e-05,
+      "loss": 0.7951,
+      "step": 223630
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 3.300455466357945e-05,
+      "loss": 0.793,
+      "step": 223640
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 3.3003793014151445e-05,
+      "loss": 0.8766,
+      "step": 223650
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 3.300303136472345e-05,
+      "loss": 0.8278,
+      "step": 223660
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 3.300226971529545e-05,
+      "loss": 0.9093,
+      "step": 223670
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 3.3001508065867445e-05,
+      "loss": 0.8613,
+      "step": 223680
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 3.300074641643944e-05,
+      "loss": 0.8176,
+      "step": 223690
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 3.2999984767011446e-05,
+      "loss": 0.8268,
+      "step": 223700
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 3.299922311758344e-05,
+      "loss": 0.905,
+      "step": 223710
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 3.299846146815544e-05,
+      "loss": 0.8255,
+      "step": 223720
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 3.299769981872744e-05,
+      "loss": 0.8327,
+      "step": 223730
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 3.2996938169299436e-05,
+      "loss": 0.7357,
+      "step": 223740
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 3.2996176519871434e-05,
+      "loss": 0.8755,
+      "step": 223750
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 3.299541487044344e-05,
+      "loss": 0.8026,
+      "step": 223760
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 3.299465322101543e-05,
+      "loss": 0.8256,
+      "step": 223770
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 3.299389157158743e-05,
+      "loss": 0.8285,
+      "step": 223780
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 3.299312992215943e-05,
+      "loss": 0.9088,
+      "step": 223790
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 3.299236827273143e-05,
+      "loss": 0.8243,
+      "step": 223800
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 3.2991606623303425e-05,
+      "loss": 0.8248,
+      "step": 223810
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 3.299084497387543e-05,
+      "loss": 0.8321,
+      "step": 223820
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 3.299008332444743e-05,
+      "loss": 0.7848,
+      "step": 223830
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 3.2989321675019424e-05,
+      "loss": 0.8298,
+      "step": 223840
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 3.298856002559142e-05,
+      "loss": 0.7646,
+      "step": 223850
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 3.298779837616342e-05,
+      "loss": 0.8426,
+      "step": 223860
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 3.2987036726735424e-05,
+      "loss": 0.8434,
+      "step": 223870
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 3.2986275077307414e-05,
+      "loss": 0.7912,
+      "step": 223880
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 3.298551342787942e-05,
+      "loss": 0.801,
+      "step": 223890
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 3.2984751778451416e-05,
+      "loss": 0.8541,
+      "step": 223900
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 3.2983990129023413e-05,
+      "loss": 0.9046,
+      "step": 223910
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 3.298322847959541e-05,
+      "loss": 0.8218,
+      "step": 223920
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 3.2982466830167415e-05,
+      "loss": 0.8527,
+      "step": 223930
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 3.2981705180739406e-05,
+      "loss": 0.8835,
+      "step": 223940
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 3.298094353131141e-05,
+      "loss": 0.8539,
+      "step": 223950
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 3.298018188188341e-05,
+      "loss": 0.7498,
+      "step": 223960
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 3.2979420232455405e-05,
+      "loss": 0.8975,
+      "step": 223970
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 3.29786585830274e-05,
+      "loss": 0.8069,
+      "step": 223980
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 3.297789693359941e-05,
+      "loss": 0.9457,
+      "step": 223990
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 3.2977135284171404e-05,
+      "loss": 0.9719,
+      "step": 224000
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 3.29763736347434e-05,
+      "loss": 0.8718,
+      "step": 224010
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 3.29756119853154e-05,
+      "loss": 0.7997,
+      "step": 224020
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 3.2974850335887397e-05,
+      "loss": 0.8872,
+      "step": 224030
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 3.29740886864594e-05,
+      "loss": 0.8978,
+      "step": 224040
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 3.29733270370314e-05,
+      "loss": 0.8129,
+      "step": 224050
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 3.2972565387603396e-05,
+      "loss": 0.8363,
+      "step": 224060
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 3.297180373817539e-05,
+      "loss": 0.8843,
+      "step": 224070
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 3.29710420887474e-05,
+      "loss": 0.7131,
+      "step": 224080
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 3.297028043931939e-05,
+      "loss": 0.8265,
+      "step": 224090
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 3.296951878989139e-05,
+      "loss": 0.926,
+      "step": 224100
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 3.296875714046339e-05,
+      "loss": 0.8667,
+      "step": 224110
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 3.296799549103539e-05,
+      "loss": 0.7249,
+      "step": 224120
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 3.2967233841607385e-05,
+      "loss": 0.7935,
+      "step": 224130
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 3.296647219217939e-05,
+      "loss": 0.8244,
+      "step": 224140
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 3.296571054275138e-05,
+      "loss": 0.8418,
+      "step": 224150
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 3.2964948893323384e-05,
+      "loss": 0.782,
+      "step": 224160
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 3.296418724389538e-05,
+      "loss": 0.8045,
+      "step": 224170
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 3.296342559446738e-05,
+      "loss": 0.8419,
+      "step": 224180
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 3.2962663945039376e-05,
+      "loss": 0.8367,
+      "step": 224190
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 3.296190229561138e-05,
+      "loss": 0.8066,
+      "step": 224200
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 3.296114064618338e-05,
+      "loss": 0.8389,
+      "step": 224210
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 3.2960378996755376e-05,
+      "loss": 0.9467,
+      "step": 224220
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 3.295961734732737e-05,
+      "loss": 0.9083,
+      "step": 224230
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 3.295885569789937e-05,
+      "loss": 0.869,
+      "step": 224240
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 3.2958094048471375e-05,
+      "loss": 0.8317,
+      "step": 224250
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 3.295733239904337e-05,
+      "loss": 0.8157,
+      "step": 224260
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 3.295657074961537e-05,
+      "loss": 0.8762,
+      "step": 224270
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 3.295580910018737e-05,
+      "loss": 0.9112,
+      "step": 224280
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 3.295504745075937e-05,
+      "loss": 0.7941,
+      "step": 224290
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 3.295428580133136e-05,
+      "loss": 0.767,
+      "step": 224300
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 3.2953524151903366e-05,
+      "loss": 0.8536,
+      "step": 224310
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 3.295276250247536e-05,
+      "loss": 0.8822,
+      "step": 224320
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 3.295200085304736e-05,
+      "loss": 0.8858,
+      "step": 224330
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 3.295123920361936e-05,
+      "loss": 0.7929,
+      "step": 224340
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 3.2950477554191356e-05,
+      "loss": 0.7665,
+      "step": 224350
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 3.2949715904763354e-05,
+      "loss": 0.8483,
+      "step": 224360
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 3.294895425533536e-05,
+      "loss": 0.8473,
+      "step": 224370
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 3.2948192605907355e-05,
+      "loss": 0.8402,
+      "step": 224380
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 3.294743095647935e-05,
+      "loss": 0.8691,
+      "step": 224390
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 3.294666930705135e-05,
+      "loss": 0.916,
+      "step": 224400
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 3.294590765762335e-05,
+      "loss": 0.816,
+      "step": 224410
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 3.294514600819535e-05,
+      "loss": 0.9273,
+      "step": 224420
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 3.294438435876735e-05,
+      "loss": 0.8539,
+      "step": 224430
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 3.294362270933935e-05,
+      "loss": 0.8022,
+      "step": 224440
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 3.2942861059911344e-05,
+      "loss": 0.8808,
+      "step": 224450
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 3.294209941048335e-05,
+      "loss": 0.7895,
+      "step": 224460
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 3.294133776105534e-05,
+      "loss": 0.7674,
+      "step": 224470
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 3.2940576111627344e-05,
+      "loss": 0.8277,
+      "step": 224480
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 3.293981446219934e-05,
+      "loss": 0.8586,
+      "step": 224490
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 3.293905281277134e-05,
+      "loss": 0.9268,
+      "step": 224500
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 3.2938291163343336e-05,
+      "loss": 0.8642,
+      "step": 224510
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 3.293752951391534e-05,
+      "loss": 0.902,
+      "step": 224520
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 3.293676786448733e-05,
+      "loss": 0.8021,
+      "step": 224530
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 3.2936006215059335e-05,
+      "loss": 0.7678,
+      "step": 224540
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 3.293524456563133e-05,
+      "loss": 0.8148,
+      "step": 224550
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 3.293448291620333e-05,
+      "loss": 0.836,
+      "step": 224560
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 3.293372126677533e-05,
+      "loss": 0.9383,
+      "step": 224570
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 3.293295961734733e-05,
+      "loss": 0.8728,
+      "step": 224580
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 3.293219796791933e-05,
+      "loss": 0.9115,
+      "step": 224590
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 3.293143631849133e-05,
+      "loss": 0.8679,
+      "step": 224600
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 3.2930674669063324e-05,
+      "loss": 0.7939,
+      "step": 224610
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 3.292991301963532e-05,
+      "loss": 0.8533,
+      "step": 224620
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 3.2929151370207326e-05,
+      "loss": 0.9217,
+      "step": 224630
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 3.292838972077932e-05,
+      "loss": 0.9331,
+      "step": 224640
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 3.292762807135132e-05,
+      "loss": 0.7329,
+      "step": 224650
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 3.292686642192332e-05,
+      "loss": 0.7739,
+      "step": 224660
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 3.292610477249532e-05,
+      "loss": 0.8618,
+      "step": 224670
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 3.292534312306731e-05,
+      "loss": 0.8536,
+      "step": 224680
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 3.292458147363932e-05,
+      "loss": 0.7904,
+      "step": 224690
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 3.2923819824211315e-05,
+      "loss": 0.8517,
+      "step": 224700
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 3.292305817478331e-05,
+      "loss": 0.7713,
+      "step": 224710
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 3.292229652535531e-05,
+      "loss": 0.7744,
+      "step": 224720
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 3.2921534875927314e-05,
+      "loss": 0.8487,
+      "step": 224730
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 3.2920773226499305e-05,
+      "loss": 0.9667,
+      "step": 224740
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 3.292001157707131e-05,
+      "loss": 0.7932,
+      "step": 224750
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 3.2919249927643306e-05,
+      "loss": 0.8596,
+      "step": 224760
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 3.2918488278215304e-05,
+      "loss": 0.8455,
+      "step": 224770
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 3.29177266287873e-05,
+      "loss": 0.7984,
+      "step": 224780
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 3.29169649793593e-05,
+      "loss": 0.8385,
+      "step": 224790
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 3.29162033299313e-05,
+      "loss": 0.8544,
+      "step": 224800
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 3.29154416805033e-05,
+      "loss": 0.7748,
+      "step": 224810
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 3.29146800310753e-05,
+      "loss": 0.9184,
+      "step": 224820
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 3.2913918381647296e-05,
+      "loss": 0.7808,
+      "step": 224830
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 3.29131567322193e-05,
+      "loss": 0.8546,
+      "step": 224840
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 3.291239508279129e-05,
+      "loss": 0.8979,
+      "step": 224850
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 3.2911633433363295e-05,
+      "loss": 0.8885,
+      "step": 224860
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 3.291087178393529e-05,
+      "loss": 0.7795,
+      "step": 224870
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 3.291011013450729e-05,
+      "loss": 0.8636,
+      "step": 224880
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 3.290934848507929e-05,
+      "loss": 0.7646,
+      "step": 224890
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 3.290858683565129e-05,
+      "loss": 0.8478,
+      "step": 224900
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 3.290782518622328e-05,
+      "loss": 0.8065,
+      "step": 224910
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 3.2907063536795286e-05,
+      "loss": 0.9013,
+      "step": 224920
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 3.2906301887367284e-05,
+      "loss": 0.8428,
+      "step": 224930
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 3.290554023793928e-05,
+      "loss": 0.7801,
+      "step": 224940
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 3.290477858851128e-05,
+      "loss": 0.8309,
+      "step": 224950
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 3.290401693908328e-05,
+      "loss": 0.7367,
+      "step": 224960
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 3.290325528965528e-05,
+      "loss": 0.8797,
+      "step": 224970
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 3.290249364022728e-05,
+      "loss": 0.8506,
+      "step": 224980
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 3.2901731990799275e-05,
+      "loss": 0.7632,
+      "step": 224990
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 3.290097034137127e-05,
+      "loss": 0.8269,
+      "step": 225000
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 3.290020869194328e-05,
+      "loss": 0.822,
+      "step": 225010
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 3.2899447042515275e-05,
+      "loss": 0.9442,
+      "step": 225020
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 3.289868539308727e-05,
+      "loss": 0.8885,
+      "step": 225030
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 3.289792374365927e-05,
+      "loss": 0.7861,
+      "step": 225040
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 3.2897162094231274e-05,
+      "loss": 0.8853,
+      "step": 225050
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 3.2896400444803264e-05,
+      "loss": 0.7829,
+      "step": 225060
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 3.289563879537527e-05,
+      "loss": 0.918,
+      "step": 225070
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 3.2894877145947266e-05,
+      "loss": 0.9052,
+      "step": 225080
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 3.2894115496519264e-05,
+      "loss": 0.8181,
+      "step": 225090
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 3.289335384709126e-05,
+      "loss": 0.8226,
+      "step": 225100
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 3.2892592197663265e-05,
+      "loss": 0.8086,
+      "step": 225110
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 3.2891830548235256e-05,
+      "loss": 0.7869,
+      "step": 225120
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 3.289106889880726e-05,
+      "loss": 0.8856,
+      "step": 225130
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 3.289030724937926e-05,
+      "loss": 0.884,
+      "step": 225140
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 3.2889545599951255e-05,
+      "loss": 0.8844,
+      "step": 225150
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 3.288878395052325e-05,
+      "loss": 0.9648,
+      "step": 225160
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 3.288802230109526e-05,
+      "loss": 0.8113,
+      "step": 225170
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 3.2887260651667254e-05,
+      "loss": 0.8089,
+      "step": 225180
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 3.288649900223925e-05,
+      "loss": 0.9043,
+      "step": 225190
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 3.288573735281125e-05,
+      "loss": 0.8744,
+      "step": 225200
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 3.288497570338325e-05,
+      "loss": 0.9588,
+      "step": 225210
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 3.288421405395525e-05,
+      "loss": 0.9379,
+      "step": 225220
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 3.288345240452725e-05,
+      "loss": 0.9147,
+      "step": 225230
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 3.2882690755099246e-05,
+      "loss": 0.8513,
+      "step": 225240
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 3.288192910567124e-05,
+      "loss": 0.7932,
+      "step": 225250
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 3.288116745624324e-05,
+      "loss": 0.881,
+      "step": 225260
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 3.288040580681524e-05,
+      "loss": 0.8784,
+      "step": 225270
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 3.287964415738724e-05,
+      "loss": 0.8987,
+      "step": 225280
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 3.287888250795923e-05,
+      "loss": 0.9611,
+      "step": 225290
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 3.287812085853124e-05,
+      "loss": 0.8834,
+      "step": 225300
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 3.2877359209103235e-05,
+      "loss": 0.9131,
+      "step": 225310
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 3.287659755967523e-05,
+      "loss": 0.8623,
+      "step": 225320
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 3.287583591024723e-05,
+      "loss": 0.8729,
+      "step": 225330
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 3.2875074260819234e-05,
+      "loss": 0.8641,
+      "step": 225340
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 3.2874312611391225e-05,
+      "loss": 0.804,
+      "step": 225350
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 3.287355096196323e-05,
+      "loss": 0.7852,
+      "step": 225360
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 3.2872789312535227e-05,
+      "loss": 0.969,
+      "step": 225370
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 3.2872027663107224e-05,
+      "loss": 0.81,
+      "step": 225380
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 3.287126601367923e-05,
+      "loss": 0.8526,
+      "step": 225390
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 3.2870504364251226e-05,
+      "loss": 0.7978,
+      "step": 225400
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 3.286974271482322e-05,
+      "loss": 0.926,
+      "step": 225410
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 3.286898106539522e-05,
+      "loss": 0.8012,
+      "step": 225420
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 3.2868219415967225e-05,
+      "loss": 0.8254,
+      "step": 225430
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 3.2867457766539216e-05,
+      "loss": 0.8604,
+      "step": 225440
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 3.286669611711122e-05,
+      "loss": 0.8563,
+      "step": 225450
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 3.286593446768322e-05,
+      "loss": 0.7933,
+      "step": 225460
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 3.2865172818255215e-05,
+      "loss": 0.9121,
+      "step": 225470
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 3.286441116882721e-05,
+      "loss": 0.9411,
+      "step": 225480
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 3.2863649519399216e-05,
+      "loss": 0.7995,
+      "step": 225490
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 3.286288786997121e-05,
+      "loss": 0.8744,
+      "step": 225500
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 3.286212622054321e-05,
+      "loss": 0.7992,
+      "step": 225510
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 3.286136457111521e-05,
+      "loss": 0.8436,
+      "step": 225520
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 3.2860602921687206e-05,
+      "loss": 0.8446,
+      "step": 225530
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 3.2859841272259204e-05,
+      "loss": 0.9258,
+      "step": 225540
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 3.285907962283121e-05,
+      "loss": 0.8978,
+      "step": 225550
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 3.28583179734032e-05,
+      "loss": 0.8209,
+      "step": 225560
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 3.28575563239752e-05,
+      "loss": 0.904,
+      "step": 225570
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 3.28567946745472e-05,
+      "loss": 0.8017,
+      "step": 225580
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 3.28560330251192e-05,
+      "loss": 0.8174,
+      "step": 225590
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 3.28552713756912e-05,
+      "loss": 0.8496,
+      "step": 225600
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 3.28545097262632e-05,
+      "loss": 0.8318,
+      "step": 225610
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 3.28537480768352e-05,
+      "loss": 0.8243,
+      "step": 225620
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 3.2852986427407195e-05,
+      "loss": 0.8709,
+      "step": 225630
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 3.28522247779792e-05,
+      "loss": 0.8293,
+      "step": 225640
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 3.285146312855119e-05,
+      "loss": 0.8547,
+      "step": 225650
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 3.2850701479123194e-05,
+      "loss": 0.8062,
+      "step": 225660
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 3.284993982969519e-05,
+      "loss": 0.8601,
+      "step": 225670
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 3.284917818026719e-05,
+      "loss": 0.8135,
+      "step": 225680
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 3.2848416530839186e-05,
+      "loss": 0.792,
+      "step": 225690
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 3.284765488141119e-05,
+      "loss": 0.8684,
+      "step": 225700
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 3.284689323198318e-05,
+      "loss": 0.8646,
+      "step": 225710
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 3.2846131582555185e-05,
+      "loss": 0.8491,
+      "step": 225720
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 3.2845369933127176e-05,
+      "loss": 0.8707,
+      "step": 225730
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 3.284460828369918e-05,
+      "loss": 0.8767,
+      "step": 225740
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 3.284384663427118e-05,
+      "loss": 0.8456,
+      "step": 225750
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 3.2843084984843175e-05,
+      "loss": 0.8646,
+      "step": 225760
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 3.284232333541518e-05,
+      "loss": 0.8572,
+      "step": 225770
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 3.284156168598718e-05,
+      "loss": 0.8102,
+      "step": 225780
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 3.2840800036559174e-05,
+      "loss": 0.7962,
+      "step": 225790
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 3.284003838713117e-05,
+      "loss": 0.8509,
+      "step": 225800
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 3.2839276737703176e-05,
+      "loss": 0.8144,
+      "step": 225810
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 3.283851508827517e-05,
+      "loss": 0.8496,
+      "step": 225820
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 3.283775343884717e-05,
+      "loss": 0.7864,
+      "step": 225830
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 3.283699178941917e-05,
+      "loss": 0.8419,
+      "step": 225840
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 3.2836230139991166e-05,
+      "loss": 0.8192,
+      "step": 225850
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 3.283546849056316e-05,
+      "loss": 0.8276,
+      "step": 225860
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 3.283470684113517e-05,
+      "loss": 0.8405,
+      "step": 225870
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 3.283394519170716e-05,
+      "loss": 0.9061,
+      "step": 225880
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 3.283318354227916e-05,
+      "loss": 0.9287,
+      "step": 225890
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 3.283242189285116e-05,
+      "loss": 1.1077,
+      "step": 225900
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 3.283166024342316e-05,
+      "loss": 0.7864,
+      "step": 225910
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 3.2830898593995155e-05,
+      "loss": 0.7679,
+      "step": 225920
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 3.283013694456716e-05,
+      "loss": 0.8244,
+      "step": 225930
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 3.282937529513915e-05,
+      "loss": 0.8381,
+      "step": 225940
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 3.2828613645711154e-05,
+      "loss": 0.8752,
+      "step": 225950
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 3.282785199628315e-05,
+      "loss": 0.8102,
+      "step": 225960
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 3.282709034685515e-05,
+      "loss": 0.8597,
+      "step": 225970
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 3.282632869742715e-05,
+      "loss": 0.8298,
+      "step": 225980
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 3.282556704799915e-05,
+      "loss": 0.861,
+      "step": 225990
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 3.282480539857115e-05,
+      "loss": 0.8144,
+      "step": 226000
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 3.2824043749143146e-05,
+      "loss": 0.807,
+      "step": 226010
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 3.282328209971515e-05,
+      "loss": 0.9294,
+      "step": 226020
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 3.282252045028714e-05,
+      "loss": 0.8379,
+      "step": 226030
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 3.2821758800859145e-05,
+      "loss": 0.8129,
+      "step": 226040
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 3.282099715143114e-05,
+      "loss": 0.861,
+      "step": 226050
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 3.282023550200314e-05,
+      "loss": 0.8792,
+      "step": 226060
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 3.281947385257514e-05,
+      "loss": 0.8647,
+      "step": 226070
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 3.281871220314714e-05,
+      "loss": 0.7902,
+      "step": 226080
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 3.281795055371913e-05,
+      "loss": 0.8133,
+      "step": 226090
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 3.2817188904291136e-05,
+      "loss": 0.7632,
+      "step": 226100
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 3.2816427254863134e-05,
+      "loss": 0.894,
+      "step": 226110
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 3.281566560543513e-05,
+      "loss": 0.7977,
+      "step": 226120
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 3.281490395600713e-05,
+      "loss": 0.8775,
+      "step": 226130
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 3.281414230657913e-05,
+      "loss": 0.9009,
+      "step": 226140
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 3.2813380657151124e-05,
+      "loss": 0.7678,
+      "step": 226150
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 3.281261900772313e-05,
+      "loss": 0.9303,
+      "step": 226160
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 3.2811857358295125e-05,
+      "loss": 0.837,
+      "step": 226170
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 3.281109570886712e-05,
+      "loss": 0.8673,
+      "step": 226180
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 3.281033405943913e-05,
+      "loss": 0.7653,
+      "step": 226190
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 3.280957241001112e-05,
+      "loss": 0.7775,
+      "step": 226200
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 3.280881076058312e-05,
+      "loss": 1.0164,
+      "step": 226210
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 3.280804911115512e-05,
+      "loss": 0.8324,
+      "step": 226220
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 3.280728746172712e-05,
+      "loss": 0.8611,
+      "step": 226230
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 3.2806525812299115e-05,
+      "loss": 0.7892,
+      "step": 226240
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 3.280576416287112e-05,
+      "loss": 0.7851,
+      "step": 226250
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 3.280500251344311e-05,
+      "loss": 0.8069,
+      "step": 226260
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 3.2804240864015114e-05,
+      "loss": 0.9322,
+      "step": 226270
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 3.280347921458711e-05,
+      "loss": 0.8566,
+      "step": 226280
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 3.280271756515911e-05,
+      "loss": 0.8293,
+      "step": 226290
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 3.2801955915731106e-05,
+      "loss": 0.8403,
+      "step": 226300
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 3.280119426630311e-05,
+      "loss": 0.8776,
+      "step": 226310
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 3.28004326168751e-05,
+      "loss": 0.8145,
+      "step": 226320
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 3.2799670967447105e-05,
+      "loss": 0.9276,
+      "step": 226330
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 3.27989093180191e-05,
+      "loss": 0.8111,
+      "step": 226340
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 3.27981476685911e-05,
+      "loss": 0.8192,
+      "step": 226350
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 3.27973860191631e-05,
+      "loss": 0.7967,
+      "step": 226360
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 3.27966243697351e-05,
+      "loss": 0.861,
+      "step": 226370
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 3.27958627203071e-05,
+      "loss": 0.8744,
+      "step": 226380
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 3.27951010708791e-05,
+      "loss": 0.8254,
+      "step": 226390
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 3.27943394214511e-05,
+      "loss": 0.8214,
+      "step": 226400
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 3.279357777202309e-05,
+      "loss": 0.8413,
+      "step": 226410
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 3.2792816122595096e-05,
+      "loss": 0.7087,
+      "step": 226420
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 3.2792054473167094e-05,
+      "loss": 0.7842,
+      "step": 226430
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 3.279129282373909e-05,
+      "loss": 0.842,
+      "step": 226440
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 3.279053117431109e-05,
+      "loss": 0.8701,
+      "step": 226450
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 3.278976952488309e-05,
+      "loss": 0.7693,
+      "step": 226460
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 3.278900787545508e-05,
+      "loss": 0.841,
+      "step": 226470
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 3.278824622602709e-05,
+      "loss": 0.8587,
+      "step": 226480
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 3.2787484576599085e-05,
+      "loss": 0.7864,
+      "step": 226490
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 3.278672292717108e-05,
+      "loss": 0.7443,
+      "step": 226500
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 3.278596127774308e-05,
+      "loss": 0.7691,
+      "step": 226510
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 3.2785199628315084e-05,
+      "loss": 0.7795,
+      "step": 226520
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 3.2784437978887075e-05,
+      "loss": 0.7853,
+      "step": 226530
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 3.278367632945908e-05,
+      "loss": 0.901,
+      "step": 226540
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 3.278291468003108e-05,
+      "loss": 0.8076,
+      "step": 226550
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 3.2782153030603074e-05,
+      "loss": 0.781,
+      "step": 226560
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 3.278139138117508e-05,
+      "loss": 0.8316,
+      "step": 226570
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 3.2780629731747076e-05,
+      "loss": 0.7926,
+      "step": 226580
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 3.277986808231907e-05,
+      "loss": 0.8526,
+      "step": 226590
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 3.277910643289107e-05,
+      "loss": 0.8776,
+      "step": 226600
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 3.2778344783463075e-05,
+      "loss": 0.7667,
+      "step": 226610
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 3.2777583134035066e-05,
+      "loss": 0.8011,
+      "step": 226620
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 3.277682148460707e-05,
+      "loss": 0.8783,
+      "step": 226630
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 3.277605983517906e-05,
+      "loss": 0.9187,
+      "step": 226640
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 3.2775298185751065e-05,
+      "loss": 0.9287,
+      "step": 226650
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 3.277453653632306e-05,
+      "loss": 0.8256,
+      "step": 226660
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 3.277377488689506e-05,
+      "loss": 0.864,
+      "step": 226670
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 3.277301323746706e-05,
+      "loss": 0.7962,
+      "step": 226680
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 3.277225158803906e-05,
+      "loss": 0.8077,
+      "step": 226690
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 3.277148993861105e-05,
+      "loss": 0.9122,
+      "step": 226700
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 3.2770728289183056e-05,
+      "loss": 0.9163,
+      "step": 226710
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 3.2769966639755054e-05,
+      "loss": 0.9104,
+      "step": 226720
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 3.276920499032705e-05,
+      "loss": 0.887,
+      "step": 226730
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 3.276844334089905e-05,
+      "loss": 0.8773,
+      "step": 226740
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 3.276768169147105e-05,
+      "loss": 0.8748,
+      "step": 226750
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 3.276692004204305e-05,
+      "loss": 0.8246,
+      "step": 226760
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 3.276615839261505e-05,
+      "loss": 0.8896,
+      "step": 226770
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 3.276539674318705e-05,
+      "loss": 0.8095,
+      "step": 226780
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 3.276463509375904e-05,
+      "loss": 0.8191,
+      "step": 226790
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 3.276387344433105e-05,
+      "loss": 0.785,
+      "step": 226800
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 3.2763111794903045e-05,
+      "loss": 0.7999,
+      "step": 226810
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 3.276235014547504e-05,
+      "loss": 0.895,
+      "step": 226820
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 3.276158849604704e-05,
+      "loss": 0.8229,
+      "step": 226830
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 3.2760826846619044e-05,
+      "loss": 0.91,
+      "step": 226840
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 3.2760065197191035e-05,
+      "loss": 0.8858,
+      "step": 226850
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 3.275930354776304e-05,
+      "loss": 0.8678,
+      "step": 226860
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 3.2758541898335036e-05,
+      "loss": 1.006,
+      "step": 226870
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 3.2757780248907034e-05,
+      "loss": 0.8016,
+      "step": 226880
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 3.275701859947903e-05,
+      "loss": 0.916,
+      "step": 226890
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 3.2756256950051035e-05,
+      "loss": 0.773,
+      "step": 226900
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 3.2755495300623026e-05,
+      "loss": 0.8368,
+      "step": 226910
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 3.275473365119503e-05,
+      "loss": 0.8788,
+      "step": 226920
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 3.275397200176703e-05,
+      "loss": 0.9294,
+      "step": 226930
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 3.2753210352339025e-05,
+      "loss": 0.7843,
+      "step": 226940
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 3.275244870291102e-05,
+      "loss": 0.8738,
+      "step": 226950
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 3.275168705348303e-05,
+      "loss": 0.8292,
+      "step": 226960
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 3.2750925404055024e-05,
+      "loss": 0.8218,
+      "step": 226970
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 3.275016375462702e-05,
+      "loss": 0.9542,
+      "step": 226980
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 3.2749402105199026e-05,
+      "loss": 0.8692,
+      "step": 226990
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 3.274864045577102e-05,
+      "loss": 0.8764,
+      "step": 227000
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 3.274787880634302e-05,
+      "loss": 0.7747,
+      "step": 227010
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 3.274711715691502e-05,
+      "loss": 0.8817,
+      "step": 227020
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 3.2746355507487016e-05,
+      "loss": 0.8733,
+      "step": 227030
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 3.2745593858059014e-05,
+      "loss": 0.8098,
+      "step": 227040
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 3.274483220863102e-05,
+      "loss": 0.9097,
+      "step": 227050
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 3.274407055920301e-05,
+      "loss": 0.8324,
+      "step": 227060
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 3.274330890977501e-05,
+      "loss": 0.8707,
+      "step": 227070
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 3.274254726034701e-05,
+      "loss": 0.8011,
+      "step": 227080
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 3.274178561091901e-05,
+      "loss": 0.8203,
+      "step": 227090
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 3.2741023961491005e-05,
+      "loss": 0.8266,
+      "step": 227100
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 3.2740262312063e-05,
+      "loss": 0.9169,
+      "step": 227110
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 3.2739500662635e-05,
+      "loss": 0.8345,
+      "step": 227120
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 3.2738739013207004e-05,
+      "loss": 0.8526,
+      "step": 227130
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 3.2737977363779e-05,
+      "loss": 0.8929,
+      "step": 227140
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 3.2737215714351e-05,
+      "loss": 0.834,
+      "step": 227150
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 3.2736454064923e-05,
+      "loss": 0.9006,
+      "step": 227160
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 3.2735692415494994e-05,
+      "loss": 0.7665,
+      "step": 227170
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 3.2734930766067e-05,
+      "loss": 0.9238,
+      "step": 227180
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 3.2734169116638996e-05,
+      "loss": 0.7637,
+      "step": 227190
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 3.273340746721099e-05,
+      "loss": 0.8935,
+      "step": 227200
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 3.273264581778299e-05,
+      "loss": 0.8363,
+      "step": 227210
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 3.2731884168354995e-05,
+      "loss": 0.858,
+      "step": 227220
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 3.2731122518926986e-05,
+      "loss": 0.7677,
+      "step": 227230
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 3.273036086949899e-05,
+      "loss": 0.9035,
+      "step": 227240
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 3.272959922007099e-05,
+      "loss": 0.8733,
+      "step": 227250
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 3.2728837570642985e-05,
+      "loss": 0.8583,
+      "step": 227260
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 3.272807592121498e-05,
+      "loss": 0.9162,
+      "step": 227270
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 3.2727314271786987e-05,
+      "loss": 0.7934,
+      "step": 227280
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 3.272655262235898e-05,
+      "loss": 0.7853,
+      "step": 227290
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 3.272579097293098e-05,
+      "loss": 0.794,
+      "step": 227300
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 3.272502932350298e-05,
+      "loss": 0.7736,
+      "step": 227310
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 3.2724267674074976e-05,
+      "loss": 0.8173,
+      "step": 227320
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 3.2723506024646974e-05,
+      "loss": 0.8495,
+      "step": 227330
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 3.272274437521898e-05,
+      "loss": 0.8029,
+      "step": 227340
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 3.2721982725790976e-05,
+      "loss": 0.7624,
+      "step": 227350
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 3.272122107636297e-05,
+      "loss": 0.87,
+      "step": 227360
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 3.272045942693498e-05,
+      "loss": 0.7695,
+      "step": 227370
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 3.271969777750697e-05,
+      "loss": 0.8325,
+      "step": 227380
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 3.271893612807897e-05,
+      "loss": 0.791,
+      "step": 227390
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 3.271817447865097e-05,
+      "loss": 0.8355,
+      "step": 227400
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 3.271741282922297e-05,
+      "loss": 0.8381,
+      "step": 227410
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 3.2716651179794965e-05,
+      "loss": 0.8235,
+      "step": 227420
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 3.271588953036697e-05,
+      "loss": 0.8197,
+      "step": 227430
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 3.271512788093896e-05,
+      "loss": 0.8835,
+      "step": 227440
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 3.2714366231510964e-05,
+      "loss": 0.8931,
+      "step": 227450
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 3.271360458208296e-05,
+      "loss": 0.8323,
+      "step": 227460
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 3.271284293265496e-05,
+      "loss": 0.8936,
+      "step": 227470
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 3.2712081283226956e-05,
+      "loss": 0.8999,
+      "step": 227480
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 3.271131963379896e-05,
+      "loss": 0.964,
+      "step": 227490
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 3.271055798437095e-05,
+      "loss": 0.7567,
+      "step": 227500
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 3.2709796334942955e-05,
+      "loss": 0.8044,
+      "step": 227510
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 3.270903468551495e-05,
+      "loss": 0.9473,
+      "step": 227520
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 3.270827303608695e-05,
+      "loss": 0.9167,
+      "step": 227530
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 3.270751138665895e-05,
+      "loss": 0.8026,
+      "step": 227540
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 3.2706749737230945e-05,
+      "loss": 0.8423,
+      "step": 227550
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 3.270598808780295e-05,
+      "loss": 0.7916,
+      "step": 227560
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 3.270522643837495e-05,
+      "loss": 0.7378,
+      "step": 227570
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 3.2704464788946944e-05,
+      "loss": 0.9205,
+      "step": 227580
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 3.270370313951894e-05,
+      "loss": 0.9087,
+      "step": 227590
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 3.2702941490090946e-05,
+      "loss": 0.9272,
+      "step": 227600
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 3.270217984066294e-05,
+      "loss": 0.8592,
+      "step": 227610
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 3.270141819123494e-05,
+      "loss": 0.8173,
+      "step": 227620
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 3.270065654180694e-05,
+      "loss": 0.8382,
+      "step": 227630
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 3.2699894892378936e-05,
+      "loss": 0.8594,
+      "step": 227640
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 3.2699133242950934e-05,
+      "loss": 0.8424,
+      "step": 227650
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 3.269837159352294e-05,
+      "loss": 0.7762,
+      "step": 227660
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 3.269760994409493e-05,
+      "loss": 0.8418,
+      "step": 227670
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 3.269684829466693e-05,
+      "loss": 0.8838,
+      "step": 227680
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 3.269608664523893e-05,
+      "loss": 0.9146,
+      "step": 227690
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 3.269532499581093e-05,
+      "loss": 0.8625,
+      "step": 227700
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 3.2694563346382925e-05,
+      "loss": 0.8496,
+      "step": 227710
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 3.269380169695493e-05,
+      "loss": 0.8013,
+      "step": 227720
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 3.269304004752693e-05,
+      "loss": 0.8279,
+      "step": 227730
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 3.2692278398098924e-05,
+      "loss": 0.8292,
+      "step": 227740
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 3.269151674867092e-05,
+      "loss": 0.7755,
+      "step": 227750
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 3.269075509924292e-05,
+      "loss": 0.855,
+      "step": 227760
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 3.2689993449814923e-05,
+      "loss": 0.8646,
+      "step": 227770
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 3.268923180038692e-05,
+      "loss": 0.8037,
+      "step": 227780
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 3.268847015095892e-05,
+      "loss": 0.9059,
+      "step": 227790
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 3.2687708501530916e-05,
+      "loss": 0.8477,
+      "step": 227800
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 3.268694685210292e-05,
+      "loss": 0.8433,
+      "step": 227810
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 3.268618520267491e-05,
+      "loss": 0.8118,
+      "step": 227820
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 3.2685423553246915e-05,
+      "loss": 0.8575,
+      "step": 227830
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 3.268466190381891e-05,
+      "loss": 0.8329,
+      "step": 227840
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 3.268390025439091e-05,
+      "loss": 0.788,
+      "step": 227850
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 3.268313860496291e-05,
+      "loss": 0.7908,
+      "step": 227860
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 3.268237695553491e-05,
+      "loss": 0.9196,
+      "step": 227870
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 3.26816153061069e-05,
+      "loss": 0.921,
+      "step": 227880
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 3.268085365667891e-05,
+      "loss": 0.7901,
+      "step": 227890
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 3.2680092007250904e-05,
+      "loss": 0.7766,
+      "step": 227900
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 3.26793303578229e-05,
+      "loss": 0.9098,
+      "step": 227910
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 3.26785687083949e-05,
+      "loss": 0.8699,
+      "step": 227920
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 3.26778070589669e-05,
+      "loss": 0.9269,
+      "step": 227930
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 3.26770454095389e-05,
+      "loss": 0.9226,
+      "step": 227940
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 3.26762837601109e-05,
+      "loss": 0.807,
+      "step": 227950
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 3.2675522110682896e-05,
+      "loss": 0.8422,
+      "step": 227960
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 3.267476046125489e-05,
+      "loss": 0.9351,
+      "step": 227970
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 3.26739988118269e-05,
+      "loss": 0.7907,
+      "step": 227980
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 3.2673237162398895e-05,
+      "loss": 0.7978,
+      "step": 227990
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 3.267247551297089e-05,
+      "loss": 0.7758,
+      "step": 228000
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 3.267171386354289e-05,
+      "loss": 0.9522,
+      "step": 228010
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 3.267095221411489e-05,
+      "loss": 0.7326,
+      "step": 228020
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 3.2670190564686885e-05,
+      "loss": 0.8115,
+      "step": 228030
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 3.266942891525889e-05,
+      "loss": 0.7802,
+      "step": 228040
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 3.266866726583088e-05,
+      "loss": 0.8832,
+      "step": 228050
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 3.2667905616402884e-05,
+      "loss": 0.9019,
+      "step": 228060
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 3.266714396697488e-05,
+      "loss": 0.9496,
+      "step": 228070
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 3.266638231754688e-05,
+      "loss": 0.7955,
+      "step": 228080
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 3.2665620668118876e-05,
+      "loss": 0.854,
+      "step": 228090
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 3.266485901869088e-05,
+      "loss": 0.9035,
+      "step": 228100
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 3.266409736926288e-05,
+      "loss": 0.8622,
+      "step": 228110
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 3.2663335719834875e-05,
+      "loss": 0.8334,
+      "step": 228120
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 3.266257407040687e-05,
+      "loss": 0.897,
+      "step": 228130
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 3.266181242097887e-05,
+      "loss": 0.8529,
+      "step": 228140
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 3.2661050771550875e-05,
+      "loss": 0.8468,
+      "step": 228150
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 3.266028912212287e-05,
+      "loss": 0.909,
+      "step": 228160
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 3.265952747269487e-05,
+      "loss": 0.8049,
+      "step": 228170
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 3.265876582326687e-05,
+      "loss": 0.8088,
+      "step": 228180
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 3.265800417383887e-05,
+      "loss": 0.8677,
+      "step": 228190
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 3.265724252441086e-05,
+      "loss": 0.7521,
+      "step": 228200
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 3.2656480874982866e-05,
+      "loss": 0.8815,
+      "step": 228210
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 3.2655719225554864e-05,
+      "loss": 0.8234,
+      "step": 228220
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 3.265495757612686e-05,
+      "loss": 0.8105,
+      "step": 228230
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 3.265419592669886e-05,
+      "loss": 0.8355,
+      "step": 228240
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 3.265343427727086e-05,
+      "loss": 0.8928,
+      "step": 228250
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 3.2652672627842854e-05,
+      "loss": 0.7894,
+      "step": 228260
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 3.265191097841486e-05,
+      "loss": 0.7664,
+      "step": 228270
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 3.2651149328986855e-05,
+      "loss": 0.8549,
+      "step": 228280
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 3.265038767955885e-05,
+      "loss": 0.7593,
+      "step": 228290
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 3.264962603013085e-05,
+      "loss": 0.8325,
+      "step": 228300
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 3.2648864380702854e-05,
+      "loss": 0.9814,
+      "step": 228310
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 3.264810273127485e-05,
+      "loss": 0.8417,
+      "step": 228320
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 3.264734108184685e-05,
+      "loss": 0.8253,
+      "step": 228330
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 3.264657943241885e-05,
+      "loss": 0.8766,
+      "step": 228340
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 3.2645817782990844e-05,
+      "loss": 0.8458,
+      "step": 228350
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 3.264505613356285e-05,
+      "loss": 0.8355,
+      "step": 228360
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 3.2644294484134846e-05,
+      "loss": 0.9102,
+      "step": 228370
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 3.2643532834706843e-05,
+      "loss": 0.815,
+      "step": 228380
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 3.264277118527884e-05,
+      "loss": 0.72,
+      "step": 228390
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 3.2642009535850845e-05,
+      "loss": 0.7957,
+      "step": 228400
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 3.2641247886422836e-05,
+      "loss": 0.8617,
+      "step": 228410
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 3.264048623699484e-05,
+      "loss": 0.7472,
+      "step": 228420
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 3.263972458756684e-05,
+      "loss": 0.9019,
+      "step": 228430
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 3.2638962938138835e-05,
+      "loss": 0.8371,
+      "step": 228440
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 3.263820128871083e-05,
+      "loss": 0.7893,
+      "step": 228450
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 3.263743963928284e-05,
+      "loss": 0.8581,
+      "step": 228460
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 3.263667798985483e-05,
+      "loss": 0.8648,
+      "step": 228470
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 3.263591634042683e-05,
+      "loss": 0.7582,
+      "step": 228480
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 3.263515469099883e-05,
+      "loss": 0.8187,
+      "step": 228490
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 3.263439304157083e-05,
+      "loss": 0.7976,
+      "step": 228500
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 3.2633631392142824e-05,
+      "loss": 0.813,
+      "step": 228510
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 3.263286974271482e-05,
+      "loss": 0.7608,
+      "step": 228520
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 3.2632108093286826e-05,
+      "loss": 0.7414,
+      "step": 228530
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 3.263134644385882e-05,
+      "loss": 0.826,
+      "step": 228540
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 3.263058479443082e-05,
+      "loss": 0.7702,
+      "step": 228550
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 3.262982314500282e-05,
+      "loss": 0.7482,
+      "step": 228560
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 3.262906149557482e-05,
+      "loss": 0.8369,
+      "step": 228570
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 3.262829984614681e-05,
+      "loss": 0.9431,
+      "step": 228580
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 3.262753819671882e-05,
+      "loss": 0.7528,
+      "step": 228590
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 3.2626776547290815e-05,
+      "loss": 0.8081,
+      "step": 228600
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 3.262601489786281e-05,
+      "loss": 0.782,
+      "step": 228610
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 3.262525324843481e-05,
+      "loss": 0.8756,
+      "step": 228620
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 3.2624491599006814e-05,
+      "loss": 0.815,
+      "step": 228630
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 3.2623729949578805e-05,
+      "loss": 0.862,
+      "step": 228640
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 3.262296830015081e-05,
+      "loss": 0.8021,
+      "step": 228650
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 3.2622206650722806e-05,
+      "loss": 0.9064,
+      "step": 228660
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 3.2621445001294804e-05,
+      "loss": 0.8296,
+      "step": 228670
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 3.26206833518668e-05,
+      "loss": 0.9071,
+      "step": 228680
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 3.2619921702438806e-05,
+      "loss": 0.8306,
+      "step": 228690
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 3.26191600530108e-05,
+      "loss": 0.8487,
+      "step": 228700
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 3.26183984035828e-05,
+      "loss": 0.8697,
+      "step": 228710
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 3.26176367541548e-05,
+      "loss": 0.8286,
+      "step": 228720
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 3.2616875104726795e-05,
+      "loss": 0.7729,
+      "step": 228730
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 3.26161134552988e-05,
+      "loss": 0.8016,
+      "step": 228740
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 3.26153518058708e-05,
+      "loss": 0.8927,
+      "step": 228750
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 3.2614590156442795e-05,
+      "loss": 0.9146,
+      "step": 228760
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 3.261382850701479e-05,
+      "loss": 0.7714,
+      "step": 228770
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 3.2613066857586796e-05,
+      "loss": 0.7574,
+      "step": 228780
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 3.261230520815879e-05,
+      "loss": 0.9977,
+      "step": 228790
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 3.261154355873079e-05,
+      "loss": 0.8143,
+      "step": 228800
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 3.261078190930279e-05,
+      "loss": 0.7958,
+      "step": 228810
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 3.2610020259874786e-05,
+      "loss": 0.8801,
+      "step": 228820
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 3.2609258610446784e-05,
+      "loss": 0.864,
+      "step": 228830
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 3.260849696101879e-05,
+      "loss": 0.8189,
+      "step": 228840
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 3.260773531159078e-05,
+      "loss": 0.7911,
+      "step": 228850
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 3.260697366216278e-05,
+      "loss": 0.8023,
+      "step": 228860
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 3.260621201273478e-05,
+      "loss": 0.7617,
+      "step": 228870
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 3.260545036330678e-05,
+      "loss": 0.7902,
+      "step": 228880
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 3.2604688713878775e-05,
+      "loss": 0.8068,
+      "step": 228890
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 3.260392706445078e-05,
+      "loss": 0.9021,
+      "step": 228900
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 3.260316541502278e-05,
+      "loss": 0.848,
+      "step": 228910
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 3.2602403765594774e-05,
+      "loss": 0.8308,
+      "step": 228920
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 3.260164211616677e-05,
+      "loss": 0.8534,
+      "step": 228930
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 3.260088046673877e-05,
+      "loss": 0.867,
+      "step": 228940
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 3.2600118817310774e-05,
+      "loss": 0.7897,
+      "step": 228950
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 3.2599357167882764e-05,
+      "loss": 0.8,
+      "step": 228960
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 3.259859551845477e-05,
+      "loss": 0.8197,
+      "step": 228970
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 3.2597833869026766e-05,
+      "loss": 0.9469,
+      "step": 228980
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 3.2597072219598763e-05,
+      "loss": 0.7592,
+      "step": 228990
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 3.259631057017076e-05,
+      "loss": 0.7872,
+      "step": 229000
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 3.2595548920742765e-05,
+      "loss": 0.8542,
+      "step": 229010
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 3.2594787271314756e-05,
+      "loss": 0.7646,
+      "step": 229020
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 3.259402562188676e-05,
+      "loss": 0.793,
+      "step": 229030
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 3.259326397245876e-05,
+      "loss": 0.7937,
+      "step": 229040
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 3.2592502323030755e-05,
+      "loss": 0.8983,
+      "step": 229050
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 3.259174067360275e-05,
+      "loss": 0.9076,
+      "step": 229060
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 3.259097902417476e-05,
+      "loss": 0.8049,
+      "step": 229070
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 3.2590217374746754e-05,
+      "loss": 0.8241,
+      "step": 229080
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 3.258945572531875e-05,
+      "loss": 0.7814,
+      "step": 229090
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 3.258869407589075e-05,
+      "loss": 0.9153,
+      "step": 229100
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 3.258793242646275e-05,
+      "loss": 0.7854,
+      "step": 229110
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 3.258717077703475e-05,
+      "loss": 0.8072,
+      "step": 229120
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 3.258640912760675e-05,
+      "loss": 0.8301,
+      "step": 229130
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 3.2585647478178746e-05,
+      "loss": 0.8729,
+      "step": 229140
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 3.258488582875074e-05,
+      "loss": 0.8459,
+      "step": 229150
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 3.258412417932275e-05,
+      "loss": 0.766,
+      "step": 229160
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 3.258336252989474e-05,
+      "loss": 0.8206,
+      "step": 229170
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 3.258260088046674e-05,
+      "loss": 1.0004,
+      "step": 229180
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 3.258183923103874e-05,
+      "loss": 0.8764,
+      "step": 229190
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 3.258107758161074e-05,
+      "loss": 0.8075,
+      "step": 229200
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 3.2580315932182735e-05,
+      "loss": 0.9782,
+      "step": 229210
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 3.257955428275474e-05,
+      "loss": 0.8889,
+      "step": 229220
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 3.257879263332673e-05,
+      "loss": 0.8551,
+      "step": 229230
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 3.2578030983898734e-05,
+      "loss": 0.7648,
+      "step": 229240
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 3.257726933447073e-05,
+      "loss": 0.9065,
+      "step": 229250
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 3.257650768504273e-05,
+      "loss": 0.8463,
+      "step": 229260
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 3.2575746035614726e-05,
+      "loss": 0.8533,
+      "step": 229270
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 3.257498438618673e-05,
+      "loss": 0.9336,
+      "step": 229280
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 3.257422273675873e-05,
+      "loss": 0.8427,
+      "step": 229290
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 3.2573461087330726e-05,
+      "loss": 0.7934,
+      "step": 229300
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 3.257269943790272e-05,
+      "loss": 0.8587,
+      "step": 229310
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 3.257193778847472e-05,
+      "loss": 0.8521,
+      "step": 229320
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 3.2571176139046725e-05,
+      "loss": 0.8221,
+      "step": 229330
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 3.257041448961872e-05,
+      "loss": 0.8469,
+      "step": 229340
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 3.256965284019072e-05,
+      "loss": 0.8239,
+      "step": 229350
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 3.256889119076272e-05,
+      "loss": 0.8537,
+      "step": 229360
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 3.256812954133472e-05,
+      "loss": 0.9374,
+      "step": 229370
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 3.256736789190671e-05,
+      "loss": 0.8665,
+      "step": 229380
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 3.2566606242478716e-05,
+      "loss": 0.8809,
+      "step": 229390
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 3.256584459305071e-05,
+      "loss": 0.658,
+      "step": 229400
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 3.256508294362271e-05,
+      "loss": 0.9328,
+      "step": 229410
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 3.256432129419471e-05,
+      "loss": 0.9305,
+      "step": 229420
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 3.2563559644766706e-05,
+      "loss": 0.8413,
+      "step": 229430
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 3.2562797995338704e-05,
+      "loss": 0.8997,
+      "step": 229440
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 3.256203634591071e-05,
+      "loss": 0.8775,
+      "step": 229450
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 3.25612746964827e-05,
+      "loss": 0.8534,
+      "step": 229460
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 3.25605130470547e-05,
+      "loss": 0.8405,
+      "step": 229470
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 3.25597513976267e-05,
+      "loss": 0.738,
+      "step": 229480
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 3.25589897481987e-05,
+      "loss": 0.848,
+      "step": 229490
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 3.25582280987707e-05,
+      "loss": 0.8449,
+      "step": 229500
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 3.25574664493427e-05,
+      "loss": 0.8419,
+      "step": 229510
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 3.25567047999147e-05,
+      "loss": 0.815,
+      "step": 229520
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 3.2555943150486694e-05,
+      "loss": 0.8045,
+      "step": 229530
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 3.25551815010587e-05,
+      "loss": 0.9014,
+      "step": 229540
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 3.255441985163069e-05,
+      "loss": 0.9174,
+      "step": 229550
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 3.2553658202202694e-05,
+      "loss": 0.8729,
+      "step": 229560
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 3.255289655277469e-05,
+      "loss": 0.8361,
+      "step": 229570
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 3.255213490334669e-05,
+      "loss": 0.8138,
+      "step": 229580
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 3.2551373253918686e-05,
+      "loss": 0.9491,
+      "step": 229590
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 3.255061160449069e-05,
+      "loss": 0.7729,
+      "step": 229600
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 3.254984995506268e-05,
+      "loss": 0.8869,
+      "step": 229610
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 3.2549088305634685e-05,
+      "loss": 0.8346,
+      "step": 229620
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 3.254832665620668e-05,
+      "loss": 0.84,
+      "step": 229630
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 3.254756500677868e-05,
+      "loss": 0.9396,
+      "step": 229640
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 3.254680335735068e-05,
+      "loss": 0.8267,
+      "step": 229650
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 3.254604170792268e-05,
+      "loss": 0.8846,
+      "step": 229660
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 3.254528005849468e-05,
+      "loss": 0.7945,
+      "step": 229670
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 3.254451840906668e-05,
+      "loss": 0.8374,
+      "step": 229680
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 3.2543756759638674e-05,
+      "loss": 0.7199,
+      "step": 229690
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 3.254299511021067e-05,
+      "loss": 0.9258,
+      "step": 229700
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 3.2542233460782676e-05,
+      "loss": 0.8288,
+      "step": 229710
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 3.2541471811354673e-05,
+      "loss": 0.9487,
+      "step": 229720
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 3.254071016192667e-05,
+      "loss": 0.8131,
+      "step": 229730
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 3.253994851249867e-05,
+      "loss": 0.873,
+      "step": 229740
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 3.253918686307067e-05,
+      "loss": 0.8411,
+      "step": 229750
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 3.253842521364266e-05,
+      "loss": 0.8454,
+      "step": 229760
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 3.253766356421467e-05,
+      "loss": 0.9295,
+      "step": 229770
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 3.2536901914786665e-05,
+      "loss": 0.9491,
+      "step": 229780
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 3.253614026535866e-05,
+      "loss": 0.8998,
+      "step": 229790
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 3.253537861593066e-05,
+      "loss": 0.9383,
+      "step": 229800
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 3.2534616966502664e-05,
+      "loss": 0.8361,
+      "step": 229810
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 3.2533855317074655e-05,
+      "loss": 0.8429,
+      "step": 229820
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 3.253309366764666e-05,
+      "loss": 0.8098,
+      "step": 229830
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 3.2532332018218657e-05,
+      "loss": 0.8087,
+      "step": 229840
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 3.2531570368790654e-05,
+      "loss": 0.7469,
+      "step": 229850
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 3.253080871936265e-05,
+      "loss": 0.7985,
+      "step": 229860
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 3.253004706993465e-05,
+      "loss": 0.7692,
+      "step": 229870
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 3.252928542050665e-05,
+      "loss": 0.832,
+      "step": 229880
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 3.252852377107865e-05,
+      "loss": 0.8685,
+      "step": 229890
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 3.252776212165065e-05,
+      "loss": 0.7611,
+      "step": 229900
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 3.2527000472222646e-05,
+      "loss": 0.9008,
+      "step": 229910
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 3.252623882279465e-05,
+      "loss": 0.8223,
+      "step": 229920
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 3.252547717336664e-05,
+      "loss": 0.8017,
+      "step": 229930
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 3.2524715523938645e-05,
+      "loss": 0.8234,
+      "step": 229940
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 3.252395387451064e-05,
+      "loss": 0.8421,
+      "step": 229950
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 3.252319222508264e-05,
+      "loss": 0.8283,
+      "step": 229960
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 3.252243057565464e-05,
+      "loss": 0.8622,
+      "step": 229970
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 3.252166892622664e-05,
+      "loss": 1.0212,
+      "step": 229980
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 3.252090727679863e-05,
+      "loss": 0.8783,
+      "step": 229990
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 3.2520145627370636e-05,
+      "loss": 0.8359,
+      "step": 230000
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 3.2519383977942634e-05,
+      "loss": 0.9262,
+      "step": 230010
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 3.251862232851463e-05,
+      "loss": 0.8155,
+      "step": 230020
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 3.251786067908663e-05,
+      "loss": 0.7449,
+      "step": 230030
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 3.251709902965863e-05,
+      "loss": 0.8242,
+      "step": 230040
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 3.2516337380230624e-05,
+      "loss": 0.8372,
+      "step": 230050
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 3.251557573080263e-05,
+      "loss": 0.9176,
+      "step": 230060
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 3.2514814081374625e-05,
+      "loss": 0.7969,
+      "step": 230070
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 3.251405243194662e-05,
+      "loss": 0.9697,
+      "step": 230080
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 3.251329078251863e-05,
+      "loss": 0.8396,
+      "step": 230090
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 3.2512529133090625e-05,
+      "loss": 0.7531,
+      "step": 230100
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 3.251176748366262e-05,
+      "loss": 0.803,
+      "step": 230110
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 3.251100583423462e-05,
+      "loss": 0.9144,
+      "step": 230120
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 3.2510244184806624e-05,
+      "loss": 0.8206,
+      "step": 230130
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 3.2509482535378614e-05,
+      "loss": 0.9212,
+      "step": 230140
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 3.250872088595062e-05,
+      "loss": 0.8483,
+      "step": 230150
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 3.2507959236522616e-05,
+      "loss": 0.8241,
+      "step": 230160
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 3.2507197587094614e-05,
+      "loss": 0.8076,
+      "step": 230170
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 3.250643593766661e-05,
+      "loss": 0.8625,
+      "step": 230180
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 3.2505674288238615e-05,
+      "loss": 0.7906,
+      "step": 230190
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 3.2504912638810606e-05,
+      "loss": 0.9871,
+      "step": 230200
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 3.250415098938261e-05,
+      "loss": 0.8107,
+      "step": 230210
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 3.250338933995461e-05,
+      "loss": 0.9068,
+      "step": 230220
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 3.2502627690526605e-05,
+      "loss": 0.7967,
+      "step": 230230
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 3.25018660410986e-05,
+      "loss": 0.7818,
+      "step": 230240
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 3.250110439167061e-05,
+      "loss": 0.9087,
+      "step": 230250
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 3.2500342742242604e-05,
+      "loss": 0.7888,
+      "step": 230260
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 3.24995810928146e-05,
+      "loss": 0.8542,
+      "step": 230270
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 3.24988194433866e-05,
+      "loss": 0.8293,
+      "step": 230280
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 3.24980577939586e-05,
+      "loss": 0.9397,
+      "step": 230290
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 3.24972961445306e-05,
+      "loss": 0.8052,
+      "step": 230300
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 3.24965344951026e-05,
+      "loss": 0.8637,
+      "step": 230310
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 3.2495772845674596e-05,
+      "loss": 0.7809,
+      "step": 230320
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 3.2495011196246593e-05,
+      "loss": 0.9613,
+      "step": 230330
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 3.249424954681859e-05,
+      "loss": 0.8614,
+      "step": 230340
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 3.249348789739059e-05,
+      "loss": 0.9345,
+      "step": 230350
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 3.249272624796259e-05,
+      "loss": 0.8365,
+      "step": 230360
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 3.249196459853458e-05,
+      "loss": 0.8049,
+      "step": 230370
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 3.249120294910659e-05,
+      "loss": 0.8164,
+      "step": 230380
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 3.2490441299678585e-05,
+      "loss": 0.8108,
+      "step": 230390
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 3.248967965025058e-05,
+      "loss": 0.963,
+      "step": 230400
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 3.248891800082258e-05,
+      "loss": 0.8011,
+      "step": 230410
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 3.2488156351394584e-05,
+      "loss": 0.7903,
+      "step": 230420
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 3.2487394701966575e-05,
+      "loss": 0.8239,
+      "step": 230430
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 3.248663305253858e-05,
+      "loss": 0.8402,
+      "step": 230440
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 3.2485871403110577e-05,
+      "loss": 0.8542,
+      "step": 230450
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 3.2485109753682574e-05,
+      "loss": 0.9573,
+      "step": 230460
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 3.248434810425458e-05,
+      "loss": 0.8855,
+      "step": 230470
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 3.2483586454826576e-05,
+      "loss": 0.8767,
+      "step": 230480
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 3.248282480539857e-05,
+      "loss": 0.8682,
+      "step": 230490
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 3.248206315597057e-05,
+      "loss": 0.8602,
+      "step": 230500
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 3.2481301506542575e-05,
+      "loss": 0.8241,
+      "step": 230510
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 3.2480539857114566e-05,
+      "loss": 0.8708,
+      "step": 230520
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 3.247977820768657e-05,
+      "loss": 0.883,
+      "step": 230530
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 3.247901655825857e-05,
+      "loss": 0.9245,
+      "step": 230540
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 3.2478254908830565e-05,
+      "loss": 0.8131,
+      "step": 230550
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 3.247749325940256e-05,
+      "loss": 0.7619,
+      "step": 230560
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 3.2476731609974567e-05,
+      "loss": 0.8105,
+      "step": 230570
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 3.247596996054656e-05,
+      "loss": 0.7818,
+      "step": 230580
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 3.247520831111856e-05,
+      "loss": 0.891,
+      "step": 230590
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 3.247444666169056e-05,
+      "loss": 0.8768,
+      "step": 230600
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 3.2473685012262556e-05,
+      "loss": 0.8374,
+      "step": 230610
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 3.2472923362834554e-05,
+      "loss": 0.7826,
+      "step": 230620
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 3.247216171340656e-05,
+      "loss": 0.8163,
+      "step": 230630
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 3.247140006397855e-05,
+      "loss": 0.8703,
+      "step": 230640
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 3.247063841455055e-05,
+      "loss": 0.7888,
+      "step": 230650
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 3.246987676512255e-05,
+      "loss": 0.9047,
+      "step": 230660
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 3.246911511569455e-05,
+      "loss": 0.8805,
+      "step": 230670
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 3.246835346626655e-05,
+      "loss": 0.8382,
+      "step": 230680
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 3.246759181683855e-05,
+      "loss": 0.9481,
+      "step": 230690
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 3.246683016741055e-05,
+      "loss": 0.8577,
+      "step": 230700
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 3.2466068517982545e-05,
+      "loss": 0.8266,
+      "step": 230710
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 3.246530686855455e-05,
+      "loss": 0.9195,
+      "step": 230720
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 3.246454521912654e-05,
+      "loss": 0.8408,
+      "step": 230730
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 3.2463783569698544e-05,
+      "loss": 0.8948,
+      "step": 230740
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 3.246302192027054e-05,
+      "loss": 0.8375,
+      "step": 230750
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 3.246226027084254e-05,
+      "loss": 0.8901,
+      "step": 230760
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 3.2461498621414536e-05,
+      "loss": 0.7106,
+      "step": 230770
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 3.2460736971986534e-05,
+      "loss": 0.8232,
+      "step": 230780
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 3.245997532255853e-05,
+      "loss": 0.7945,
+      "step": 230790
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 3.2459213673130535e-05,
+      "loss": 0.8163,
+      "step": 230800
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 3.2458452023702526e-05,
+      "loss": 0.8409,
+      "step": 230810
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 3.245769037427453e-05,
+      "loss": 0.8847,
+      "step": 230820
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 3.245692872484653e-05,
+      "loss": 0.7169,
+      "step": 230830
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 3.2456167075418525e-05,
+      "loss": 0.7142,
+      "step": 230840
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 3.245540542599052e-05,
+      "loss": 0.8111,
+      "step": 230850
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 3.245464377656253e-05,
+      "loss": 0.8269,
+      "step": 230860
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 3.2453882127134524e-05,
+      "loss": 0.8414,
+      "step": 230870
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 3.245312047770652e-05,
+      "loss": 0.8801,
+      "step": 230880
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 3.2452358828278526e-05,
+      "loss": 0.8811,
+      "step": 230890
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 3.245159717885052e-05,
+      "loss": 0.8476,
+      "step": 230900
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 3.245083552942252e-05,
+      "loss": 0.8395,
+      "step": 230910
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 3.245007387999452e-05,
+      "loss": 0.8475,
+      "step": 230920
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 3.2449312230566516e-05,
+      "loss": 0.8822,
+      "step": 230930
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 3.2448550581138513e-05,
+      "loss": 0.7779,
+      "step": 230940
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 3.244778893171052e-05,
+      "loss": 0.8689,
+      "step": 230950
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 3.244702728228251e-05,
+      "loss": 0.8678,
+      "step": 230960
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 3.244626563285451e-05,
+      "loss": 0.8701,
+      "step": 230970
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 3.244550398342651e-05,
+      "loss": 0.9148,
+      "step": 230980
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 3.244474233399851e-05,
+      "loss": 0.7919,
+      "step": 230990
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 3.2443980684570505e-05,
+      "loss": 0.9134,
+      "step": 231000
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 3.244321903514251e-05,
+      "loss": 0.854,
+      "step": 231010
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 3.24424573857145e-05,
+      "loss": 0.8894,
+      "step": 231020
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 3.2441695736286504e-05,
+      "loss": 0.789,
+      "step": 231030
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 3.24409340868585e-05,
+      "loss": 0.8026,
+      "step": 231040
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 3.24401724374305e-05,
+      "loss": 0.8576,
+      "step": 231050
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 3.24394107880025e-05,
+      "loss": 0.9365,
+      "step": 231060
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 3.24386491385745e-05,
+      "loss": 0.769,
+      "step": 231070
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 3.24378874891465e-05,
+      "loss": 0.8138,
+      "step": 231080
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 3.2437125839718496e-05,
+      "loss": 0.8092,
+      "step": 231090
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 3.24363641902905e-05,
+      "loss": 0.7769,
+      "step": 231100
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 3.243560254086249e-05,
+      "loss": 0.9618,
+      "step": 231110
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 3.2434840891434495e-05,
+      "loss": 0.9113,
+      "step": 231120
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 3.243407924200649e-05,
+      "loss": 0.7892,
+      "step": 231130
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 3.243331759257849e-05,
+      "loss": 0.8738,
+      "step": 231140
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 3.243255594315049e-05,
+      "loss": 0.848,
+      "step": 231150
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 3.243179429372249e-05,
+      "loss": 0.7586,
+      "step": 231160
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 3.243103264429448e-05,
+      "loss": 0.9075,
+      "step": 231170
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 3.2430270994866487e-05,
+      "loss": 0.7899,
+      "step": 231180
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 3.2429509345438484e-05,
+      "loss": 0.8307,
+      "step": 231190
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 3.242874769601048e-05,
+      "loss": 0.9619,
+      "step": 231200
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 3.242798604658248e-05,
+      "loss": 0.8527,
+      "step": 231210
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 3.242722439715448e-05,
+      "loss": 0.8547,
+      "step": 231220
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 3.2426462747726474e-05,
+      "loss": 0.8499,
+      "step": 231230
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 3.242570109829848e-05,
+      "loss": 0.8107,
+      "step": 231240
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 3.2424939448870476e-05,
+      "loss": 0.8421,
+      "step": 231250
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 3.242417779944247e-05,
+      "loss": 0.84,
+      "step": 231260
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 3.242341615001448e-05,
+      "loss": 0.8262,
+      "step": 231270
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 3.242265450058647e-05,
+      "loss": 0.8182,
+      "step": 231280
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 3.242189285115847e-05,
+      "loss": 0.7872,
+      "step": 231290
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 3.242113120173047e-05,
+      "loss": 0.8847,
+      "step": 231300
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 3.242036955230247e-05,
+      "loss": 0.8,
+      "step": 231310
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 3.2419607902874465e-05,
+      "loss": 0.7757,
+      "step": 231320
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 3.241884625344647e-05,
+      "loss": 0.9184,
+      "step": 231330
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 3.241808460401846e-05,
+      "loss": 0.8286,
+      "step": 231340
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 3.2417322954590464e-05,
+      "loss": 0.7729,
+      "step": 231350
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 3.241656130516246e-05,
+      "loss": 0.8005,
+      "step": 231360
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 3.241579965573446e-05,
+      "loss": 0.8819,
+      "step": 231370
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 3.2415038006306456e-05,
+      "loss": 0.8732,
+      "step": 231380
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 3.241427635687846e-05,
+      "loss": 0.9285,
+      "step": 231390
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 3.241351470745045e-05,
+      "loss": 0.9107,
+      "step": 231400
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 3.2412753058022455e-05,
+      "loss": 0.8136,
+      "step": 231410
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 3.241199140859445e-05,
+      "loss": 0.7653,
+      "step": 231420
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 3.241122975916645e-05,
+      "loss": 0.7991,
+      "step": 231430
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 3.241046810973845e-05,
+      "loss": 1.0183,
+      "step": 231440
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 3.240970646031045e-05,
+      "loss": 0.82,
+      "step": 231450
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 3.240894481088245e-05,
+      "loss": 0.7351,
+      "step": 231460
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 3.240818316145445e-05,
+      "loss": 0.888,
+      "step": 231470
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 3.240742151202645e-05,
+      "loss": 0.884,
+      "step": 231480
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 3.240665986259844e-05,
+      "loss": 0.894,
+      "step": 231490
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 3.2405898213170446e-05,
+      "loss": 0.8869,
+      "step": 231500
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 3.2405136563742444e-05,
+      "loss": 0.8253,
+      "step": 231510
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 3.240437491431444e-05,
+      "loss": 0.9248,
+      "step": 231520
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 3.240361326488644e-05,
+      "loss": 0.9346,
+      "step": 231530
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 3.240285161545844e-05,
+      "loss": 0.8088,
+      "step": 231540
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 3.2402089966030433e-05,
+      "loss": 0.8614,
+      "step": 231550
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 3.240132831660244e-05,
+      "loss": 0.9401,
+      "step": 231560
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 3.2400566667174435e-05,
+      "loss": 0.8109,
+      "step": 231570
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 3.239980501774643e-05,
+      "loss": 0.7732,
+      "step": 231580
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 3.239904336831843e-05,
+      "loss": 0.8979,
+      "step": 231590
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 3.2398281718890434e-05,
+      "loss": 0.753,
+      "step": 231600
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 3.2397520069462425e-05,
+      "loss": 0.7263,
+      "step": 231610
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 3.239675842003443e-05,
+      "loss": 0.8189,
+      "step": 231620
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 3.239599677060643e-05,
+      "loss": 0.8932,
+      "step": 231630
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 3.2395235121178424e-05,
+      "loss": 0.8858,
+      "step": 231640
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 3.239447347175042e-05,
+      "loss": 0.7733,
+      "step": 231650
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 3.2393711822322426e-05,
+      "loss": 0.7902,
+      "step": 231660
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 3.239295017289442e-05,
+      "loss": 0.874,
+      "step": 231670
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 3.239218852346642e-05,
+      "loss": 0.8892,
+      "step": 231680
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 3.2391426874038425e-05,
+      "loss": 0.8827,
+      "step": 231690
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 3.2390665224610416e-05,
+      "loss": 0.8056,
+      "step": 231700
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 3.238990357518242e-05,
+      "loss": 0.8978,
+      "step": 231710
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 3.238914192575441e-05,
+      "loss": 0.8291,
+      "step": 231720
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 3.2388380276326415e-05,
+      "loss": 0.7987,
+      "step": 231730
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 3.238761862689841e-05,
+      "loss": 0.8709,
+      "step": 231740
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 3.238685697747041e-05,
+      "loss": 0.7757,
+      "step": 231750
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 3.238609532804241e-05,
+      "loss": 0.8779,
+      "step": 231760
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 3.238533367861441e-05,
+      "loss": 0.8369,
+      "step": 231770
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 3.23845720291864e-05,
+      "loss": 0.8302,
+      "step": 231780
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 3.2383810379758407e-05,
+      "loss": 0.8941,
+      "step": 231790
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 3.2383048730330404e-05,
+      "loss": 0.9089,
+      "step": 231800
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 3.23822870809024e-05,
+      "loss": 0.8289,
+      "step": 231810
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 3.23815254314744e-05,
+      "loss": 0.847,
+      "step": 231820
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 3.23807637820464e-05,
+      "loss": 0.8537,
+      "step": 231830
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 3.23800021326184e-05,
+      "loss": 0.8947,
+      "step": 231840
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 3.23792404831904e-05,
+      "loss": 0.958,
+      "step": 231850
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 3.23784788337624e-05,
+      "loss": 0.8907,
+      "step": 231860
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 3.237771718433439e-05,
+      "loss": 0.7898,
+      "step": 231870
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 3.23769555349064e-05,
+      "loss": 0.8941,
+      "step": 231880
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 3.2376193885478395e-05,
+      "loss": 0.8227,
+      "step": 231890
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 3.237543223605039e-05,
+      "loss": 0.8856,
+      "step": 231900
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 3.237467058662239e-05,
+      "loss": 0.8591,
+      "step": 231910
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 3.2373908937194394e-05,
+      "loss": 0.8287,
+      "step": 231920
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 3.2373147287766385e-05,
+      "loss": 0.7784,
+      "step": 231930
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 3.237238563833839e-05,
+      "loss": 0.8858,
+      "step": 231940
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 3.2371623988910386e-05,
+      "loss": 0.8282,
+      "step": 231950
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 3.2370862339482384e-05,
+      "loss": 0.8508,
+      "step": 231960
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 3.237010069005438e-05,
+      "loss": 0.8395,
+      "step": 231970
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 3.2369339040626386e-05,
+      "loss": 0.9002,
+      "step": 231980
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 3.2368577391198376e-05,
+      "loss": 0.8158,
+      "step": 231990
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 3.236781574177038e-05,
+      "loss": 0.8542,
+      "step": 232000
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 3.236705409234238e-05,
+      "loss": 0.9097,
+      "step": 232010
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 3.2366292442914375e-05,
+      "loss": 0.8417,
+      "step": 232020
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 3.236553079348637e-05,
+      "loss": 0.77,
+      "step": 232030
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 3.236476914405838e-05,
+      "loss": 0.8433,
+      "step": 232040
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 3.2364007494630375e-05,
+      "loss": 0.7945,
+      "step": 232050
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 3.236324584520237e-05,
+      "loss": 0.8889,
+      "step": 232060
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 3.2362484195774376e-05,
+      "loss": 0.8769,
+      "step": 232070
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 3.236172254634637e-05,
+      "loss": 0.8928,
+      "step": 232080
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 3.236096089691837e-05,
+      "loss": 0.8221,
+      "step": 232090
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 3.236019924749037e-05,
+      "loss": 0.7865,
+      "step": 232100
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 3.2359437598062366e-05,
+      "loss": 0.8991,
+      "step": 232110
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 3.2358675948634364e-05,
+      "loss": 0.917,
+      "step": 232120
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 3.235791429920637e-05,
+      "loss": 0.9542,
+      "step": 232130
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 3.235715264977836e-05,
+      "loss": 0.9073,
+      "step": 232140
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 3.235639100035036e-05,
+      "loss": 0.9237,
+      "step": 232150
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 3.2355629350922353e-05,
+      "loss": 0.8667,
+      "step": 232160
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 3.235486770149436e-05,
+      "loss": 0.7744,
+      "step": 232170
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 3.2354106052066355e-05,
+      "loss": 0.82,
+      "step": 232180
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 3.235334440263835e-05,
+      "loss": 0.8304,
+      "step": 232190
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 3.235258275321035e-05,
+      "loss": 0.7811,
+      "step": 232200
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 3.2351821103782354e-05,
+      "loss": 0.9308,
+      "step": 232210
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 3.235105945435435e-05,
+      "loss": 0.8445,
+      "step": 232220
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 3.235029780492635e-05,
+      "loss": 0.8243,
+      "step": 232230
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 3.234953615549835e-05,
+      "loss": 0.7599,
+      "step": 232240
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 3.2348774506070344e-05,
+      "loss": 0.7991,
+      "step": 232250
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 3.234801285664235e-05,
+      "loss": 0.9111,
+      "step": 232260
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 3.2347251207214346e-05,
+      "loss": 0.9445,
+      "step": 232270
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 3.2346489557786343e-05,
+      "loss": 0.8637,
+      "step": 232280
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 3.234572790835834e-05,
+      "loss": 0.9514,
+      "step": 232290
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 3.2344966258930345e-05,
+      "loss": 0.858,
+      "step": 232300
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 3.2344204609502336e-05,
+      "loss": 0.772,
+      "step": 232310
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 3.234344296007434e-05,
+      "loss": 0.9254,
+      "step": 232320
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 3.234268131064634e-05,
+      "loss": 0.864,
+      "step": 232330
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 3.2341919661218335e-05,
+      "loss": 0.7304,
+      "step": 232340
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 3.234115801179033e-05,
+      "loss": 0.9386,
+      "step": 232350
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 3.234039636236234e-05,
+      "loss": 0.7931,
+      "step": 232360
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 3.233963471293433e-05,
+      "loss": 0.8153,
+      "step": 232370
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 3.233887306350633e-05,
+      "loss": 0.8073,
+      "step": 232380
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 3.233811141407833e-05,
+      "loss": 0.7662,
+      "step": 232390
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 3.2337349764650327e-05,
+      "loss": 0.8364,
+      "step": 232400
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 3.2336588115222324e-05,
+      "loss": 0.8542,
+      "step": 232410
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 3.233582646579433e-05,
+      "loss": 0.8453,
+      "step": 232420
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 3.2335064816366326e-05,
+      "loss": 1.0966,
+      "step": 232430
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 3.233430316693832e-05,
+      "loss": 0.8639,
+      "step": 232440
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 3.233354151751032e-05,
+      "loss": 0.8237,
+      "step": 232450
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 3.233277986808232e-05,
+      "loss": 0.9016,
+      "step": 232460
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 3.233201821865432e-05,
+      "loss": 0.8325,
+      "step": 232470
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 3.233125656922632e-05,
+      "loss": 0.899,
+      "step": 232480
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 3.233049491979832e-05,
+      "loss": 0.7843,
+      "step": 232490
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 3.2329733270370315e-05,
+      "loss": 0.9116,
+      "step": 232500
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 3.232897162094232e-05,
+      "loss": 0.9135,
+      "step": 232510
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 3.232820997151431e-05,
+      "loss": 0.8555,
+      "step": 232520
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 3.2327448322086314e-05,
+      "loss": 0.9229,
+      "step": 232530
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 3.232668667265831e-05,
+      "loss": 0.8707,
+      "step": 232540
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 3.232592502323031e-05,
+      "loss": 0.8553,
+      "step": 232550
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 3.2325163373802306e-05,
+      "loss": 0.7967,
+      "step": 232560
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 3.232440172437431e-05,
+      "loss": 0.9847,
+      "step": 232570
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 3.23236400749463e-05,
+      "loss": 0.8482,
+      "step": 232580
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 3.2322878425518306e-05,
+      "loss": 0.7505,
+      "step": 232590
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 3.23221167760903e-05,
+      "loss": 0.8832,
+      "step": 232600
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 3.23213551266623e-05,
+      "loss": 0.8188,
+      "step": 232610
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 3.23205934772343e-05,
+      "loss": 0.8925,
+      "step": 232620
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 3.2319831827806295e-05,
+      "loss": 0.9537,
+      "step": 232630
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 3.23190701783783e-05,
+      "loss": 0.8466,
+      "step": 232640
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 3.23183085289503e-05,
+      "loss": 0.8086,
+      "step": 232650
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 3.2317546879522295e-05,
+      "loss": 0.8977,
+      "step": 232660
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 3.231678523009429e-05,
+      "loss": 0.8763,
+      "step": 232670
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 3.2316023580666296e-05,
+      "loss": 0.7984,
+      "step": 232680
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 3.231526193123829e-05,
+      "loss": 0.8654,
+      "step": 232690
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 3.231450028181029e-05,
+      "loss": 0.7563,
+      "step": 232700
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 3.231373863238229e-05,
+      "loss": 0.8346,
+      "step": 232710
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 3.2312976982954286e-05,
+      "loss": 0.7783,
+      "step": 232720
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 3.2312215333526284e-05,
+      "loss": 0.846,
+      "step": 232730
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 3.231145368409829e-05,
+      "loss": 0.8338,
+      "step": 232740
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 3.231069203467028e-05,
+      "loss": 0.8666,
+      "step": 232750
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 3.230993038524228e-05,
+      "loss": 0.8505,
+      "step": 232760
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 3.230916873581428e-05,
+      "loss": 0.824,
+      "step": 232770
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 3.230840708638628e-05,
+      "loss": 0.9036,
+      "step": 232780
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 3.2307645436958275e-05,
+      "loss": 0.9059,
+      "step": 232790
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 3.230688378753028e-05,
+      "loss": 0.8153,
+      "step": 232800
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 3.230612213810228e-05,
+      "loss": 0.8002,
+      "step": 232810
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 3.2305360488674274e-05,
+      "loss": 0.7928,
+      "step": 232820
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 3.230459883924627e-05,
+      "loss": 0.7988,
+      "step": 232830
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 3.230383718981827e-05,
+      "loss": 0.8783,
+      "step": 232840
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 3.2303075540390274e-05,
+      "loss": 1.0643,
+      "step": 232850
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 3.230231389096227e-05,
+      "loss": 1.0083,
+      "step": 232860
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 3.230155224153427e-05,
+      "loss": 0.8329,
+      "step": 232870
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 3.2300790592106266e-05,
+      "loss": 0.9066,
+      "step": 232880
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 3.230002894267827e-05,
+      "loss": 0.8014,
+      "step": 232890
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 3.229926729325026e-05,
+      "loss": 0.9104,
+      "step": 232900
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 3.2298505643822265e-05,
+      "loss": 0.8203,
+      "step": 232910
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 3.229774399439426e-05,
+      "loss": 0.9029,
+      "step": 232920
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 3.229698234496626e-05,
+      "loss": 0.9857,
+      "step": 232930
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 3.229622069553826e-05,
+      "loss": 0.8106,
+      "step": 232940
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 3.229545904611026e-05,
+      "loss": 0.853,
+      "step": 232950
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 3.229469739668225e-05,
+      "loss": 0.8069,
+      "step": 232960
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 3.229393574725426e-05,
+      "loss": 0.8422,
+      "step": 232970
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 3.2293174097826254e-05,
+      "loss": 0.8052,
+      "step": 232980
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 3.229241244839825e-05,
+      "loss": 0.8393,
+      "step": 232990
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 3.229165079897025e-05,
+      "loss": 0.8686,
+      "step": 233000
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 3.229088914954225e-05,
+      "loss": 0.8077,
+      "step": 233010
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 3.229012750011425e-05,
+      "loss": 0.8427,
+      "step": 233020
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 3.228936585068625e-05,
+      "loss": 0.8301,
+      "step": 233030
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 3.2288604201258246e-05,
+      "loss": 0.8408,
+      "step": 233040
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 3.228784255183024e-05,
+      "loss": 0.801,
+      "step": 233050
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 3.228708090240225e-05,
+      "loss": 0.955,
+      "step": 233060
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 3.2286319252974245e-05,
+      "loss": 0.8623,
+      "step": 233070
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 3.228555760354624e-05,
+      "loss": 0.9181,
+      "step": 233080
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 3.228479595411824e-05,
+      "loss": 0.71,
+      "step": 233090
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 3.228403430469024e-05,
+      "loss": 0.8781,
+      "step": 233100
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 3.2283272655262235e-05,
+      "loss": 0.7922,
+      "step": 233110
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 3.228251100583424e-05,
+      "loss": 0.788,
+      "step": 233120
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 3.228174935640623e-05,
+      "loss": 0.9293,
+      "step": 233130
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 3.2280987706978234e-05,
+      "loss": 0.8095,
+      "step": 233140
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 3.228022605755023e-05,
+      "loss": 0.8364,
+      "step": 233150
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 3.227946440812223e-05,
+      "loss": 0.7114,
+      "step": 233160
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 3.2278702758694226e-05,
+      "loss": 0.9151,
+      "step": 233170
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 3.227794110926623e-05,
+      "loss": 0.7032,
+      "step": 233180
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 3.227717945983823e-05,
+      "loss": 0.8364,
+      "step": 233190
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 3.2276417810410226e-05,
+      "loss": 0.7758,
+      "step": 233200
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 3.227565616098222e-05,
+      "loss": 0.8561,
+      "step": 233210
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 3.227489451155422e-05,
+      "loss": 0.9467,
+      "step": 233220
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 3.2274132862126225e-05,
+      "loss": 0.8337,
+      "step": 233230
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 3.227337121269822e-05,
+      "loss": 0.8567,
+      "step": 233240
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 3.227260956327022e-05,
+      "loss": 0.8528,
+      "step": 233250
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 3.227184791384222e-05,
+      "loss": 0.8148,
+      "step": 233260
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 3.227108626441422e-05,
+      "loss": 0.9346,
+      "step": 233270
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 3.227032461498621e-05,
+      "loss": 0.9486,
+      "step": 233280
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 3.2269562965558216e-05,
+      "loss": 0.8706,
+      "step": 233290
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 3.2268801316130214e-05,
+      "loss": 0.8172,
+      "step": 233300
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 3.226803966670221e-05,
+      "loss": 0.8206,
+      "step": 233310
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 3.226727801727421e-05,
+      "loss": 0.8436,
+      "step": 233320
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 3.226651636784621e-05,
+      "loss": 0.8299,
+      "step": 233330
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 3.2265754718418204e-05,
+      "loss": 0.733,
+      "step": 233340
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 3.226499306899021e-05,
+      "loss": 0.8782,
+      "step": 233350
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 3.2264231419562205e-05,
+      "loss": 0.7529,
+      "step": 233360
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 3.22634697701342e-05,
+      "loss": 0.8301,
+      "step": 233370
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 3.22627081207062e-05,
+      "loss": 0.9344,
+      "step": 233380
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 3.2261946471278205e-05,
+      "loss": 0.8835,
+      "step": 233390
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 3.22611848218502e-05,
+      "loss": 0.7698,
+      "step": 233400
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 3.22604231724222e-05,
+      "loss": 0.7752,
+      "step": 233410
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 3.22596615229942e-05,
+      "loss": 0.8302,
+      "step": 233420
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 3.2258899873566194e-05,
+      "loss": 0.8666,
+      "step": 233430
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 3.22581382241382e-05,
+      "loss": 0.7898,
+      "step": 233440
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 3.2257376574710196e-05,
+      "loss": 0.8237,
+      "step": 233450
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 3.2256614925282194e-05,
+      "loss": 0.9494,
+      "step": 233460
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 3.225585327585419e-05,
+      "loss": 0.8765,
+      "step": 233470
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 3.2255091626426195e-05,
+      "loss": 0.8277,
+      "step": 233480
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 3.2254329976998186e-05,
+      "loss": 0.9498,
+      "step": 233490
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 3.225356832757019e-05,
+      "loss": 0.8911,
+      "step": 233500
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 3.225280667814219e-05,
+      "loss": 0.8245,
+      "step": 233510
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 3.2252045028714185e-05,
+      "loss": 0.8409,
+      "step": 233520
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 3.225128337928618e-05,
+      "loss": 0.8649,
+      "step": 233530
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 3.225052172985818e-05,
+      "loss": 0.9299,
+      "step": 233540
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 3.224976008043018e-05,
+      "loss": 0.9259,
+      "step": 233550
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 3.224899843100218e-05,
+      "loss": 0.7693,
+      "step": 233560
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 3.224823678157418e-05,
+      "loss": 0.8145,
+      "step": 233570
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 3.224747513214618e-05,
+      "loss": 0.8293,
+      "step": 233580
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 3.2246713482718174e-05,
+      "loss": 0.8184,
+      "step": 233590
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 3.224595183329017e-05,
+      "loss": 0.8171,
+      "step": 233600
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 3.2245190183862176e-05,
+      "loss": 0.8772,
+      "step": 233610
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 3.224442853443417e-05,
+      "loss": 0.8249,
+      "step": 233620
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 3.224366688500617e-05,
+      "loss": 0.835,
+      "step": 233630
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 3.224290523557817e-05,
+      "loss": 0.8299,
+      "step": 233640
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 3.224214358615017e-05,
+      "loss": 0.7666,
+      "step": 233650
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 3.224138193672216e-05,
+      "loss": 0.8857,
+      "step": 233660
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 3.224062028729417e-05,
+      "loss": 0.9592,
+      "step": 233670
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 3.2239858637866165e-05,
+      "loss": 0.8787,
+      "step": 233680
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 3.223909698843816e-05,
+      "loss": 0.8346,
+      "step": 233690
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 3.223833533901016e-05,
+      "loss": 0.9523,
+      "step": 233700
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 3.2237573689582164e-05,
+      "loss": 0.8719,
+      "step": 233710
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 3.2236812040154155e-05,
+      "loss": 0.8517,
+      "step": 233720
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 3.223605039072616e-05,
+      "loss": 0.7704,
+      "step": 233730
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 3.2235288741298157e-05,
+      "loss": 0.8381,
+      "step": 233740
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 3.2234527091870154e-05,
+      "loss": 0.8456,
+      "step": 233750
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 3.223376544244215e-05,
+      "loss": 0.8884,
+      "step": 233760
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 3.2233003793014156e-05,
+      "loss": 0.8458,
+      "step": 233770
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 3.223224214358615e-05,
+      "loss": 0.8758,
+      "step": 233780
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 3.223148049415815e-05,
+      "loss": 0.8143,
+      "step": 233790
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 3.223071884473015e-05,
+      "loss": 0.7827,
+      "step": 233800
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 3.2229957195302146e-05,
+      "loss": 0.8659,
+      "step": 233810
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 3.222919554587415e-05,
+      "loss": 0.7896,
+      "step": 233820
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 3.222843389644615e-05,
+      "loss": 0.706,
+      "step": 233830
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 3.2227672247018145e-05,
+      "loss": 0.8698,
+      "step": 233840
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 3.222691059759014e-05,
+      "loss": 0.8467,
+      "step": 233850
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 3.2226148948162146e-05,
+      "loss": 0.7706,
+      "step": 233860
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 3.222538729873414e-05,
+      "loss": 0.7864,
+      "step": 233870
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 3.222462564930614e-05,
+      "loss": 0.8747,
+      "step": 233880
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 3.222386399987814e-05,
+      "loss": 0.9558,
+      "step": 233890
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 3.2223102350450136e-05,
+      "loss": 0.8399,
+      "step": 233900
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 3.2222340701022134e-05,
+      "loss": 0.8835,
+      "step": 233910
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 3.222157905159414e-05,
+      "loss": 0.9382,
+      "step": 233920
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 3.222081740216613e-05,
+      "loss": 0.8095,
+      "step": 233930
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 3.222005575273813e-05,
+      "loss": 0.8905,
+      "step": 233940
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 3.221929410331013e-05,
+      "loss": 0.8698,
+      "step": 233950
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 3.221853245388213e-05,
+      "loss": 0.895,
+      "step": 233960
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 3.2217770804454125e-05,
+      "loss": 0.8521,
+      "step": 233970
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 3.221700915502613e-05,
+      "loss": 0.811,
+      "step": 233980
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 3.221624750559813e-05,
+      "loss": 0.8071,
+      "step": 233990
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 3.2215485856170125e-05,
+      "loss": 0.8582,
+      "step": 234000
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 3.221472420674212e-05,
+      "loss": 0.8156,
+      "step": 234010
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 3.221396255731412e-05,
+      "loss": 0.87,
+      "step": 234020
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 3.2213200907886124e-05,
+      "loss": 0.814,
+      "step": 234030
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 3.2212439258458114e-05,
+      "loss": 0.9102,
+      "step": 234040
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 3.221167760903012e-05,
+      "loss": 0.8625,
+      "step": 234050
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 3.2210915959602116e-05,
+      "loss": 0.8657,
+      "step": 234060
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 3.2210154310174114e-05,
+      "loss": 0.7895,
+      "step": 234070
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 3.220939266074611e-05,
+      "loss": 0.8748,
+      "step": 234080
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 3.2208631011318115e-05,
+      "loss": 0.8177,
+      "step": 234090
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 3.2207869361890106e-05,
+      "loss": 0.8622,
+      "step": 234100
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 3.220710771246211e-05,
+      "loss": 0.9306,
+      "step": 234110
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 3.220634606303411e-05,
+      "loss": 0.7957,
+      "step": 234120
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 3.2205584413606105e-05,
+      "loss": 0.9117,
+      "step": 234130
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 3.22048227641781e-05,
+      "loss": 0.93,
+      "step": 234140
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 3.220406111475011e-05,
+      "loss": 0.8685,
+      "step": 234150
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 3.2203299465322104e-05,
+      "loss": 0.8278,
+      "step": 234160
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 3.22025378158941e-05,
+      "loss": 0.8806,
+      "step": 234170
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 3.22017761664661e-05,
+      "loss": 0.8905,
+      "step": 234180
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 3.22010145170381e-05,
+      "loss": 0.9256,
+      "step": 234190
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 3.22002528676101e-05,
+      "loss": 0.8769,
+      "step": 234200
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 3.21994912181821e-05,
+      "loss": 0.8329,
+      "step": 234210
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 3.2198729568754096e-05,
+      "loss": 0.9168,
+      "step": 234220
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 3.219796791932609e-05,
+      "loss": 0.8276,
+      "step": 234230
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 3.21972062698981e-05,
+      "loss": 0.8782,
+      "step": 234240
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 3.219644462047009e-05,
+      "loss": 0.8317,
+      "step": 234250
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 3.219568297104209e-05,
+      "loss": 0.7546,
+      "step": 234260
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 3.219492132161409e-05,
+      "loss": 0.8173,
+      "step": 234270
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 3.219415967218609e-05,
+      "loss": 0.8358,
+      "step": 234280
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 3.2193398022758085e-05,
+      "loss": 0.8308,
+      "step": 234290
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 3.219263637333009e-05,
+      "loss": 0.802,
+      "step": 234300
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 3.219187472390208e-05,
+      "loss": 0.7974,
+      "step": 234310
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 3.2191113074474084e-05,
+      "loss": 0.9349,
+      "step": 234320
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 3.219035142504608e-05,
+      "loss": 0.8843,
+      "step": 234330
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 3.218958977561808e-05,
+      "loss": 0.8159,
+      "step": 234340
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 3.2188828126190077e-05,
+      "loss": 0.8326,
+      "step": 234350
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 3.218806647676208e-05,
+      "loss": 0.7596,
+      "step": 234360
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 3.218730482733408e-05,
+      "loss": 0.91,
+      "step": 234370
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 3.2186543177906076e-05,
+      "loss": 0.8708,
+      "step": 234380
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 3.218578152847807e-05,
+      "loss": 0.8867,
+      "step": 234390
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 3.218501987905007e-05,
+      "loss": 0.895,
+      "step": 234400
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 3.2184258229622075e-05,
+      "loss": 0.8024,
+      "step": 234410
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 3.218349658019407e-05,
+      "loss": 0.8869,
+      "step": 234420
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 3.218273493076607e-05,
+      "loss": 0.8088,
+      "step": 234430
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 3.218197328133807e-05,
+      "loss": 0.8704,
+      "step": 234440
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 3.218121163191007e-05,
+      "loss": 0.8417,
+      "step": 234450
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 3.218044998248206e-05,
+      "loss": 0.8324,
+      "step": 234460
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 3.2179688333054066e-05,
+      "loss": 0.8173,
+      "step": 234470
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 3.217892668362606e-05,
+      "loss": 0.8297,
+      "step": 234480
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 3.217816503419806e-05,
+      "loss": 0.8856,
+      "step": 234490
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 3.217740338477006e-05,
+      "loss": 0.7956,
+      "step": 234500
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 3.2176641735342056e-05,
+      "loss": 0.8295,
+      "step": 234510
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 3.2175880085914054e-05,
+      "loss": 0.8106,
+      "step": 234520
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 3.217511843648606e-05,
+      "loss": 0.8461,
+      "step": 234530
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 3.217435678705805e-05,
+      "loss": 0.8703,
+      "step": 234540
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 3.217359513763005e-05,
+      "loss": 0.8863,
+      "step": 234550
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 3.217283348820205e-05,
+      "loss": 0.8971,
+      "step": 234560
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 3.217207183877405e-05,
+      "loss": 0.8734,
+      "step": 234570
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 3.217131018934605e-05,
+      "loss": 0.8785,
+      "step": 234580
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 3.217054853991805e-05,
+      "loss": 0.9411,
+      "step": 234590
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 3.216978689049005e-05,
+      "loss": 0.8427,
+      "step": 234600
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 3.2169025241062045e-05,
+      "loss": 0.7753,
+      "step": 234610
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 3.216826359163405e-05,
+      "loss": 0.8382,
+      "step": 234620
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 3.216750194220604e-05,
+      "loss": 0.8988,
+      "step": 234630
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 3.2166740292778044e-05,
+      "loss": 0.7753,
+      "step": 234640
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 3.216597864335004e-05,
+      "loss": 0.8381,
+      "step": 234650
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 3.216521699392204e-05,
+      "loss": 0.7642,
+      "step": 234660
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 3.2164455344494036e-05,
+      "loss": 0.716,
+      "step": 234670
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 3.216369369506604e-05,
+      "loss": 0.8199,
+      "step": 234680
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 3.216293204563803e-05,
+      "loss": 0.929,
+      "step": 234690
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 3.2162170396210035e-05,
+      "loss": 0.8469,
+      "step": 234700
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 3.216140874678203e-05,
+      "loss": 0.8187,
+      "step": 234710
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 3.216064709735403e-05,
+      "loss": 0.9049,
+      "step": 234720
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 3.215988544792603e-05,
+      "loss": 0.8692,
+      "step": 234730
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 3.215912379849803e-05,
+      "loss": 0.9449,
+      "step": 234740
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 3.215836214907002e-05,
+      "loss": 0.8857,
+      "step": 234750
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 3.215760049964203e-05,
+      "loss": 0.8644,
+      "step": 234760
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 3.2156838850214024e-05,
+      "loss": 0.8594,
+      "step": 234770
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 3.215607720078602e-05,
+      "loss": 0.8229,
+      "step": 234780
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 3.2155315551358026e-05,
+      "loss": 0.8104,
+      "step": 234790
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 3.2154553901930024e-05,
+      "loss": 0.8592,
+      "step": 234800
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 3.215379225250202e-05,
+      "loss": 0.7597,
+      "step": 234810
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 3.215303060307402e-05,
+      "loss": 0.9204,
+      "step": 234820
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 3.215226895364602e-05,
+      "loss": 0.8284,
+      "step": 234830
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 3.215150730421801e-05,
+      "loss": 0.7871,
+      "step": 234840
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 3.215074565479002e-05,
+      "loss": 0.843,
+      "step": 234850
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 3.2149984005362015e-05,
+      "loss": 0.8523,
+      "step": 234860
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 3.214922235593401e-05,
+      "loss": 0.817,
+      "step": 234870
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 3.214846070650601e-05,
+      "loss": 0.9342,
+      "step": 234880
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 3.2147699057078014e-05,
+      "loss": 0.8106,
+      "step": 234890
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 3.2146937407650005e-05,
+      "loss": 0.8554,
+      "step": 234900
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 3.214617575822201e-05,
+      "loss": 0.8037,
+      "step": 234910
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 3.214541410879401e-05,
+      "loss": 0.8341,
+      "step": 234920
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 3.2144652459366004e-05,
+      "loss": 0.8551,
+      "step": 234930
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 3.2143890809938e-05,
+      "loss": 0.928,
+      "step": 234940
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 3.214312916051e-05,
+      "loss": 0.803,
+      "step": 234950
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 3.2142367511082e-05,
+      "loss": 0.869,
+      "step": 234960
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 3.2141605861654e-05,
+      "loss": 0.828,
+      "step": 234970
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 3.2140844212226e-05,
+      "loss": 0.8603,
+      "step": 234980
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 3.2140082562797996e-05,
+      "loss": 0.8941,
+      "step": 234990
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 3.213932091337e-05,
+      "loss": 0.8911,
+      "step": 235000
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 3.213855926394199e-05,
+      "loss": 0.7672,
+      "step": 235010
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 3.2137797614513995e-05,
+      "loss": 0.7695,
+      "step": 235020
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 3.213703596508599e-05,
+      "loss": 0.9098,
+      "step": 235030
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 3.213627431565799e-05,
+      "loss": 0.9166,
+      "step": 235040
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 3.213551266622999e-05,
+      "loss": 0.9546,
+      "step": 235050
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 3.213475101680199e-05,
+      "loss": 0.8189,
+      "step": 235060
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 3.213398936737398e-05,
+      "loss": 0.8494,
+      "step": 235070
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 3.2133227717945986e-05,
+      "loss": 0.7767,
+      "step": 235080
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 3.2132466068517984e-05,
+      "loss": 0.8985,
+      "step": 235090
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 3.213170441908998e-05,
+      "loss": 0.8294,
+      "step": 235100
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 3.213094276966198e-05,
+      "loss": 0.818,
+      "step": 235110
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 3.213018112023398e-05,
+      "loss": 0.8381,
+      "step": 235120
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 3.2129419470805974e-05,
+      "loss": 0.9045,
+      "step": 235130
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 3.212865782137798e-05,
+      "loss": 0.9142,
+      "step": 235140
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 3.2127896171949975e-05,
+      "loss": 0.9638,
+      "step": 235150
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 3.212713452252197e-05,
+      "loss": 0.8664,
+      "step": 235160
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 3.212637287309398e-05,
+      "loss": 0.7806,
+      "step": 235170
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 3.2125611223665975e-05,
+      "loss": 0.7715,
+      "step": 235180
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 3.212484957423797e-05,
+      "loss": 0.8653,
+      "step": 235190
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 3.212408792480997e-05,
+      "loss": 0.8017,
+      "step": 235200
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 3.2123326275381974e-05,
+      "loss": 0.796,
+      "step": 235210
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 3.2122564625953965e-05,
+      "loss": 0.8413,
+      "step": 235220
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 3.212180297652597e-05,
+      "loss": 0.8096,
+      "step": 235230
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 3.2121041327097966e-05,
+      "loss": 0.851,
+      "step": 235240
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 3.2120279677669964e-05,
+      "loss": 0.8605,
+      "step": 235250
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 3.211951802824196e-05,
+      "loss": 0.8841,
+      "step": 235260
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 3.2118756378813965e-05,
+      "loss": 0.9591,
+      "step": 235270
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 3.2117994729385956e-05,
+      "loss": 0.8366,
+      "step": 235280
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 3.211723307995796e-05,
+      "loss": 0.8024,
+      "step": 235290
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 3.211647143052996e-05,
+      "loss": 0.8215,
+      "step": 235300
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 3.2115709781101955e-05,
+      "loss": 0.8613,
+      "step": 235310
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 3.211494813167395e-05,
+      "loss": 0.8916,
+      "step": 235320
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 3.211418648224596e-05,
+      "loss": 0.7685,
+      "step": 235330
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 3.211342483281795e-05,
+      "loss": 0.8837,
+      "step": 235340
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 3.211266318338995e-05,
+      "loss": 0.8257,
+      "step": 235350
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 3.211190153396195e-05,
+      "loss": 0.8919,
+      "step": 235360
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 3.211113988453395e-05,
+      "loss": 0.9163,
+      "step": 235370
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 3.211037823510595e-05,
+      "loss": 0.9328,
+      "step": 235380
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 3.210961658567794e-05,
+      "loss": 0.9098,
+      "step": 235390
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 3.2108854936249946e-05,
+      "loss": 0.8898,
+      "step": 235400
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 3.2108093286821944e-05,
+      "loss": 0.8592,
+      "step": 235410
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 3.210733163739394e-05,
+      "loss": 0.8172,
+      "step": 235420
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 3.210656998796594e-05,
+      "loss": 0.8622,
+      "step": 235430
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 3.210580833853794e-05,
+      "loss": 0.8479,
+      "step": 235440
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 3.210504668910993e-05,
+      "loss": 0.7877,
+      "step": 235450
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 3.210428503968194e-05,
+      "loss": 0.7916,
+      "step": 235460
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 3.2103523390253935e-05,
+      "loss": 0.8604,
+      "step": 235470
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 3.210276174082593e-05,
+      "loss": 0.7968,
+      "step": 235480
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 3.210200009139793e-05,
+      "loss": 0.8414,
+      "step": 235490
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 3.2101238441969934e-05,
+      "loss": 0.8791,
+      "step": 235500
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 3.2100476792541925e-05,
+      "loss": 0.8122,
+      "step": 235510
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 3.209971514311393e-05,
+      "loss": 0.862,
+      "step": 235520
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 3.209895349368593e-05,
+      "loss": 0.8725,
+      "step": 235530
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 3.2098191844257924e-05,
+      "loss": 0.8493,
+      "step": 235540
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 3.209743019482992e-05,
+      "loss": 0.8135,
+      "step": 235550
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 3.2096668545401926e-05,
+      "loss": 0.8792,
+      "step": 235560
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 3.209590689597392e-05,
+      "loss": 0.8627,
+      "step": 235570
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 3.209514524654592e-05,
+      "loss": 0.8237,
+      "step": 235580
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 3.2094383597117925e-05,
+      "loss": 0.8927,
+      "step": 235590
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 3.2093621947689916e-05,
+      "loss": 0.885,
+      "step": 235600
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 3.209286029826192e-05,
+      "loss": 0.7848,
+      "step": 235610
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 3.209209864883392e-05,
+      "loss": 0.7585,
+      "step": 235620
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 3.2091336999405915e-05,
+      "loss": 0.8307,
+      "step": 235630
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 3.209057534997791e-05,
+      "loss": 0.8804,
+      "step": 235640
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 3.2089813700549917e-05,
+      "loss": 0.7835,
+      "step": 235650
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 3.208905205112191e-05,
+      "loss": 0.8795,
+      "step": 235660
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 3.208829040169391e-05,
+      "loss": 0.957,
+      "step": 235670
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 3.208752875226591e-05,
+      "loss": 0.7873,
+      "step": 235680
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 3.2086767102837906e-05,
+      "loss": 0.7826,
+      "step": 235690
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 3.2086005453409904e-05,
+      "loss": 0.8301,
+      "step": 235700
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 3.208524380398191e-05,
+      "loss": 0.8319,
+      "step": 235710
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 3.20844821545539e-05,
+      "loss": 0.8033,
+      "step": 235720
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 3.20837205051259e-05,
+      "loss": 0.9277,
+      "step": 235730
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 3.20829588556979e-05,
+      "loss": 0.7785,
+      "step": 235740
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 3.20821972062699e-05,
+      "loss": 0.8849,
+      "step": 235750
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 3.20814355568419e-05,
+      "loss": 0.7884,
+      "step": 235760
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 3.20806739074139e-05,
+      "loss": 0.8613,
+      "step": 235770
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 3.20799122579859e-05,
+      "loss": 0.7393,
+      "step": 235780
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 3.2079150608557895e-05,
+      "loss": 0.8646,
+      "step": 235790
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 3.20783889591299e-05,
+      "loss": 0.7811,
+      "step": 235800
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 3.207762730970189e-05,
+      "loss": 0.7939,
+      "step": 235810
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 3.2076865660273894e-05,
+      "loss": 0.8537,
+      "step": 235820
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 3.207610401084589e-05,
+      "loss": 0.7698,
+      "step": 235830
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 3.207534236141789e-05,
+      "loss": 0.861,
+      "step": 235840
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 3.2074580711989886e-05,
+      "loss": 0.752,
+      "step": 235850
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 3.2073819062561884e-05,
+      "loss": 0.9845,
+      "step": 235860
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 3.207305741313388e-05,
+      "loss": 0.8802,
+      "step": 235870
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 3.2072295763705885e-05,
+      "loss": 0.8503,
+      "step": 235880
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 3.2071534114277876e-05,
+      "loss": 0.7929,
+      "step": 235890
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 3.207077246484988e-05,
+      "loss": 0.8584,
+      "step": 235900
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 3.207001081542188e-05,
+      "loss": 0.861,
+      "step": 235910
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 3.2069249165993875e-05,
+      "loss": 0.9166,
+      "step": 235920
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 3.206848751656587e-05,
+      "loss": 0.823,
+      "step": 235930
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 3.206772586713788e-05,
+      "loss": 0.8091,
+      "step": 235940
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 3.2066964217709874e-05,
+      "loss": 0.8179,
+      "step": 235950
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 3.206620256828187e-05,
+      "loss": 0.8052,
+      "step": 235960
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 3.2065440918853876e-05,
+      "loss": 0.8657,
+      "step": 235970
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 3.206467926942587e-05,
+      "loss": 0.7857,
+      "step": 235980
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 3.206391761999787e-05,
+      "loss": 0.864,
+      "step": 235990
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 3.206315597056987e-05,
+      "loss": 0.9204,
+      "step": 236000
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 3.2062394321141866e-05,
+      "loss": 0.8037,
+      "step": 236010
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 3.2061632671713864e-05,
+      "loss": 0.8767,
+      "step": 236020
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 3.206087102228587e-05,
+      "loss": 0.8905,
+      "step": 236030
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 3.206010937285786e-05,
+      "loss": 0.8079,
+      "step": 236040
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 3.205934772342986e-05,
+      "loss": 0.8743,
+      "step": 236050
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 3.205858607400186e-05,
+      "loss": 0.9525,
+      "step": 236060
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 3.205782442457386e-05,
+      "loss": 0.7974,
+      "step": 236070
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 3.2057062775145855e-05,
+      "loss": 0.8748,
+      "step": 236080
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 3.205630112571786e-05,
+      "loss": 0.7944,
+      "step": 236090
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 3.205553947628985e-05,
+      "loss": 0.8163,
+      "step": 236100
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 3.2054777826861854e-05,
+      "loss": 0.8325,
+      "step": 236110
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 3.205401617743385e-05,
+      "loss": 0.8642,
+      "step": 236120
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 3.205325452800585e-05,
+      "loss": 0.8625,
+      "step": 236130
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 3.205249287857785e-05,
+      "loss": 0.8659,
+      "step": 236140
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 3.205173122914985e-05,
+      "loss": 0.889,
+      "step": 236150
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 3.205096957972185e-05,
+      "loss": 0.9282,
+      "step": 236160
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 3.2050207930293846e-05,
+      "loss": 0.7879,
+      "step": 236170
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 3.204944628086585e-05,
+      "loss": 0.9227,
+      "step": 236180
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 3.204868463143784e-05,
+      "loss": 0.7916,
+      "step": 236190
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 3.2047922982009845e-05,
+      "loss": 0.8337,
+      "step": 236200
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 3.204716133258184e-05,
+      "loss": 0.8062,
+      "step": 236210
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 3.204639968315384e-05,
+      "loss": 0.8865,
+      "step": 236220
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 3.204563803372584e-05,
+      "loss": 1.0005,
+      "step": 236230
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 3.204487638429784e-05,
+      "loss": 0.8516,
+      "step": 236240
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 3.204411473486983e-05,
+      "loss": 0.9233,
+      "step": 236250
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 3.2043353085441837e-05,
+      "loss": 0.8962,
+      "step": 236260
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 3.2042591436013834e-05,
+      "loss": 0.8498,
+      "step": 236270
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 3.204182978658583e-05,
+      "loss": 0.8549,
+      "step": 236280
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 3.204106813715783e-05,
+      "loss": 0.872,
+      "step": 236290
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 3.204030648772983e-05,
+      "loss": 0.9173,
+      "step": 236300
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 3.2039544838301824e-05,
+      "loss": 0.8973,
+      "step": 236310
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 3.203878318887383e-05,
+      "loss": 0.8316,
+      "step": 236320
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 3.2038021539445826e-05,
+      "loss": 0.8455,
+      "step": 236330
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 3.203725989001782e-05,
+      "loss": 0.847,
+      "step": 236340
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 3.203649824058982e-05,
+      "loss": 0.7968,
+      "step": 236350
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 3.203573659116182e-05,
+      "loss": 0.7546,
+      "step": 236360
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 3.203497494173382e-05,
+      "loss": 0.7866,
+      "step": 236370
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 3.203421329230582e-05,
+      "loss": 0.8533,
+      "step": 236380
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 3.203345164287782e-05,
+      "loss": 0.8007,
+      "step": 236390
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 3.2032689993449815e-05,
+      "loss": 0.8543,
+      "step": 236400
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 3.203192834402182e-05,
+      "loss": 0.9141,
+      "step": 236410
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 3.203116669459381e-05,
+      "loss": 0.9321,
+      "step": 236420
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 3.2030405045165814e-05,
+      "loss": 0.9722,
+      "step": 236430
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 3.202964339573781e-05,
+      "loss": 0.8295,
+      "step": 236440
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 3.202888174630981e-05,
+      "loss": 0.8123,
+      "step": 236450
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 3.2028120096881806e-05,
+      "loss": 0.9078,
+      "step": 236460
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 3.202735844745381e-05,
+      "loss": 0.9439,
+      "step": 236470
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 3.20265967980258e-05,
+      "loss": 0.8819,
+      "step": 236480
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 3.2025835148597805e-05,
+      "loss": 0.8928,
+      "step": 236490
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 3.20250734991698e-05,
+      "loss": 0.8438,
+      "step": 236500
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 3.20243118497418e-05,
+      "loss": 0.8835,
+      "step": 236510
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 3.20235502003138e-05,
+      "loss": 0.9025,
+      "step": 236520
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 3.20227885508858e-05,
+      "loss": 0.9775,
+      "step": 236530
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 3.20220269014578e-05,
+      "loss": 0.881,
+      "step": 236540
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 3.20212652520298e-05,
+      "loss": 0.8317,
+      "step": 236550
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 3.20205036026018e-05,
+      "loss": 0.9847,
+      "step": 236560
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 3.201974195317379e-05,
+      "loss": 0.7579,
+      "step": 236570
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 3.2018980303745796e-05,
+      "loss": 0.8162,
+      "step": 236580
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 3.2018218654317794e-05,
+      "loss": 0.84,
+      "step": 236590
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 3.201745700488979e-05,
+      "loss": 0.8378,
+      "step": 236600
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 3.201669535546179e-05,
+      "loss": 0.8114,
+      "step": 236610
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 3.201593370603379e-05,
+      "loss": 0.8861,
+      "step": 236620
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 3.2015172056605784e-05,
+      "loss": 0.9632,
+      "step": 236630
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 3.201441040717779e-05,
+      "loss": 0.9696,
+      "step": 236640
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 3.2013648757749785e-05,
+      "loss": 0.8623,
+      "step": 236650
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 3.201288710832178e-05,
+      "loss": 0.8394,
+      "step": 236660
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 3.201212545889378e-05,
+      "loss": 0.8091,
+      "step": 236670
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 3.2011363809465784e-05,
+      "loss": 0.819,
+      "step": 236680
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 3.2010602160037775e-05,
+      "loss": 0.8292,
+      "step": 236690
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 3.200984051060978e-05,
+      "loss": 0.8706,
+      "step": 236700
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 3.200907886118178e-05,
+      "loss": 0.7414,
+      "step": 236710
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 3.2008317211753774e-05,
+      "loss": 0.8449,
+      "step": 236720
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 3.200755556232577e-05,
+      "loss": 0.8069,
+      "step": 236730
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 3.2006793912897776e-05,
+      "loss": 0.9099,
+      "step": 236740
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 3.2006032263469773e-05,
+      "loss": 0.8855,
+      "step": 236750
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 3.200527061404177e-05,
+      "loss": 0.8566,
+      "step": 236760
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 3.200450896461377e-05,
+      "loss": 0.8632,
+      "step": 236770
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 3.2003747315185766e-05,
+      "loss": 0.8427,
+      "step": 236780
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 3.200298566575777e-05,
+      "loss": 0.9328,
+      "step": 236790
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 3.200222401632976e-05,
+      "loss": 0.8924,
+      "step": 236800
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 3.2001462366901765e-05,
+      "loss": 0.805,
+      "step": 236810
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 3.200070071747376e-05,
+      "loss": 0.8764,
+      "step": 236820
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 3.199993906804576e-05,
+      "loss": 0.8091,
+      "step": 236830
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 3.199917741861776e-05,
+      "loss": 0.9085,
+      "step": 236840
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 3.199841576918976e-05,
+      "loss": 0.7436,
+      "step": 236850
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 3.199765411976175e-05,
+      "loss": 0.9166,
+      "step": 236860
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 3.199689247033376e-05,
+      "loss": 0.8496,
+      "step": 236870
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 3.1996130820905754e-05,
+      "loss": 0.9384,
+      "step": 236880
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 3.199536917147775e-05,
+      "loss": 0.7444,
+      "step": 236890
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 3.199460752204975e-05,
+      "loss": 0.9388,
+      "step": 236900
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 3.199384587262175e-05,
+      "loss": 0.8029,
+      "step": 236910
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 3.199308422319375e-05,
+      "loss": 0.8006,
+      "step": 236920
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 3.199232257376575e-05,
+      "loss": 0.8888,
+      "step": 236930
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 3.1991560924337746e-05,
+      "loss": 0.9342,
+      "step": 236940
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 3.199079927490974e-05,
+      "loss": 0.8306,
+      "step": 236950
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 3.199003762548175e-05,
+      "loss": 0.864,
+      "step": 236960
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 3.1989275976053745e-05,
+      "loss": 0.8672,
+      "step": 236970
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 3.198851432662574e-05,
+      "loss": 0.9429,
+      "step": 236980
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 3.198775267719774e-05,
+      "loss": 0.7806,
+      "step": 236990
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 3.1986991027769744e-05,
+      "loss": 0.7592,
+      "step": 237000
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 3.1986229378341735e-05,
+      "loss": 0.7647,
+      "step": 237010
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 3.198546772891374e-05,
+      "loss": 0.9647,
+      "step": 237020
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 3.1984706079485736e-05,
+      "loss": 0.8798,
+      "step": 237030
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 3.1983944430057734e-05,
+      "loss": 0.7893,
+      "step": 237040
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 3.198318278062973e-05,
+      "loss": 0.806,
+      "step": 237050
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 3.1982421131201736e-05,
+      "loss": 0.8026,
+      "step": 237060
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 3.1981659481773726e-05,
+      "loss": 0.8448,
+      "step": 237070
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 3.198089783234573e-05,
+      "loss": 0.765,
+      "step": 237080
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 3.198013618291773e-05,
+      "loss": 0.9052,
+      "step": 237090
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 3.1979374533489725e-05,
+      "loss": 0.9351,
+      "step": 237100
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 3.197861288406172e-05,
+      "loss": 0.8821,
+      "step": 237110
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 3.197785123463373e-05,
+      "loss": 0.8291,
+      "step": 237120
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 3.1977089585205725e-05,
+      "loss": 0.825,
+      "step": 237130
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 3.197632793577772e-05,
+      "loss": 0.7829,
+      "step": 237140
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 3.197556628634972e-05,
+      "loss": 0.8642,
+      "step": 237150
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 3.197480463692172e-05,
+      "loss": 0.7886,
+      "step": 237160
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 3.197404298749372e-05,
+      "loss": 0.8478,
+      "step": 237170
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 3.197328133806572e-05,
+      "loss": 0.8817,
+      "step": 237180
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 3.1972519688637716e-05,
+      "loss": 0.8265,
+      "step": 237190
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 3.1971758039209714e-05,
+      "loss": 0.7958,
+      "step": 237200
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 3.197099638978172e-05,
+      "loss": 0.8482,
+      "step": 237210
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 3.197023474035371e-05,
+      "loss": 0.852,
+      "step": 237220
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 3.196947309092571e-05,
+      "loss": 0.8046,
+      "step": 237230
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 3.1968711441497704e-05,
+      "loss": 0.8142,
+      "step": 237240
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 3.196794979206971e-05,
+      "loss": 0.9309,
+      "step": 237250
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 3.1967188142641705e-05,
+      "loss": 0.7937,
+      "step": 237260
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 3.19664264932137e-05,
+      "loss": 0.8028,
+      "step": 237270
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 3.19656648437857e-05,
+      "loss": 0.8206,
+      "step": 237280
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 3.1964903194357704e-05,
+      "loss": 0.8396,
+      "step": 237290
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 3.19641415449297e-05,
+      "loss": 0.7619,
+      "step": 237300
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 3.19633798955017e-05,
+      "loss": 0.8886,
+      "step": 237310
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 3.19626182460737e-05,
+      "loss": 0.7792,
+      "step": 237320
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 3.1961856596645694e-05,
+      "loss": 0.8819,
+      "step": 237330
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 3.19610949472177e-05,
+      "loss": 0.8502,
+      "step": 237340
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 3.1960333297789696e-05,
+      "loss": 0.913,
+      "step": 237350
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 3.1959571648361693e-05,
+      "loss": 0.8941,
+      "step": 237360
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 3.195880999893369e-05,
+      "loss": 0.7833,
+      "step": 237370
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 3.1958048349505695e-05,
+      "loss": 0.8808,
+      "step": 237380
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 3.1957286700077686e-05,
+      "loss": 0.869,
+      "step": 237390
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 3.195652505064969e-05,
+      "loss": 0.8179,
+      "step": 237400
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 3.195576340122169e-05,
+      "loss": 0.7877,
+      "step": 237410
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 3.1955001751793685e-05,
+      "loss": 0.8356,
+      "step": 237420
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 3.195424010236568e-05,
+      "loss": 0.8172,
+      "step": 237430
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 3.195347845293769e-05,
+      "loss": 0.7613,
+      "step": 237440
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 3.195271680350968e-05,
+      "loss": 0.957,
+      "step": 237450
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 3.195195515408168e-05,
+      "loss": 0.8695,
+      "step": 237460
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 3.195119350465368e-05,
+      "loss": 0.936,
+      "step": 237470
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 3.195043185522568e-05,
+      "loss": 0.8329,
+      "step": 237480
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 3.1949670205797674e-05,
+      "loss": 0.779,
+      "step": 237490
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 3.194890855636968e-05,
+      "loss": 0.8375,
+      "step": 237500
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 3.1948146906941676e-05,
+      "loss": 0.8986,
+      "step": 237510
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 3.194738525751367e-05,
+      "loss": 0.8517,
+      "step": 237520
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 3.194662360808567e-05,
+      "loss": 0.8422,
+      "step": 237530
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 3.194586195865767e-05,
+      "loss": 0.7799,
+      "step": 237540
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 3.194510030922967e-05,
+      "loss": 0.867,
+      "step": 237550
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 3.194433865980167e-05,
+      "loss": 0.8309,
+      "step": 237560
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 3.194357701037367e-05,
+      "loss": 0.8185,
+      "step": 237570
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 3.1942815360945665e-05,
+      "loss": 0.8384,
+      "step": 237580
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 3.194205371151767e-05,
+      "loss": 0.8528,
+      "step": 237590
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 3.194129206208966e-05,
+      "loss": 0.8804,
+      "step": 237600
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 3.1940530412661664e-05,
+      "loss": 0.8611,
+      "step": 237610
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 3.193976876323366e-05,
+      "loss": 0.8678,
+      "step": 237620
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 3.193900711380566e-05,
+      "loss": 0.9489,
+      "step": 237630
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 3.1938245464377656e-05,
+      "loss": 0.8627,
+      "step": 237640
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 3.193748381494966e-05,
+      "loss": 0.8385,
+      "step": 237650
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 3.193672216552165e-05,
+      "loss": 0.7315,
+      "step": 237660
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 3.1935960516093656e-05,
+      "loss": 0.879,
+      "step": 237670
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 3.193519886666565e-05,
+      "loss": 0.7702,
+      "step": 237680
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 3.193443721723765e-05,
+      "loss": 0.8118,
+      "step": 237690
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 3.193367556780965e-05,
+      "loss": 0.8209,
+      "step": 237700
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 3.1932913918381645e-05,
+      "loss": 0.941,
+      "step": 237710
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 3.193215226895365e-05,
+      "loss": 0.8396,
+      "step": 237720
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 3.193139061952565e-05,
+      "loss": 0.9299,
+      "step": 237730
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 3.1930628970097645e-05,
+      "loss": 0.8823,
+      "step": 237740
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 3.192986732066964e-05,
+      "loss": 0.7973,
+      "step": 237750
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 3.1929105671241646e-05,
+      "loss": 0.8876,
+      "step": 237760
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 3.192834402181364e-05,
+      "loss": 0.7745,
+      "step": 237770
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 3.192758237238564e-05,
+      "loss": 0.8332,
+      "step": 237780
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 3.192682072295764e-05,
+      "loss": 0.7812,
+      "step": 237790
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 3.1926059073529636e-05,
+      "loss": 0.8979,
+      "step": 237800
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 3.1925297424101634e-05,
+      "loss": 0.9834,
+      "step": 237810
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 3.192453577467364e-05,
+      "loss": 0.8615,
+      "step": 237820
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 3.192377412524563e-05,
+      "loss": 0.9171,
+      "step": 237830
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 3.192301247581763e-05,
+      "loss": 0.8826,
+      "step": 237840
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 3.192225082638963e-05,
+      "loss": 0.7709,
+      "step": 237850
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 3.192148917696163e-05,
+      "loss": 0.7862,
+      "step": 237860
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 3.1920727527533625e-05,
+      "loss": 0.8451,
+      "step": 237870
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 3.191996587810563e-05,
+      "loss": 0.9167,
+      "step": 237880
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 3.191920422867763e-05,
+      "loss": 0.8859,
+      "step": 237890
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 3.1918442579249624e-05,
+      "loss": 0.8268,
+      "step": 237900
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 3.191768092982162e-05,
+      "loss": 0.8133,
+      "step": 237910
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 3.191691928039362e-05,
+      "loss": 0.8356,
+      "step": 237920
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 3.1916157630965624e-05,
+      "loss": 0.862,
+      "step": 237930
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 3.191539598153762e-05,
+      "loss": 0.8565,
+      "step": 237940
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 3.191463433210962e-05,
+      "loss": 0.8446,
+      "step": 237950
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 3.1913872682681616e-05,
+      "loss": 0.8099,
+      "step": 237960
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 3.191311103325362e-05,
+      "loss": 0.8082,
+      "step": 237970
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 3.191234938382561e-05,
+      "loss": 0.8415,
+      "step": 237980
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 3.1911587734397615e-05,
+      "loss": 0.7961,
+      "step": 237990
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 3.191082608496961e-05,
+      "loss": 0.8321,
+      "step": 238000
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 3.191006443554161e-05,
+      "loss": 0.8482,
+      "step": 238010
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 3.190930278611361e-05,
+      "loss": 0.7717,
+      "step": 238020
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 3.190854113668561e-05,
+      "loss": 0.8732,
+      "step": 238030
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 3.19077794872576e-05,
+      "loss": 0.8326,
+      "step": 238040
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 3.190701783782961e-05,
+      "loss": 1.0214,
+      "step": 238050
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 3.1906256188401604e-05,
+      "loss": 0.9087,
+      "step": 238060
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 3.19054945389736e-05,
+      "loss": 0.9083,
+      "step": 238070
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 3.19047328895456e-05,
+      "loss": 0.8273,
+      "step": 238080
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 3.1903971240117603e-05,
+      "loss": 0.8494,
+      "step": 238090
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 3.19032095906896e-05,
+      "loss": 0.8999,
+      "step": 238100
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 3.19024479412616e-05,
+      "loss": 0.8783,
+      "step": 238110
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 3.1901686291833596e-05,
+      "loss": 0.8407,
+      "step": 238120
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 3.190092464240559e-05,
+      "loss": 0.8911,
+      "step": 238130
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 3.19001629929776e-05,
+      "loss": 0.873,
+      "step": 238140
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 3.189940134354959e-05,
+      "loss": 0.7526,
+      "step": 238150
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 3.189863969412159e-05,
+      "loss": 0.8058,
+      "step": 238160
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 3.189787804469359e-05,
+      "loss": 0.7449,
+      "step": 238170
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 3.189711639526559e-05,
+      "loss": 1.0934,
+      "step": 238180
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 3.1896354745837585e-05,
+      "loss": 0.8795,
+      "step": 238190
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 3.189559309640959e-05,
+      "loss": 0.8494,
+      "step": 238200
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 3.189483144698158e-05,
+      "loss": 0.9111,
+      "step": 238210
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 3.1894069797553584e-05,
+      "loss": 0.8511,
+      "step": 238220
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 3.189330814812558e-05,
+      "loss": 0.8739,
+      "step": 238230
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 3.189254649869758e-05,
+      "loss": 0.853,
+      "step": 238240
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 3.1891784849269576e-05,
+      "loss": 0.9703,
+      "step": 238250
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 3.189102319984158e-05,
+      "loss": 0.9519,
+      "step": 238260
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 3.189026155041358e-05,
+      "loss": 0.9245,
+      "step": 238270
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 3.1889499900985576e-05,
+      "loss": 0.8725,
+      "step": 238280
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 3.188873825155757e-05,
+      "loss": 0.856,
+      "step": 238290
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 3.188797660212957e-05,
+      "loss": 0.8253,
+      "step": 238300
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 3.1887214952701575e-05,
+      "loss": 0.9148,
+      "step": 238310
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 3.188645330327357e-05,
+      "loss": 0.8524,
+      "step": 238320
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 3.188569165384557e-05,
+      "loss": 0.8262,
+      "step": 238330
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 3.188493000441757e-05,
+      "loss": 0.8453,
+      "step": 238340
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 3.188416835498957e-05,
+      "loss": 0.8937,
+      "step": 238350
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 3.188340670556156e-05,
+      "loss": 0.8679,
+      "step": 238360
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 3.1882645056133566e-05,
+      "loss": 0.8529,
+      "step": 238370
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 3.1881883406705564e-05,
+      "loss": 0.7932,
+      "step": 238380
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 3.188112175727756e-05,
+      "loss": 0.8676,
+      "step": 238390
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 3.188036010784956e-05,
+      "loss": 0.9062,
+      "step": 238400
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 3.187959845842156e-05,
+      "loss": 0.8178,
+      "step": 238410
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 3.1878836808993554e-05,
+      "loss": 0.8695,
+      "step": 238420
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 3.187807515956556e-05,
+      "loss": 0.9087,
+      "step": 238430
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 3.1877313510137555e-05,
+      "loss": 0.8629,
+      "step": 238440
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 3.187655186070955e-05,
+      "loss": 0.7867,
+      "step": 238450
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 3.187579021128155e-05,
+      "loss": 0.829,
+      "step": 238460
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 3.1875028561853555e-05,
+      "loss": 0.8536,
+      "step": 238470
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 3.187426691242555e-05,
+      "loss": 0.8585,
+      "step": 238480
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 3.187350526299755e-05,
+      "loss": 0.8121,
+      "step": 238490
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 3.187274361356955e-05,
+      "loss": 0.8648,
+      "step": 238500
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 3.1871981964141544e-05,
+      "loss": 0.7837,
+      "step": 238510
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 3.187122031471355e-05,
+      "loss": 0.9076,
+      "step": 238520
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 3.1870458665285546e-05,
+      "loss": 0.9047,
+      "step": 238530
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 3.1869697015857544e-05,
+      "loss": 0.7658,
+      "step": 238540
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 3.186893536642954e-05,
+      "loss": 0.8036,
+      "step": 238550
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 3.1868173717001545e-05,
+      "loss": 0.9556,
+      "step": 238560
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 3.1867412067573536e-05,
+      "loss": 0.8887,
+      "step": 238570
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 3.186665041814554e-05,
+      "loss": 0.8942,
+      "step": 238580
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 3.186588876871754e-05,
+      "loss": 0.8368,
+      "step": 238590
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 3.1865127119289535e-05,
+      "loss": 0.7952,
+      "step": 238600
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 3.186436546986153e-05,
+      "loss": 0.7935,
+      "step": 238610
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 3.186360382043353e-05,
+      "loss": 0.8585,
+      "step": 238620
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 3.186284217100553e-05,
+      "loss": 1.0133,
+      "step": 238630
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 3.186208052157753e-05,
+      "loss": 0.9088,
+      "step": 238640
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 3.186131887214953e-05,
+      "loss": 0.7948,
+      "step": 238650
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 3.186055722272153e-05,
+      "loss": 0.9378,
+      "step": 238660
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 3.1859795573293524e-05,
+      "loss": 0.944,
+      "step": 238670
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 3.185903392386552e-05,
+      "loss": 0.8194,
+      "step": 238680
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 3.1858272274437526e-05,
+      "loss": 0.8612,
+      "step": 238690
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 3.1857510625009523e-05,
+      "loss": 0.87,
+      "step": 238700
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 3.185674897558152e-05,
+      "loss": 0.8047,
+      "step": 238710
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 3.185598732615352e-05,
+      "loss": 0.8653,
+      "step": 238720
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 3.185522567672552e-05,
+      "loss": 0.8754,
+      "step": 238730
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 3.185446402729751e-05,
+      "loss": 0.7568,
+      "step": 238740
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 3.185370237786952e-05,
+      "loss": 0.9642,
+      "step": 238750
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 3.1852940728441515e-05,
+      "loss": 0.8395,
+      "step": 238760
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 3.185217907901351e-05,
+      "loss": 0.8943,
+      "step": 238770
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 3.185141742958551e-05,
+      "loss": 0.8677,
+      "step": 238780
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 3.1850655780157514e-05,
+      "loss": 0.8951,
+      "step": 238790
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 3.1849894130729505e-05,
+      "loss": 0.8772,
+      "step": 238800
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 3.184913248130151e-05,
+      "loss": 0.8789,
+      "step": 238810
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 3.1848370831873507e-05,
+      "loss": 0.8204,
+      "step": 238820
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 3.1847609182445504e-05,
+      "loss": 0.8672,
+      "step": 238830
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 3.18468475330175e-05,
+      "loss": 0.8585,
+      "step": 238840
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 3.1846085883589506e-05,
+      "loss": 0.8767,
+      "step": 238850
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 3.18453242341615e-05,
+      "loss": 0.8635,
+      "step": 238860
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 3.18445625847335e-05,
+      "loss": 0.814,
+      "step": 238870
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 3.18438009353055e-05,
+      "loss": 0.7515,
+      "step": 238880
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 3.1843039285877496e-05,
+      "loss": 0.8449,
+      "step": 238890
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 3.18422776364495e-05,
+      "loss": 0.895,
+      "step": 238900
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 3.18415159870215e-05,
+      "loss": 0.7905,
+      "step": 238910
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 3.1840754337593495e-05,
+      "loss": 0.9213,
+      "step": 238920
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 3.183999268816549e-05,
+      "loss": 0.8029,
+      "step": 238930
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 3.1839231038737497e-05,
+      "loss": 0.8529,
+      "step": 238940
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 3.183846938930949e-05,
+      "loss": 0.8361,
+      "step": 238950
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 3.183770773988149e-05,
+      "loss": 0.8577,
+      "step": 238960
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 3.183694609045349e-05,
+      "loss": 0.793,
+      "step": 238970
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 3.1836184441025486e-05,
+      "loss": 0.998,
+      "step": 238980
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 3.1835422791597484e-05,
+      "loss": 0.8805,
+      "step": 238990
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 3.183466114216949e-05,
+      "loss": 0.8925,
+      "step": 239000
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 3.183389949274148e-05,
+      "loss": 0.8448,
+      "step": 239010
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 3.183313784331348e-05,
+      "loss": 0.8427,
+      "step": 239020
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 3.183237619388548e-05,
+      "loss": 0.909,
+      "step": 239030
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 3.183161454445748e-05,
+      "loss": 0.8115,
+      "step": 239040
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 3.1830852895029475e-05,
+      "loss": 0.8576,
+      "step": 239050
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 3.183009124560148e-05,
+      "loss": 0.9555,
+      "step": 239060
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 3.182932959617348e-05,
+      "loss": 0.8802,
+      "step": 239070
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 3.1828567946745475e-05,
+      "loss": 0.8419,
+      "step": 239080
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 3.182780629731747e-05,
+      "loss": 0.8704,
+      "step": 239090
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 3.182704464788947e-05,
+      "loss": 0.8236,
+      "step": 239100
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 3.1826282998461474e-05,
+      "loss": 0.9496,
+      "step": 239110
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 3.1825521349033464e-05,
+      "loss": 0.9309,
+      "step": 239120
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 3.182475969960547e-05,
+      "loss": 0.8922,
+      "step": 239130
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 3.1823998050177466e-05,
+      "loss": 0.7861,
+      "step": 239140
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 3.1823236400749464e-05,
+      "loss": 0.9271,
+      "step": 239150
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 3.182247475132146e-05,
+      "loss": 0.7073,
+      "step": 239160
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 3.1821713101893465e-05,
+      "loss": 0.8761,
+      "step": 239170
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 3.1820951452465456e-05,
+      "loss": 0.8933,
+      "step": 239180
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 3.182018980303746e-05,
+      "loss": 0.8105,
+      "step": 239190
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 3.181942815360946e-05,
+      "loss": 0.8192,
+      "step": 239200
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 3.1818666504181455e-05,
+      "loss": 0.8702,
+      "step": 239210
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 3.181790485475345e-05,
+      "loss": 0.8217,
+      "step": 239220
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 3.181714320532546e-05,
+      "loss": 0.8814,
+      "step": 239230
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 3.181638155589745e-05,
+      "loss": 0.7828,
+      "step": 239240
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 3.181561990646945e-05,
+      "loss": 0.7834,
+      "step": 239250
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 3.181485825704145e-05,
+      "loss": 0.8082,
+      "step": 239260
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 3.181409660761345e-05,
+      "loss": 0.8457,
+      "step": 239270
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 3.181333495818545e-05,
+      "loss": 0.8414,
+      "step": 239280
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 3.181257330875745e-05,
+      "loss": 0.9377,
+      "step": 239290
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 3.1811811659329446e-05,
+      "loss": 0.8851,
+      "step": 239300
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 3.1811050009901443e-05,
+      "loss": 0.8762,
+      "step": 239310
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 3.181028836047345e-05,
+      "loss": 0.8986,
+      "step": 239320
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 3.180952671104544e-05,
+      "loss": 0.846,
+      "step": 239330
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 3.180876506161744e-05,
+      "loss": 0.9804,
+      "step": 239340
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 3.180800341218944e-05,
+      "loss": 0.8604,
+      "step": 239350
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 3.180724176276144e-05,
+      "loss": 0.8729,
+      "step": 239360
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 3.1806480113333435e-05,
+      "loss": 0.9223,
+      "step": 239370
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 3.180571846390544e-05,
+      "loss": 0.7841,
+      "step": 239380
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 3.180495681447743e-05,
+      "loss": 0.8706,
+      "step": 239390
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 3.1804195165049434e-05,
+      "loss": 0.8216,
+      "step": 239400
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 3.180343351562143e-05,
+      "loss": 0.8312,
+      "step": 239410
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 3.180267186619343e-05,
+      "loss": 0.8436,
+      "step": 239420
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 3.1801910216765427e-05,
+      "loss": 0.8101,
+      "step": 239430
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 3.180114856733743e-05,
+      "loss": 0.8793,
+      "step": 239440
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 3.180038691790943e-05,
+      "loss": 0.8492,
+      "step": 239450
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 3.1799625268481426e-05,
+      "loss": 0.8479,
+      "step": 239460
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 3.179886361905342e-05,
+      "loss": 0.8675,
+      "step": 239470
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 3.179810196962542e-05,
+      "loss": 0.8534,
+      "step": 239480
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 3.1797340320197425e-05,
+      "loss": 0.9361,
+      "step": 239490
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 3.179657867076942e-05,
+      "loss": 0.8813,
+      "step": 239500
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 3.179581702134142e-05,
+      "loss": 0.9088,
+      "step": 239510
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 3.179505537191342e-05,
+      "loss": 0.7167,
+      "step": 239520
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 3.179429372248542e-05,
+      "loss": 0.8634,
+      "step": 239530
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 3.179353207305741e-05,
+      "loss": 0.8731,
+      "step": 239540
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 3.1792770423629417e-05,
+      "loss": 0.9396,
+      "step": 239550
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 3.179200877420141e-05,
+      "loss": 0.8969,
+      "step": 239560
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 3.179124712477341e-05,
+      "loss": 0.7814,
+      "step": 239570
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 3.179048547534541e-05,
+      "loss": 0.9052,
+      "step": 239580
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 3.1789723825917406e-05,
+      "loss": 0.8468,
+      "step": 239590
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 3.1788962176489404e-05,
+      "loss": 0.8908,
+      "step": 239600
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 3.178820052706141e-05,
+      "loss": 0.8633,
+      "step": 239610
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 3.17874388776334e-05,
+      "loss": 0.8545,
+      "step": 239620
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 3.17866772282054e-05,
+      "loss": 0.8112,
+      "step": 239630
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 3.17859155787774e-05,
+      "loss": 0.8481,
+      "step": 239640
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 3.17851539293494e-05,
+      "loss": 0.8009,
+      "step": 239650
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 3.17843922799214e-05,
+      "loss": 0.8221,
+      "step": 239660
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 3.17836306304934e-05,
+      "loss": 0.9808,
+      "step": 239670
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 3.17828689810654e-05,
+      "loss": 0.8141,
+      "step": 239680
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 3.1782107331637395e-05,
+      "loss": 0.8884,
+      "step": 239690
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 3.17813456822094e-05,
+      "loss": 0.7186,
+      "step": 239700
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 3.178058403278139e-05,
+      "loss": 0.8705,
+      "step": 239710
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 3.1779822383353394e-05,
+      "loss": 0.8446,
+      "step": 239720
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 3.177906073392539e-05,
+      "loss": 0.8993,
+      "step": 239730
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 3.177829908449739e-05,
+      "loss": 0.8612,
+      "step": 239740
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 3.1777537435069386e-05,
+      "loss": 0.9106,
+      "step": 239750
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 3.177677578564139e-05,
+      "loss": 0.8431,
+      "step": 239760
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 3.177601413621338e-05,
+      "loss": 0.7605,
+      "step": 239770
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 3.1775252486785385e-05,
+      "loss": 0.8727,
+      "step": 239780
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 3.177449083735738e-05,
+      "loss": 0.8274,
+      "step": 239790
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 3.177372918792938e-05,
+      "loss": 0.83,
+      "step": 239800
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 3.177296753850138e-05,
+      "loss": 0.7898,
+      "step": 239810
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 3.177220588907338e-05,
+      "loss": 0.8236,
+      "step": 239820
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 3.177144423964537e-05,
+      "loss": 0.8801,
+      "step": 239830
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 3.177068259021738e-05,
+      "loss": 0.8533,
+      "step": 239840
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 3.1769920940789374e-05,
+      "loss": 0.928,
+      "step": 239850
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 3.176915929136137e-05,
+      "loss": 0.7772,
+      "step": 239860
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 3.1768397641933376e-05,
+      "loss": 0.7878,
+      "step": 239870
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 3.1767635992505374e-05,
+      "loss": 0.83,
+      "step": 239880
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 3.176687434307737e-05,
+      "loss": 0.7869,
+      "step": 239890
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 3.176611269364937e-05,
+      "loss": 0.8511,
+      "step": 239900
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 3.176535104422137e-05,
+      "loss": 0.8426,
+      "step": 239910
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 3.1764589394793363e-05,
+      "loss": 0.9558,
+      "step": 239920
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 3.176382774536537e-05,
+      "loss": 0.9268,
+      "step": 239930
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 3.1763066095937365e-05,
+      "loss": 0.8932,
+      "step": 239940
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 3.176230444650936e-05,
+      "loss": 0.8611,
+      "step": 239950
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 3.176154279708136e-05,
+      "loss": 0.825,
+      "step": 239960
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 3.1760781147653364e-05,
+      "loss": 0.74,
+      "step": 239970
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 3.1760019498225355e-05,
+      "loss": 0.8829,
+      "step": 239980
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 3.175925784879736e-05,
+      "loss": 0.8727,
+      "step": 239990
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 3.175849619936935e-05,
+      "loss": 0.9004,
+      "step": 240000
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 3.1757734549941354e-05,
+      "loss": 0.8352,
+      "step": 240010
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 3.175697290051335e-05,
+      "loss": 0.877,
+      "step": 240020
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 3.175621125108535e-05,
+      "loss": 0.9222,
+      "step": 240030
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 3.1755449601657347e-05,
+      "loss": 0.8485,
+      "step": 240040
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 3.175468795222935e-05,
+      "loss": 0.9047,
+      "step": 240050
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 3.175392630280135e-05,
+      "loss": 0.8677,
+      "step": 240060
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 3.1753164653373346e-05,
+      "loss": 0.7704,
+      "step": 240070
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 3.175240300394535e-05,
+      "loss": 0.8206,
+      "step": 240080
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 3.175164135451734e-05,
+      "loss": 0.8542,
+      "step": 240090
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 3.1750879705089345e-05,
+      "loss": 0.8465,
+      "step": 240100
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 3.175011805566134e-05,
+      "loss": 0.8016,
+      "step": 240110
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 3.174935640623334e-05,
+      "loss": 0.7964,
+      "step": 240120
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 3.174859475680534e-05,
+      "loss": 0.8115,
+      "step": 240130
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 3.174783310737734e-05,
+      "loss": 0.9145,
+      "step": 240140
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 3.174707145794933e-05,
+      "loss": 1.0145,
+      "step": 240150
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 3.1746309808521337e-05,
+      "loss": 0.9326,
+      "step": 240160
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 3.1745548159093334e-05,
+      "loss": 0.8687,
+      "step": 240170
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 3.174478650966533e-05,
+      "loss": 0.8305,
+      "step": 240180
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 3.174402486023733e-05,
+      "loss": 0.792,
+      "step": 240190
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 3.174326321080933e-05,
+      "loss": 0.8647,
+      "step": 240200
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 3.1742501561381324e-05,
+      "loss": 0.9546,
+      "step": 240210
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 3.174173991195333e-05,
+      "loss": 0.8092,
+      "step": 240220
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 3.1740978262525326e-05,
+      "loss": 0.8278,
+      "step": 240230
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 3.174021661309732e-05,
+      "loss": 0.8333,
+      "step": 240240
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 3.173945496366933e-05,
+      "loss": 0.8936,
+      "step": 240250
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 3.1738693314241325e-05,
+      "loss": 0.828,
+      "step": 240260
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 3.173793166481332e-05,
+      "loss": 0.7966,
+      "step": 240270
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 3.173717001538532e-05,
+      "loss": 0.8085,
+      "step": 240280
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 3.1736408365957324e-05,
+      "loss": 0.837,
+      "step": 240290
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 3.1735646716529315e-05,
+      "loss": 0.9905,
+      "step": 240300
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 3.173488506710132e-05,
+      "loss": 0.8496,
+      "step": 240310
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 3.1734123417673316e-05,
+      "loss": 0.8584,
+      "step": 240320
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 3.1733361768245314e-05,
+      "loss": 0.906,
+      "step": 240330
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 3.173260011881731e-05,
+      "loss": 0.8555,
+      "step": 240340
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 3.1731838469389316e-05,
+      "loss": 0.8567,
+      "step": 240350
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 3.1731076819961306e-05,
+      "loss": 0.8266,
+      "step": 240360
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 3.173031517053331e-05,
+      "loss": 0.8068,
+      "step": 240370
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 3.172955352110531e-05,
+      "loss": 0.9784,
+      "step": 240380
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 3.1728791871677305e-05,
+      "loss": 0.8555,
+      "step": 240390
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 3.17280302222493e-05,
+      "loss": 0.8299,
+      "step": 240400
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 3.172726857282131e-05,
+      "loss": 0.851,
+      "step": 240410
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 3.17265069233933e-05,
+      "loss": 0.8936,
+      "step": 240420
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 3.17257452739653e-05,
+      "loss": 0.797,
+      "step": 240430
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 3.17249836245373e-05,
+      "loss": 0.8203,
+      "step": 240440
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 3.17242219751093e-05,
+      "loss": 0.866,
+      "step": 240450
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 3.17234603256813e-05,
+      "loss": 0.8586,
+      "step": 240460
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 3.172269867625329e-05,
+      "loss": 0.8983,
+      "step": 240470
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 3.1721937026825296e-05,
+      "loss": 0.7914,
+      "step": 240480
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 3.1721175377397294e-05,
+      "loss": 0.8451,
+      "step": 240490
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 3.172041372796929e-05,
+      "loss": 0.8816,
+      "step": 240500
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 3.171965207854129e-05,
+      "loss": 0.9792,
+      "step": 240510
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 3.171889042911329e-05,
+      "loss": 0.8338,
+      "step": 240520
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 3.1718128779685283e-05,
+      "loss": 0.7811,
+      "step": 240530
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 3.171736713025729e-05,
+      "loss": 0.848,
+      "step": 240540
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 3.1716605480829285e-05,
+      "loss": 0.7689,
+      "step": 240550
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 3.171584383140128e-05,
+      "loss": 0.9128,
+      "step": 240560
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 3.171508218197328e-05,
+      "loss": 0.8702,
+      "step": 240570
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 3.1714320532545284e-05,
+      "loss": 0.7291,
+      "step": 240580
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 3.1713558883117275e-05,
+      "loss": 0.8373,
+      "step": 240590
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 3.171279723368928e-05,
+      "loss": 0.8351,
+      "step": 240600
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 3.171203558426128e-05,
+      "loss": 0.7823,
+      "step": 240610
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 3.1711273934833274e-05,
+      "loss": 0.8538,
+      "step": 240620
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 3.171051228540527e-05,
+      "loss": 0.8542,
+      "step": 240630
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 3.1709750635977276e-05,
+      "loss": 0.8008,
+      "step": 240640
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 3.1708988986549273e-05,
+      "loss": 0.8292,
+      "step": 240650
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 3.170822733712127e-05,
+      "loss": 0.881,
+      "step": 240660
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 3.1707465687693275e-05,
+      "loss": 0.9512,
+      "step": 240670
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 3.1706704038265266e-05,
+      "loss": 0.7633,
+      "step": 240680
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 3.170594238883727e-05,
+      "loss": 0.8661,
+      "step": 240690
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 3.170518073940927e-05,
+      "loss": 0.8462,
+      "step": 240700
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 3.1704419089981265e-05,
+      "loss": 0.8573,
+      "step": 240710
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 3.170365744055326e-05,
+      "loss": 0.746,
+      "step": 240720
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 3.170289579112527e-05,
+      "loss": 0.8645,
+      "step": 240730
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 3.170213414169726e-05,
+      "loss": 0.9186,
+      "step": 240740
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 3.170137249226926e-05,
+      "loss": 0.8207,
+      "step": 240750
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 3.170061084284126e-05,
+      "loss": 0.8654,
+      "step": 240760
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 3.1699849193413257e-05,
+      "loss": 0.8292,
+      "step": 240770
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 3.1699087543985254e-05,
+      "loss": 0.8824,
+      "step": 240780
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 3.169832589455726e-05,
+      "loss": 0.8365,
+      "step": 240790
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 3.169756424512925e-05,
+      "loss": 0.7553,
+      "step": 240800
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 3.169680259570125e-05,
+      "loss": 0.8255,
+      "step": 240810
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 3.169604094627325e-05,
+      "loss": 0.8373,
+      "step": 240820
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 3.169527929684525e-05,
+      "loss": 0.7509,
+      "step": 240830
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 3.1694517647417246e-05,
+      "loss": 0.7804,
+      "step": 240840
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 3.169375599798925e-05,
+      "loss": 0.8294,
+      "step": 240850
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 3.169299434856125e-05,
+      "loss": 0.9043,
+      "step": 240860
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 3.1692232699133245e-05,
+      "loss": 0.8486,
+      "step": 240870
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 3.169147104970525e-05,
+      "loss": 0.8056,
+      "step": 240880
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 3.169070940027724e-05,
+      "loss": 0.8169,
+      "step": 240890
+    },
+    {
+      "epoch": 11.0,
+      "eval_cer": 0.6203485152092113,
+      "eval_em": 0.007568590350047304,
+      "eval_f1": 0.007568590350047304,
+      "eval_loss": 0.8935345411300659,
+      "eval_runtime": 1010.8334,
+      "eval_samples_per_second": 10.457,
+      "eval_steps_per_second": 1.308,
+      "eval_wer": 0.9924314096499527,
+      "step": 240897
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 3.1689947750849244e-05,
+      "loss": 0.8359,
+      "step": 240900
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 3.168918610142124e-05,
+      "loss": 0.8584,
+      "step": 240910
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 3.168842445199324e-05,
+      "loss": 0.7108,
+      "step": 240920
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 3.1687662802565236e-05,
+      "loss": 0.7579,
+      "step": 240930
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 3.1686901153137234e-05,
+      "loss": 0.8364,
+      "step": 240940
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 3.168613950370923e-05,
+      "loss": 0.8522,
+      "step": 240950
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 3.1685377854281236e-05,
+      "loss": 0.9445,
+      "step": 240960
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 3.1684616204853226e-05,
+      "loss": 0.9391,
+      "step": 240970
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 3.168385455542523e-05,
+      "loss": 0.8572,
+      "step": 240980
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 3.168309290599723e-05,
+      "loss": 0.8192,
+      "step": 240990
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 3.1682331256569225e-05,
+      "loss": 0.7784,
+      "step": 241000
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 3.168156960714122e-05,
+      "loss": 0.9331,
+      "step": 241010
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 3.168080795771323e-05,
+      "loss": 0.7767,
+      "step": 241020
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 3.1680046308285225e-05,
+      "loss": 0.8103,
+      "step": 241030
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 3.167928465885722e-05,
+      "loss": 0.9256,
+      "step": 241040
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 3.1678523009429226e-05,
+      "loss": 0.8119,
+      "step": 241050
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 3.167776136000122e-05,
+      "loss": 0.7982,
+      "step": 241060
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 3.167699971057322e-05,
+      "loss": 0.831,
+      "step": 241070
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 3.167623806114522e-05,
+      "loss": 0.7322,
+      "step": 241080
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 3.1675476411717216e-05,
+      "loss": 0.8826,
+      "step": 241090
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 3.1674714762289214e-05,
+      "loss": 0.8098,
+      "step": 241100
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 3.167395311286122e-05,
+      "loss": 0.7332,
+      "step": 241110
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 3.167319146343321e-05,
+      "loss": 0.9271,
+      "step": 241120
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 3.167242981400521e-05,
+      "loss": 0.7474,
+      "step": 241130
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 3.167166816457721e-05,
+      "loss": 0.7637,
+      "step": 241140
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 3.167090651514921e-05,
+      "loss": 0.962,
+      "step": 241150
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 3.1670144865721205e-05,
+      "loss": 0.7569,
+      "step": 241160
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 3.166938321629321e-05,
+      "loss": 0.9008,
+      "step": 241170
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 3.16686215668652e-05,
+      "loss": 0.742,
+      "step": 241180
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 3.1667859917437204e-05,
+      "loss": 0.8293,
+      "step": 241190
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 3.16670982680092e-05,
+      "loss": 0.8746,
+      "step": 241200
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 3.16663366185812e-05,
+      "loss": 0.7576,
+      "step": 241210
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 3.16655749691532e-05,
+      "loss": 0.7915,
+      "step": 241220
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 3.16648133197252e-05,
+      "loss": 0.969,
+      "step": 241230
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 3.16640516702972e-05,
+      "loss": 0.8467,
+      "step": 241240
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 3.1663290020869196e-05,
+      "loss": 0.8121,
+      "step": 241250
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 3.16625283714412e-05,
+      "loss": 0.8295,
+      "step": 241260
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 3.166176672201319e-05,
+      "loss": 0.794,
+      "step": 241270
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 3.1661005072585195e-05,
+      "loss": 0.8669,
+      "step": 241280
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 3.166024342315719e-05,
+      "loss": 0.8015,
+      "step": 241290
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 3.165948177372919e-05,
+      "loss": 0.7575,
+      "step": 241300
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 3.165872012430119e-05,
+      "loss": 0.8482,
+      "step": 241310
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 3.165795847487319e-05,
+      "loss": 0.8285,
+      "step": 241320
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 3.165719682544518e-05,
+      "loss": 0.8434,
+      "step": 241330
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 3.165643517601719e-05,
+      "loss": 0.8421,
+      "step": 241340
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 3.1655673526589184e-05,
+      "loss": 0.8541,
+      "step": 241350
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 3.165491187716118e-05,
+      "loss": 0.8394,
+      "step": 241360
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 3.165415022773318e-05,
+      "loss": 0.8263,
+      "step": 241370
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 3.1653388578305177e-05,
+      "loss": 0.8489,
+      "step": 241380
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 3.1652626928877174e-05,
+      "loss": 0.7573,
+      "step": 241390
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 3.165186527944918e-05,
+      "loss": 0.7745,
+      "step": 241400
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 3.1651103630021176e-05,
+      "loss": 0.8748,
+      "step": 241410
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 3.165034198059317e-05,
+      "loss": 0.8497,
+      "step": 241420
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 3.164958033116517e-05,
+      "loss": 0.8462,
+      "step": 241430
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 3.164881868173717e-05,
+      "loss": 0.891,
+      "step": 241440
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 3.164805703230917e-05,
+      "loss": 0.8247,
+      "step": 241450
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 3.164729538288117e-05,
+      "loss": 0.9336,
+      "step": 241460
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 3.164653373345317e-05,
+      "loss": 0.9087,
+      "step": 241470
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 3.1645772084025165e-05,
+      "loss": 0.725,
+      "step": 241480
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 3.164501043459717e-05,
+      "loss": 0.7115,
+      "step": 241490
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 3.164424878516916e-05,
+      "loss": 0.842,
+      "step": 241500
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 3.1643487135741164e-05,
+      "loss": 0.823,
+      "step": 241510
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 3.164272548631316e-05,
+      "loss": 0.9159,
+      "step": 241520
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 3.164196383688516e-05,
+      "loss": 0.7811,
+      "step": 241530
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 3.1641202187457156e-05,
+      "loss": 0.7638,
+      "step": 241540
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 3.164044053802916e-05,
+      "loss": 0.8231,
+      "step": 241550
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 3.163967888860115e-05,
+      "loss": 0.8055,
+      "step": 241560
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 3.1638917239173156e-05,
+      "loss": 0.7474,
+      "step": 241570
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 3.163815558974515e-05,
+      "loss": 0.8932,
+      "step": 241580
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 3.163739394031715e-05,
+      "loss": 0.776,
+      "step": 241590
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 3.163663229088915e-05,
+      "loss": 0.8656,
+      "step": 241600
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 3.163587064146115e-05,
+      "loss": 0.7387,
+      "step": 241610
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 3.163510899203315e-05,
+      "loss": 0.7914,
+      "step": 241620
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 3.163434734260515e-05,
+      "loss": 0.7958,
+      "step": 241630
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 3.1633585693177145e-05,
+      "loss": 0.7844,
+      "step": 241640
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 3.163282404374914e-05,
+      "loss": 0.7111,
+      "step": 241650
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 3.1632062394321146e-05,
+      "loss": 0.8092,
+      "step": 241660
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 3.1631300744893144e-05,
+      "loss": 0.82,
+      "step": 241670
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 3.163053909546514e-05,
+      "loss": 0.7717,
+      "step": 241680
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 3.162977744603714e-05,
+      "loss": 0.8566,
+      "step": 241690
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 3.162901579660914e-05,
+      "loss": 0.6811,
+      "step": 241700
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 3.1628254147181134e-05,
+      "loss": 0.8609,
+      "step": 241710
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 3.162749249775314e-05,
+      "loss": 0.8414,
+      "step": 241720
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 3.1626730848325135e-05,
+      "loss": 0.8233,
+      "step": 241730
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 3.162596919889713e-05,
+      "loss": 0.7893,
+      "step": 241740
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 3.162520754946913e-05,
+      "loss": 0.9066,
+      "step": 241750
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 3.1624445900041135e-05,
+      "loss": 0.8619,
+      "step": 241760
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 3.1623684250613125e-05,
+      "loss": 0.945,
+      "step": 241770
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 3.162292260118513e-05,
+      "loss": 0.7756,
+      "step": 241780
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 3.162216095175713e-05,
+      "loss": 0.9067,
+      "step": 241790
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 3.1621399302329124e-05,
+      "loss": 0.8516,
+      "step": 241800
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 3.162063765290112e-05,
+      "loss": 0.8436,
+      "step": 241810
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 3.1619876003473126e-05,
+      "loss": 0.7937,
+      "step": 241820
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 3.1619114354045124e-05,
+      "loss": 0.8312,
+      "step": 241830
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 3.161835270461712e-05,
+      "loss": 0.8906,
+      "step": 241840
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 3.161759105518912e-05,
+      "loss": 0.8352,
+      "step": 241850
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 3.1616829405761116e-05,
+      "loss": 0.7399,
+      "step": 241860
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 3.161606775633312e-05,
+      "loss": 0.8418,
+      "step": 241870
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 3.161530610690511e-05,
+      "loss": 0.7891,
+      "step": 241880
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 3.1614544457477115e-05,
+      "loss": 0.8057,
+      "step": 241890
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 3.161378280804911e-05,
+      "loss": 0.7652,
+      "step": 241900
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 3.161302115862111e-05,
+      "loss": 0.8196,
+      "step": 241910
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 3.161225950919311e-05,
+      "loss": 0.8585,
+      "step": 241920
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 3.161149785976511e-05,
+      "loss": 0.7971,
+      "step": 241930
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 3.16107362103371e-05,
+      "loss": 0.8043,
+      "step": 241940
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 3.160997456090911e-05,
+      "loss": 0.8591,
+      "step": 241950
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 3.1609212911481104e-05,
+      "loss": 0.7969,
+      "step": 241960
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 3.16084512620531e-05,
+      "loss": 0.8971,
+      "step": 241970
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 3.16076896126251e-05,
+      "loss": 0.8642,
+      "step": 241980
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 3.16069279631971e-05,
+      "loss": 0.8264,
+      "step": 241990
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 3.16061663137691e-05,
+      "loss": 0.8154,
+      "step": 242000
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 3.16054046643411e-05,
+      "loss": 0.8535,
+      "step": 242010
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 3.1604643014913096e-05,
+      "loss": 0.782,
+      "step": 242020
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 3.160388136548509e-05,
+      "loss": 0.7712,
+      "step": 242030
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 3.16031197160571e-05,
+      "loss": 0.7995,
+      "step": 242040
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 3.1602358066629095e-05,
+      "loss": 0.7231,
+      "step": 242050
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 3.160159641720109e-05,
+      "loss": 0.8438,
+      "step": 242060
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 3.160083476777309e-05,
+      "loss": 0.8089,
+      "step": 242070
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 3.1600073118345094e-05,
+      "loss": 0.84,
+      "step": 242080
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 3.1599311468917085e-05,
+      "loss": 0.8301,
+      "step": 242090
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 3.159854981948909e-05,
+      "loss": 0.8363,
+      "step": 242100
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 3.1597788170061086e-05,
+      "loss": 0.7877,
+      "step": 242110
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 3.1597026520633084e-05,
+      "loss": 0.791,
+      "step": 242120
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 3.159626487120508e-05,
+      "loss": 0.8479,
+      "step": 242130
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 3.1595503221777086e-05,
+      "loss": 0.8882,
+      "step": 242140
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 3.1594741572349076e-05,
+      "loss": 0.8428,
+      "step": 242150
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 3.159397992292108e-05,
+      "loss": 0.8888,
+      "step": 242160
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 3.159321827349308e-05,
+      "loss": 0.7719,
+      "step": 242170
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 3.1592456624065076e-05,
+      "loss": 0.776,
+      "step": 242180
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 3.159169497463707e-05,
+      "loss": 0.8058,
+      "step": 242190
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 3.159093332520908e-05,
+      "loss": 0.8157,
+      "step": 242200
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 3.1590171675781075e-05,
+      "loss": 0.8225,
+      "step": 242210
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 3.158941002635307e-05,
+      "loss": 0.75,
+      "step": 242220
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 3.158864837692507e-05,
+      "loss": 0.7658,
+      "step": 242230
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 3.158788672749707e-05,
+      "loss": 0.7322,
+      "step": 242240
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 3.158712507806907e-05,
+      "loss": 0.8608,
+      "step": 242250
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 3.158636342864107e-05,
+      "loss": 0.807,
+      "step": 242260
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 3.1585601779213066e-05,
+      "loss": 0.8166,
+      "step": 242270
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 3.1584840129785064e-05,
+      "loss": 0.8478,
+      "step": 242280
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 3.158407848035707e-05,
+      "loss": 0.8263,
+      "step": 242290
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 3.158331683092906e-05,
+      "loss": 0.8021,
+      "step": 242300
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 3.158255518150106e-05,
+      "loss": 0.9196,
+      "step": 242310
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 3.1581793532073054e-05,
+      "loss": 0.7328,
+      "step": 242320
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 3.158103188264506e-05,
+      "loss": 0.7172,
+      "step": 242330
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 3.1580270233217055e-05,
+      "loss": 0.8663,
+      "step": 242340
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 3.157950858378905e-05,
+      "loss": 0.9549,
+      "step": 242350
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 3.157874693436105e-05,
+      "loss": 0.7855,
+      "step": 242360
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 3.1577985284933055e-05,
+      "loss": 0.704,
+      "step": 242370
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 3.157722363550505e-05,
+      "loss": 0.7771,
+      "step": 242380
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 3.157646198607705e-05,
+      "loss": 0.8504,
+      "step": 242390
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 3.157570033664905e-05,
+      "loss": 0.7395,
+      "step": 242400
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 3.1574938687221044e-05,
+      "loss": 0.8071,
+      "step": 242410
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 3.157417703779305e-05,
+      "loss": 0.8565,
+      "step": 242420
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 3.1573415388365046e-05,
+      "loss": 0.8735,
+      "step": 242430
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 3.1572653738937044e-05,
+      "loss": 0.7509,
+      "step": 242440
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 3.157189208950904e-05,
+      "loss": 0.9038,
+      "step": 242450
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 3.1571130440081045e-05,
+      "loss": 0.8348,
+      "step": 242460
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 3.1570368790653036e-05,
+      "loss": 0.7451,
+      "step": 242470
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 3.156960714122504e-05,
+      "loss": 0.7697,
+      "step": 242480
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 3.156884549179704e-05,
+      "loss": 0.8114,
+      "step": 242490
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 3.1568083842369035e-05,
+      "loss": 0.7224,
+      "step": 242500
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 3.156732219294103e-05,
+      "loss": 0.8408,
+      "step": 242510
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 3.156656054351304e-05,
+      "loss": 0.8131,
+      "step": 242520
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 3.156579889408503e-05,
+      "loss": 0.7368,
+      "step": 242530
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 3.156503724465703e-05,
+      "loss": 0.7489,
+      "step": 242540
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 3.156427559522903e-05,
+      "loss": 0.7418,
+      "step": 242550
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 3.156351394580103e-05,
+      "loss": 0.8675,
+      "step": 242560
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 3.1562752296373024e-05,
+      "loss": 0.768,
+      "step": 242570
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 3.156199064694503e-05,
+      "loss": 0.8788,
+      "step": 242580
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 3.1561228997517026e-05,
+      "loss": 0.7667,
+      "step": 242590
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 3.156046734808902e-05,
+      "loss": 0.7849,
+      "step": 242600
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 3.155970569866102e-05,
+      "loss": 0.7872,
+      "step": 242610
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 3.155894404923302e-05,
+      "loss": 0.8391,
+      "step": 242620
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 3.155818239980502e-05,
+      "loss": 0.7361,
+      "step": 242630
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 3.155742075037702e-05,
+      "loss": 0.8319,
+      "step": 242640
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 3.155665910094902e-05,
+      "loss": 0.7749,
+      "step": 242650
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 3.1555897451521015e-05,
+      "loss": 0.824,
+      "step": 242660
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 3.155513580209302e-05,
+      "loss": 0.8186,
+      "step": 242670
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 3.155437415266501e-05,
+      "loss": 0.7194,
+      "step": 242680
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 3.1553612503237014e-05,
+      "loss": 0.7449,
+      "step": 242690
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 3.155285085380901e-05,
+      "loss": 0.7283,
+      "step": 242700
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 3.155208920438101e-05,
+      "loss": 0.7768,
+      "step": 242710
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 3.1551327554953007e-05,
+      "loss": 0.8418,
+      "step": 242720
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 3.155056590552501e-05,
+      "loss": 0.8282,
+      "step": 242730
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 3.1549804256097e-05,
+      "loss": 0.8435,
+      "step": 242740
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 3.1549042606669006e-05,
+      "loss": 0.8135,
+      "step": 242750
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 3.1548280957241e-05,
+      "loss": 0.8435,
+      "step": 242760
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 3.1547519307813e-05,
+      "loss": 0.8654,
+      "step": 242770
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 3.1546757658385e-05,
+      "loss": 0.7202,
+      "step": 242780
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 3.1545996008956996e-05,
+      "loss": 0.8173,
+      "step": 242790
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 3.1545234359529e-05,
+      "loss": 0.7738,
+      "step": 242800
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 3.1544472710101e-05,
+      "loss": 0.7575,
+      "step": 242810
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 3.1543711060672995e-05,
+      "loss": 0.8516,
+      "step": 242820
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 3.154294941124499e-05,
+      "loss": 0.8197,
+      "step": 242830
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 3.1542187761816996e-05,
+      "loss": 0.9168,
+      "step": 242840
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 3.154142611238899e-05,
+      "loss": 0.7896,
+      "step": 242850
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 3.154066446296099e-05,
+      "loss": 0.9448,
+      "step": 242860
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 3.153990281353299e-05,
+      "loss": 0.8627,
+      "step": 242870
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 3.1539141164104986e-05,
+      "loss": 0.7698,
+      "step": 242880
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 3.1538379514676984e-05,
+      "loss": 0.8011,
+      "step": 242890
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 3.153761786524899e-05,
+      "loss": 0.8028,
+      "step": 242900
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 3.153685621582098e-05,
+      "loss": 0.8279,
+      "step": 242910
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 3.153609456639298e-05,
+      "loss": 0.804,
+      "step": 242920
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 3.153533291696498e-05,
+      "loss": 0.7493,
+      "step": 242930
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 3.153457126753698e-05,
+      "loss": 0.81,
+      "step": 242940
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 3.1533809618108975e-05,
+      "loss": 0.8506,
+      "step": 242950
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 3.153304796868098e-05,
+      "loss": 0.7905,
+      "step": 242960
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 3.153228631925298e-05,
+      "loss": 0.7299,
+      "step": 242970
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 3.1531524669824975e-05,
+      "loss": 0.7366,
+      "step": 242980
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 3.153076302039697e-05,
+      "loss": 0.8702,
+      "step": 242990
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 3.153000137096897e-05,
+      "loss": 0.8437,
+      "step": 243000
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 3.1529239721540974e-05,
+      "loss": 0.7845,
+      "step": 243010
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 3.152847807211297e-05,
+      "loss": 0.8027,
+      "step": 243020
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 3.152771642268497e-05,
+      "loss": 0.93,
+      "step": 243030
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 3.1526954773256966e-05,
+      "loss": 0.8123,
+      "step": 243040
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 3.152619312382897e-05,
+      "loss": 0.7944,
+      "step": 243050
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 3.152543147440096e-05,
+      "loss": 0.8761,
+      "step": 243060
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 3.1524669824972965e-05,
+      "loss": 0.9835,
+      "step": 243070
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 3.152390817554496e-05,
+      "loss": 0.8822,
+      "step": 243080
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 3.152314652611696e-05,
+      "loss": 0.83,
+      "step": 243090
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 3.152238487668896e-05,
+      "loss": 0.7742,
+      "step": 243100
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 3.152162322726096e-05,
+      "loss": 0.8183,
+      "step": 243110
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 3.152086157783295e-05,
+      "loss": 0.8847,
+      "step": 243120
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 3.152009992840496e-05,
+      "loss": 0.6989,
+      "step": 243130
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 3.1519338278976954e-05,
+      "loss": 0.8782,
+      "step": 243140
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 3.151857662954895e-05,
+      "loss": 0.7814,
+      "step": 243150
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 3.151781498012095e-05,
+      "loss": 0.7753,
+      "step": 243160
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 3.1517053330692953e-05,
+      "loss": 0.8072,
+      "step": 243170
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 3.151629168126495e-05,
+      "loss": 0.8729,
+      "step": 243180
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 3.151553003183695e-05,
+      "loss": 0.7416,
+      "step": 243190
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 3.1514768382408946e-05,
+      "loss": 0.8621,
+      "step": 243200
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 3.151400673298094e-05,
+      "loss": 0.801,
+      "step": 243210
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 3.151324508355295e-05,
+      "loss": 0.7979,
+      "step": 243220
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 3.151248343412494e-05,
+      "loss": 0.8937,
+      "step": 243230
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 3.151172178469694e-05,
+      "loss": 0.9408,
+      "step": 243240
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 3.151096013526894e-05,
+      "loss": 0.7753,
+      "step": 243250
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 3.151019848584094e-05,
+      "loss": 0.8583,
+      "step": 243260
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 3.1509436836412935e-05,
+      "loss": 0.8077,
+      "step": 243270
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 3.150867518698494e-05,
+      "loss": 0.7585,
+      "step": 243280
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 3.150791353755693e-05,
+      "loss": 0.9108,
+      "step": 243290
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 3.1507151888128934e-05,
+      "loss": 0.8608,
+      "step": 243300
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 3.150639023870093e-05,
+      "loss": 0.831,
+      "step": 243310
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 3.150562858927293e-05,
+      "loss": 0.8501,
+      "step": 243320
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 3.1504866939844927e-05,
+      "loss": 0.6943,
+      "step": 243330
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 3.150410529041693e-05,
+      "loss": 0.772,
+      "step": 243340
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 3.150334364098893e-05,
+      "loss": 0.8379,
+      "step": 243350
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 3.1502581991560926e-05,
+      "loss": 0.8542,
+      "step": 243360
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 3.150182034213292e-05,
+      "loss": 0.7812,
+      "step": 243370
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 3.150105869270492e-05,
+      "loss": 0.7921,
+      "step": 243380
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 3.1500297043276925e-05,
+      "loss": 0.9032,
+      "step": 243390
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 3.149953539384892e-05,
+      "loss": 0.8338,
+      "step": 243400
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 3.149877374442092e-05,
+      "loss": 0.8362,
+      "step": 243410
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 3.149801209499292e-05,
+      "loss": 0.8266,
+      "step": 243420
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 3.149725044556492e-05,
+      "loss": 0.8747,
+      "step": 243430
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 3.149648879613691e-05,
+      "loss": 0.7609,
+      "step": 243440
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 3.1495727146708916e-05,
+      "loss": 0.8686,
+      "step": 243450
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 3.1494965497280914e-05,
+      "loss": 0.788,
+      "step": 243460
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 3.149420384785291e-05,
+      "loss": 0.8371,
+      "step": 243470
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 3.149344219842491e-05,
+      "loss": 0.8623,
+      "step": 243480
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 3.149268054899691e-05,
+      "loss": 0.7847,
+      "step": 243490
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 3.1491918899568904e-05,
+      "loss": 0.8378,
+      "step": 243500
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 3.149115725014091e-05,
+      "loss": 0.7585,
+      "step": 243510
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 3.1490395600712905e-05,
+      "loss": 0.8112,
+      "step": 243520
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 3.14896339512849e-05,
+      "loss": 0.7809,
+      "step": 243530
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 3.14888723018569e-05,
+      "loss": 0.8669,
+      "step": 243540
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 3.1488110652428905e-05,
+      "loss": 0.7866,
+      "step": 243550
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 3.14873490030009e-05,
+      "loss": 0.8278,
+      "step": 243560
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 3.14865873535729e-05,
+      "loss": 0.8453,
+      "step": 243570
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 3.14858257041449e-05,
+      "loss": 0.7872,
+      "step": 243580
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 3.1485064054716895e-05,
+      "loss": 0.82,
+      "step": 243590
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 3.14843024052889e-05,
+      "loss": 0.734,
+      "step": 243600
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 3.1483540755860896e-05,
+      "loss": 0.7625,
+      "step": 243610
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 3.1482779106432894e-05,
+      "loss": 0.7295,
+      "step": 243620
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 3.148201745700489e-05,
+      "loss": 0.8618,
+      "step": 243630
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 3.1481255807576895e-05,
+      "loss": 0.8839,
+      "step": 243640
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 3.1480494158148886e-05,
+      "loss": 0.8215,
+      "step": 243650
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 3.147973250872089e-05,
+      "loss": 0.7835,
+      "step": 243660
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 3.147897085929289e-05,
+      "loss": 0.8241,
+      "step": 243670
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 3.1478209209864885e-05,
+      "loss": 0.7724,
+      "step": 243680
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 3.147744756043688e-05,
+      "loss": 0.7814,
+      "step": 243690
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 3.147668591100888e-05,
+      "loss": 0.7876,
+      "step": 243700
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 3.147592426158088e-05,
+      "loss": 0.8625,
+      "step": 243710
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 3.147516261215288e-05,
+      "loss": 0.7673,
+      "step": 243720
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 3.147440096272487e-05,
+      "loss": 0.8501,
+      "step": 243730
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 3.147363931329688e-05,
+      "loss": 0.8379,
+      "step": 243740
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 3.1472877663868874e-05,
+      "loss": 0.8753,
+      "step": 243750
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 3.147211601444087e-05,
+      "loss": 0.7954,
+      "step": 243760
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 3.1471354365012876e-05,
+      "loss": 0.795,
+      "step": 243770
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 3.1470592715584874e-05,
+      "loss": 0.8025,
+      "step": 243780
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 3.146983106615687e-05,
+      "loss": 0.8797,
+      "step": 243790
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 3.146906941672887e-05,
+      "loss": 0.7649,
+      "step": 243800
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 3.146830776730087e-05,
+      "loss": 0.7501,
+      "step": 243810
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 3.146754611787286e-05,
+      "loss": 0.819,
+      "step": 243820
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 3.146678446844487e-05,
+      "loss": 0.7973,
+      "step": 243830
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 3.1466022819016865e-05,
+      "loss": 0.8248,
+      "step": 243840
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 3.146526116958886e-05,
+      "loss": 0.7822,
+      "step": 243850
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 3.146449952016086e-05,
+      "loss": 0.7778,
+      "step": 243860
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 3.1463737870732864e-05,
+      "loss": 0.8433,
+      "step": 243870
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 3.1462976221304855e-05,
+      "loss": 0.8748,
+      "step": 243880
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 3.146221457187686e-05,
+      "loss": 0.7946,
+      "step": 243890
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 3.146145292244886e-05,
+      "loss": 0.8417,
+      "step": 243900
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 3.1460691273020854e-05,
+      "loss": 0.824,
+      "step": 243910
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 3.145992962359285e-05,
+      "loss": 0.8239,
+      "step": 243920
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 3.1459167974164856e-05,
+      "loss": 0.8315,
+      "step": 243930
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 3.1458406324736847e-05,
+      "loss": 0.9533,
+      "step": 243940
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 3.145764467530885e-05,
+      "loss": 0.8135,
+      "step": 243950
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 3.145688302588085e-05,
+      "loss": 0.8809,
+      "step": 243960
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 3.1456121376452846e-05,
+      "loss": 0.9038,
+      "step": 243970
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 3.145535972702485e-05,
+      "loss": 0.8514,
+      "step": 243980
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 3.145459807759685e-05,
+      "loss": 0.8,
+      "step": 243990
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 3.1453836428168845e-05,
+      "loss": 0.849,
+      "step": 244000
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 3.145307477874084e-05,
+      "loss": 0.8262,
+      "step": 244010
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 3.1452313129312847e-05,
+      "loss": 0.8549,
+      "step": 244020
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 3.145155147988484e-05,
+      "loss": 0.853,
+      "step": 244030
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 3.145078983045684e-05,
+      "loss": 0.8293,
+      "step": 244040
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 3.145002818102884e-05,
+      "loss": 0.7952,
+      "step": 244050
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 3.1449266531600836e-05,
+      "loss": 0.8279,
+      "step": 244060
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 3.1448504882172834e-05,
+      "loss": 0.8308,
+      "step": 244070
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 3.144774323274484e-05,
+      "loss": 0.8506,
+      "step": 244080
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 3.144698158331683e-05,
+      "loss": 0.8039,
+      "step": 244090
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 3.144621993388883e-05,
+      "loss": 0.8165,
+      "step": 244100
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 3.144545828446083e-05,
+      "loss": 0.8451,
+      "step": 244110
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 3.144469663503283e-05,
+      "loss": 0.8245,
+      "step": 244120
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 3.1443934985604826e-05,
+      "loss": 0.8217,
+      "step": 244130
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 3.144317333617683e-05,
+      "loss": 0.7225,
+      "step": 244140
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 3.144241168674883e-05,
+      "loss": 0.9314,
+      "step": 244150
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 3.1441650037320825e-05,
+      "loss": 0.8116,
+      "step": 244160
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 3.144088838789282e-05,
+      "loss": 0.784,
+      "step": 244170
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 3.144012673846482e-05,
+      "loss": 0.8235,
+      "step": 244180
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 3.1439365089036824e-05,
+      "loss": 0.8346,
+      "step": 244190
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 3.1438603439608815e-05,
+      "loss": 0.7517,
+      "step": 244200
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 3.143784179018082e-05,
+      "loss": 0.731,
+      "step": 244210
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 3.1437080140752816e-05,
+      "loss": 0.8848,
+      "step": 244220
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 3.1436318491324814e-05,
+      "loss": 0.8197,
+      "step": 244230
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 3.143555684189681e-05,
+      "loss": 0.8382,
+      "step": 244240
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 3.1434795192468815e-05,
+      "loss": 0.8111,
+      "step": 244250
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 3.1434033543040806e-05,
+      "loss": 0.915,
+      "step": 244260
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 3.143327189361281e-05,
+      "loss": 0.8236,
+      "step": 244270
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 3.143251024418481e-05,
+      "loss": 0.8706,
+      "step": 244280
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 3.1431748594756805e-05,
+      "loss": 0.8456,
+      "step": 244290
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 3.14309869453288e-05,
+      "loss": 0.8048,
+      "step": 244300
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 3.143022529590081e-05,
+      "loss": 0.7034,
+      "step": 244310
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 3.14294636464728e-05,
+      "loss": 0.8689,
+      "step": 244320
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 3.14287019970448e-05,
+      "loss": 0.8098,
+      "step": 244330
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 3.14279403476168e-05,
+      "loss": 0.9895,
+      "step": 244340
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 3.14271786981888e-05,
+      "loss": 0.9059,
+      "step": 244350
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 3.14264170487608e-05,
+      "loss": 0.8302,
+      "step": 244360
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 3.14256553993328e-05,
+      "loss": 0.847,
+      "step": 244370
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 3.1424893749904796e-05,
+      "loss": 0.9083,
+      "step": 244380
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 3.1424132100476794e-05,
+      "loss": 0.7659,
+      "step": 244390
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 3.14233704510488e-05,
+      "loss": 0.9056,
+      "step": 244400
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 3.142260880162079e-05,
+      "loss": 0.7852,
+      "step": 244410
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 3.142184715219279e-05,
+      "loss": 0.8424,
+      "step": 244420
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 3.142108550276479e-05,
+      "loss": 0.8055,
+      "step": 244430
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 3.142032385333679e-05,
+      "loss": 0.7451,
+      "step": 244440
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 3.1419562203908785e-05,
+      "loss": 0.7796,
+      "step": 244450
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 3.141880055448079e-05,
+      "loss": 0.8262,
+      "step": 244460
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 3.141803890505278e-05,
+      "loss": 0.7949,
+      "step": 244470
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 3.1417277255624784e-05,
+      "loss": 0.7794,
+      "step": 244480
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 3.141651560619678e-05,
+      "loss": 0.853,
+      "step": 244490
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 3.141575395676878e-05,
+      "loss": 0.8154,
+      "step": 244500
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 3.141499230734078e-05,
+      "loss": 0.85,
+      "step": 244510
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 3.141423065791278e-05,
+      "loss": 0.8268,
+      "step": 244520
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 3.141346900848477e-05,
+      "loss": 0.7897,
+      "step": 244530
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 3.1412707359056776e-05,
+      "loss": 0.6952,
+      "step": 244540
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 3.141194570962877e-05,
+      "loss": 0.7518,
+      "step": 244550
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 3.141118406020077e-05,
+      "loss": 0.8737,
+      "step": 244560
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 3.1410422410772775e-05,
+      "loss": 0.8587,
+      "step": 244570
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 3.140966076134477e-05,
+      "loss": 0.8028,
+      "step": 244580
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 3.140889911191677e-05,
+      "loss": 0.7863,
+      "step": 244590
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 3.140813746248877e-05,
+      "loss": 0.764,
+      "step": 244600
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 3.1407375813060765e-05,
+      "loss": 0.7935,
+      "step": 244610
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 3.140661416363276e-05,
+      "loss": 0.8482,
+      "step": 244620
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 3.1405852514204767e-05,
+      "loss": 0.7941,
+      "step": 244630
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 3.140509086477676e-05,
+      "loss": 0.8732,
+      "step": 244640
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 3.140432921534876e-05,
+      "loss": 0.729,
+      "step": 244650
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 3.140356756592076e-05,
+      "loss": 0.7723,
+      "step": 244660
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 3.1402805916492756e-05,
+      "loss": 0.754,
+      "step": 244670
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 3.1402044267064754e-05,
+      "loss": 0.8738,
+      "step": 244680
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 3.140128261763676e-05,
+      "loss": 0.7604,
+      "step": 244690
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 3.140052096820875e-05,
+      "loss": 0.7574,
+      "step": 244700
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 3.139975931878075e-05,
+      "loss": 1.0321,
+      "step": 244710
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 3.139899766935275e-05,
+      "loss": 0.8746,
+      "step": 244720
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 3.139823601992475e-05,
+      "loss": 0.7762,
+      "step": 244730
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 3.1397474370496746e-05,
+      "loss": 0.8181,
+      "step": 244740
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 3.139671272106875e-05,
+      "loss": 0.9045,
+      "step": 244750
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 3.139595107164075e-05,
+      "loss": 0.8733,
+      "step": 244760
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 3.1395189422212745e-05,
+      "loss": 0.8441,
+      "step": 244770
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 3.139442777278475e-05,
+      "loss": 0.8502,
+      "step": 244780
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 3.139366612335674e-05,
+      "loss": 0.853,
+      "step": 244790
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 3.1392904473928744e-05,
+      "loss": 0.8798,
+      "step": 244800
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 3.139214282450074e-05,
+      "loss": 0.83,
+      "step": 244810
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 3.139138117507274e-05,
+      "loss": 0.8062,
+      "step": 244820
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 3.1390619525644736e-05,
+      "loss": 0.8822,
+      "step": 244830
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 3.138985787621674e-05,
+      "loss": 0.8544,
+      "step": 244840
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 3.138909622678873e-05,
+      "loss": 0.8014,
+      "step": 244850
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 3.1388334577360735e-05,
+      "loss": 0.8439,
+      "step": 244860
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 3.138757292793273e-05,
+      "loss": 0.8337,
+      "step": 244870
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 3.138681127850473e-05,
+      "loss": 0.7987,
+      "step": 244880
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 3.138604962907673e-05,
+      "loss": 0.7488,
+      "step": 244890
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 3.138528797964873e-05,
+      "loss": 0.8758,
+      "step": 244900
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 3.138452633022072e-05,
+      "loss": 0.7894,
+      "step": 244910
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 3.138376468079273e-05,
+      "loss": 0.8087,
+      "step": 244920
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 3.1383003031364724e-05,
+      "loss": 0.8517,
+      "step": 244930
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 3.138224138193672e-05,
+      "loss": 0.7141,
+      "step": 244940
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 3.1381479732508726e-05,
+      "loss": 0.8027,
+      "step": 244950
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 3.1380718083080724e-05,
+      "loss": 0.7532,
+      "step": 244960
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 3.137995643365272e-05,
+      "loss": 0.8683,
+      "step": 244970
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 3.137919478422472e-05,
+      "loss": 0.8147,
+      "step": 244980
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 3.137843313479672e-05,
+      "loss": 0.8486,
+      "step": 244990
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 3.1377671485368714e-05,
+      "loss": 0.8122,
+      "step": 245000
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 3.137690983594072e-05,
+      "loss": 0.8092,
+      "step": 245010
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 3.1376148186512715e-05,
+      "loss": 0.7377,
+      "step": 245020
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 3.137538653708471e-05,
+      "loss": 0.8736,
+      "step": 245030
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 3.137462488765671e-05,
+      "loss": 0.7871,
+      "step": 245040
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 3.1373863238228714e-05,
+      "loss": 0.7763,
+      "step": 245050
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 3.1373101588800705e-05,
+      "loss": 0.7966,
+      "step": 245060
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 3.137233993937271e-05,
+      "loss": 0.761,
+      "step": 245070
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 3.13715782899447e-05,
+      "loss": 0.8757,
+      "step": 245080
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 3.1370816640516704e-05,
+      "loss": 0.8316,
+      "step": 245090
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 3.13700549910887e-05,
+      "loss": 0.762,
+      "step": 245100
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 3.13692933416607e-05,
+      "loss": 0.8995,
+      "step": 245110
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 3.13685316922327e-05,
+      "loss": 0.8032,
+      "step": 245120
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 3.13677700428047e-05,
+      "loss": 0.7505,
+      "step": 245130
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 3.13670083933767e-05,
+      "loss": 0.8032,
+      "step": 245140
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 3.1366246743948696e-05,
+      "loss": 0.754,
+      "step": 245150
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 3.13654850945207e-05,
+      "loss": 0.9667,
+      "step": 245160
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 3.136472344509269e-05,
+      "loss": 0.8636,
+      "step": 245170
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 3.1363961795664695e-05,
+      "loss": 0.9348,
+      "step": 245180
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 3.136320014623669e-05,
+      "loss": 0.8279,
+      "step": 245190
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 3.136243849680869e-05,
+      "loss": 0.8548,
+      "step": 245200
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 3.136167684738069e-05,
+      "loss": 0.7613,
+      "step": 245210
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 3.136091519795269e-05,
+      "loss": 0.8079,
+      "step": 245220
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 3.136015354852468e-05,
+      "loss": 0.7473,
+      "step": 245230
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 3.1359391899096687e-05,
+      "loss": 0.8709,
+      "step": 245240
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 3.1358630249668684e-05,
+      "loss": 0.8771,
+      "step": 245250
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 3.135786860024068e-05,
+      "loss": 0.7553,
+      "step": 245260
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 3.135710695081268e-05,
+      "loss": 0.8368,
+      "step": 245270
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 3.135634530138468e-05,
+      "loss": 0.86,
+      "step": 245280
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 3.1355583651956674e-05,
+      "loss": 0.7453,
+      "step": 245290
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 3.135482200252868e-05,
+      "loss": 0.7889,
+      "step": 245300
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 3.1354060353100676e-05,
+      "loss": 0.8092,
+      "step": 245310
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 3.135329870367267e-05,
+      "loss": 0.8278,
+      "step": 245320
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 3.135253705424467e-05,
+      "loss": 0.8654,
+      "step": 245330
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 3.1351775404816675e-05,
+      "loss": 0.8614,
+      "step": 245340
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 3.135101375538867e-05,
+      "loss": 0.7616,
+      "step": 245350
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 3.135025210596067e-05,
+      "loss": 0.9594,
+      "step": 245360
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 3.1349490456532674e-05,
+      "loss": 0.7674,
+      "step": 245370
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 3.1348728807104665e-05,
+      "loss": 0.7385,
+      "step": 245380
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 3.134796715767667e-05,
+      "loss": 0.7719,
+      "step": 245390
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 3.1347205508248666e-05,
+      "loss": 0.8182,
+      "step": 245400
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 3.1346443858820664e-05,
+      "loss": 0.9208,
+      "step": 245410
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 3.134568220939266e-05,
+      "loss": 0.7439,
+      "step": 245420
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 3.1344920559964666e-05,
+      "loss": 0.8806,
+      "step": 245430
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 3.1344158910536656e-05,
+      "loss": 0.7943,
+      "step": 245440
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 3.134339726110866e-05,
+      "loss": 0.9061,
+      "step": 245450
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 3.134263561168066e-05,
+      "loss": 0.808,
+      "step": 245460
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 3.1341873962252655e-05,
+      "loss": 0.7439,
+      "step": 245470
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 3.134111231282465e-05,
+      "loss": 0.844,
+      "step": 245480
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 3.134035066339666e-05,
+      "loss": 0.8859,
+      "step": 245490
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 3.133958901396865e-05,
+      "loss": 0.7968,
+      "step": 245500
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 3.133882736454065e-05,
+      "loss": 0.8256,
+      "step": 245510
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 3.133806571511265e-05,
+      "loss": 0.7936,
+      "step": 245520
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 3.133730406568465e-05,
+      "loss": 0.8436,
+      "step": 245530
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 3.133654241625665e-05,
+      "loss": 0.7955,
+      "step": 245540
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 3.133578076682864e-05,
+      "loss": 0.8351,
+      "step": 245550
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 3.1335019117400646e-05,
+      "loss": 0.8382,
+      "step": 245560
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 3.1334257467972644e-05,
+      "loss": 0.7893,
+      "step": 245570
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 3.133349581854464e-05,
+      "loss": 0.8154,
+      "step": 245580
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 3.133273416911664e-05,
+      "loss": 0.8017,
+      "step": 245590
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 3.133197251968864e-05,
+      "loss": 0.8613,
+      "step": 245600
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 3.1331210870260634e-05,
+      "loss": 0.7987,
+      "step": 245610
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 3.133044922083264e-05,
+      "loss": 0.8091,
+      "step": 245620
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 3.1329687571404635e-05,
+      "loss": 0.7611,
+      "step": 245630
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 3.132892592197663e-05,
+      "loss": 0.7354,
+      "step": 245640
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 3.132816427254863e-05,
+      "loss": 0.8396,
+      "step": 245650
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 3.1327402623120634e-05,
+      "loss": 0.8253,
+      "step": 245660
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 3.1326640973692625e-05,
+      "loss": 0.8058,
+      "step": 245670
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 3.132587932426463e-05,
+      "loss": 0.7877,
+      "step": 245680
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 3.132511767483663e-05,
+      "loss": 0.8965,
+      "step": 245690
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 3.1324356025408624e-05,
+      "loss": 0.7505,
+      "step": 245700
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 3.132359437598062e-05,
+      "loss": 0.8436,
+      "step": 245710
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 3.1322832726552626e-05,
+      "loss": 0.8587,
+      "step": 245720
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 3.1322071077124623e-05,
+      "loss": 0.7504,
+      "step": 245730
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 3.132130942769662e-05,
+      "loss": 0.7458,
+      "step": 245740
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 3.1320547778268625e-05,
+      "loss": 0.8017,
+      "step": 245750
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 3.1319786128840616e-05,
+      "loss": 0.8516,
+      "step": 245760
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 3.131902447941262e-05,
+      "loss": 0.7552,
+      "step": 245770
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 3.131826282998462e-05,
+      "loss": 0.8394,
+      "step": 245780
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 3.1317501180556615e-05,
+      "loss": 0.8254,
+      "step": 245790
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 3.131673953112861e-05,
+      "loss": 0.7857,
+      "step": 245800
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 3.131597788170062e-05,
+      "loss": 0.8453,
+      "step": 245810
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 3.131521623227261e-05,
+      "loss": 0.7183,
+      "step": 245820
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 3.131445458284461e-05,
+      "loss": 0.8405,
+      "step": 245830
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 3.131369293341661e-05,
+      "loss": 0.833,
+      "step": 245840
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 3.131293128398861e-05,
+      "loss": 0.8375,
+      "step": 245850
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 3.1312169634560604e-05,
+      "loss": 0.7895,
+      "step": 245860
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 3.131140798513261e-05,
+      "loss": 0.8063,
+      "step": 245870
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 3.13106463357046e-05,
+      "loss": 0.8246,
+      "step": 245880
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 3.13098846862766e-05,
+      "loss": 0.8046,
+      "step": 245890
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 3.13091230368486e-05,
+      "loss": 0.9101,
+      "step": 245900
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 3.13083613874206e-05,
+      "loss": 0.8648,
+      "step": 245910
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 3.1307599737992596e-05,
+      "loss": 0.7411,
+      "step": 245920
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 3.13068380885646e-05,
+      "loss": 0.9013,
+      "step": 245930
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 3.13060764391366e-05,
+      "loss": 0.8091,
+      "step": 245940
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 3.1305314789708595e-05,
+      "loss": 0.826,
+      "step": 245950
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 3.13045531402806e-05,
+      "loss": 0.7267,
+      "step": 245960
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 3.130379149085259e-05,
+      "loss": 0.8864,
+      "step": 245970
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 3.1303029841424594e-05,
+      "loss": 0.7756,
+      "step": 245980
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 3.1302268191996585e-05,
+      "loss": 0.8679,
+      "step": 245990
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 3.130150654256859e-05,
+      "loss": 0.8129,
+      "step": 246000
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 3.1300744893140586e-05,
+      "loss": 0.8019,
+      "step": 246010
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 3.1299983243712584e-05,
+      "loss": 0.8285,
+      "step": 246020
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 3.129922159428458e-05,
+      "loss": 0.8582,
+      "step": 246030
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 3.1298459944856586e-05,
+      "loss": 0.762,
+      "step": 246040
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 3.1297698295428576e-05,
+      "loss": 0.7945,
+      "step": 246050
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 3.129693664600058e-05,
+      "loss": 0.8541,
+      "step": 246060
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 3.129617499657258e-05,
+      "loss": 0.8494,
+      "step": 246070
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 3.1295413347144575e-05,
+      "loss": 0.8404,
+      "step": 246080
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 3.129465169771657e-05,
+      "loss": 0.7634,
+      "step": 246090
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 3.129389004828858e-05,
+      "loss": 0.8613,
+      "step": 246100
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 3.1293128398860575e-05,
+      "loss": 0.793,
+      "step": 246110
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 3.129236674943257e-05,
+      "loss": 0.9263,
+      "step": 246120
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 3.129160510000457e-05,
+      "loss": 0.8968,
+      "step": 246130
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 3.129084345057657e-05,
+      "loss": 0.7892,
+      "step": 246140
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 3.129008180114857e-05,
+      "loss": 0.7429,
+      "step": 246150
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 3.128932015172057e-05,
+      "loss": 0.7889,
+      "step": 246160
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 3.1288558502292566e-05,
+      "loss": 0.8817,
+      "step": 246170
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 3.1287796852864564e-05,
+      "loss": 0.7806,
+      "step": 246180
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 3.128703520343657e-05,
+      "loss": 0.9367,
+      "step": 246190
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 3.128627355400856e-05,
+      "loss": 0.8556,
+      "step": 246200
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 3.128551190458056e-05,
+      "loss": 0.7584,
+      "step": 246210
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 3.128475025515256e-05,
+      "loss": 0.8131,
+      "step": 246220
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 3.128398860572456e-05,
+      "loss": 0.7483,
+      "step": 246230
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 3.1283226956296555e-05,
+      "loss": 0.7577,
+      "step": 246240
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 3.128246530686856e-05,
+      "loss": 0.9055,
+      "step": 246250
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 3.128170365744055e-05,
+      "loss": 0.8855,
+      "step": 246260
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 3.1280942008012554e-05,
+      "loss": 0.877,
+      "step": 246270
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 3.128018035858455e-05,
+      "loss": 0.7667,
+      "step": 246280
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 3.127941870915655e-05,
+      "loss": 0.8529,
+      "step": 246290
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 3.127865705972855e-05,
+      "loss": 0.7752,
+      "step": 246300
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 3.127789541030055e-05,
+      "loss": 0.8079,
+      "step": 246310
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 3.127713376087255e-05,
+      "loss": 0.8091,
+      "step": 246320
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 3.1276372111444546e-05,
+      "loss": 0.8261,
+      "step": 246330
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 3.127561046201655e-05,
+      "loss": 0.8448,
+      "step": 246340
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 3.127484881258854e-05,
+      "loss": 0.7969,
+      "step": 246350
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 3.1274087163160545e-05,
+      "loss": 0.8829,
+      "step": 246360
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 3.127332551373254e-05,
+      "loss": 0.7656,
+      "step": 246370
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 3.127256386430454e-05,
+      "loss": 0.8123,
+      "step": 246380
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 3.127180221487654e-05,
+      "loss": 0.792,
+      "step": 246390
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 3.127104056544854e-05,
+      "loss": 0.8158,
+      "step": 246400
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 3.127027891602053e-05,
+      "loss": 0.7268,
+      "step": 246410
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 3.126951726659254e-05,
+      "loss": 0.9328,
+      "step": 246420
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 3.1268755617164534e-05,
+      "loss": 0.7963,
+      "step": 246430
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 3.126799396773653e-05,
+      "loss": 0.84,
+      "step": 246440
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 3.126723231830853e-05,
+      "loss": 0.8331,
+      "step": 246450
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 3.126647066888053e-05,
+      "loss": 0.8201,
+      "step": 246460
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 3.1265709019452524e-05,
+      "loss": 0.8565,
+      "step": 246470
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 3.126494737002453e-05,
+      "loss": 0.792,
+      "step": 246480
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 3.1264185720596526e-05,
+      "loss": 0.8099,
+      "step": 246490
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 3.126342407116852e-05,
+      "loss": 0.8424,
+      "step": 246500
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 3.126266242174052e-05,
+      "loss": 0.8071,
+      "step": 246510
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 3.126190077231252e-05,
+      "loss": 0.7946,
+      "step": 246520
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 3.126113912288452e-05,
+      "loss": 0.7742,
+      "step": 246530
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 3.126037747345652e-05,
+      "loss": 0.7132,
+      "step": 246540
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 3.125961582402852e-05,
+      "loss": 0.769,
+      "step": 246550
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 3.1258854174600515e-05,
+      "loss": 0.7647,
+      "step": 246560
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 3.125809252517252e-05,
+      "loss": 0.8873,
+      "step": 246570
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 3.125733087574451e-05,
+      "loss": 0.8421,
+      "step": 246580
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 3.1256569226316514e-05,
+      "loss": 0.7606,
+      "step": 246590
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 3.125580757688851e-05,
+      "loss": 0.8073,
+      "step": 246600
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 3.125504592746051e-05,
+      "loss": 0.8713,
+      "step": 246610
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 3.1254284278032506e-05,
+      "loss": 0.9312,
+      "step": 246620
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 3.125352262860451e-05,
+      "loss": 0.8241,
+      "step": 246630
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 3.12527609791765e-05,
+      "loss": 0.8545,
+      "step": 246640
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 3.1251999329748506e-05,
+      "loss": 0.704,
+      "step": 246650
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 3.12512376803205e-05,
+      "loss": 0.8553,
+      "step": 246660
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 3.12504760308925e-05,
+      "loss": 0.7587,
+      "step": 246670
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 3.12497143814645e-05,
+      "loss": 0.813,
+      "step": 246680
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 3.12489527320365e-05,
+      "loss": 0.8824,
+      "step": 246690
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 3.12481910826085e-05,
+      "loss": 0.8151,
+      "step": 246700
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 3.12474294331805e-05,
+      "loss": 0.7686,
+      "step": 246710
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 3.1246667783752495e-05,
+      "loss": 0.7407,
+      "step": 246720
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 3.124590613432449e-05,
+      "loss": 0.8126,
+      "step": 246730
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 3.1245144484896496e-05,
+      "loss": 0.758,
+      "step": 246740
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 3.1244382835468494e-05,
+      "loss": 0.8688,
+      "step": 246750
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 3.124362118604049e-05,
+      "loss": 0.9234,
+      "step": 246760
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 3.124285953661249e-05,
+      "loss": 0.7657,
+      "step": 246770
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 3.124209788718449e-05,
+      "loss": 0.8445,
+      "step": 246780
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 3.1241336237756484e-05,
+      "loss": 0.89,
+      "step": 246790
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 3.124057458832849e-05,
+      "loss": 0.9168,
+      "step": 246800
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 3.1239812938900485e-05,
+      "loss": 0.8822,
+      "step": 246810
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 3.123905128947248e-05,
+      "loss": 0.8468,
+      "step": 246820
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 3.123828964004448e-05,
+      "loss": 0.8485,
+      "step": 246830
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 3.1237527990616485e-05,
+      "loss": 0.8057,
+      "step": 246840
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 3.1236766341188475e-05,
+      "loss": 0.8071,
+      "step": 246850
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 3.123600469176048e-05,
+      "loss": 0.7975,
+      "step": 246860
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 3.123524304233248e-05,
+      "loss": 0.6968,
+      "step": 246870
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 3.1234481392904474e-05,
+      "loss": 0.8307,
+      "step": 246880
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 3.123371974347647e-05,
+      "loss": 0.8275,
+      "step": 246890
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 3.1232958094048476e-05,
+      "loss": 0.8206,
+      "step": 246900
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 3.1232196444620474e-05,
+      "loss": 0.8562,
+      "step": 246910
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 3.123143479519247e-05,
+      "loss": 0.7771,
+      "step": 246920
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 3.123067314576447e-05,
+      "loss": 0.8672,
+      "step": 246930
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 3.1229911496336466e-05,
+      "loss": 0.8318,
+      "step": 246940
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 3.122914984690847e-05,
+      "loss": 0.7564,
+      "step": 246950
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 3.122838819748046e-05,
+      "loss": 0.8077,
+      "step": 246960
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 3.1227626548052465e-05,
+      "loss": 0.9374,
+      "step": 246970
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 3.122686489862446e-05,
+      "loss": 0.8605,
+      "step": 246980
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 3.122610324919646e-05,
+      "loss": 0.8553,
+      "step": 246990
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 3.122534159976846e-05,
+      "loss": 0.8265,
+      "step": 247000
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 3.122457995034046e-05,
+      "loss": 0.8382,
+      "step": 247010
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 3.122381830091245e-05,
+      "loss": 0.7972,
+      "step": 247020
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 3.122305665148446e-05,
+      "loss": 0.9174,
+      "step": 247030
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 3.1222295002056454e-05,
+      "loss": 0.86,
+      "step": 247040
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 3.122153335262845e-05,
+      "loss": 0.7795,
+      "step": 247050
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 3.122077170320045e-05,
+      "loss": 0.8438,
+      "step": 247060
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 3.1220010053772453e-05,
+      "loss": 0.902,
+      "step": 247070
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 3.121924840434445e-05,
+      "loss": 0.8693,
+      "step": 247080
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 3.121848675491645e-05,
+      "loss": 0.8446,
+      "step": 247090
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 3.1217725105488446e-05,
+      "loss": 0.8386,
+      "step": 247100
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 3.121696345606044e-05,
+      "loss": 0.7491,
+      "step": 247110
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 3.121620180663245e-05,
+      "loss": 0.8729,
+      "step": 247120
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 3.1215440157204445e-05,
+      "loss": 1.016,
+      "step": 247130
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 3.121467850777644e-05,
+      "loss": 0.8719,
+      "step": 247140
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 3.121391685834844e-05,
+      "loss": 0.8929,
+      "step": 247150
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 3.1213155208920444e-05,
+      "loss": 0.7056,
+      "step": 247160
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 3.1212393559492435e-05,
+      "loss": 0.8798,
+      "step": 247170
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 3.121163191006444e-05,
+      "loss": 0.8264,
+      "step": 247180
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 3.1210870260636437e-05,
+      "loss": 0.9052,
+      "step": 247190
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 3.1210108611208434e-05,
+      "loss": 0.8143,
+      "step": 247200
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 3.120934696178043e-05,
+      "loss": 0.8981,
+      "step": 247210
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 3.1208585312352436e-05,
+      "loss": 0.8035,
+      "step": 247220
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 3.1207823662924426e-05,
+      "loss": 0.817,
+      "step": 247230
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 3.120706201349643e-05,
+      "loss": 0.8614,
+      "step": 247240
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 3.120630036406843e-05,
+      "loss": 0.8163,
+      "step": 247250
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 3.1205538714640426e-05,
+      "loss": 0.6858,
+      "step": 247260
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 3.120477706521242e-05,
+      "loss": 0.8081,
+      "step": 247270
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 3.120401541578443e-05,
+      "loss": 0.7737,
+      "step": 247280
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 3.1203253766356425e-05,
+      "loss": 0.8575,
+      "step": 247290
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 3.120249211692842e-05,
+      "loss": 0.8343,
+      "step": 247300
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 3.120173046750042e-05,
+      "loss": 0.8001,
+      "step": 247310
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 3.120096881807242e-05,
+      "loss": 0.8814,
+      "step": 247320
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 3.120020716864442e-05,
+      "loss": 0.8686,
+      "step": 247330
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 3.119944551921642e-05,
+      "loss": 0.8171,
+      "step": 247340
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 3.1198683869788416e-05,
+      "loss": 0.8206,
+      "step": 247350
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 3.1197922220360414e-05,
+      "loss": 0.7475,
+      "step": 247360
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 3.119716057093241e-05,
+      "loss": 1.0184,
+      "step": 247370
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 3.119639892150441e-05,
+      "loss": 0.8445,
+      "step": 247380
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 3.119563727207641e-05,
+      "loss": 0.8879,
+      "step": 247390
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 3.1194875622648404e-05,
+      "loss": 0.8234,
+      "step": 247400
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 3.119411397322041e-05,
+      "loss": 0.8269,
+      "step": 247410
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 3.1193352323792405e-05,
+      "loss": 0.8437,
+      "step": 247420
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 3.11925906743644e-05,
+      "loss": 0.8558,
+      "step": 247430
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 3.11918290249364e-05,
+      "loss": 0.8327,
+      "step": 247440
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 3.1191067375508405e-05,
+      "loss": 0.8831,
+      "step": 247450
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 3.11903057260804e-05,
+      "loss": 0.8439,
+      "step": 247460
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 3.11895440766524e-05,
+      "loss": 0.8757,
+      "step": 247470
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 3.11887824272244e-05,
+      "loss": 0.8036,
+      "step": 247480
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 3.1188020777796394e-05,
+      "loss": 0.8746,
+      "step": 247490
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 3.11872591283684e-05,
+      "loss": 0.7969,
+      "step": 247500
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 3.1186497478940396e-05,
+      "loss": 0.7941,
+      "step": 247510
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 3.1185735829512394e-05,
+      "loss": 0.8893,
+      "step": 247520
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 3.118497418008439e-05,
+      "loss": 0.7973,
+      "step": 247530
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 3.1184212530656395e-05,
+      "loss": 0.8358,
+      "step": 247540
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 3.1183450881228386e-05,
+      "loss": 0.9419,
+      "step": 247550
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 3.118268923180039e-05,
+      "loss": 0.8919,
+      "step": 247560
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 3.118192758237239e-05,
+      "loss": 0.7761,
+      "step": 247570
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 3.1181165932944385e-05,
+      "loss": 0.8488,
+      "step": 247580
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 3.118040428351638e-05,
+      "loss": 0.7847,
+      "step": 247590
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 3.117964263408839e-05,
+      "loss": 0.8455,
+      "step": 247600
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 3.117888098466038e-05,
+      "loss": 0.866,
+      "step": 247610
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 3.117811933523238e-05,
+      "loss": 0.8509,
+      "step": 247620
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 3.117735768580438e-05,
+      "loss": 0.8632,
+      "step": 247630
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 3.117659603637638e-05,
+      "loss": 0.873,
+      "step": 247640
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 3.1175834386948374e-05,
+      "loss": 0.8895,
+      "step": 247650
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 3.117507273752038e-05,
+      "loss": 0.8268,
+      "step": 247660
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 3.1174311088092376e-05,
+      "loss": 0.8821,
+      "step": 247670
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 3.1173549438664373e-05,
+      "loss": 0.8204,
+      "step": 247680
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 3.117278778923637e-05,
+      "loss": 0.886,
+      "step": 247690
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 3.117202613980837e-05,
+      "loss": 0.8494,
+      "step": 247700
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 3.117126449038037e-05,
+      "loss": 0.7654,
+      "step": 247710
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 3.117050284095237e-05,
+      "loss": 0.874,
+      "step": 247720
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 3.116974119152437e-05,
+      "loss": 0.7696,
+      "step": 247730
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 3.1168979542096365e-05,
+      "loss": 0.7786,
+      "step": 247740
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 3.116821789266837e-05,
+      "loss": 0.7866,
+      "step": 247750
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 3.116745624324036e-05,
+      "loss": 0.8162,
+      "step": 247760
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 3.1166694593812364e-05,
+      "loss": 0.8819,
+      "step": 247770
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 3.116593294438436e-05,
+      "loss": 0.9013,
+      "step": 247780
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 3.116517129495636e-05,
+      "loss": 0.8136,
+      "step": 247790
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 3.1164409645528357e-05,
+      "loss": 0.7495,
+      "step": 247800
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 3.116364799610036e-05,
+      "loss": 0.7916,
+      "step": 247810
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 3.116288634667235e-05,
+      "loss": 0.7493,
+      "step": 247820
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 3.1162124697244356e-05,
+      "loss": 0.9047,
+      "step": 247830
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 3.116136304781635e-05,
+      "loss": 0.9067,
+      "step": 247840
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 3.116060139838835e-05,
+      "loss": 0.9568,
+      "step": 247850
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 3.115983974896035e-05,
+      "loss": 0.7777,
+      "step": 247860
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 3.1159078099532346e-05,
+      "loss": 0.7903,
+      "step": 247870
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 3.115831645010435e-05,
+      "loss": 0.8296,
+      "step": 247880
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 3.115755480067635e-05,
+      "loss": 0.8614,
+      "step": 247890
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 3.1156793151248345e-05,
+      "loss": 0.9586,
+      "step": 247900
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 3.115603150182034e-05,
+      "loss": 0.7897,
+      "step": 247910
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 3.1155269852392347e-05,
+      "loss": 0.848,
+      "step": 247920
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 3.115450820296434e-05,
+      "loss": 0.7796,
+      "step": 247930
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 3.115374655353634e-05,
+      "loss": 0.8485,
+      "step": 247940
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 3.115298490410834e-05,
+      "loss": 0.7393,
+      "step": 247950
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 3.1152223254680336e-05,
+      "loss": 0.7736,
+      "step": 247960
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 3.1151461605252334e-05,
+      "loss": 0.8443,
+      "step": 247970
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 3.115069995582434e-05,
+      "loss": 0.8171,
+      "step": 247980
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 3.114993830639633e-05,
+      "loss": 0.8154,
+      "step": 247990
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 3.114917665696833e-05,
+      "loss": 0.889,
+      "step": 248000
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 3.114841500754033e-05,
+      "loss": 0.9352,
+      "step": 248010
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 3.114765335811233e-05,
+      "loss": 0.8049,
+      "step": 248020
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 3.1146891708684325e-05,
+      "loss": 0.8148,
+      "step": 248030
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 3.114613005925633e-05,
+      "loss": 0.8226,
+      "step": 248040
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 3.114536840982833e-05,
+      "loss": 0.8383,
+      "step": 248050
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 3.1144606760400325e-05,
+      "loss": 0.802,
+      "step": 248060
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 3.114384511097232e-05,
+      "loss": 0.859,
+      "step": 248070
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 3.114308346154432e-05,
+      "loss": 0.7737,
+      "step": 248080
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 3.1142321812116324e-05,
+      "loss": 0.9091,
+      "step": 248090
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 3.114156016268832e-05,
+      "loss": 0.7465,
+      "step": 248100
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 3.114079851326032e-05,
+      "loss": 0.8294,
+      "step": 248110
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 3.1140036863832316e-05,
+      "loss": 0.8341,
+      "step": 248120
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 3.113927521440432e-05,
+      "loss": 0.9702,
+      "step": 248130
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 3.113851356497631e-05,
+      "loss": 0.8996,
+      "step": 248140
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 3.1137751915548315e-05,
+      "loss": 0.8633,
+      "step": 248150
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 3.113699026612031e-05,
+      "loss": 0.8316,
+      "step": 248160
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 3.113622861669231e-05,
+      "loss": 0.8388,
+      "step": 248170
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 3.113546696726431e-05,
+      "loss": 0.8331,
+      "step": 248180
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 3.113470531783631e-05,
+      "loss": 0.7476,
+      "step": 248190
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 3.11339436684083e-05,
+      "loss": 0.8803,
+      "step": 248200
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 3.113318201898031e-05,
+      "loss": 0.9058,
+      "step": 248210
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 3.1132420369552304e-05,
+      "loss": 0.8773,
+      "step": 248220
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 3.11316587201243e-05,
+      "loss": 0.8521,
+      "step": 248230
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 3.11308970706963e-05,
+      "loss": 0.8075,
+      "step": 248240
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 3.1130135421268304e-05,
+      "loss": 0.7717,
+      "step": 248250
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 3.11293737718403e-05,
+      "loss": 0.8368,
+      "step": 248260
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 3.11286121224123e-05,
+      "loss": 0.8316,
+      "step": 248270
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 3.1127850472984296e-05,
+      "loss": 0.9232,
+      "step": 248280
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 3.1127088823556293e-05,
+      "loss": 0.7394,
+      "step": 248290
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 3.11263271741283e-05,
+      "loss": 0.907,
+      "step": 248300
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 3.112556552470029e-05,
+      "loss": 0.8937,
+      "step": 248310
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 3.112480387527229e-05,
+      "loss": 0.8344,
+      "step": 248320
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 3.112404222584429e-05,
+      "loss": 0.7385,
+      "step": 248330
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 3.112328057641629e-05,
+      "loss": 0.881,
+      "step": 248340
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 3.1122518926988285e-05,
+      "loss": 0.8102,
+      "step": 248350
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 3.112175727756029e-05,
+      "loss": 0.8109,
+      "step": 248360
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 3.112099562813228e-05,
+      "loss": 0.8556,
+      "step": 248370
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 3.1120233978704284e-05,
+      "loss": 0.729,
+      "step": 248380
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 3.111947232927628e-05,
+      "loss": 0.7811,
+      "step": 248390
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 3.111871067984828e-05,
+      "loss": 0.8938,
+      "step": 248400
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 3.1117949030420277e-05,
+      "loss": 0.8171,
+      "step": 248410
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 3.111718738099228e-05,
+      "loss": 0.8651,
+      "step": 248420
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 3.111642573156427e-05,
+      "loss": 0.8875,
+      "step": 248430
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 3.1115664082136276e-05,
+      "loss": 0.8144,
+      "step": 248440
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 3.111490243270827e-05,
+      "loss": 0.8036,
+      "step": 248450
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 3.111414078328027e-05,
+      "loss": 0.811,
+      "step": 248460
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 3.1113379133852275e-05,
+      "loss": 0.8279,
+      "step": 248470
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 3.111261748442427e-05,
+      "loss": 0.8459,
+      "step": 248480
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 3.111185583499627e-05,
+      "loss": 0.9328,
+      "step": 248490
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 3.111109418556827e-05,
+      "loss": 0.8294,
+      "step": 248500
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 3.111033253614027e-05,
+      "loss": 0.7836,
+      "step": 248510
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 3.110957088671226e-05,
+      "loss": 0.7941,
+      "step": 248520
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 3.1108809237284267e-05,
+      "loss": 0.8756,
+      "step": 248530
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 3.1108047587856264e-05,
+      "loss": 0.8484,
+      "step": 248540
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 3.110728593842826e-05,
+      "loss": 0.7825,
+      "step": 248550
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 3.110652428900026e-05,
+      "loss": 0.8591,
+      "step": 248560
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 3.110576263957226e-05,
+      "loss": 0.8356,
+      "step": 248570
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 3.1105000990144254e-05,
+      "loss": 0.7993,
+      "step": 248580
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 3.110423934071626e-05,
+      "loss": 0.7936,
+      "step": 248590
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 3.1103477691288256e-05,
+      "loss": 0.7855,
+      "step": 248600
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 3.110271604186025e-05,
+      "loss": 0.8633,
+      "step": 248610
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 3.110195439243225e-05,
+      "loss": 0.8161,
+      "step": 248620
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 3.1101192743004255e-05,
+      "loss": 0.8406,
+      "step": 248630
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 3.110043109357625e-05,
+      "loss": 0.6882,
+      "step": 248640
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 3.109966944414825e-05,
+      "loss": 0.8135,
+      "step": 248650
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 3.109890779472025e-05,
+      "loss": 0.9174,
+      "step": 248660
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 3.1098146145292245e-05,
+      "loss": 0.7913,
+      "step": 248670
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 3.109738449586425e-05,
+      "loss": 0.8293,
+      "step": 248680
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 3.1096622846436246e-05,
+      "loss": 0.7926,
+      "step": 248690
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 3.1095861197008244e-05,
+      "loss": 0.836,
+      "step": 248700
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 3.109509954758024e-05,
+      "loss": 0.841,
+      "step": 248710
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 3.1094337898152245e-05,
+      "loss": 0.8167,
+      "step": 248720
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 3.1093576248724236e-05,
+      "loss": 0.8149,
+      "step": 248730
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 3.109281459929624e-05,
+      "loss": 0.8106,
+      "step": 248740
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 3.109205294986824e-05,
+      "loss": 1.0094,
+      "step": 248750
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 3.1091291300440235e-05,
+      "loss": 0.8339,
+      "step": 248760
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 3.109052965101223e-05,
+      "loss": 0.8202,
+      "step": 248770
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 3.108976800158423e-05,
+      "loss": 0.7929,
+      "step": 248780
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 3.108900635215623e-05,
+      "loss": 0.7478,
+      "step": 248790
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 3.108824470272823e-05,
+      "loss": 0.866,
+      "step": 248800
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 3.108748305330022e-05,
+      "loss": 0.779,
+      "step": 248810
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 3.108672140387223e-05,
+      "loss": 0.8079,
+      "step": 248820
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 3.1085959754444224e-05,
+      "loss": 0.8168,
+      "step": 248830
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 3.108519810501622e-05,
+      "loss": 0.8945,
+      "step": 248840
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 3.1084436455588226e-05,
+      "loss": 0.9021,
+      "step": 248850
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 3.1083674806160224e-05,
+      "loss": 0.8273,
+      "step": 248860
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 3.108291315673222e-05,
+      "loss": 0.7281,
+      "step": 248870
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 3.108215150730422e-05,
+      "loss": 0.7999,
+      "step": 248880
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 3.108138985787622e-05,
+      "loss": 0.8308,
+      "step": 248890
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 3.1080628208448213e-05,
+      "loss": 0.9545,
+      "step": 248900
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 3.107986655902022e-05,
+      "loss": 0.7528,
+      "step": 248910
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 3.1079104909592215e-05,
+      "loss": 0.763,
+      "step": 248920
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 3.107834326016421e-05,
+      "loss": 0.7773,
+      "step": 248930
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 3.107758161073621e-05,
+      "loss": 0.7649,
+      "step": 248940
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 3.1076819961308214e-05,
+      "loss": 0.8081,
+      "step": 248950
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 3.1076058311880205e-05,
+      "loss": 0.8791,
+      "step": 248960
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 3.107529666245221e-05,
+      "loss": 0.911,
+      "step": 248970
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 3.107453501302421e-05,
+      "loss": 0.801,
+      "step": 248980
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 3.1073773363596204e-05,
+      "loss": 0.804,
+      "step": 248990
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 3.10730117141682e-05,
+      "loss": 0.8954,
+      "step": 249000
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 3.1072250064740206e-05,
+      "loss": 0.7646,
+      "step": 249010
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 3.1071488415312197e-05,
+      "loss": 0.8433,
+      "step": 249020
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 3.10707267658842e-05,
+      "loss": 0.769,
+      "step": 249030
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 3.10699651164562e-05,
+      "loss": 0.8107,
+      "step": 249040
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 3.1069203467028196e-05,
+      "loss": 0.8335,
+      "step": 249050
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 3.10684418176002e-05,
+      "loss": 0.8974,
+      "step": 249060
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 3.10676801681722e-05,
+      "loss": 0.8095,
+      "step": 249070
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 3.1066918518744195e-05,
+      "loss": 0.8213,
+      "step": 249080
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 3.106615686931619e-05,
+      "loss": 0.8325,
+      "step": 249090
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 3.10653952198882e-05,
+      "loss": 0.7738,
+      "step": 249100
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 3.106463357046019e-05,
+      "loss": 0.7747,
+      "step": 249110
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 3.106387192103219e-05,
+      "loss": 0.7944,
+      "step": 249120
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 3.106311027160419e-05,
+      "loss": 0.7981,
+      "step": 249130
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 3.1062348622176187e-05,
+      "loss": 0.7437,
+      "step": 249140
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 3.1061586972748184e-05,
+      "loss": 0.8357,
+      "step": 249150
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 3.106082532332019e-05,
+      "loss": 0.9221,
+      "step": 249160
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 3.106006367389218e-05,
+      "loss": 0.7518,
+      "step": 249170
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 3.105930202446418e-05,
+      "loss": 0.7922,
+      "step": 249180
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 3.105854037503618e-05,
+      "loss": 0.8246,
+      "step": 249190
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 3.105777872560818e-05,
+      "loss": 0.8052,
+      "step": 249200
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 3.1057017076180176e-05,
+      "loss": 0.9042,
+      "step": 249210
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 3.105625542675217e-05,
+      "loss": 0.8325,
+      "step": 249220
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 3.105549377732417e-05,
+      "loss": 0.8255,
+      "step": 249230
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 3.1054732127896175e-05,
+      "loss": 0.7573,
+      "step": 249240
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 3.105397047846817e-05,
+      "loss": 0.902,
+      "step": 249250
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 3.105320882904017e-05,
+      "loss": 0.9181,
+      "step": 249260
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 3.1052447179612174e-05,
+      "loss": 0.8101,
+      "step": 249270
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 3.1051685530184165e-05,
+      "loss": 0.7729,
+      "step": 249280
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 3.105092388075617e-05,
+      "loss": 0.9269,
+      "step": 249290
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 3.1050162231328166e-05,
+      "loss": 0.7455,
+      "step": 249300
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 3.1049400581900164e-05,
+      "loss": 0.7845,
+      "step": 249310
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 3.104863893247216e-05,
+      "loss": 0.8499,
+      "step": 249320
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 3.1047877283044166e-05,
+      "loss": 0.8621,
+      "step": 249330
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 3.1047115633616156e-05,
+      "loss": 0.7643,
+      "step": 249340
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 3.104635398418816e-05,
+      "loss": 0.7854,
+      "step": 249350
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 3.104559233476016e-05,
+      "loss": 0.7438,
+      "step": 249360
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 3.1044830685332155e-05,
+      "loss": 0.7624,
+      "step": 249370
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 3.104406903590415e-05,
+      "loss": 0.8874,
+      "step": 249380
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 3.104330738647616e-05,
+      "loss": 0.8085,
+      "step": 249390
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 3.104254573704815e-05,
+      "loss": 0.7376,
+      "step": 249400
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 3.104178408762015e-05,
+      "loss": 0.8307,
+      "step": 249410
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 3.104102243819215e-05,
+      "loss": 0.8445,
+      "step": 249420
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 3.104026078876415e-05,
+      "loss": 0.8088,
+      "step": 249430
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 3.103949913933615e-05,
+      "loss": 0.7711,
+      "step": 249440
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 3.103873748990815e-05,
+      "loss": 0.8468,
+      "step": 249450
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 3.1037975840480146e-05,
+      "loss": 0.9492,
+      "step": 249460
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 3.1037214191052144e-05,
+      "loss": 0.8521,
+      "step": 249470
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 3.103645254162415e-05,
+      "loss": 0.7879,
+      "step": 249480
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 3.103569089219614e-05,
+      "loss": 0.8682,
+      "step": 249490
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 3.103492924276814e-05,
+      "loss": 0.8194,
+      "step": 249500
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 3.103416759334014e-05,
+      "loss": 0.8215,
+      "step": 249510
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 3.103340594391214e-05,
+      "loss": 0.8091,
+      "step": 249520
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 3.1032644294484135e-05,
+      "loss": 0.8551,
+      "step": 249530
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 3.103188264505614e-05,
+      "loss": 0.8176,
+      "step": 249540
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 3.103112099562813e-05,
+      "loss": 0.8542,
+      "step": 249550
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 3.1030359346200134e-05,
+      "loss": 0.7554,
+      "step": 249560
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 3.102959769677213e-05,
+      "loss": 0.9451,
+      "step": 249570
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 3.102883604734413e-05,
+      "loss": 0.9354,
+      "step": 249580
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 3.102807439791613e-05,
+      "loss": 0.8156,
+      "step": 249590
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 3.102731274848813e-05,
+      "loss": 0.8737,
+      "step": 249600
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 3.102655109906012e-05,
+      "loss": 0.7929,
+      "step": 249610
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 3.1025789449632126e-05,
+      "loss": 0.8778,
+      "step": 249620
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 3.1025027800204123e-05,
+      "loss": 0.8382,
+      "step": 249630
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 3.102426615077612e-05,
+      "loss": 0.8335,
+      "step": 249640
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 3.1023504501348125e-05,
+      "loss": 0.7865,
+      "step": 249650
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 3.102274285192012e-05,
+      "loss": 0.8834,
+      "step": 249660
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 3.102198120249212e-05,
+      "loss": 0.782,
+      "step": 249670
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 3.102121955306412e-05,
+      "loss": 0.8087,
+      "step": 249680
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 3.1020457903636115e-05,
+      "loss": 0.8864,
+      "step": 249690
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 3.101969625420811e-05,
+      "loss": 0.8707,
+      "step": 249700
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 3.101893460478012e-05,
+      "loss": 0.768,
+      "step": 249710
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 3.101817295535211e-05,
+      "loss": 0.7702,
+      "step": 249720
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 3.101741130592411e-05,
+      "loss": 0.8542,
+      "step": 249730
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 3.101664965649611e-05,
+      "loss": 0.8285,
+      "step": 249740
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 3.1015888007068107e-05,
+      "loss": 0.9425,
+      "step": 249750
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 3.1015126357640104e-05,
+      "loss": 0.8696,
+      "step": 249760
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 3.101436470821211e-05,
+      "loss": 0.756,
+      "step": 249770
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 3.10136030587841e-05,
+      "loss": 0.8847,
+      "step": 249780
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 3.10128414093561e-05,
+      "loss": 0.8358,
+      "step": 249790
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 3.10120797599281e-05,
+      "loss": 0.7298,
+      "step": 249800
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 3.10113181105001e-05,
+      "loss": 0.8669,
+      "step": 249810
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 3.1010556461072096e-05,
+      "loss": 0.839,
+      "step": 249820
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 3.10097948116441e-05,
+      "loss": 0.7889,
+      "step": 249830
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 3.10090331622161e-05,
+      "loss": 0.8613,
+      "step": 249840
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 3.1008271512788095e-05,
+      "loss": 0.7484,
+      "step": 249850
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 3.10075098633601e-05,
+      "loss": 0.8956,
+      "step": 249860
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 3.100674821393209e-05,
+      "loss": 0.8175,
+      "step": 249870
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 3.1005986564504094e-05,
+      "loss": 0.8964,
+      "step": 249880
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 3.100522491507609e-05,
+      "loss": 0.9278,
+      "step": 249890
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 3.100446326564809e-05,
+      "loss": 0.8602,
+      "step": 249900
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 3.1003701616220086e-05,
+      "loss": 0.7223,
+      "step": 249910
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 3.100293996679209e-05,
+      "loss": 0.8173,
+      "step": 249920
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 3.100217831736408e-05,
+      "loss": 0.8073,
+      "step": 249930
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 3.1001416667936086e-05,
+      "loss": 0.8605,
+      "step": 249940
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 3.100065501850808e-05,
+      "loss": 0.7985,
+      "step": 249950
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 3.099989336908008e-05,
+      "loss": 0.8862,
+      "step": 249960
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 3.099913171965208e-05,
+      "loss": 0.8301,
+      "step": 249970
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 3.099837007022408e-05,
+      "loss": 0.7656,
+      "step": 249980
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 3.099760842079607e-05,
+      "loss": 0.8332,
+      "step": 249990
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 3.099684677136808e-05,
+      "loss": 0.7704,
+      "step": 250000
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 3.0996085121940075e-05,
+      "loss": 0.8211,
+      "step": 250010
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 3.099532347251207e-05,
+      "loss": 0.8837,
+      "step": 250020
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 3.099456182308407e-05,
+      "loss": 0.7732,
+      "step": 250030
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 3.0993800173656074e-05,
+      "loss": 0.786,
+      "step": 250040
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 3.099303852422807e-05,
+      "loss": 0.8596,
+      "step": 250050
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 3.099227687480007e-05,
+      "loss": 0.8474,
+      "step": 250060
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 3.099151522537207e-05,
+      "loss": 0.7958,
+      "step": 250070
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 3.0990753575944064e-05,
+      "loss": 0.7968,
+      "step": 250080
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 3.098999192651607e-05,
+      "loss": 0.7879,
+      "step": 250090
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 3.0989230277088065e-05,
+      "loss": 0.7775,
+      "step": 250100
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 3.098846862766006e-05,
+      "loss": 0.8358,
+      "step": 250110
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 3.098770697823206e-05,
+      "loss": 0.872,
+      "step": 250120
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 3.0986945328804064e-05,
+      "loss": 0.8224,
+      "step": 250130
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 3.0986183679376055e-05,
+      "loss": 0.9168,
+      "step": 250140
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 3.098542202994806e-05,
+      "loss": 0.818,
+      "step": 250150
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 3.098466038052005e-05,
+      "loss": 0.8242,
+      "step": 250160
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 3.0983898731092054e-05,
+      "loss": 0.8143,
+      "step": 250170
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 3.098313708166405e-05,
+      "loss": 0.8658,
+      "step": 250180
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 3.098237543223605e-05,
+      "loss": 0.7404,
+      "step": 250190
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 3.098161378280805e-05,
+      "loss": 0.8642,
+      "step": 250200
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 3.098085213338005e-05,
+      "loss": 0.8769,
+      "step": 250210
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 3.098009048395205e-05,
+      "loss": 0.8617,
+      "step": 250220
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 3.0979328834524046e-05,
+      "loss": 0.8097,
+      "step": 250230
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 3.097856718509605e-05,
+      "loss": 0.9109,
+      "step": 250240
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 3.097780553566804e-05,
+      "loss": 0.8247,
+      "step": 250250
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 3.0977043886240045e-05,
+      "loss": 0.8863,
+      "step": 250260
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 3.097628223681204e-05,
+      "loss": 0.9038,
+      "step": 250270
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 3.097552058738404e-05,
+      "loss": 0.8203,
+      "step": 250280
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 3.097475893795604e-05,
+      "loss": 0.8188,
+      "step": 250290
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 3.097399728852804e-05,
+      "loss": 0.8151,
+      "step": 250300
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 3.097323563910003e-05,
+      "loss": 0.7722,
+      "step": 250310
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 3.097247398967204e-05,
+      "loss": 0.8718,
+      "step": 250320
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 3.0971712340244034e-05,
+      "loss": 0.8068,
+      "step": 250330
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 3.097095069081603e-05,
+      "loss": 0.8899,
+      "step": 250340
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 3.097018904138803e-05,
+      "loss": 0.8708,
+      "step": 250350
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 3.096942739196003e-05,
+      "loss": 0.7817,
+      "step": 250360
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 3.0968665742532024e-05,
+      "loss": 0.8718,
+      "step": 250370
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 3.096790409310403e-05,
+      "loss": 0.8249,
+      "step": 250380
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 3.0967142443676026e-05,
+      "loss": 0.8866,
+      "step": 250390
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 3.096638079424802e-05,
+      "loss": 0.7963,
+      "step": 250400
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 3.096561914482002e-05,
+      "loss": 0.8179,
+      "step": 250410
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 3.0964857495392025e-05,
+      "loss": 0.7935,
+      "step": 250420
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 3.096409584596402e-05,
+      "loss": 0.7805,
+      "step": 250430
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 3.096333419653602e-05,
+      "loss": 0.8191,
+      "step": 250440
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 3.0962572547108024e-05,
+      "loss": 0.8334,
+      "step": 250450
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 3.0961810897680015e-05,
+      "loss": 0.8717,
+      "step": 250460
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 3.096104924825202e-05,
+      "loss": 0.7573,
+      "step": 250470
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 3.0960287598824016e-05,
+      "loss": 0.8975,
+      "step": 250480
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 3.0959525949396014e-05,
+      "loss": 0.8258,
+      "step": 250490
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 3.095876429996801e-05,
+      "loss": 0.7975,
+      "step": 250500
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 3.0958002650540016e-05,
+      "loss": 0.8185,
+      "step": 250510
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 3.0957241001112006e-05,
+      "loss": 0.7298,
+      "step": 250520
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 3.095647935168401e-05,
+      "loss": 0.8168,
+      "step": 250530
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 3.095571770225601e-05,
+      "loss": 0.9057,
+      "step": 250540
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 3.0954956052828006e-05,
+      "loss": 0.8288,
+      "step": 250550
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 3.09541944034e-05,
+      "loss": 0.8028,
+      "step": 250560
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 3.095343275397201e-05,
+      "loss": 0.8768,
+      "step": 250570
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 3.0952671104544e-05,
+      "loss": 0.8383,
+      "step": 250580
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 3.0951909455116e-05,
+      "loss": 0.7793,
+      "step": 250590
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 3.0951147805688e-05,
+      "loss": 0.8305,
+      "step": 250600
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 3.095038615626e-05,
+      "loss": 0.7729,
+      "step": 250610
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 3.0949624506831995e-05,
+      "loss": 0.8419,
+      "step": 250620
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 3.094886285740399e-05,
+      "loss": 0.8866,
+      "step": 250630
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 3.0948101207975996e-05,
+      "loss": 0.8183,
+      "step": 250640
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 3.0947339558547994e-05,
+      "loss": 0.8225,
+      "step": 250650
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 3.094657790911999e-05,
+      "loss": 0.8418,
+      "step": 250660
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 3.094581625969199e-05,
+      "loss": 0.846,
+      "step": 250670
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 3.094505461026399e-05,
+      "loss": 0.7471,
+      "step": 250680
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 3.0944292960835984e-05,
+      "loss": 0.8338,
+      "step": 250690
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 3.094353131140799e-05,
+      "loss": 0.8005,
+      "step": 250700
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 3.0942769661979985e-05,
+      "loss": 0.8594,
+      "step": 250710
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 3.094200801255198e-05,
+      "loss": 0.7892,
+      "step": 250720
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 3.094124636312398e-05,
+      "loss": 0.808,
+      "step": 250730
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 3.0940484713695985e-05,
+      "loss": 0.8558,
+      "step": 250740
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 3.0939723064267975e-05,
+      "loss": 0.7958,
+      "step": 250750
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 3.093896141483998e-05,
+      "loss": 0.8233,
+      "step": 250760
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 3.093819976541198e-05,
+      "loss": 0.9167,
+      "step": 250770
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 3.0937438115983974e-05,
+      "loss": 0.7808,
+      "step": 250780
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 3.093667646655597e-05,
+      "loss": 0.8696,
+      "step": 250790
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 3.0935914817127976e-05,
+      "loss": 0.8291,
+      "step": 250800
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 3.0935153167699974e-05,
+      "loss": 0.8013,
+      "step": 250810
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 3.093439151827197e-05,
+      "loss": 0.776,
+      "step": 250820
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 3.093362986884397e-05,
+      "loss": 0.9013,
+      "step": 250830
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 3.0932868219415966e-05,
+      "loss": 0.8565,
+      "step": 250840
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 3.093210656998797e-05,
+      "loss": 0.7815,
+      "step": 250850
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 3.093134492055997e-05,
+      "loss": 0.8784,
+      "step": 250860
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 3.0930583271131965e-05,
+      "loss": 0.8238,
+      "step": 250870
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 3.092982162170396e-05,
+      "loss": 0.869,
+      "step": 250880
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 3.092905997227597e-05,
+      "loss": 0.8498,
+      "step": 250890
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 3.092829832284796e-05,
+      "loss": 0.8231,
+      "step": 250900
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 3.092753667341996e-05,
+      "loss": 0.7008,
+      "step": 250910
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 3.092677502399196e-05,
+      "loss": 0.7903,
+      "step": 250920
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 3.092601337456396e-05,
+      "loss": 0.7917,
+      "step": 250930
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 3.0925251725135954e-05,
+      "loss": 0.844,
+      "step": 250940
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 3.092449007570796e-05,
+      "loss": 0.8798,
+      "step": 250950
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 3.092372842627995e-05,
+      "loss": 0.7995,
+      "step": 250960
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 3.092296677685195e-05,
+      "loss": 0.8436,
+      "step": 250970
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 3.092220512742395e-05,
+      "loss": 0.8407,
+      "step": 250980
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 3.092144347799595e-05,
+      "loss": 0.8133,
+      "step": 250990
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 3.0920681828567946e-05,
+      "loss": 0.7067,
+      "step": 251000
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 3.091992017913995e-05,
+      "loss": 0.9018,
+      "step": 251010
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 3.091915852971195e-05,
+      "loss": 0.7808,
+      "step": 251020
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 3.0918396880283945e-05,
+      "loss": 0.8978,
+      "step": 251030
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 3.091763523085595e-05,
+      "loss": 0.85,
+      "step": 251040
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 3.091687358142794e-05,
+      "loss": 0.8951,
+      "step": 251050
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 3.0916111931999944e-05,
+      "loss": 0.8705,
+      "step": 251060
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 3.0915350282571935e-05,
+      "loss": 0.8536,
+      "step": 251070
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 3.091458863314394e-05,
+      "loss": 0.8901,
+      "step": 251080
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 3.0913826983715936e-05,
+      "loss": 0.8093,
+      "step": 251090
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 3.0913065334287934e-05,
+      "loss": 0.8349,
+      "step": 251100
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 3.091230368485993e-05,
+      "loss": 0.9096,
+      "step": 251110
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 3.0911542035431936e-05,
+      "loss": 0.8588,
+      "step": 251120
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 3.0910780386003926e-05,
+      "loss": 0.8291,
+      "step": 251130
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 3.091001873657593e-05,
+      "loss": 0.868,
+      "step": 251140
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 3.090925708714793e-05,
+      "loss": 0.7954,
+      "step": 251150
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 3.0908495437719926e-05,
+      "loss": 0.801,
+      "step": 251160
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 3.090773378829192e-05,
+      "loss": 0.9077,
+      "step": 251170
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 3.090697213886393e-05,
+      "loss": 0.9066,
+      "step": 251180
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 3.0906210489435925e-05,
+      "loss": 0.7683,
+      "step": 251190
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 3.090544884000792e-05,
+      "loss": 0.7793,
+      "step": 251200
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 3.090468719057992e-05,
+      "loss": 0.808,
+      "step": 251210
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 3.090392554115192e-05,
+      "loss": 0.8388,
+      "step": 251220
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 3.090316389172392e-05,
+      "loss": 0.7293,
+      "step": 251230
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 3.090240224229592e-05,
+      "loss": 0.8436,
+      "step": 251240
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 3.0901640592867916e-05,
+      "loss": 0.8651,
+      "step": 251250
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 3.0900878943439914e-05,
+      "loss": 0.8524,
+      "step": 251260
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 3.090011729401192e-05,
+      "loss": 0.869,
+      "step": 251270
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 3.089935564458391e-05,
+      "loss": 0.8214,
+      "step": 251280
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 3.089859399515591e-05,
+      "loss": 0.8909,
+      "step": 251290
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 3.089783234572791e-05,
+      "loss": 0.9659,
+      "step": 251300
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 3.089707069629991e-05,
+      "loss": 0.8341,
+      "step": 251310
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 3.0896309046871905e-05,
+      "loss": 0.7996,
+      "step": 251320
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 3.089554739744391e-05,
+      "loss": 0.7577,
+      "step": 251330
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 3.08947857480159e-05,
+      "loss": 0.8949,
+      "step": 251340
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 3.0894024098587905e-05,
+      "loss": 0.8523,
+      "step": 251350
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 3.08932624491599e-05,
+      "loss": 0.7908,
+      "step": 251360
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 3.08925007997319e-05,
+      "loss": 0.7742,
+      "step": 251370
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 3.08917391503039e-05,
+      "loss": 0.8563,
+      "step": 251380
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 3.08909775008759e-05,
+      "loss": 0.8243,
+      "step": 251390
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 3.08902158514479e-05,
+      "loss": 0.848,
+      "step": 251400
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 3.0889454202019896e-05,
+      "loss": 0.8077,
+      "step": 251410
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 3.0888692552591894e-05,
+      "loss": 0.8373,
+      "step": 251420
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 3.088793090316389e-05,
+      "loss": 0.7161,
+      "step": 251430
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 3.0887169253735895e-05,
+      "loss": 0.8477,
+      "step": 251440
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 3.088640760430789e-05,
+      "loss": 0.7334,
+      "step": 251450
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 3.088564595487989e-05,
+      "loss": 0.7835,
+      "step": 251460
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 3.088488430545189e-05,
+      "loss": 0.846,
+      "step": 251470
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 3.088412265602389e-05,
+      "loss": 0.8797,
+      "step": 251480
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 3.088336100659588e-05,
+      "loss": 0.8553,
+      "step": 251490
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 3.088259935716789e-05,
+      "loss": 0.825,
+      "step": 251500
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 3.0881837707739884e-05,
+      "loss": 0.8675,
+      "step": 251510
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 3.088107605831188e-05,
+      "loss": 0.7915,
+      "step": 251520
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 3.088031440888388e-05,
+      "loss": 0.846,
+      "step": 251530
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 3.087955275945588e-05,
+      "loss": 0.7808,
+      "step": 251540
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 3.0878791110027874e-05,
+      "loss": 0.8881,
+      "step": 251550
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 3.087802946059988e-05,
+      "loss": 0.84,
+      "step": 251560
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 3.0877267811171876e-05,
+      "loss": 0.9294,
+      "step": 251570
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 3.087650616174387e-05,
+      "loss": 0.8064,
+      "step": 251580
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 3.087574451231587e-05,
+      "loss": 0.7632,
+      "step": 251590
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 3.087498286288787e-05,
+      "loss": 0.872,
+      "step": 251600
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 3.087422121345987e-05,
+      "loss": 0.8121,
+      "step": 251610
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 3.087345956403187e-05,
+      "loss": 0.8313,
+      "step": 251620
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 3.087269791460387e-05,
+      "loss": 0.7872,
+      "step": 251630
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 3.0871936265175865e-05,
+      "loss": 0.836,
+      "step": 251640
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 3.087117461574787e-05,
+      "loss": 0.8225,
+      "step": 251650
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 3.087041296631986e-05,
+      "loss": 0.9251,
+      "step": 251660
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 3.0869651316891864e-05,
+      "loss": 0.8246,
+      "step": 251670
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 3.086888966746386e-05,
+      "loss": 0.9939,
+      "step": 251680
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 3.086812801803586e-05,
+      "loss": 0.7902,
+      "step": 251690
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 3.0867366368607857e-05,
+      "loss": 0.7847,
+      "step": 251700
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 3.086660471917986e-05,
+      "loss": 0.8485,
+      "step": 251710
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 3.086584306975185e-05,
+      "loss": 0.7925,
+      "step": 251720
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 3.0865081420323856e-05,
+      "loss": 0.7607,
+      "step": 251730
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 3.086431977089585e-05,
+      "loss": 0.8481,
+      "step": 251740
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 3.086355812146785e-05,
+      "loss": 0.8412,
+      "step": 251750
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 3.086279647203985e-05,
+      "loss": 0.7838,
+      "step": 251760
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 3.086203482261185e-05,
+      "loss": 0.8407,
+      "step": 251770
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 3.086127317318385e-05,
+      "loss": 0.8608,
+      "step": 251780
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 3.086051152375585e-05,
+      "loss": 0.8306,
+      "step": 251790
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 3.0859749874327845e-05,
+      "loss": 0.9333,
+      "step": 251800
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 3.085898822489984e-05,
+      "loss": 0.755,
+      "step": 251810
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 3.0858226575471846e-05,
+      "loss": 0.9299,
+      "step": 251820
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 3.0857464926043844e-05,
+      "loss": 0.8099,
+      "step": 251830
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 3.085670327661584e-05,
+      "loss": 0.8222,
+      "step": 251840
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 3.085594162718784e-05,
+      "loss": 0.8401,
+      "step": 251850
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 3.085517997775984e-05,
+      "loss": 0.7317,
+      "step": 251860
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 3.0854418328331834e-05,
+      "loss": 0.7767,
+      "step": 251870
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 3.085365667890384e-05,
+      "loss": 0.8204,
+      "step": 251880
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 3.0852895029475835e-05,
+      "loss": 0.793,
+      "step": 251890
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 3.085213338004783e-05,
+      "loss": 0.8702,
+      "step": 251900
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 3.085137173061983e-05,
+      "loss": 0.7993,
+      "step": 251910
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 3.0850610081191835e-05,
+      "loss": 0.76,
+      "step": 251920
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 3.0849848431763825e-05,
+      "loss": 0.8133,
+      "step": 251930
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 3.084908678233583e-05,
+      "loss": 0.8362,
+      "step": 251940
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 3.084832513290783e-05,
+      "loss": 0.9144,
+      "step": 251950
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 3.0847563483479825e-05,
+      "loss": 0.8545,
+      "step": 251960
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 3.084680183405182e-05,
+      "loss": 0.8528,
+      "step": 251970
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 3.084604018462382e-05,
+      "loss": 0.8084,
+      "step": 251980
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 3.0845278535195824e-05,
+      "loss": 0.7975,
+      "step": 251990
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 3.084451688576782e-05,
+      "loss": 0.8299,
+      "step": 252000
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 3.084375523633982e-05,
+      "loss": 0.7355,
+      "step": 252010
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 3.0842993586911816e-05,
+      "loss": 0.8118,
+      "step": 252020
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 3.084223193748382e-05,
+      "loss": 0.7833,
+      "step": 252030
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 3.084147028805581e-05,
+      "loss": 0.8697,
+      "step": 252040
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 3.0840708638627815e-05,
+      "loss": 0.8335,
+      "step": 252050
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 3.083994698919981e-05,
+      "loss": 0.7694,
+      "step": 252060
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 3.083918533977181e-05,
+      "loss": 0.8493,
+      "step": 252070
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 3.083842369034381e-05,
+      "loss": 0.8138,
+      "step": 252080
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 3.083766204091581e-05,
+      "loss": 0.812,
+      "step": 252090
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 3.08369003914878e-05,
+      "loss": 0.8316,
+      "step": 252100
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 3.083613874205981e-05,
+      "loss": 0.8217,
+      "step": 252110
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 3.0835377092631804e-05,
+      "loss": 0.8878,
+      "step": 252120
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 3.08346154432038e-05,
+      "loss": 0.7675,
+      "step": 252130
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 3.08338537937758e-05,
+      "loss": 0.8066,
+      "step": 252140
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 3.0833092144347803e-05,
+      "loss": 0.898,
+      "step": 252150
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 3.08323304949198e-05,
+      "loss": 0.7789,
+      "step": 252160
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 3.08315688454918e-05,
+      "loss": 0.8349,
+      "step": 252170
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 3.0830807196063796e-05,
+      "loss": 0.8152,
+      "step": 252180
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 3.083004554663579e-05,
+      "loss": 0.828,
+      "step": 252190
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 3.08292838972078e-05,
+      "loss": 0.8371,
+      "step": 252200
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 3.0828522247779795e-05,
+      "loss": 0.8368,
+      "step": 252210
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 3.082776059835179e-05,
+      "loss": 0.9104,
+      "step": 252220
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 3.082699894892379e-05,
+      "loss": 0.8694,
+      "step": 252230
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 3.0826237299495794e-05,
+      "loss": 0.8282,
+      "step": 252240
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 3.0825475650067785e-05,
+      "loss": 0.7455,
+      "step": 252250
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 3.082471400063979e-05,
+      "loss": 0.8788,
+      "step": 252260
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 3.082395235121179e-05,
+      "loss": 0.8753,
+      "step": 252270
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 3.0823190701783784e-05,
+      "loss": 0.784,
+      "step": 252280
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 3.082242905235578e-05,
+      "loss": 0.8581,
+      "step": 252290
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 3.0821667402927786e-05,
+      "loss": 0.7973,
+      "step": 252300
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 3.0820905753499777e-05,
+      "loss": 0.8327,
+      "step": 252310
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 3.082014410407178e-05,
+      "loss": 0.8332,
+      "step": 252320
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 3.081938245464378e-05,
+      "loss": 0.8801,
+      "step": 252330
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 3.0818620805215776e-05,
+      "loss": 0.7971,
+      "step": 252340
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 3.081785915578777e-05,
+      "loss": 0.8193,
+      "step": 252350
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 3.081709750635978e-05,
+      "loss": 0.8035,
+      "step": 252360
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 3.0816335856931775e-05,
+      "loss": 0.8514,
+      "step": 252370
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 3.081557420750377e-05,
+      "loss": 0.8829,
+      "step": 252380
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 3.081481255807577e-05,
+      "loss": 0.8351,
+      "step": 252390
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 3.081405090864777e-05,
+      "loss": 0.8691,
+      "step": 252400
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 3.081328925921977e-05,
+      "loss": 0.7922,
+      "step": 252410
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 3.081252760979177e-05,
+      "loss": 0.8246,
+      "step": 252420
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 3.0811765960363766e-05,
+      "loss": 0.848,
+      "step": 252430
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 3.0811004310935764e-05,
+      "loss": 0.7939,
+      "step": 252440
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 3.081024266150776e-05,
+      "loss": 0.84,
+      "step": 252450
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 3.080948101207976e-05,
+      "loss": 0.8504,
+      "step": 252460
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 3.080871936265176e-05,
+      "loss": 0.8946,
+      "step": 252470
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 3.0807957713223754e-05,
+      "loss": 0.859,
+      "step": 252480
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 3.080719606379576e-05,
+      "loss": 0.8536,
+      "step": 252490
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 3.0806434414367755e-05,
+      "loss": 0.8565,
+      "step": 252500
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 3.080567276493975e-05,
+      "loss": 0.7612,
+      "step": 252510
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 3.080491111551175e-05,
+      "loss": 0.8225,
+      "step": 252520
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 3.0804149466083755e-05,
+      "loss": 0.8286,
+      "step": 252530
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 3.080338781665575e-05,
+      "loss": 0.7925,
+      "step": 252540
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 3.080262616722775e-05,
+      "loss": 0.8397,
+      "step": 252550
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 3.080186451779975e-05,
+      "loss": 0.9048,
+      "step": 252560
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 3.0801102868371745e-05,
+      "loss": 0.7377,
+      "step": 252570
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 3.080034121894375e-05,
+      "loss": 0.9439,
+      "step": 252580
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 3.0799579569515746e-05,
+      "loss": 0.8642,
+      "step": 252590
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 3.0798817920087744e-05,
+      "loss": 0.9636,
+      "step": 252600
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 3.079805627065974e-05,
+      "loss": 0.7551,
+      "step": 252610
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 3.0797294621231745e-05,
+      "loss": 0.8658,
+      "step": 252620
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 3.0796532971803736e-05,
+      "loss": 0.8487,
+      "step": 252630
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 3.079577132237574e-05,
+      "loss": 0.8064,
+      "step": 252640
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 3.079500967294774e-05,
+      "loss": 0.8133,
+      "step": 252650
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 3.0794248023519735e-05,
+      "loss": 0.8403,
+      "step": 252660
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 3.079348637409173e-05,
+      "loss": 0.7669,
+      "step": 252670
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 3.079272472466374e-05,
+      "loss": 0.8725,
+      "step": 252680
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 3.079196307523573e-05,
+      "loss": 0.7787,
+      "step": 252690
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 3.079120142580773e-05,
+      "loss": 0.7992,
+      "step": 252700
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 3.079043977637973e-05,
+      "loss": 0.9453,
+      "step": 252710
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 3.078967812695173e-05,
+      "loss": 0.8886,
+      "step": 252720
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 3.0788916477523724e-05,
+      "loss": 0.8247,
+      "step": 252730
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 3.078815482809573e-05,
+      "loss": 0.8107,
+      "step": 252740
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 3.0787393178667726e-05,
+      "loss": 0.789,
+      "step": 252750
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 3.0786631529239724e-05,
+      "loss": 0.8939,
+      "step": 252760
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 3.078586987981172e-05,
+      "loss": 0.7883,
+      "step": 252770
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 3.078510823038372e-05,
+      "loss": 0.9119,
+      "step": 252780
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 3.078434658095572e-05,
+      "loss": 0.8495,
+      "step": 252790
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 3.078358493152772e-05,
+      "loss": 0.9085,
+      "step": 252800
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 3.078282328209972e-05,
+      "loss": 0.7691,
+      "step": 252810
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 3.0782061632671715e-05,
+      "loss": 0.8177,
+      "step": 252820
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 3.078129998324372e-05,
+      "loss": 0.8728,
+      "step": 252830
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 3.078053833381571e-05,
+      "loss": 0.8872,
+      "step": 252840
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 3.0779776684387714e-05,
+      "loss": 0.8596,
+      "step": 252850
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 3.077901503495971e-05,
+      "loss": 0.7471,
+      "step": 252860
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 3.077825338553171e-05,
+      "loss": 0.8762,
+      "step": 252870
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 3.077749173610371e-05,
+      "loss": 0.8283,
+      "step": 252880
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 3.077673008667571e-05,
+      "loss": 0.8687,
+      "step": 252890
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 3.07759684372477e-05,
+      "loss": 0.8008,
+      "step": 252900
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 3.0775206787819706e-05,
+      "loss": 0.8951,
+      "step": 252910
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 3.0774445138391697e-05,
+      "loss": 0.7686,
+      "step": 252920
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 3.07736834889637e-05,
+      "loss": 0.8665,
+      "step": 252930
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 3.07729218395357e-05,
+      "loss": 0.8233,
+      "step": 252940
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 3.0772160190107696e-05,
+      "loss": 0.8403,
+      "step": 252950
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 3.07713985406797e-05,
+      "loss": 0.827,
+      "step": 252960
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 3.07706368912517e-05,
+      "loss": 0.7873,
+      "step": 252970
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 3.0769875241823695e-05,
+      "loss": 0.8495,
+      "step": 252980
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 3.076911359239569e-05,
+      "loss": 0.8459,
+      "step": 252990
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 3.0768351942967697e-05,
+      "loss": 0.9322,
+      "step": 253000
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 3.076759029353969e-05,
+      "loss": 0.903,
+      "step": 253010
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 3.076682864411169e-05,
+      "loss": 0.88,
+      "step": 253020
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 3.076606699468369e-05,
+      "loss": 0.8967,
+      "step": 253030
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 3.0765305345255686e-05,
+      "loss": 0.8296,
+      "step": 253040
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 3.0764543695827684e-05,
+      "loss": 0.793,
+      "step": 253050
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 3.076378204639969e-05,
+      "loss": 0.7412,
+      "step": 253060
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 3.076302039697168e-05,
+      "loss": 0.711,
+      "step": 253070
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 3.076225874754368e-05,
+      "loss": 0.7354,
+      "step": 253080
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 3.076149709811568e-05,
+      "loss": 0.8055,
+      "step": 253090
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 3.076073544868768e-05,
+      "loss": 0.7344,
+      "step": 253100
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 3.0759973799259676e-05,
+      "loss": 0.8528,
+      "step": 253110
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 3.075921214983168e-05,
+      "loss": 0.8705,
+      "step": 253120
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 3.075845050040367e-05,
+      "loss": 0.793,
+      "step": 253130
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 3.0757688850975675e-05,
+      "loss": 0.8224,
+      "step": 253140
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 3.075692720154767e-05,
+      "loss": 0.9205,
+      "step": 253150
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 3.075616555211967e-05,
+      "loss": 0.7914,
+      "step": 253160
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 3.0755403902691674e-05,
+      "loss": 0.7869,
+      "step": 253170
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 3.075464225326367e-05,
+      "loss": 0.7365,
+      "step": 253180
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 3.075388060383567e-05,
+      "loss": 0.8753,
+      "step": 253190
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 3.0753118954407666e-05,
+      "loss": 0.8206,
+      "step": 253200
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 3.075235730497967e-05,
+      "loss": 0.8426,
+      "step": 253210
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 3.075159565555166e-05,
+      "loss": 0.8171,
+      "step": 253220
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 3.0750834006123665e-05,
+      "loss": 0.8115,
+      "step": 253230
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 3.075007235669566e-05,
+      "loss": 0.8373,
+      "step": 253240
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 3.074931070726766e-05,
+      "loss": 0.8732,
+      "step": 253250
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 3.074854905783966e-05,
+      "loss": 0.8541,
+      "step": 253260
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 3.074778740841166e-05,
+      "loss": 0.8095,
+      "step": 253270
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 3.074702575898365e-05,
+      "loss": 0.8405,
+      "step": 253280
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 3.074626410955566e-05,
+      "loss": 0.8235,
+      "step": 253290
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 3.0745502460127654e-05,
+      "loss": 0.8388,
+      "step": 253300
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 3.074474081069965e-05,
+      "loss": 0.8229,
+      "step": 253310
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 3.074397916127165e-05,
+      "loss": 0.9137,
+      "step": 253320
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 3.0743217511843654e-05,
+      "loss": 0.7109,
+      "step": 253330
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 3.074245586241565e-05,
+      "loss": 0.8746,
+      "step": 253340
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 3.074169421298765e-05,
+      "loss": 0.8889,
+      "step": 253350
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 3.0740932563559646e-05,
+      "loss": 0.9204,
+      "step": 253360
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 3.0740170914131644e-05,
+      "loss": 0.8372,
+      "step": 253370
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 3.073940926470365e-05,
+      "loss": 0.8224,
+      "step": 253380
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 3.073864761527564e-05,
+      "loss": 0.8376,
+      "step": 253390
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 3.073788596584764e-05,
+      "loss": 0.9068,
+      "step": 253400
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 3.073712431641964e-05,
+      "loss": 0.8044,
+      "step": 253410
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 3.073636266699164e-05,
+      "loss": 0.8428,
+      "step": 253420
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 3.0735601017563635e-05,
+      "loss": 0.7913,
+      "step": 253430
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 3.073483936813564e-05,
+      "loss": 0.7534,
+      "step": 253440
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 3.073407771870763e-05,
+      "loss": 0.8154,
+      "step": 253450
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 3.0733316069279634e-05,
+      "loss": 0.8657,
+      "step": 253460
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 3.073255441985163e-05,
+      "loss": 0.7452,
+      "step": 253470
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 3.073179277042363e-05,
+      "loss": 0.8441,
+      "step": 253480
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 3.073103112099563e-05,
+      "loss": 0.8814,
+      "step": 253490
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 3.073026947156763e-05,
+      "loss": 0.8859,
+      "step": 253500
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 3.072950782213962e-05,
+      "loss": 0.7567,
+      "step": 253510
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 3.0728746172711626e-05,
+      "loss": 0.852,
+      "step": 253520
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 3.072798452328362e-05,
+      "loss": 0.7822,
+      "step": 253530
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 3.072722287385562e-05,
+      "loss": 0.8016,
+      "step": 253540
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 3.0726461224427625e-05,
+      "loss": 0.8123,
+      "step": 253550
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 3.072569957499962e-05,
+      "loss": 0.8562,
+      "step": 253560
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 3.072493792557162e-05,
+      "loss": 0.7578,
+      "step": 253570
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 3.072417627614362e-05,
+      "loss": 0.8263,
+      "step": 253580
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 3.072341462671562e-05,
+      "loss": 0.8986,
+      "step": 253590
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 3.072265297728761e-05,
+      "loss": 0.8562,
+      "step": 253600
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 3.0721891327859617e-05,
+      "loss": 0.8256,
+      "step": 253610
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 3.0721129678431614e-05,
+      "loss": 0.8569,
+      "step": 253620
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 3.072036802900361e-05,
+      "loss": 0.7894,
+      "step": 253630
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 3.071960637957561e-05,
+      "loss": 0.8344,
+      "step": 253640
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 3.071884473014761e-05,
+      "loss": 0.7852,
+      "step": 253650
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 3.0718083080719604e-05,
+      "loss": 0.8619,
+      "step": 253660
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 3.071732143129161e-05,
+      "loss": 0.8913,
+      "step": 253670
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 3.0716559781863606e-05,
+      "loss": 0.9166,
+      "step": 253680
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 3.07157981324356e-05,
+      "loss": 0.756,
+      "step": 253690
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 3.07150364830076e-05,
+      "loss": 0.8295,
+      "step": 253700
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 3.0714274833579605e-05,
+      "loss": 0.9077,
+      "step": 253710
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 3.0713513184151596e-05,
+      "loss": 0.8495,
+      "step": 253720
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 3.07127515347236e-05,
+      "loss": 0.82,
+      "step": 253730
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 3.07119898852956e-05,
+      "loss": 0.7868,
+      "step": 253740
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 3.0711228235867595e-05,
+      "loss": 0.8952,
+      "step": 253750
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 3.07104665864396e-05,
+      "loss": 0.7891,
+      "step": 253760
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 3.0709704937011596e-05,
+      "loss": 0.8565,
+      "step": 253770
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 3.0708943287583594e-05,
+      "loss": 0.807,
+      "step": 253780
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 3.070818163815559e-05,
+      "loss": 0.8665,
+      "step": 253790
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 3.0707419988727596e-05,
+      "loss": 0.8702,
+      "step": 253800
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 3.0706658339299586e-05,
+      "loss": 0.7857,
+      "step": 253810
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 3.070589668987159e-05,
+      "loss": 0.8575,
+      "step": 253820
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 3.070513504044358e-05,
+      "loss": 0.8209,
+      "step": 253830
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 3.0704373391015585e-05,
+      "loss": 0.6954,
+      "step": 253840
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 3.070361174158758e-05,
+      "loss": 0.8425,
+      "step": 253850
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 3.070285009215958e-05,
+      "loss": 0.7416,
+      "step": 253860
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 3.070208844273158e-05,
+      "loss": 0.9224,
+      "step": 253870
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 3.070132679330358e-05,
+      "loss": 0.8363,
+      "step": 253880
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 3.070056514387557e-05,
+      "loss": 0.7606,
+      "step": 253890
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 3.069980349444758e-05,
+      "loss": 0.8341,
+      "step": 253900
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 3.0699041845019574e-05,
+      "loss": 0.8288,
+      "step": 253910
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 3.069828019559157e-05,
+      "loss": 0.7706,
+      "step": 253920
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 3.0697518546163576e-05,
+      "loss": 0.8014,
+      "step": 253930
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 3.0696756896735574e-05,
+      "loss": 0.7906,
+      "step": 253940
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 3.069599524730757e-05,
+      "loss": 0.8207,
+      "step": 253950
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 3.069523359787957e-05,
+      "loss": 0.8723,
+      "step": 253960
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 3.069447194845157e-05,
+      "loss": 0.8681,
+      "step": 253970
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 3.0693710299023564e-05,
+      "loss": 0.7605,
+      "step": 253980
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 3.069294864959557e-05,
+      "loss": 0.8292,
+      "step": 253990
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 3.0692187000167565e-05,
+      "loss": 0.8086,
+      "step": 254000
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 3.069142535073956e-05,
+      "loss": 0.9097,
+      "step": 254010
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 3.069066370131156e-05,
+      "loss": 0.8109,
+      "step": 254020
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 3.0689902051883564e-05,
+      "loss": 0.7804,
+      "step": 254030
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 3.0689140402455555e-05,
+      "loss": 0.846,
+      "step": 254040
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 3.068837875302756e-05,
+      "loss": 0.9094,
+      "step": 254050
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 3.068761710359956e-05,
+      "loss": 0.7727,
+      "step": 254060
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 3.0686855454171554e-05,
+      "loss": 0.8823,
+      "step": 254070
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 3.068609380474355e-05,
+      "loss": 0.8193,
+      "step": 254080
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 3.0685332155315556e-05,
+      "loss": 0.8402,
+      "step": 254090
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 3.068457050588755e-05,
+      "loss": 0.8976,
+      "step": 254100
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 3.068380885645955e-05,
+      "loss": 0.8191,
+      "step": 254110
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 3.068304720703155e-05,
+      "loss": 0.7399,
+      "step": 254120
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 3.0682285557603546e-05,
+      "loss": 0.8696,
+      "step": 254130
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 3.068152390817555e-05,
+      "loss": 0.7927,
+      "step": 254140
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 3.068076225874755e-05,
+      "loss": 0.8982,
+      "step": 254150
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 3.0680000609319545e-05,
+      "loss": 0.7395,
+      "step": 254160
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 3.067923895989154e-05,
+      "loss": 0.9192,
+      "step": 254170
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 3.067847731046355e-05,
+      "loss": 0.7753,
+      "step": 254180
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 3.067771566103554e-05,
+      "loss": 0.7927,
+      "step": 254190
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 3.067695401160754e-05,
+      "loss": 0.8976,
+      "step": 254200
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 3.067619236217954e-05,
+      "loss": 0.8703,
+      "step": 254210
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 3.067543071275154e-05,
+      "loss": 0.8485,
+      "step": 254220
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 3.0674669063323534e-05,
+      "loss": 0.8051,
+      "step": 254230
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 3.067390741389554e-05,
+      "loss": 0.8551,
+      "step": 254240
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 3.067314576446753e-05,
+      "loss": 0.831,
+      "step": 254250
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 3.067238411503953e-05,
+      "loss": 0.8361,
+      "step": 254260
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 3.067162246561153e-05,
+      "loss": 0.7578,
+      "step": 254270
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 3.067086081618353e-05,
+      "loss": 0.7269,
+      "step": 254280
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 3.0670099166755526e-05,
+      "loss": 0.8804,
+      "step": 254290
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 3.066933751732752e-05,
+      "loss": 0.8588,
+      "step": 254300
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 3.066857586789952e-05,
+      "loss": 0.9239,
+      "step": 254310
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 3.0667814218471525e-05,
+      "loss": 0.8004,
+      "step": 254320
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 3.066705256904352e-05,
+      "loss": 0.7851,
+      "step": 254330
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 3.066629091961552e-05,
+      "loss": 0.877,
+      "step": 254340
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 3.0665529270187524e-05,
+      "loss": 0.8199,
+      "step": 254350
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 3.0664767620759515e-05,
+      "loss": 0.8443,
+      "step": 254360
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 3.066400597133152e-05,
+      "loss": 0.7465,
+      "step": 254370
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 3.0663244321903516e-05,
+      "loss": 0.7939,
+      "step": 254380
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 3.0662482672475514e-05,
+      "loss": 0.9398,
+      "step": 254390
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 3.066172102304751e-05,
+      "loss": 0.8788,
+      "step": 254400
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 3.0660959373619516e-05,
+      "loss": 0.7419,
+      "step": 254410
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 3.0660197724191506e-05,
+      "loss": 0.8289,
+      "step": 254420
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 3.065943607476351e-05,
+      "loss": 0.886,
+      "step": 254430
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 3.065867442533551e-05,
+      "loss": 0.7585,
+      "step": 254440
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 3.0657912775907505e-05,
+      "loss": 0.8168,
+      "step": 254450
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 3.06571511264795e-05,
+      "loss": 0.8186,
+      "step": 254460
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 3.065638947705151e-05,
+      "loss": 0.8488,
+      "step": 254470
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 3.06556278276235e-05,
+      "loss": 0.7385,
+      "step": 254480
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 3.06548661781955e-05,
+      "loss": 0.8607,
+      "step": 254490
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 3.06541045287675e-05,
+      "loss": 0.7816,
+      "step": 254500
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 3.06533428793395e-05,
+      "loss": 0.8285,
+      "step": 254510
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 3.0652581229911494e-05,
+      "loss": 0.8033,
+      "step": 254520
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 3.06518195804835e-05,
+      "loss": 0.8391,
+      "step": 254530
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 3.0651057931055496e-05,
+      "loss": 0.8456,
+      "step": 254540
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 3.0650296281627494e-05,
+      "loss": 0.9124,
+      "step": 254550
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 3.06495346321995e-05,
+      "loss": 0.8974,
+      "step": 254560
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 3.064877298277149e-05,
+      "loss": 0.8264,
+      "step": 254570
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 3.064801133334349e-05,
+      "loss": 0.6994,
+      "step": 254580
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 3.064724968391549e-05,
+      "loss": 0.7461,
+      "step": 254590
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 3.064648803448749e-05,
+      "loss": 0.7964,
+      "step": 254600
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 3.0645726385059485e-05,
+      "loss": 0.8266,
+      "step": 254610
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 3.064496473563149e-05,
+      "loss": 0.7316,
+      "step": 254620
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 3.064420308620348e-05,
+      "loss": 0.8335,
+      "step": 254630
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 3.0643441436775484e-05,
+      "loss": 0.9152,
+      "step": 254640
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 3.064267978734748e-05,
+      "loss": 0.8624,
+      "step": 254650
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 3.064191813791948e-05,
+      "loss": 0.7942,
+      "step": 254660
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 3.064115648849148e-05,
+      "loss": 0.8694,
+      "step": 254670
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 3.064039483906348e-05,
+      "loss": 0.8411,
+      "step": 254680
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 3.063963318963547e-05,
+      "loss": 0.9386,
+      "step": 254690
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 3.0638871540207476e-05,
+      "loss": 0.854,
+      "step": 254700
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 3.0638109890779473e-05,
+      "loss": 0.8122,
+      "step": 254710
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 3.063734824135147e-05,
+      "loss": 0.8533,
+      "step": 254720
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 3.0636586591923475e-05,
+      "loss": 0.8439,
+      "step": 254730
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 3.063582494249547e-05,
+      "loss": 0.7987,
+      "step": 254740
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 3.063506329306747e-05,
+      "loss": 0.811,
+      "step": 254750
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 3.063430164363947e-05,
+      "loss": 0.827,
+      "step": 254760
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 3.0633539994211465e-05,
+      "loss": 1.0487,
+      "step": 254770
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 3.063277834478346e-05,
+      "loss": 0.7608,
+      "step": 254780
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 3.063201669535547e-05,
+      "loss": 0.784,
+      "step": 254790
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 3.063125504592746e-05,
+      "loss": 0.7801,
+      "step": 254800
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 3.063049339649946e-05,
+      "loss": 0.8194,
+      "step": 254810
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 3.062973174707146e-05,
+      "loss": 0.8574,
+      "step": 254820
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 3.062897009764346e-05,
+      "loss": 0.8699,
+      "step": 254830
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 3.0628208448215454e-05,
+      "loss": 0.8607,
+      "step": 254840
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 3.062744679878746e-05,
+      "loss": 0.806,
+      "step": 254850
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 3.062668514935945e-05,
+      "loss": 0.7552,
+      "step": 254860
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 3.062592349993145e-05,
+      "loss": 0.8456,
+      "step": 254870
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 3.062516185050345e-05,
+      "loss": 0.7986,
+      "step": 254880
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 3.062440020107545e-05,
+      "loss": 0.7829,
+      "step": 254890
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 3.0623638551647446e-05,
+      "loss": 0.8414,
+      "step": 254900
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 3.062287690221945e-05,
+      "loss": 0.8083,
+      "step": 254910
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 3.062211525279145e-05,
+      "loss": 0.8321,
+      "step": 254920
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 3.0621353603363445e-05,
+      "loss": 0.8504,
+      "step": 254930
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 3.062059195393545e-05,
+      "loss": 0.8884,
+      "step": 254940
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 3.061983030450744e-05,
+      "loss": 0.8255,
+      "step": 254950
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 3.0619068655079444e-05,
+      "loss": 0.8324,
+      "step": 254960
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 3.061830700565144e-05,
+      "loss": 0.9778,
+      "step": 254970
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 3.061754535622344e-05,
+      "loss": 0.8766,
+      "step": 254980
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 3.0616783706795436e-05,
+      "loss": 0.6998,
+      "step": 254990
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 3.061602205736744e-05,
+      "loss": 0.8193,
+      "step": 255000
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 3.061526040793943e-05,
+      "loss": 0.8812,
+      "step": 255010
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 3.0614498758511436e-05,
+      "loss": 0.8542,
+      "step": 255020
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 3.061373710908343e-05,
+      "loss": 0.8504,
+      "step": 255030
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 3.061297545965543e-05,
+      "loss": 0.8653,
+      "step": 255040
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 3.061221381022743e-05,
+      "loss": 0.7115,
+      "step": 255050
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 3.061145216079943e-05,
+      "loss": 0.8234,
+      "step": 255060
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 3.061069051137142e-05,
+      "loss": 0.7741,
+      "step": 255070
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 3.060992886194343e-05,
+      "loss": 0.8038,
+      "step": 255080
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 3.0609167212515425e-05,
+      "loss": 0.8491,
+      "step": 255090
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 3.060840556308742e-05,
+      "loss": 0.7609,
+      "step": 255100
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 3.060764391365942e-05,
+      "loss": 0.7568,
+      "step": 255110
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 3.0606882264231424e-05,
+      "loss": 0.7586,
+      "step": 255120
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 3.060612061480342e-05,
+      "loss": 0.8391,
+      "step": 255130
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 3.060535896537542e-05,
+      "loss": 0.8975,
+      "step": 255140
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 3.060459731594742e-05,
+      "loss": 0.804,
+      "step": 255150
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 3.0603835666519414e-05,
+      "loss": 0.8409,
+      "step": 255160
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 3.060307401709142e-05,
+      "loss": 0.7656,
+      "step": 255170
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 3.0602312367663415e-05,
+      "loss": 0.8111,
+      "step": 255180
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 3.060155071823541e-05,
+      "loss": 0.7989,
+      "step": 255190
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 3.060078906880741e-05,
+      "loss": 0.7966,
+      "step": 255200
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 3.060002741937941e-05,
+      "loss": 0.7951,
+      "step": 255210
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 3.0599265769951405e-05,
+      "loss": 0.8213,
+      "step": 255220
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 3.059850412052341e-05,
+      "loss": 0.8461,
+      "step": 255230
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 3.05977424710954e-05,
+      "loss": 0.7682,
+      "step": 255240
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 3.0596980821667404e-05,
+      "loss": 0.7721,
+      "step": 255250
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 3.05962191722394e-05,
+      "loss": 0.8524,
+      "step": 255260
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 3.05954575228114e-05,
+      "loss": 0.9193,
+      "step": 255270
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 3.05946958733834e-05,
+      "loss": 0.7603,
+      "step": 255280
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 3.05939342239554e-05,
+      "loss": 0.8425,
+      "step": 255290
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 3.05931725745274e-05,
+      "loss": 0.8925,
+      "step": 255300
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 3.0592410925099396e-05,
+      "loss": 0.8135,
+      "step": 255310
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 3.0591649275671393e-05,
+      "loss": 0.7873,
+      "step": 255320
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 3.059088762624339e-05,
+      "loss": 0.8302,
+      "step": 255330
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 3.0590125976815395e-05,
+      "loss": 0.7819,
+      "step": 255340
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 3.058936432738739e-05,
+      "loss": 0.8682,
+      "step": 255350
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 3.058860267795939e-05,
+      "loss": 0.7318,
+      "step": 255360
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 3.058784102853139e-05,
+      "loss": 0.8075,
+      "step": 255370
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 3.058707937910339e-05,
+      "loss": 0.7792,
+      "step": 255380
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 3.058631772967538e-05,
+      "loss": 0.7891,
+      "step": 255390
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 3.058555608024739e-05,
+      "loss": 0.794,
+      "step": 255400
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 3.0584794430819384e-05,
+      "loss": 0.7835,
+      "step": 255410
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 3.058403278139138e-05,
+      "loss": 0.7553,
+      "step": 255420
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 3.058327113196338e-05,
+      "loss": 0.8514,
+      "step": 255430
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 3.0582509482535383e-05,
+      "loss": 0.8384,
+      "step": 255440
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 3.0581747833107374e-05,
+      "loss": 0.7762,
+      "step": 255450
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 3.058098618367938e-05,
+      "loss": 0.8162,
+      "step": 255460
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 3.0580224534251376e-05,
+      "loss": 0.8119,
+      "step": 255470
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 3.057946288482337e-05,
+      "loss": 0.8786,
+      "step": 255480
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 3.057870123539537e-05,
+      "loss": 0.8963,
+      "step": 255490
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 3.0577939585967375e-05,
+      "loss": 0.7909,
+      "step": 255500
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 3.057717793653937e-05,
+      "loss": 0.8184,
+      "step": 255510
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 3.057641628711137e-05,
+      "loss": 0.8829,
+      "step": 255520
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 3.0575654637683374e-05,
+      "loss": 0.8292,
+      "step": 255530
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 3.0574892988255365e-05,
+      "loss": 0.9031,
+      "step": 255540
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 3.057413133882737e-05,
+      "loss": 0.8255,
+      "step": 255550
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 3.0573369689399367e-05,
+      "loss": 0.7791,
+      "step": 255560
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 3.0572608039971364e-05,
+      "loss": 0.7501,
+      "step": 255570
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 3.057184639054336e-05,
+      "loss": 0.8042,
+      "step": 255580
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 3.0571084741115366e-05,
+      "loss": 0.904,
+      "step": 255590
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 3.0570323091687356e-05,
+      "loss": 0.7951,
+      "step": 255600
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 3.056956144225936e-05,
+      "loss": 0.9468,
+      "step": 255610
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 3.056879979283136e-05,
+      "loss": 0.7669,
+      "step": 255620
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 3.0568038143403356e-05,
+      "loss": 0.8209,
+      "step": 255630
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 3.056727649397535e-05,
+      "loss": 0.7994,
+      "step": 255640
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 3.056651484454736e-05,
+      "loss": 0.9496,
+      "step": 255650
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 3.056575319511935e-05,
+      "loss": 0.9773,
+      "step": 255660
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 3.056499154569135e-05,
+      "loss": 0.9436,
+      "step": 255670
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 3.056422989626335e-05,
+      "loss": 0.8995,
+      "step": 255680
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 3.056346824683535e-05,
+      "loss": 0.8276,
+      "step": 255690
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 3.0562706597407345e-05,
+      "loss": 0.73,
+      "step": 255700
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 3.056194494797934e-05,
+      "loss": 0.7936,
+      "step": 255710
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 3.0561183298551346e-05,
+      "loss": 0.8377,
+      "step": 255720
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 3.0560421649123344e-05,
+      "loss": 0.8126,
+      "step": 255730
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 3.055965999969534e-05,
+      "loss": 0.7451,
+      "step": 255740
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 3.055889835026734e-05,
+      "loss": 0.8001,
+      "step": 255750
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 3.055813670083934e-05,
+      "loss": 0.8421,
+      "step": 255760
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 3.0557375051411334e-05,
+      "loss": 0.816,
+      "step": 255770
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 3.055661340198334e-05,
+      "loss": 0.7916,
+      "step": 255780
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 3.0555851752555335e-05,
+      "loss": 0.8272,
+      "step": 255790
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 3.055509010312733e-05,
+      "loss": 0.8466,
+      "step": 255800
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 3.055432845369933e-05,
+      "loss": 0.9214,
+      "step": 255810
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 3.0553566804271335e-05,
+      "loss": 0.7705,
+      "step": 255820
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 3.0552805154843325e-05,
+      "loss": 0.7901,
+      "step": 255830
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 3.055204350541533e-05,
+      "loss": 0.8462,
+      "step": 255840
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 3.055128185598733e-05,
+      "loss": 0.8969,
+      "step": 255850
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 3.0550520206559324e-05,
+      "loss": 0.8594,
+      "step": 255860
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 3.054975855713132e-05,
+      "loss": 0.8911,
+      "step": 255870
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 3.0548996907703326e-05,
+      "loss": 0.9021,
+      "step": 255880
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 3.0548235258275324e-05,
+      "loss": 0.9109,
+      "step": 255890
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 3.054747360884732e-05,
+      "loss": 0.8102,
+      "step": 255900
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 3.054671195941932e-05,
+      "loss": 0.8328,
+      "step": 255910
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 3.0545950309991316e-05,
+      "loss": 0.8777,
+      "step": 255920
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 3.054518866056332e-05,
+      "loss": 0.8441,
+      "step": 255930
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 3.054442701113532e-05,
+      "loss": 0.701,
+      "step": 255940
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 3.0543665361707315e-05,
+      "loss": 0.89,
+      "step": 255950
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 3.054290371227931e-05,
+      "loss": 0.8666,
+      "step": 255960
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 3.054214206285132e-05,
+      "loss": 0.8128,
+      "step": 255970
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 3.054138041342331e-05,
+      "loss": 0.7962,
+      "step": 255980
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 3.054061876399531e-05,
+      "loss": 0.7573,
+      "step": 255990
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 3.053985711456731e-05,
+      "loss": 0.8184,
+      "step": 256000
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 3.053909546513931e-05,
+      "loss": 0.8289,
+      "step": 256010
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 3.0538333815711304e-05,
+      "loss": 0.8467,
+      "step": 256020
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 3.053757216628331e-05,
+      "loss": 0.8245,
+      "step": 256030
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 3.05368105168553e-05,
+      "loss": 0.8856,
+      "step": 256040
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 3.0536048867427303e-05,
+      "loss": 0.9291,
+      "step": 256050
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 3.05352872179993e-05,
+      "loss": 0.7997,
+      "step": 256060
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 3.05345255685713e-05,
+      "loss": 0.8004,
+      "step": 256070
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 3.0533763919143296e-05,
+      "loss": 0.8399,
+      "step": 256080
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 3.05330022697153e-05,
+      "loss": 0.8153,
+      "step": 256090
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 3.05322406202873e-05,
+      "loss": 0.8483,
+      "step": 256100
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 3.0531478970859295e-05,
+      "loss": 0.855,
+      "step": 256110
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 3.053071732143129e-05,
+      "loss": 0.8602,
+      "step": 256120
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 3.052995567200329e-05,
+      "loss": 0.9278,
+      "step": 256130
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 3.0529194022575294e-05,
+      "loss": 0.7991,
+      "step": 256140
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 3.0528432373147285e-05,
+      "loss": 0.8423,
+      "step": 256150
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 3.052767072371929e-05,
+      "loss": 0.8704,
+      "step": 256160
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 3.0526909074291287e-05,
+      "loss": 0.8373,
+      "step": 256170
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 3.0526147424863284e-05,
+      "loss": 0.7725,
+      "step": 256180
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 3.052538577543528e-05,
+      "loss": 0.8704,
+      "step": 256190
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 3.0524624126007286e-05,
+      "loss": 0.8803,
+      "step": 256200
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 3.0523862476579276e-05,
+      "loss": 0.8397,
+      "step": 256210
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 3.052310082715128e-05,
+      "loss": 0.854,
+      "step": 256220
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 3.052233917772328e-05,
+      "loss": 0.7309,
+      "step": 256230
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 3.0521577528295276e-05,
+      "loss": 0.9047,
+      "step": 256240
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 3.052081587886727e-05,
+      "loss": 0.8689,
+      "step": 256250
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 3.052005422943928e-05,
+      "loss": 0.8494,
+      "step": 256260
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 3.0519292580011275e-05,
+      "loss": 0.8317,
+      "step": 256270
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 3.051853093058327e-05,
+      "loss": 0.8671,
+      "step": 256280
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 3.051776928115527e-05,
+      "loss": 0.8835,
+      "step": 256290
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 3.0517007631727267e-05,
+      "loss": 0.8437,
+      "step": 256300
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 3.0516245982299268e-05,
+      "loss": 0.8568,
+      "step": 256310
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 3.051548433287127e-05,
+      "loss": 0.7268,
+      "step": 256320
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 3.0514722683443263e-05,
+      "loss": 0.8935,
+      "step": 256330
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 3.0513961034015264e-05,
+      "loss": 0.872,
+      "step": 256340
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 3.0513199384587265e-05,
+      "loss": 0.8641,
+      "step": 256350
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 3.0512437735159262e-05,
+      "loss": 0.7853,
+      "step": 256360
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 3.051167608573126e-05,
+      "loss": 0.9384,
+      "step": 256370
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 3.051091443630326e-05,
+      "loss": 0.8374,
+      "step": 256380
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 3.0510152786875258e-05,
+      "loss": 0.757,
+      "step": 256390
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 3.050939113744726e-05,
+      "loss": 0.8242,
+      "step": 256400
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 3.050862948801926e-05,
+      "loss": 0.829,
+      "step": 256410
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 3.0507867838591254e-05,
+      "loss": 0.8119,
+      "step": 256420
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 3.0507106189163255e-05,
+      "loss": 0.8001,
+      "step": 256430
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 3.0506344539735255e-05,
+      "loss": 0.8238,
+      "step": 256440
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 3.050558289030725e-05,
+      "loss": 0.8758,
+      "step": 256450
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 3.050482124087925e-05,
+      "loss": 0.8703,
+      "step": 256460
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 3.050405959145125e-05,
+      "loss": 0.9152,
+      "step": 256470
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 3.0503297942023245e-05,
+      "loss": 0.8207,
+      "step": 256480
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 3.0502536292595246e-05,
+      "loss": 0.9317,
+      "step": 256490
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 3.0501774643167247e-05,
+      "loss": 0.8396,
+      "step": 256500
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 3.050101299373924e-05,
+      "loss": 0.8079,
+      "step": 256510
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 3.0500251344311242e-05,
+      "loss": 0.842,
+      "step": 256520
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 3.0499489694883243e-05,
+      "loss": 0.9022,
+      "step": 256530
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 3.0498728045455237e-05,
+      "loss": 0.8932,
+      "step": 256540
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 3.0497966396027238e-05,
+      "loss": 0.8616,
+      "step": 256550
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 3.049720474659924e-05,
+      "loss": 0.8221,
+      "step": 256560
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 3.0496443097171236e-05,
+      "loss": 0.777,
+      "step": 256570
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 3.0495681447743234e-05,
+      "loss": 0.9118,
+      "step": 256580
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 3.049491979831523e-05,
+      "loss": 0.849,
+      "step": 256590
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 3.0494158148887232e-05,
+      "loss": 0.8275,
+      "step": 256600
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 3.0493396499459233e-05,
+      "loss": 0.8542,
+      "step": 256610
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 3.0492634850031227e-05,
+      "loss": 0.7583,
+      "step": 256620
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 3.0491873200603228e-05,
+      "loss": 0.8124,
+      "step": 256630
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 3.049111155117523e-05,
+      "loss": 0.8035,
+      "step": 256640
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 3.0490349901747223e-05,
+      "loss": 0.8082,
+      "step": 256650
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 3.0489588252319223e-05,
+      "loss": 0.8831,
+      "step": 256660
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 3.0488826602891224e-05,
+      "loss": 0.8055,
+      "step": 256670
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 3.048806495346322e-05,
+      "loss": 0.8901,
+      "step": 256680
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 3.048730330403522e-05,
+      "loss": 0.7731,
+      "step": 256690
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 3.048654165460722e-05,
+      "loss": 0.8709,
+      "step": 256700
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 3.0485780005179214e-05,
+      "loss": 0.8567,
+      "step": 256710
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 3.0485018355751215e-05,
+      "loss": 0.7792,
+      "step": 256720
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 3.0484256706323216e-05,
+      "loss": 0.8905,
+      "step": 256730
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 3.048349505689521e-05,
+      "loss": 0.8791,
+      "step": 256740
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 3.048273340746721e-05,
+      "loss": 0.7976,
+      "step": 256750
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 3.048197175803921e-05,
+      "loss": 0.8579,
+      "step": 256760
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 3.048121010861121e-05,
+      "loss": 0.855,
+      "step": 256770
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 3.048044845918321e-05,
+      "loss": 0.9842,
+      "step": 256780
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 3.0479686809755207e-05,
+      "loss": 0.8866,
+      "step": 256790
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 3.0478925160327205e-05,
+      "loss": 0.8882,
+      "step": 256800
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 3.0478163510899206e-05,
+      "loss": 0.7876,
+      "step": 256810
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 3.0477401861471207e-05,
+      "loss": 0.8639,
+      "step": 256820
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 3.04766402120432e-05,
+      "loss": 0.8129,
+      "step": 256830
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 3.04758785626152e-05,
+      "loss": 0.8029,
+      "step": 256840
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 3.0475116913187202e-05,
+      "loss": 0.7592,
+      "step": 256850
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 3.0474355263759196e-05,
+      "loss": 0.7873,
+      "step": 256860
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 3.0473593614331197e-05,
+      "loss": 0.8595,
+      "step": 256870
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 3.0472831964903198e-05,
+      "loss": 0.8733,
+      "step": 256880
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 3.0472070315475192e-05,
+      "loss": 0.778,
+      "step": 256890
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 3.0471308666047193e-05,
+      "loss": 0.8858,
+      "step": 256900
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 3.0470547016619194e-05,
+      "loss": 0.926,
+      "step": 256910
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 3.0469785367191188e-05,
+      "loss": 0.7931,
+      "step": 256920
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 3.046902371776319e-05,
+      "loss": 0.8493,
+      "step": 256930
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 3.046826206833519e-05,
+      "loss": 0.8811,
+      "step": 256940
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 3.0467500418907184e-05,
+      "loss": 0.8431,
+      "step": 256950
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 3.0466738769479185e-05,
+      "loss": 0.7107,
+      "step": 256960
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 3.0465977120051186e-05,
+      "loss": 0.7339,
+      "step": 256970
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 3.0465215470623183e-05,
+      "loss": 0.8158,
+      "step": 256980
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 3.0464453821195184e-05,
+      "loss": 0.8113,
+      "step": 256990
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 3.046369217176718e-05,
+      "loss": 0.8501,
+      "step": 257000
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 3.046293052233918e-05,
+      "loss": 0.8863,
+      "step": 257010
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 3.046216887291118e-05,
+      "loss": 0.8832,
+      "step": 257020
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 3.046140722348318e-05,
+      "loss": 0.9237,
+      "step": 257030
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 3.0460645574055175e-05,
+      "loss": 0.8092,
+      "step": 257040
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 3.0459883924627175e-05,
+      "loss": 0.8928,
+      "step": 257050
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 3.045912227519917e-05,
+      "loss": 0.8403,
+      "step": 257060
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 3.045836062577117e-05,
+      "loss": 0.8278,
+      "step": 257070
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 3.045759897634317e-05,
+      "loss": 0.8465,
+      "step": 257080
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 3.0456837326915165e-05,
+      "loss": 0.9248,
+      "step": 257090
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 3.0456075677487166e-05,
+      "loss": 0.7673,
+      "step": 257100
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 3.0455314028059167e-05,
+      "loss": 0.8643,
+      "step": 257110
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 3.045455237863116e-05,
+      "loss": 0.7814,
+      "step": 257120
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 3.0453790729203162e-05,
+      "loss": 0.7915,
+      "step": 257130
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 3.0453029079775163e-05,
+      "loss": 1.0306,
+      "step": 257140
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 3.045226743034716e-05,
+      "loss": 0.8253,
+      "step": 257150
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 3.045150578091916e-05,
+      "loss": 0.8818,
+      "step": 257160
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 3.045074413149116e-05,
+      "loss": 0.8794,
+      "step": 257170
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 3.0449982482063156e-05,
+      "loss": 0.8276,
+      "step": 257180
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 3.0449220832635157e-05,
+      "loss": 0.8254,
+      "step": 257190
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 3.0448459183207158e-05,
+      "loss": 0.7093,
+      "step": 257200
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 3.0447697533779152e-05,
+      "loss": 0.9234,
+      "step": 257210
+    },
+    {
+      "epoch": 11.75,
+      "learning_rate": 3.0446935884351153e-05,
+      "loss": 0.8895,
+      "step": 257220
+    },
+    {
+      "epoch": 11.75,
+      "learning_rate": 3.0446174234923154e-05,
+      "loss": 0.7911,
+      "step": 257230
+    },
+    {
+      "epoch": 11.75,
+      "learning_rate": 3.0445412585495148e-05,
+      "loss": 0.9826,
+      "step": 257240
+    },
+    {
+      "epoch": 11.75,
+      "learning_rate": 3.044465093606715e-05,
+      "loss": 0.859,
+      "step": 257250
+    },
+    {
+      "epoch": 11.75,
+      "learning_rate": 3.044388928663915e-05,
+      "loss": 0.8306,
+      "step": 257260
+    },
+    {
+      "epoch": 11.75,
+      "learning_rate": 3.0443127637211143e-05,
+      "loss": 0.8829,
+      "step": 257270
+    },
+    {
+      "epoch": 11.75,
+      "learning_rate": 3.0442365987783144e-05,
+      "loss": 0.8938,
+      "step": 257280
+    },
+    {
+      "epoch": 11.75,
+      "learning_rate": 3.0441604338355145e-05,
+      "loss": 0.8085,
+      "step": 257290
+    },
+    {
+      "epoch": 11.75,
+      "learning_rate": 3.044084268892714e-05,
+      "loss": 0.7476,
+      "step": 257300
+    },
+    {
+      "epoch": 11.75,
+      "learning_rate": 3.044008103949914e-05,
+      "loss": 0.808,
+      "step": 257310
+    },
+    {
+      "epoch": 11.75,
+      "learning_rate": 3.043931939007114e-05,
+      "loss": 0.8761,
+      "step": 257320
+    },
+    {
+      "epoch": 11.75,
+      "learning_rate": 3.0438557740643135e-05,
+      "loss": 0.8182,
+      "step": 257330
+    },
+    {
+      "epoch": 11.75,
+      "learning_rate": 3.0437796091215136e-05,
+      "loss": 0.7447,
+      "step": 257340
+    },
+    {
+      "epoch": 11.75,
+      "learning_rate": 3.0437034441787137e-05,
+      "loss": 0.8866,
+      "step": 257350
+    },
+    {
+      "epoch": 11.75,
+      "learning_rate": 3.0436272792359134e-05,
+      "loss": 0.8111,
+      "step": 257360
+    },
+    {
+      "epoch": 11.75,
+      "learning_rate": 3.0435511142931135e-05,
+      "loss": 0.8279,
+      "step": 257370
+    },
+    {
+      "epoch": 11.75,
+      "learning_rate": 3.0434749493503133e-05,
+      "loss": 0.821,
+      "step": 257380
+    },
+    {
+      "epoch": 11.75,
+      "learning_rate": 3.043398784407513e-05,
+      "loss": 0.8284,
+      "step": 257390
+    },
+    {
+      "epoch": 11.75,
+      "learning_rate": 3.043322619464713e-05,
+      "loss": 0.7732,
+      "step": 257400
+    },
+    {
+      "epoch": 11.75,
+      "learning_rate": 3.0432464545219132e-05,
+      "loss": 0.83,
+      "step": 257410
+    },
+    {
+      "epoch": 11.75,
+      "learning_rate": 3.0431702895791126e-05,
+      "loss": 0.7009,
+      "step": 257420
+    },
+    {
+      "epoch": 11.75,
+      "learning_rate": 3.0430941246363127e-05,
+      "loss": 0.7796,
+      "step": 257430
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 3.0430179596935127e-05,
+      "loss": 0.7859,
+      "step": 257440
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 3.042941794750712e-05,
+      "loss": 0.8475,
+      "step": 257450
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 3.0428656298079122e-05,
+      "loss": 0.8416,
+      "step": 257460
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 3.0427894648651123e-05,
+      "loss": 0.7529,
+      "step": 257470
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 3.0427132999223117e-05,
+      "loss": 0.7639,
+      "step": 257480
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 3.0426371349795118e-05,
+      "loss": 0.8808,
+      "step": 257490
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 3.042560970036712e-05,
+      "loss": 0.889,
+      "step": 257500
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 3.0424848050939113e-05,
+      "loss": 0.8241,
+      "step": 257510
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 3.0424086401511114e-05,
+      "loss": 0.8817,
+      "step": 257520
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 3.042332475208311e-05,
+      "loss": 0.8738,
+      "step": 257530
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 3.042256310265511e-05,
+      "loss": 0.7695,
+      "step": 257540
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 3.042180145322711e-05,
+      "loss": 0.7886,
+      "step": 257550
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 3.0421039803799107e-05,
+      "loss": 0.8113,
+      "step": 257560
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 3.0420278154371108e-05,
+      "loss": 0.8194,
+      "step": 257570
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 3.041951650494311e-05,
+      "loss": 0.8571,
+      "step": 257580
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 3.0418754855515103e-05,
+      "loss": 0.9272,
+      "step": 257590
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 3.0417993206087104e-05,
+      "loss": 0.8161,
+      "step": 257600
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 3.0417231556659105e-05,
+      "loss": 0.8745,
+      "step": 257610
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 3.04164699072311e-05,
+      "loss": 0.8989,
+      "step": 257620
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 3.04157082578031e-05,
+      "loss": 0.7841,
+      "step": 257630
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 3.04149466083751e-05,
+      "loss": 0.8461,
+      "step": 257640
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 3.0414184958947095e-05,
+      "loss": 0.7592,
+      "step": 257650
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 3.0413423309519095e-05,
+      "loss": 0.8814,
+      "step": 257660
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 3.0412661660091096e-05,
+      "loss": 0.8081,
+      "step": 257670
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 3.041190001066309e-05,
+      "loss": 0.8902,
+      "step": 257680
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 3.041113836123509e-05,
+      "loss": 0.776,
+      "step": 257690
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 3.0410376711807092e-05,
+      "loss": 0.7967,
+      "step": 257700
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 3.0409615062379086e-05,
+      "loss": 0.8226,
+      "step": 257710
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 3.0408853412951087e-05,
+      "loss": 0.89,
+      "step": 257720
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 3.0408091763523088e-05,
+      "loss": 0.7967,
+      "step": 257730
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 3.0407330114095085e-05,
+      "loss": 0.8585,
+      "step": 257740
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 3.0406568464667086e-05,
+      "loss": 0.8903,
+      "step": 257750
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 3.0405806815239084e-05,
+      "loss": 0.8296,
+      "step": 257760
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 3.040504516581108e-05,
+      "loss": 0.9608,
+      "step": 257770
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 3.0404283516383082e-05,
+      "loss": 0.7648,
+      "step": 257780
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 3.0403521866955083e-05,
+      "loss": 0.8429,
+      "step": 257790
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 3.0402760217527077e-05,
+      "loss": 0.8386,
+      "step": 257800
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 3.0401998568099078e-05,
+      "loss": 0.7553,
+      "step": 257810
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 3.040123691867108e-05,
+      "loss": 0.7877,
+      "step": 257820
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 3.0400475269243073e-05,
+      "loss": 0.7538,
+      "step": 257830
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 3.0399713619815074e-05,
+      "loss": 0.8171,
+      "step": 257840
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 3.0398951970387074e-05,
+      "loss": 0.7904,
+      "step": 257850
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 3.039819032095907e-05,
+      "loss": 0.8291,
+      "step": 257860
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 3.039742867153107e-05,
+      "loss": 0.9564,
+      "step": 257870
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 3.039666702210307e-05,
+      "loss": 0.8685,
+      "step": 257880
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 3.0395905372675064e-05,
+      "loss": 0.8586,
+      "step": 257890
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 3.0395143723247065e-05,
+      "loss": 0.7886,
+      "step": 257900
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 3.0394382073819066e-05,
+      "loss": 0.8467,
+      "step": 257910
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 3.039362042439106e-05,
+      "loss": 0.766,
+      "step": 257920
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 3.039285877496306e-05,
+      "loss": 0.8873,
+      "step": 257930
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 3.0392097125535062e-05,
+      "loss": 0.8184,
+      "step": 257940
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 3.039133547610706e-05,
+      "loss": 0.8252,
+      "step": 257950
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 3.039057382667906e-05,
+      "loss": 0.8796,
+      "step": 257960
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 3.0389812177251058e-05,
+      "loss": 0.8539,
+      "step": 257970
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 3.0389050527823055e-05,
+      "loss": 0.8308,
+      "step": 257980
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 3.0388288878395056e-05,
+      "loss": 0.8206,
+      "step": 257990
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 3.038752722896705e-05,
+      "loss": 0.946,
+      "step": 258000
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 3.038676557953905e-05,
+      "loss": 0.8302,
+      "step": 258010
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 3.0386003930111052e-05,
+      "loss": 0.7995,
+      "step": 258020
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 3.0385242280683046e-05,
+      "loss": 0.803,
+      "step": 258030
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 3.0384480631255047e-05,
+      "loss": 0.8842,
+      "step": 258040
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 3.0383718981827047e-05,
+      "loss": 0.8117,
+      "step": 258050
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 3.038295733239904e-05,
+      "loss": 0.7841,
+      "step": 258060
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 3.0382195682971042e-05,
+      "loss": 0.8203,
+      "step": 258070
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 3.0381434033543043e-05,
+      "loss": 0.8402,
+      "step": 258080
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 3.0380672384115037e-05,
+      "loss": 0.8287,
+      "step": 258090
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 3.0379910734687038e-05,
+      "loss": 0.7788,
+      "step": 258100
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 3.037914908525904e-05,
+      "loss": 0.7704,
+      "step": 258110
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 3.0378387435831037e-05,
+      "loss": 0.8482,
+      "step": 258120
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 3.0377625786403034e-05,
+      "loss": 0.9301,
+      "step": 258130
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 3.0376864136975035e-05,
+      "loss": 0.8034,
+      "step": 258140
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 3.0376102487547032e-05,
+      "loss": 0.8294,
+      "step": 258150
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 3.0375340838119033e-05,
+      "loss": 0.9423,
+      "step": 258160
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 3.0374579188691034e-05,
+      "loss": 0.9263,
+      "step": 258170
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 3.0373817539263028e-05,
+      "loss": 0.879,
+      "step": 258180
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 3.037305588983503e-05,
+      "loss": 0.8164,
+      "step": 258190
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 3.037229424040703e-05,
+      "loss": 0.8223,
+      "step": 258200
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 3.0371532590979024e-05,
+      "loss": 0.7872,
+      "step": 258210
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 3.0370770941551025e-05,
+      "loss": 0.7735,
+      "step": 258220
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 3.0370009292123026e-05,
+      "loss": 0.8796,
+      "step": 258230
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 3.036924764269502e-05,
+      "loss": 0.882,
+      "step": 258240
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 3.036848599326702e-05,
+      "loss": 0.929,
+      "step": 258250
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 3.036772434383902e-05,
+      "loss": 0.867,
+      "step": 258260
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 3.0366962694411015e-05,
+      "loss": 0.752,
+      "step": 258270
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 3.0366201044983016e-05,
+      "loss": 0.8579,
+      "step": 258280
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 3.0365439395555017e-05,
+      "loss": 0.7636,
+      "step": 258290
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 3.036467774612701e-05,
+      "loss": 0.7953,
+      "step": 258300
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 3.0363916096699012e-05,
+      "loss": 0.8178,
+      "step": 258310
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 3.0363154447271013e-05,
+      "loss": 0.802,
+      "step": 258320
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 3.036239279784301e-05,
+      "loss": 0.884,
+      "step": 258330
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 3.0361631148415008e-05,
+      "loss": 0.8589,
+      "step": 258340
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 3.036086949898701e-05,
+      "loss": 0.8803,
+      "step": 258350
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 3.0360107849559006e-05,
+      "loss": 0.8326,
+      "step": 258360
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 3.0359346200131007e-05,
+      "loss": 0.9109,
+      "step": 258370
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 3.0358584550703008e-05,
+      "loss": 0.8201,
+      "step": 258380
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 3.0357822901275002e-05,
+      "loss": 0.8864,
+      "step": 258390
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 3.0357061251847003e-05,
+      "loss": 0.7605,
+      "step": 258400
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 3.0356299602419004e-05,
+      "loss": 0.8015,
+      "step": 258410
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 3.0355537952990998e-05,
+      "loss": 0.8708,
+      "step": 258420
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 3.0354776303563e-05,
+      "loss": 0.879,
+      "step": 258430
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 3.0354014654134993e-05,
+      "loss": 0.8489,
+      "step": 258440
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 3.0353253004706994e-05,
+      "loss": 0.8683,
+      "step": 258450
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 3.0352491355278994e-05,
+      "loss": 0.8734,
+      "step": 258460
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 3.035172970585099e-05,
+      "loss": 0.8395,
+      "step": 258470
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 3.035096805642299e-05,
+      "loss": 0.819,
+      "step": 258480
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 3.035020640699499e-05,
+      "loss": 0.8339,
+      "step": 258490
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 3.0349444757566988e-05,
+      "loss": 0.8582,
+      "step": 258500
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 3.0348683108138985e-05,
+      "loss": 0.8658,
+      "step": 258510
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 3.0347921458710986e-05,
+      "loss": 0.8595,
+      "step": 258520
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 3.0347159809282983e-05,
+      "loss": 0.7372,
+      "step": 258530
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 3.0346398159854984e-05,
+      "loss": 0.8048,
+      "step": 258540
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 3.0345636510426985e-05,
+      "loss": 0.8304,
+      "step": 258550
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 3.034487486099898e-05,
+      "loss": 0.8382,
+      "step": 258560
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 3.034411321157098e-05,
+      "loss": 0.8021,
+      "step": 258570
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 3.034335156214298e-05,
+      "loss": 0.8735,
+      "step": 258580
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 3.0342589912714975e-05,
+      "loss": 0.8342,
+      "step": 258590
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 3.0341828263286976e-05,
+      "loss": 0.8663,
+      "step": 258600
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 3.0341066613858977e-05,
+      "loss": 0.7762,
+      "step": 258610
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 3.034030496443097e-05,
+      "loss": 0.7887,
+      "step": 258620
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 3.0339543315002972e-05,
+      "loss": 0.9241,
+      "step": 258630
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 3.0338781665574973e-05,
+      "loss": 0.7874,
+      "step": 258640
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 3.0338020016146967e-05,
+      "loss": 0.8426,
+      "step": 258650
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 3.0337258366718968e-05,
+      "loss": 0.8269,
+      "step": 258660
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 3.033649671729097e-05,
+      "loss": 0.8389,
+      "step": 258670
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 3.0335735067862962e-05,
+      "loss": 0.7742,
+      "step": 258680
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 3.0334973418434963e-05,
+      "loss": 0.799,
+      "step": 258690
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 3.0334211769006964e-05,
+      "loss": 0.8611,
+      "step": 258700
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 3.033345011957896e-05,
+      "loss": 0.7778,
+      "step": 258710
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 3.033268847015096e-05,
+      "loss": 0.8903,
+      "step": 258720
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 3.033192682072296e-05,
+      "loss": 0.8027,
+      "step": 258730
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 3.0331165171294957e-05,
+      "loss": 0.8545,
+      "step": 258740
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 3.0330403521866958e-05,
+      "loss": 0.7276,
+      "step": 258750
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 3.032964187243896e-05,
+      "loss": 0.7625,
+      "step": 258760
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 3.0328880223010953e-05,
+      "loss": 0.8197,
+      "step": 258770
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 3.0328118573582954e-05,
+      "loss": 0.7987,
+      "step": 258780
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 3.0327356924154955e-05,
+      "loss": 0.9335,
+      "step": 258790
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 3.032659527472695e-05,
+      "loss": 0.8618,
+      "step": 258800
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 3.032583362529895e-05,
+      "loss": 0.7863,
+      "step": 258810
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 3.032507197587095e-05,
+      "loss": 0.8386,
+      "step": 258820
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 3.0324310326442945e-05,
+      "loss": 0.883,
+      "step": 258830
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 3.0323548677014946e-05,
+      "loss": 0.8044,
+      "step": 258840
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 3.0322787027586946e-05,
+      "loss": 0.7671,
+      "step": 258850
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 3.032202537815894e-05,
+      "loss": 0.9155,
+      "step": 258860
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 3.032126372873094e-05,
+      "loss": 0.7663,
+      "step": 258870
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 3.0320502079302942e-05,
+      "loss": 0.8325,
+      "step": 258880
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 3.0319740429874936e-05,
+      "loss": 0.7977,
+      "step": 258890
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 3.0318978780446937e-05,
+      "loss": 0.7981,
+      "step": 258900
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 3.0318217131018935e-05,
+      "loss": 0.8359,
+      "step": 258910
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 3.0317455481590936e-05,
+      "loss": 0.7872,
+      "step": 258920
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 3.0316693832162933e-05,
+      "loss": 0.8867,
+      "step": 258930
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 3.031593218273493e-05,
+      "loss": 0.7843,
+      "step": 258940
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 3.031517053330693e-05,
+      "loss": 0.79,
+      "step": 258950
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 3.0314408883878932e-05,
+      "loss": 0.7609,
+      "step": 258960
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 3.0313647234450926e-05,
+      "loss": 0.8861,
+      "step": 258970
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 3.0312885585022927e-05,
+      "loss": 0.8195,
+      "step": 258980
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 3.0312123935594928e-05,
+      "loss": 0.7517,
+      "step": 258990
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 3.0311362286166922e-05,
+      "loss": 0.7835,
+      "step": 259000
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 3.0310600636738923e-05,
+      "loss": 0.8034,
+      "step": 259010
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 3.0309838987310924e-05,
+      "loss": 0.857,
+      "step": 259020
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 3.0309077337882918e-05,
+      "loss": 0.8175,
+      "step": 259030
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 3.030831568845492e-05,
+      "loss": 0.9029,
+      "step": 259040
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 3.030755403902692e-05,
+      "loss": 0.8481,
+      "step": 259050
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 3.0306792389598914e-05,
+      "loss": 0.8529,
+      "step": 259060
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 3.0306030740170914e-05,
+      "loss": 0.8206,
+      "step": 259070
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 3.0305269090742915e-05,
+      "loss": 0.776,
+      "step": 259080
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 3.030450744131491e-05,
+      "loss": 0.8454,
+      "step": 259090
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 3.030374579188691e-05,
+      "loss": 0.8099,
+      "step": 259100
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 3.030298414245891e-05,
+      "loss": 0.8471,
+      "step": 259110
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 3.030222249303091e-05,
+      "loss": 0.7952,
+      "step": 259120
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 3.030146084360291e-05,
+      "loss": 0.9179,
+      "step": 259130
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 3.0300699194174907e-05,
+      "loss": 0.8678,
+      "step": 259140
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 3.0299937544746904e-05,
+      "loss": 0.8271,
+      "step": 259150
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 3.0299175895318905e-05,
+      "loss": 0.8478,
+      "step": 259160
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 3.0298414245890906e-05,
+      "loss": 0.8784,
+      "step": 259170
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 3.02976525964629e-05,
+      "loss": 0.7755,
+      "step": 259180
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 3.02968909470349e-05,
+      "loss": 0.7952,
+      "step": 259190
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 3.0296129297606902e-05,
+      "loss": 0.832,
+      "step": 259200
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 3.0295367648178896e-05,
+      "loss": 0.8179,
+      "step": 259210
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 3.0294605998750897e-05,
+      "loss": 0.8326,
+      "step": 259220
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 3.0293844349322898e-05,
+      "loss": 0.8574,
+      "step": 259230
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 3.0293082699894892e-05,
+      "loss": 0.7771,
+      "step": 259240
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 3.0292321050466893e-05,
+      "loss": 0.808,
+      "step": 259250
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 3.0291559401038893e-05,
+      "loss": 0.8118,
+      "step": 259260
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 3.0290797751610888e-05,
+      "loss": 0.9582,
+      "step": 259270
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 3.029003610218289e-05,
+      "loss": 0.8236,
+      "step": 259280
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 3.028927445275489e-05,
+      "loss": 0.8403,
+      "step": 259290
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 3.0288512803326887e-05,
+      "loss": 0.8215,
+      "step": 259300
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 3.0287751153898884e-05,
+      "loss": 0.818,
+      "step": 259310
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 3.0286989504470885e-05,
+      "loss": 0.8877,
+      "step": 259320
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 3.0286227855042882e-05,
+      "loss": 0.7691,
+      "step": 259330
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 3.0285466205614883e-05,
+      "loss": 0.7845,
+      "step": 259340
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 3.0284704556186884e-05,
+      "loss": 0.7932,
+      "step": 259350
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 3.0283942906758878e-05,
+      "loss": 0.841,
+      "step": 259360
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 3.028318125733088e-05,
+      "loss": 0.891,
+      "step": 259370
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 3.0282419607902873e-05,
+      "loss": 0.8065,
+      "step": 259380
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 3.0281657958474874e-05,
+      "loss": 0.837,
+      "step": 259390
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 3.0280896309046875e-05,
+      "loss": 0.811,
+      "step": 259400
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 3.028013465961887e-05,
+      "loss": 0.8603,
+      "step": 259410
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 3.027937301019087e-05,
+      "loss": 0.745,
+      "step": 259420
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 3.027861136076287e-05,
+      "loss": 0.9292,
+      "step": 259430
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 3.0277849711334865e-05,
+      "loss": 0.829,
+      "step": 259440
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 3.0277088061906866e-05,
+      "loss": 0.7846,
+      "step": 259450
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 3.0276326412478866e-05,
+      "loss": 0.8872,
+      "step": 259460
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 3.027556476305086e-05,
+      "loss": 0.7772,
+      "step": 259470
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 3.027480311362286e-05,
+      "loss": 0.7288,
+      "step": 259480
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 3.0274041464194862e-05,
+      "loss": 0.8832,
+      "step": 259490
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 3.027327981476686e-05,
+      "loss": 0.749,
+      "step": 259500
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 3.027251816533886e-05,
+      "loss": 0.833,
+      "step": 259510
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 3.0271756515910858e-05,
+      "loss": 0.8536,
+      "step": 259520
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 3.0270994866482856e-05,
+      "loss": 0.7935,
+      "step": 259530
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 3.0270233217054856e-05,
+      "loss": 0.871,
+      "step": 259540
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 3.0269471567626857e-05,
+      "loss": 0.7609,
+      "step": 259550
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 3.026870991819885e-05,
+      "loss": 0.8433,
+      "step": 259560
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 3.0267948268770852e-05,
+      "loss": 0.8211,
+      "step": 259570
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 3.0267186619342853e-05,
+      "loss": 0.8014,
+      "step": 259580
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 3.0266424969914847e-05,
+      "loss": 0.8897,
+      "step": 259590
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 3.0265663320486848e-05,
+      "loss": 0.7661,
+      "step": 259600
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 3.026490167105885e-05,
+      "loss": 0.835,
+      "step": 259610
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 3.0264140021630843e-05,
+      "loss": 0.8709,
+      "step": 259620
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 3.0263378372202844e-05,
+      "loss": 0.774,
+      "step": 259630
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 3.0262616722774845e-05,
+      "loss": 0.6935,
+      "step": 259640
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 3.026185507334684e-05,
+      "loss": 0.8337,
+      "step": 259650
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 3.026109342391884e-05,
+      "loss": 0.7978,
+      "step": 259660
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 3.026033177449084e-05,
+      "loss": 0.9515,
+      "step": 259670
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 3.0259570125062834e-05,
+      "loss": 0.8831,
+      "step": 259680
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 3.0258808475634835e-05,
+      "loss": 0.7733,
+      "step": 259690
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 3.0258046826206836e-05,
+      "loss": 0.8423,
+      "step": 259700
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 3.0257285176778834e-05,
+      "loss": 0.8685,
+      "step": 259710
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 3.0256523527350835e-05,
+      "loss": 0.8378,
+      "step": 259720
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 3.0255761877922832e-05,
+      "loss": 0.8561,
+      "step": 259730
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 3.025500022849483e-05,
+      "loss": 0.7592,
+      "step": 259740
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 3.025423857906683e-05,
+      "loss": 0.84,
+      "step": 259750
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 3.025347692963883e-05,
+      "loss": 0.9056,
+      "step": 259760
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 3.0252715280210825e-05,
+      "loss": 0.9056,
+      "step": 259770
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 3.0251953630782826e-05,
+      "loss": 0.818,
+      "step": 259780
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 3.0251191981354827e-05,
+      "loss": 0.8519,
+      "step": 259790
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 3.025043033192682e-05,
+      "loss": 0.8436,
+      "step": 259800
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 3.0249668682498822e-05,
+      "loss": 0.7995,
+      "step": 259810
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 3.0248907033070816e-05,
+      "loss": 0.8635,
+      "step": 259820
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 3.0248145383642817e-05,
+      "loss": 0.892,
+      "step": 259830
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 3.0247383734214818e-05,
+      "loss": 0.9296,
+      "step": 259840
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 3.0246622084786812e-05,
+      "loss": 0.8098,
+      "step": 259850
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 3.0245860435358813e-05,
+      "loss": 0.7781,
+      "step": 259860
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 3.0245098785930813e-05,
+      "loss": 0.7876,
+      "step": 259870
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 3.024433713650281e-05,
+      "loss": 0.8007,
+      "step": 259880
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 3.024357548707481e-05,
+      "loss": 0.8701,
+      "step": 259890
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 3.024281383764681e-05,
+      "loss": 0.8323,
+      "step": 259900
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 3.0242052188218807e-05,
+      "loss": 0.866,
+      "step": 259910
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 3.0241290538790808e-05,
+      "loss": 0.8404,
+      "step": 259920
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 3.024052888936281e-05,
+      "loss": 0.8441,
+      "step": 259930
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 3.0239767239934802e-05,
+      "loss": 0.7898,
+      "step": 259940
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 3.0239005590506803e-05,
+      "loss": 0.8423,
+      "step": 259950
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 3.0238243941078804e-05,
+      "loss": 0.9191,
+      "step": 259960
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 3.0237482291650798e-05,
+      "loss": 0.8711,
+      "step": 259970
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 3.02367206422228e-05,
+      "loss": 0.927,
+      "step": 259980
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 3.02359589927948e-05,
+      "loss": 0.8294,
+      "step": 259990
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 3.0235197343366794e-05,
+      "loss": 0.7826,
+      "step": 260000
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 3.0234435693938795e-05,
+      "loss": 0.8367,
+      "step": 260010
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 3.0233674044510796e-05,
+      "loss": 0.8922,
+      "step": 260020
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 3.023291239508279e-05,
+      "loss": 0.8252,
+      "step": 260030
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 3.023215074565479e-05,
+      "loss": 0.853,
+      "step": 260040
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 3.023138909622679e-05,
+      "loss": 0.8208,
+      "step": 260050
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 3.0230627446798786e-05,
+      "loss": 0.7893,
+      "step": 260060
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 3.0229865797370787e-05,
+      "loss": 0.7918,
+      "step": 260070
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 3.0229104147942787e-05,
+      "loss": 0.8082,
+      "step": 260080
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 3.0228342498514785e-05,
+      "loss": 0.8075,
+      "step": 260090
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 3.0227580849086786e-05,
+      "loss": 0.9092,
+      "step": 260100
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 3.0226819199658783e-05,
+      "loss": 0.7556,
+      "step": 260110
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 3.022605755023078e-05,
+      "loss": 0.88,
+      "step": 260120
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 3.022529590080278e-05,
+      "loss": 0.8579,
+      "step": 260130
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 3.0224534251374782e-05,
+      "loss": 0.834,
+      "step": 260140
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 3.0223772601946776e-05,
+      "loss": 0.7778,
+      "step": 260150
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 3.0223010952518777e-05,
+      "loss": 0.8207,
+      "step": 260160
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 3.0222249303090778e-05,
+      "loss": 0.7533,
+      "step": 260170
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 3.0221487653662772e-05,
+      "loss": 0.8969,
+      "step": 260180
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 3.0220726004234773e-05,
+      "loss": 0.8355,
+      "step": 260190
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 3.0219964354806774e-05,
+      "loss": 0.7972,
+      "step": 260200
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 3.0219202705378768e-05,
+      "loss": 0.8638,
+      "step": 260210
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 3.021844105595077e-05,
+      "loss": 0.7121,
+      "step": 260220
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 3.021767940652277e-05,
+      "loss": 0.8684,
+      "step": 260230
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 3.0216917757094764e-05,
+      "loss": 0.8171,
+      "step": 260240
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 3.0216156107666765e-05,
+      "loss": 0.8103,
+      "step": 260250
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 3.0215394458238765e-05,
+      "loss": 0.8556,
+      "step": 260260
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 3.021463280881076e-05,
+      "loss": 0.8277,
+      "step": 260270
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 3.021387115938276e-05,
+      "loss": 0.7712,
+      "step": 260280
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 3.0213109509954758e-05,
+      "loss": 0.828,
+      "step": 260290
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 3.021234786052676e-05,
+      "loss": 0.9927,
+      "step": 260300
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 3.021158621109876e-05,
+      "loss": 0.8441,
+      "step": 260310
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 3.0210824561670754e-05,
+      "loss": 0.8305,
+      "step": 260320
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 3.0210062912242755e-05,
+      "loss": 0.9,
+      "step": 260330
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 3.0209301262814755e-05,
+      "loss": 0.8217,
+      "step": 260340
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 3.020853961338675e-05,
+      "loss": 0.7794,
+      "step": 260350
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 3.020777796395875e-05,
+      "loss": 0.8598,
+      "step": 260360
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 3.020701631453075e-05,
+      "loss": 0.8224,
+      "step": 260370
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 3.0206254665102745e-05,
+      "loss": 0.7949,
+      "step": 260380
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 3.0205493015674746e-05,
+      "loss": 0.7735,
+      "step": 260390
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 3.0204731366246747e-05,
+      "loss": 0.8531,
+      "step": 260400
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 3.020396971681874e-05,
+      "loss": 0.8424,
+      "step": 260410
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 3.0203208067390742e-05,
+      "loss": 0.7925,
+      "step": 260420
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 3.0202446417962743e-05,
+      "loss": 0.8429,
+      "step": 260430
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 3.0201684768534737e-05,
+      "loss": 0.9232,
+      "step": 260440
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 3.0200923119106738e-05,
+      "loss": 0.7899,
+      "step": 260450
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 3.020016146967874e-05,
+      "loss": 0.9389,
+      "step": 260460
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 3.0199399820250736e-05,
+      "loss": 0.8971,
+      "step": 260470
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 3.0198638170822733e-05,
+      "loss": 0.824,
+      "step": 260480
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 3.0197876521394734e-05,
+      "loss": 0.8034,
+      "step": 260490
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 3.0197114871966732e-05,
+      "loss": 0.8295,
+      "step": 260500
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 3.0196353222538733e-05,
+      "loss": 0.7904,
+      "step": 260510
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 3.0195591573110733e-05,
+      "loss": 0.8067,
+      "step": 260520
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 3.0194829923682728e-05,
+      "loss": 0.8224,
+      "step": 260530
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 3.019406827425473e-05,
+      "loss": 0.8142,
+      "step": 260540
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 3.019330662482673e-05,
+      "loss": 0.8416,
+      "step": 260550
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 3.0192544975398723e-05,
+      "loss": 0.835,
+      "step": 260560
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 3.0191783325970724e-05,
+      "loss": 0.854,
+      "step": 260570
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 3.0191021676542725e-05,
+      "loss": 0.7794,
+      "step": 260580
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 3.019026002711472e-05,
+      "loss": 0.8978,
+      "step": 260590
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 3.018949837768672e-05,
+      "loss": 0.7962,
+      "step": 260600
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 3.018873672825872e-05,
+      "loss": 0.8255,
+      "step": 260610
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 3.0187975078830715e-05,
+      "loss": 0.7819,
+      "step": 260620
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 3.0187213429402716e-05,
+      "loss": 0.7314,
+      "step": 260630
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 3.0186451779974717e-05,
+      "loss": 0.8818,
+      "step": 260640
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 3.018569013054671e-05,
+      "loss": 0.8142,
+      "step": 260650
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 3.018492848111871e-05,
+      "loss": 0.9104,
+      "step": 260660
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 3.0184166831690712e-05,
+      "loss": 0.855,
+      "step": 260670
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 3.018340518226271e-05,
+      "loss": 0.9068,
+      "step": 260680
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 3.0182643532834707e-05,
+      "loss": 0.7563,
+      "step": 260690
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 3.0181881883406708e-05,
+      "loss": 1.0044,
+      "step": 260700
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 3.0181120233978706e-05,
+      "loss": 0.8665,
+      "step": 260710
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 3.0180358584550707e-05,
+      "loss": 0.8182,
+      "step": 260720
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 3.0179596935122707e-05,
+      "loss": 0.7781,
+      "step": 260730
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 3.01788352856947e-05,
+      "loss": 0.9134,
+      "step": 260740
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 3.0178073636266702e-05,
+      "loss": 0.7469,
+      "step": 260750
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 3.0177311986838696e-05,
+      "loss": 0.7976,
+      "step": 260760
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 3.0176550337410697e-05,
+      "loss": 0.7882,
+      "step": 260770
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 3.0175788687982698e-05,
+      "loss": 0.8314,
+      "step": 260780
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 3.0175027038554692e-05,
+      "loss": 0.8354,
+      "step": 260790
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 3.0174265389126693e-05,
+      "loss": 0.8045,
+      "step": 260800
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 3.0173503739698694e-05,
+      "loss": 0.7774,
+      "step": 260810
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 3.0172742090270688e-05,
+      "loss": 0.8202,
+      "step": 260820
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 3.017198044084269e-05,
+      "loss": 0.7598,
+      "step": 260830
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 3.017121879141469e-05,
+      "loss": 0.7163,
+      "step": 260840
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 3.0170457141986687e-05,
+      "loss": 0.8475,
+      "step": 260850
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 3.0169695492558685e-05,
+      "loss": 0.9251,
+      "step": 260860
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 3.0168933843130685e-05,
+      "loss": 0.9491,
+      "step": 260870
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 3.0168172193702683e-05,
+      "loss": 0.8881,
+      "step": 260880
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 3.0167410544274684e-05,
+      "loss": 0.8048,
+      "step": 260890
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 3.0166648894846685e-05,
+      "loss": 0.9304,
+      "step": 260900
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 3.016588724541868e-05,
+      "loss": 0.8363,
+      "step": 260910
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 3.016512559599068e-05,
+      "loss": 0.8176,
+      "step": 260920
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 3.016436394656268e-05,
+      "loss": 0.8857,
+      "step": 260930
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 3.0163602297134675e-05,
+      "loss": 0.819,
+      "step": 260940
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 3.0162840647706675e-05,
+      "loss": 0.8407,
+      "step": 260950
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 3.0162078998278676e-05,
+      "loss": 0.9146,
+      "step": 260960
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 3.016131734885067e-05,
+      "loss": 0.8971,
+      "step": 260970
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 3.016055569942267e-05,
+      "loss": 0.7339,
+      "step": 260980
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 3.0159794049994672e-05,
+      "loss": 0.7673,
+      "step": 260990
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 3.0159032400566666e-05,
+      "loss": 0.7408,
+      "step": 261000
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 3.0158270751138667e-05,
+      "loss": 0.826,
+      "step": 261010
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 3.0157509101710668e-05,
+      "loss": 0.8157,
+      "step": 261020
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 3.0156747452282662e-05,
+      "loss": 0.7769,
+      "step": 261030
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 3.0155985802854663e-05,
+      "loss": 0.8504,
+      "step": 261040
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 3.0155224153426664e-05,
+      "loss": 0.7874,
+      "step": 261050
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 3.015446250399866e-05,
+      "loss": 0.8021,
+      "step": 261060
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 3.015370085457066e-05,
+      "loss": 0.8686,
+      "step": 261070
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 3.015293920514266e-05,
+      "loss": 0.7661,
+      "step": 261080
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 3.0152177555714657e-05,
+      "loss": 0.8156,
+      "step": 261090
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 3.0151415906286658e-05,
+      "loss": 0.863,
+      "step": 261100
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 3.015065425685866e-05,
+      "loss": 0.8161,
+      "step": 261110
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 3.0149892607430653e-05,
+      "loss": 0.815,
+      "step": 261120
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 3.0149130958002654e-05,
+      "loss": 0.7246,
+      "step": 261130
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 3.0148369308574654e-05,
+      "loss": 0.7075,
+      "step": 261140
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 3.014760765914665e-05,
+      "loss": 0.8889,
+      "step": 261150
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 3.014684600971865e-05,
+      "loss": 0.6811,
+      "step": 261160
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 3.014608436029065e-05,
+      "loss": 0.8193,
+      "step": 261170
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 3.0145322710862644e-05,
+      "loss": 0.8251,
+      "step": 261180
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 3.0144561061434645e-05,
+      "loss": 0.924,
+      "step": 261190
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 3.014379941200664e-05,
+      "loss": 0.7699,
+      "step": 261200
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 3.014303776257864e-05,
+      "loss": 0.8063,
+      "step": 261210
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 3.014227611315064e-05,
+      "loss": 0.7669,
+      "step": 261220
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 3.0141514463722635e-05,
+      "loss": 0.8458,
+      "step": 261230
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 3.0140752814294636e-05,
+      "loss": 0.7396,
+      "step": 261240
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 3.0139991164866637e-05,
+      "loss": 0.7576,
+      "step": 261250
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 3.0139229515438634e-05,
+      "loss": 0.7844,
+      "step": 261260
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 3.0138467866010635e-05,
+      "loss": 0.8079,
+      "step": 261270
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 3.0137706216582632e-05,
+      "loss": 0.8358,
+      "step": 261280
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 3.013694456715463e-05,
+      "loss": 0.7894,
+      "step": 261290
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 3.013618291772663e-05,
+      "loss": 0.7717,
+      "step": 261300
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 3.013542126829863e-05,
+      "loss": 0.8414,
+      "step": 261310
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 3.0134659618870626e-05,
+      "loss": 0.8439,
+      "step": 261320
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 3.0133897969442627e-05,
+      "loss": 0.8327,
+      "step": 261330
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 3.0133136320014627e-05,
+      "loss": 0.8366,
+      "step": 261340
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 3.013237467058662e-05,
+      "loss": 0.8842,
+      "step": 261350
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 3.0131613021158622e-05,
+      "loss": 0.9168,
+      "step": 261360
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 3.0130851371730623e-05,
+      "loss": 0.7538,
+      "step": 261370
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 3.0130089722302617e-05,
+      "loss": 0.9188,
+      "step": 261380
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 3.0129328072874618e-05,
+      "loss": 0.687,
+      "step": 261390
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 3.012856642344662e-05,
+      "loss": 0.7974,
+      "step": 261400
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 3.0127804774018613e-05,
+      "loss": 0.7935,
+      "step": 261410
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 3.0127043124590614e-05,
+      "loss": 0.8945,
+      "step": 261420
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 3.0126281475162615e-05,
+      "loss": 0.8628,
+      "step": 261430
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 3.012551982573461e-05,
+      "loss": 0.9564,
+      "step": 261440
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 3.012475817630661e-05,
+      "loss": 0.892,
+      "step": 261450
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 3.012399652687861e-05,
+      "loss": 0.8314,
+      "step": 261460
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 3.0123234877450608e-05,
+      "loss": 0.7267,
+      "step": 261470
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 3.012247322802261e-05,
+      "loss": 0.8794,
+      "step": 261480
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 3.0121711578594606e-05,
+      "loss": 0.7901,
+      "step": 261490
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 3.0120949929166604e-05,
+      "loss": 0.8405,
+      "step": 261500
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 3.0120188279738605e-05,
+      "loss": 0.9028,
+      "step": 261510
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 3.0119426630310606e-05,
+      "loss": 0.8438,
+      "step": 261520
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 3.01186649808826e-05,
+      "loss": 0.833,
+      "step": 261530
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 3.01179033314546e-05,
+      "loss": 0.8128,
+      "step": 261540
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 3.01171416820266e-05,
+      "loss": 0.8606,
+      "step": 261550
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 3.0116380032598595e-05,
+      "loss": 0.8485,
+      "step": 261560
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 3.0115618383170596e-05,
+      "loss": 0.8593,
+      "step": 261570
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 3.0114856733742597e-05,
+      "loss": 0.8585,
+      "step": 261580
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 3.011409508431459e-05,
+      "loss": 0.765,
+      "step": 261590
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 3.0113333434886592e-05,
+      "loss": 0.869,
+      "step": 261600
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 3.0112571785458593e-05,
+      "loss": 0.8031,
+      "step": 261610
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 3.0111810136030587e-05,
+      "loss": 0.7399,
+      "step": 261620
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 3.0111048486602588e-05,
+      "loss": 0.9378,
+      "step": 261630
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 3.011028683717459e-05,
+      "loss": 0.8322,
+      "step": 261640
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 3.0109525187746586e-05,
+      "loss": 0.8305,
+      "step": 261650
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 3.0108763538318584e-05,
+      "loss": 0.812,
+      "step": 261660
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 3.010800188889058e-05,
+      "loss": 0.8449,
+      "step": 261670
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 3.0107240239462582e-05,
+      "loss": 0.8589,
+      "step": 261680
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 3.0106478590034583e-05,
+      "loss": 0.8558,
+      "step": 261690
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 3.0105716940606577e-05,
+      "loss": 0.8249,
+      "step": 261700
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 3.0104955291178578e-05,
+      "loss": 0.7858,
+      "step": 261710
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 3.010419364175058e-05,
+      "loss": 0.8937,
+      "step": 261720
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 3.0103431992322573e-05,
+      "loss": 0.8227,
+      "step": 261730
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 3.0102670342894574e-05,
+      "loss": 0.8407,
+      "step": 261740
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 3.0101908693466574e-05,
+      "loss": 0.849,
+      "step": 261750
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 3.010114704403857e-05,
+      "loss": 0.8502,
+      "step": 261760
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 3.010038539461057e-05,
+      "loss": 0.8362,
+      "step": 261770
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 3.009962374518257e-05,
+      "loss": 0.9271,
+      "step": 261780
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 3.0098862095754564e-05,
+      "loss": 0.7652,
+      "step": 261790
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 3.0098100446326565e-05,
+      "loss": 0.8571,
+      "step": 261800
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 3.0097338796898566e-05,
+      "loss": 0.853,
+      "step": 261810
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 3.009657714747056e-05,
+      "loss": 0.8357,
+      "step": 261820
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 3.009581549804256e-05,
+      "loss": 0.7457,
+      "step": 261830
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 3.0095053848614562e-05,
+      "loss": 0.8915,
+      "step": 261840
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 3.009429219918656e-05,
+      "loss": 0.8838,
+      "step": 261850
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 3.009353054975856e-05,
+      "loss": 0.761,
+      "step": 261860
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 3.0092768900330558e-05,
+      "loss": 0.8078,
+      "step": 261870
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 3.0092007250902555e-05,
+      "loss": 0.7693,
+      "step": 261880
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 3.0091245601474556e-05,
+      "loss": 0.8114,
+      "step": 261890
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 3.0090483952046557e-05,
+      "loss": 0.8077,
+      "step": 261900
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 3.008972230261855e-05,
+      "loss": 1.0044,
+      "step": 261910
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 3.008896065319055e-05,
+      "loss": 0.8594,
+      "step": 261920
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 3.0088199003762552e-05,
+      "loss": 0.8471,
+      "step": 261930
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 3.0087437354334547e-05,
+      "loss": 0.8266,
+      "step": 261940
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 3.0086675704906547e-05,
+      "loss": 0.8274,
+      "step": 261950
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 3.0085914055478548e-05,
+      "loss": 0.8202,
+      "step": 261960
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 3.0085152406050542e-05,
+      "loss": 0.8268,
+      "step": 261970
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 3.0084390756622543e-05,
+      "loss": 0.8171,
+      "step": 261980
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 3.0083629107194544e-05,
+      "loss": 0.8256,
+      "step": 261990
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 3.0082867457766538e-05,
+      "loss": 0.7504,
+      "step": 262000
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 3.008210580833854e-05,
+      "loss": 0.8446,
+      "step": 262010
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 3.008134415891054e-05,
+      "loss": 0.9241,
+      "step": 262020
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 3.0080582509482534e-05,
+      "loss": 0.8257,
+      "step": 262030
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 3.0079820860054535e-05,
+      "loss": 0.8111,
+      "step": 262040
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 3.0079059210626536e-05,
+      "loss": 0.7639,
+      "step": 262050
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 3.0078297561198533e-05,
+      "loss": 0.7989,
+      "step": 262060
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 3.0077535911770534e-05,
+      "loss": 0.8749,
+      "step": 262070
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 3.007677426234253e-05,
+      "loss": 0.7547,
+      "step": 262080
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 3.007601261291453e-05,
+      "loss": 0.7705,
+      "step": 262090
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 3.007525096348653e-05,
+      "loss": 0.7912,
+      "step": 262100
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 3.007448931405853e-05,
+      "loss": 0.7728,
+      "step": 262110
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 3.0073727664630525e-05,
+      "loss": 0.9685,
+      "step": 262120
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 3.0072966015202526e-05,
+      "loss": 0.7968,
+      "step": 262130
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 3.007220436577452e-05,
+      "loss": 0.8172,
+      "step": 262140
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 3.007144271634652e-05,
+      "loss": 0.8585,
+      "step": 262150
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 3.007068106691852e-05,
+      "loss": 0.8353,
+      "step": 262160
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 3.0069919417490515e-05,
+      "loss": 0.8713,
+      "step": 262170
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 3.0069157768062516e-05,
+      "loss": 0.8612,
+      "step": 262180
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 3.0068396118634517e-05,
+      "loss": 0.8821,
+      "step": 262190
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 3.006763446920651e-05,
+      "loss": 0.8902,
+      "step": 262200
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 3.0066872819778512e-05,
+      "loss": 0.8095,
+      "step": 262210
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 3.0066111170350513e-05,
+      "loss": 0.8275,
+      "step": 262220
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 3.006534952092251e-05,
+      "loss": 0.7689,
+      "step": 262230
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 3.0064587871494508e-05,
+      "loss": 0.7923,
+      "step": 262240
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 3.006382622206651e-05,
+      "loss": 0.8118,
+      "step": 262250
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 3.0063064572638506e-05,
+      "loss": 0.88,
+      "step": 262260
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 3.0062302923210507e-05,
+      "loss": 0.7812,
+      "step": 262270
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 3.0061541273782508e-05,
+      "loss": 0.8968,
+      "step": 262280
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 3.0060779624354502e-05,
+      "loss": 0.7804,
+      "step": 262290
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 3.0060017974926503e-05,
+      "loss": 0.8547,
+      "step": 262300
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 3.0059256325498504e-05,
+      "loss": 0.805,
+      "step": 262310
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 3.0058494676070498e-05,
+      "loss": 0.8695,
+      "step": 262320
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 3.00577330266425e-05,
+      "loss": 0.861,
+      "step": 262330
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 3.00569713772145e-05,
+      "loss": 0.8312,
+      "step": 262340
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 3.0056209727786494e-05,
+      "loss": 0.8238,
+      "step": 262350
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 3.0055448078358494e-05,
+      "loss": 0.8524,
+      "step": 262360
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 3.0054686428930495e-05,
+      "loss": 0.8418,
+      "step": 262370
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 3.005392477950249e-05,
+      "loss": 0.7882,
+      "step": 262380
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 3.005316313007449e-05,
+      "loss": 0.7886,
+      "step": 262390
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 3.005240148064649e-05,
+      "loss": 0.8249,
+      "step": 262400
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 3.0051639831218485e-05,
+      "loss": 0.8283,
+      "step": 262410
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 3.0050878181790486e-05,
+      "loss": 0.7923,
+      "step": 262420
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 3.0050116532362487e-05,
+      "loss": 0.8887,
+      "step": 262430
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 3.0049354882934484e-05,
+      "loss": 0.7528,
+      "step": 262440
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 3.0048593233506485e-05,
+      "loss": 0.8527,
+      "step": 262450
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 3.0047831584078483e-05,
+      "loss": 0.8748,
+      "step": 262460
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 3.004706993465048e-05,
+      "loss": 0.8551,
+      "step": 262470
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 3.004630828522248e-05,
+      "loss": 0.8124,
+      "step": 262480
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 3.0045546635794482e-05,
+      "loss": 0.8385,
+      "step": 262490
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 3.0044784986366476e-05,
+      "loss": 0.7471,
+      "step": 262500
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 3.0044023336938477e-05,
+      "loss": 0.7949,
+      "step": 262510
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 3.0043261687510478e-05,
+      "loss": 0.861,
+      "step": 262520
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 3.004250003808247e-05,
+      "loss": 0.855,
+      "step": 262530
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 3.0041738388654472e-05,
+      "loss": 0.7392,
+      "step": 262540
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 3.0040976739226473e-05,
+      "loss": 0.8675,
+      "step": 262550
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 3.0040215089798467e-05,
+      "loss": 0.7791,
+      "step": 262560
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 3.0039453440370468e-05,
+      "loss": 0.833,
+      "step": 262570
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 3.003869179094247e-05,
+      "loss": 0.8935,
+      "step": 262580
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 3.0037930141514463e-05,
+      "loss": 0.9235,
+      "step": 262590
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 3.0037168492086464e-05,
+      "loss": 0.8697,
+      "step": 262600
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 3.003640684265846e-05,
+      "loss": 0.8883,
+      "step": 262610
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 3.003564519323046e-05,
+      "loss": 0.7412,
+      "step": 262620
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 3.003488354380246e-05,
+      "loss": 0.972,
+      "step": 262630
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 3.0034121894374457e-05,
+      "loss": 0.8083,
+      "step": 262640
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 3.0033360244946458e-05,
+      "loss": 0.7995,
+      "step": 262650
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 3.003259859551846e-05,
+      "loss": 0.8871,
+      "step": 262660
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 3.0031836946090453e-05,
+      "loss": 0.7741,
+      "step": 262670
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 3.0031075296662454e-05,
+      "loss": 0.8385,
+      "step": 262680
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 3.0030313647234455e-05,
+      "loss": 0.8094,
+      "step": 262690
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 3.002955199780645e-05,
+      "loss": 0.8271,
+      "step": 262700
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 3.002879034837845e-05,
+      "loss": 0.7485,
+      "step": 262710
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 3.002802869895045e-05,
+      "loss": 0.9047,
+      "step": 262720
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 3.0027267049522445e-05,
+      "loss": 0.9081,
+      "step": 262730
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 3.0026505400094446e-05,
+      "loss": 0.7976,
+      "step": 262740
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 3.0025743750666446e-05,
+      "loss": 0.7743,
+      "step": 262750
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 3.002498210123844e-05,
+      "loss": 0.7775,
+      "step": 262760
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 3.002422045181044e-05,
+      "loss": 0.9015,
+      "step": 262770
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 3.0023458802382442e-05,
+      "loss": 0.8406,
+      "step": 262780
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 3.0022697152954436e-05,
+      "loss": 0.8384,
+      "step": 262790
+    },
+    {
+      "epoch": 12.0,
+      "eval_cer": 0.6203951637631656,
+      "eval_em": 0.007568590350047304,
+      "eval_f1": 0.007568590350047304,
+      "eval_loss": 0.8934775590896606,
+      "eval_runtime": 1007.4125,
+      "eval_samples_per_second": 10.492,
+      "eval_steps_per_second": 1.312,
+      "eval_wer": 0.9924314096499527,
+      "step": 262797
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 3.0021935503526437e-05,
+      "loss": 0.8089,
+      "step": 262800
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 3.0021173854098438e-05,
+      "loss": 0.77,
+      "step": 262810
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 3.0020412204670435e-05,
+      "loss": 0.7486,
+      "step": 262820
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 3.0019650555242433e-05,
+      "loss": 0.8641,
+      "step": 262830
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 3.0018888905814434e-05,
+      "loss": 0.8727,
+      "step": 262840
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 3.001812725638643e-05,
+      "loss": 0.8312,
+      "step": 262850
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 3.0017365606958432e-05,
+      "loss": 0.7515,
+      "step": 262860
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 3.0016603957530433e-05,
+      "loss": 0.7276,
+      "step": 262870
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 3.0015842308102427e-05,
+      "loss": 0.8326,
+      "step": 262880
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 3.0015080658674428e-05,
+      "loss": 0.8361,
+      "step": 262890
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 3.001431900924643e-05,
+      "loss": 0.7764,
+      "step": 262900
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 3.0013557359818423e-05,
+      "loss": 0.7775,
+      "step": 262910
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 3.0012795710390424e-05,
+      "loss": 0.7949,
+      "step": 262920
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 3.0012034060962425e-05,
+      "loss": 0.8347,
+      "step": 262930
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 3.001127241153442e-05,
+      "loss": 0.8623,
+      "step": 262940
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 3.001051076210642e-05,
+      "loss": 0.7264,
+      "step": 262950
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 3.000974911267842e-05,
+      "loss": 0.8205,
+      "step": 262960
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 3.0008987463250414e-05,
+      "loss": 0.76,
+      "step": 262970
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 3.0008225813822415e-05,
+      "loss": 0.7907,
+      "step": 262980
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 3.0007464164394416e-05,
+      "loss": 0.864,
+      "step": 262990
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 3.000670251496641e-05,
+      "loss": 0.8381,
+      "step": 263000
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 3.000594086553841e-05,
+      "loss": 0.7286,
+      "step": 263010
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 3.0005179216110412e-05,
+      "loss": 0.8286,
+      "step": 263020
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 3.000441756668241e-05,
+      "loss": 0.8181,
+      "step": 263030
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 3.0003655917254407e-05,
+      "loss": 0.8412,
+      "step": 263040
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 3.0002894267826404e-05,
+      "loss": 0.8307,
+      "step": 263050
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 3.0002132618398405e-05,
+      "loss": 0.8022,
+      "step": 263060
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 3.0001370968970406e-05,
+      "loss": 0.7473,
+      "step": 263070
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 3.00006093195424e-05,
+      "loss": 0.7941,
+      "step": 263080
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 2.99998476701144e-05,
+      "loss": 0.9181,
+      "step": 263090
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 2.9999086020686402e-05,
+      "loss": 0.8016,
+      "step": 263100
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 2.9998324371258396e-05,
+      "loss": 0.8527,
+      "step": 263110
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 2.9997562721830397e-05,
+      "loss": 0.7836,
+      "step": 263120
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 2.9996801072402398e-05,
+      "loss": 0.8303,
+      "step": 263130
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 2.999603942297439e-05,
+      "loss": 0.8089,
+      "step": 263140
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 2.9995277773546393e-05,
+      "loss": 0.8337,
+      "step": 263150
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 2.9994516124118393e-05,
+      "loss": 0.7909,
+      "step": 263160
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 2.9993754474690387e-05,
+      "loss": 0.7611,
+      "step": 263170
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 2.9992992825262388e-05,
+      "loss": 0.9041,
+      "step": 263180
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 2.999223117583439e-05,
+      "loss": 0.8366,
+      "step": 263190
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 2.9991469526406387e-05,
+      "loss": 0.8562,
+      "step": 263200
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 2.9990707876978384e-05,
+      "loss": 0.7907,
+      "step": 263210
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 2.9989946227550385e-05,
+      "loss": 0.7711,
+      "step": 263220
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 2.9989184578122382e-05,
+      "loss": 0.7458,
+      "step": 263230
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 2.9988422928694383e-05,
+      "loss": 0.7135,
+      "step": 263240
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 2.9987661279266384e-05,
+      "loss": 0.8358,
+      "step": 263250
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 2.9986899629838378e-05,
+      "loss": 0.7489,
+      "step": 263260
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 2.998613798041038e-05,
+      "loss": 0.7779,
+      "step": 263270
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 2.998537633098238e-05,
+      "loss": 0.7928,
+      "step": 263280
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 2.9984614681554374e-05,
+      "loss": 0.8468,
+      "step": 263290
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 2.9983853032126375e-05,
+      "loss": 0.8357,
+      "step": 263300
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 2.9983091382698376e-05,
+      "loss": 0.7251,
+      "step": 263310
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 2.998232973327037e-05,
+      "loss": 0.8522,
+      "step": 263320
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 2.998156808384237e-05,
+      "loss": 0.7051,
+      "step": 263330
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 2.998080643441437e-05,
+      "loss": 0.7916,
+      "step": 263340
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 2.9980044784986366e-05,
+      "loss": 0.7528,
+      "step": 263350
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 2.9979283135558366e-05,
+      "loss": 0.7955,
+      "step": 263360
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 2.9978521486130367e-05,
+      "loss": 0.73,
+      "step": 263370
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 2.997775983670236e-05,
+      "loss": 0.7406,
+      "step": 263380
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 2.9976998187274362e-05,
+      "loss": 0.7881,
+      "step": 263390
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 2.9976236537846363e-05,
+      "loss": 0.7497,
+      "step": 263400
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 2.997547488841836e-05,
+      "loss": 0.6968,
+      "step": 263410
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 2.9974713238990358e-05,
+      "loss": 0.8604,
+      "step": 263420
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 2.997395158956236e-05,
+      "loss": 0.7284,
+      "step": 263430
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 2.9973189940134356e-05,
+      "loss": 0.7578,
+      "step": 263440
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 2.9972428290706357e-05,
+      "loss": 0.7841,
+      "step": 263450
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 2.9971666641278358e-05,
+      "loss": 0.7401,
+      "step": 263460
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 2.9970904991850352e-05,
+      "loss": 0.8017,
+      "step": 263470
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 2.9970143342422353e-05,
+      "loss": 0.8569,
+      "step": 263480
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 2.9969381692994354e-05,
+      "loss": 0.8191,
+      "step": 263490
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 2.9968620043566348e-05,
+      "loss": 0.8003,
+      "step": 263500
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 2.996785839413835e-05,
+      "loss": 0.8478,
+      "step": 263510
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 2.9967096744710343e-05,
+      "loss": 0.8041,
+      "step": 263520
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 2.9966335095282344e-05,
+      "loss": 0.9471,
+      "step": 263530
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 2.9965573445854345e-05,
+      "loss": 0.7991,
+      "step": 263540
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 2.996481179642634e-05,
+      "loss": 0.8037,
+      "step": 263550
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 2.996405014699834e-05,
+      "loss": 0.7303,
+      "step": 263560
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 2.996328849757034e-05,
+      "loss": 0.764,
+      "step": 263570
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 2.9962526848142334e-05,
+      "loss": 0.8398,
+      "step": 263580
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 2.9961765198714335e-05,
+      "loss": 0.7091,
+      "step": 263590
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 2.9961003549286336e-05,
+      "loss": 0.7634,
+      "step": 263600
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 2.9960241899858334e-05,
+      "loss": 0.8176,
+      "step": 263610
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 2.9959480250430334e-05,
+      "loss": 0.8327,
+      "step": 263620
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 2.9958718601002332e-05,
+      "loss": 0.8694,
+      "step": 263630
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 2.995795695157433e-05,
+      "loss": 0.8999,
+      "step": 263640
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 2.995719530214633e-05,
+      "loss": 0.8728,
+      "step": 263650
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 2.995643365271833e-05,
+      "loss": 0.7762,
+      "step": 263660
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 2.9955672003290325e-05,
+      "loss": 0.8083,
+      "step": 263670
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 2.9954910353862326e-05,
+      "loss": 0.7244,
+      "step": 263680
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 2.9954148704434327e-05,
+      "loss": 0.7815,
+      "step": 263690
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 2.995338705500632e-05,
+      "loss": 0.8267,
+      "step": 263700
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 2.9952625405578322e-05,
+      "loss": 0.8833,
+      "step": 263710
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 2.9951863756150323e-05,
+      "loss": 0.7531,
+      "step": 263720
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 2.9951102106722317e-05,
+      "loss": 0.7904,
+      "step": 263730
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 2.9950340457294318e-05,
+      "loss": 0.7261,
+      "step": 263740
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 2.994957880786632e-05,
+      "loss": 0.901,
+      "step": 263750
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 2.9948817158438313e-05,
+      "loss": 0.8291,
+      "step": 263760
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 2.9948055509010313e-05,
+      "loss": 0.778,
+      "step": 263770
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 2.9947293859582314e-05,
+      "loss": 0.7468,
+      "step": 263780
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 2.994653221015431e-05,
+      "loss": 0.8927,
+      "step": 263790
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 2.994577056072631e-05,
+      "loss": 0.825,
+      "step": 263800
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 2.994500891129831e-05,
+      "loss": 0.8959,
+      "step": 263810
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 2.9944247261870307e-05,
+      "loss": 0.8488,
+      "step": 263820
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 2.994348561244231e-05,
+      "loss": 0.82,
+      "step": 263830
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 2.9942723963014306e-05,
+      "loss": 0.9774,
+      "step": 263840
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 2.9941962313586303e-05,
+      "loss": 0.7668,
+      "step": 263850
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 2.9941200664158304e-05,
+      "loss": 0.7496,
+      "step": 263860
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 2.9940439014730305e-05,
+      "loss": 0.8287,
+      "step": 263870
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 2.99396773653023e-05,
+      "loss": 0.883,
+      "step": 263880
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 2.99389157158743e-05,
+      "loss": 0.703,
+      "step": 263890
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 2.99381540664463e-05,
+      "loss": 0.8457,
+      "step": 263900
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 2.9937392417018295e-05,
+      "loss": 0.7775,
+      "step": 263910
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 2.9936630767590296e-05,
+      "loss": 0.8359,
+      "step": 263920
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 2.9935869118162297e-05,
+      "loss": 0.7887,
+      "step": 263930
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 2.993510746873429e-05,
+      "loss": 0.716,
+      "step": 263940
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 2.993434581930629e-05,
+      "loss": 0.8011,
+      "step": 263950
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 2.9933584169878292e-05,
+      "loss": 0.8855,
+      "step": 263960
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 2.9932822520450286e-05,
+      "loss": 0.7568,
+      "step": 263970
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 2.9932060871022287e-05,
+      "loss": 0.8659,
+      "step": 263980
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 2.9931299221594285e-05,
+      "loss": 0.7878,
+      "step": 263990
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 2.9930537572166286e-05,
+      "loss": 0.8243,
+      "step": 264000
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 2.9929775922738283e-05,
+      "loss": 0.7282,
+      "step": 264010
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 2.992901427331028e-05,
+      "loss": 0.7956,
+      "step": 264020
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 2.992825262388228e-05,
+      "loss": 0.7224,
+      "step": 264030
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 2.9927490974454282e-05,
+      "loss": 0.7715,
+      "step": 264040
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 2.9926729325026276e-05,
+      "loss": 0.8146,
+      "step": 264050
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 2.9925967675598277e-05,
+      "loss": 0.7593,
+      "step": 264060
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 2.9925206026170278e-05,
+      "loss": 0.8589,
+      "step": 264070
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 2.9924444376742272e-05,
+      "loss": 0.9236,
+      "step": 264080
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 2.9923682727314273e-05,
+      "loss": 0.744,
+      "step": 264090
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 2.9922921077886274e-05,
+      "loss": 0.791,
+      "step": 264100
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 2.9922159428458268e-05,
+      "loss": 0.844,
+      "step": 264110
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 2.992139777903027e-05,
+      "loss": 0.9335,
+      "step": 264120
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 2.992063612960227e-05,
+      "loss": 0.7955,
+      "step": 264130
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 2.9919874480174264e-05,
+      "loss": 0.7448,
+      "step": 264140
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 2.9919112830746265e-05,
+      "loss": 0.7757,
+      "step": 264150
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 2.9918351181318265e-05,
+      "loss": 0.8293,
+      "step": 264160
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 2.991758953189026e-05,
+      "loss": 0.8565,
+      "step": 264170
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 2.991682788246226e-05,
+      "loss": 0.8447,
+      "step": 264180
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 2.991606623303426e-05,
+      "loss": 0.8219,
+      "step": 264190
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 2.991530458360626e-05,
+      "loss": 0.7913,
+      "step": 264200
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 2.991454293417826e-05,
+      "loss": 0.8666,
+      "step": 264210
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 2.9913781284750257e-05,
+      "loss": 0.7246,
+      "step": 264220
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 2.9913019635322254e-05,
+      "loss": 0.8788,
+      "step": 264230
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 2.9912257985894255e-05,
+      "loss": 0.8676,
+      "step": 264240
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 2.9911496336466256e-05,
+      "loss": 0.8134,
+      "step": 264250
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 2.991073468703825e-05,
+      "loss": 0.744,
+      "step": 264260
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 2.990997303761025e-05,
+      "loss": 0.7845,
+      "step": 264270
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 2.9909211388182252e-05,
+      "loss": 0.7679,
+      "step": 264280
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 2.9908449738754246e-05,
+      "loss": 0.7876,
+      "step": 264290
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 2.9907688089326247e-05,
+      "loss": 0.7164,
+      "step": 264300
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 2.9906926439898248e-05,
+      "loss": 0.7553,
+      "step": 264310
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 2.9906164790470242e-05,
+      "loss": 0.8875,
+      "step": 264320
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 2.9905403141042243e-05,
+      "loss": 0.6919,
+      "step": 264330
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 2.9904641491614244e-05,
+      "loss": 0.8536,
+      "step": 264340
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 2.9903879842186238e-05,
+      "loss": 0.711,
+      "step": 264350
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 2.990311819275824e-05,
+      "loss": 0.7998,
+      "step": 264360
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 2.990235654333024e-05,
+      "loss": 0.8795,
+      "step": 264370
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 2.9901594893902233e-05,
+      "loss": 0.7972,
+      "step": 264380
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 2.9900833244474234e-05,
+      "loss": 0.7446,
+      "step": 264390
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 2.9900071595046235e-05,
+      "loss": 0.7694,
+      "step": 264400
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 2.9899309945618233e-05,
+      "loss": 0.799,
+      "step": 264410
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 2.9898548296190233e-05,
+      "loss": 0.7524,
+      "step": 264420
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 2.9897786646762227e-05,
+      "loss": 0.8466,
+      "step": 264430
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 2.989702499733423e-05,
+      "loss": 0.804,
+      "step": 264440
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 2.989626334790623e-05,
+      "loss": 0.8495,
+      "step": 264450
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 2.9895501698478223e-05,
+      "loss": 0.7778,
+      "step": 264460
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 2.9894740049050224e-05,
+      "loss": 0.8004,
+      "step": 264470
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 2.9893978399622225e-05,
+      "loss": 0.774,
+      "step": 264480
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 2.989321675019422e-05,
+      "loss": 0.7801,
+      "step": 264490
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 2.989245510076622e-05,
+      "loss": 0.8168,
+      "step": 264500
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 2.989169345133822e-05,
+      "loss": 0.7773,
+      "step": 264510
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 2.9890931801910215e-05,
+      "loss": 0.7705,
+      "step": 264520
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 2.9890170152482216e-05,
+      "loss": 0.779,
+      "step": 264530
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 2.9889408503054217e-05,
+      "loss": 0.8287,
+      "step": 264540
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 2.988864685362621e-05,
+      "loss": 0.8195,
+      "step": 264550
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 2.988788520419821e-05,
+      "loss": 0.9145,
+      "step": 264560
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 2.9887123554770212e-05,
+      "loss": 0.7919,
+      "step": 264570
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 2.988636190534221e-05,
+      "loss": 0.9256,
+      "step": 264580
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 2.9885600255914207e-05,
+      "loss": 0.7838,
+      "step": 264590
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 2.9884838606486208e-05,
+      "loss": 0.814,
+      "step": 264600
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 2.9884076957058206e-05,
+      "loss": 0.7974,
+      "step": 264610
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 2.9883315307630206e-05,
+      "loss": 0.7183,
+      "step": 264620
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 2.9882553658202207e-05,
+      "loss": 0.7696,
+      "step": 264630
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 2.98817920087742e-05,
+      "loss": 0.8385,
+      "step": 264640
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 2.9881030359346202e-05,
+      "loss": 0.8556,
+      "step": 264650
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 2.9880268709918203e-05,
+      "loss": 0.7988,
+      "step": 264660
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 2.9879507060490197e-05,
+      "loss": 0.7664,
+      "step": 264670
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 2.9878745411062198e-05,
+      "loss": 0.8432,
+      "step": 264680
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 2.98779837616342e-05,
+      "loss": 0.8196,
+      "step": 264690
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 2.9877222112206193e-05,
+      "loss": 0.7706,
+      "step": 264700
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 2.9876460462778194e-05,
+      "loss": 0.9081,
+      "step": 264710
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 2.9875698813350195e-05,
+      "loss": 0.7658,
+      "step": 264720
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 2.987493716392219e-05,
+      "loss": 0.7671,
+      "step": 264730
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 2.987417551449419e-05,
+      "loss": 0.8902,
+      "step": 264740
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 2.987341386506619e-05,
+      "loss": 0.8043,
+      "step": 264750
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 2.9872652215638185e-05,
+      "loss": 0.8558,
+      "step": 264760
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 2.9871890566210185e-05,
+      "loss": 0.851,
+      "step": 264770
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 2.9871128916782186e-05,
+      "loss": 0.763,
+      "step": 264780
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 2.9870367267354184e-05,
+      "loss": 0.8623,
+      "step": 264790
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 2.9869605617926185e-05,
+      "loss": 0.762,
+      "step": 264800
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 2.9868843968498182e-05,
+      "loss": 0.7216,
+      "step": 264810
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 2.986808231907018e-05,
+      "loss": 0.8005,
+      "step": 264820
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 2.986732066964218e-05,
+      "loss": 0.7966,
+      "step": 264830
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 2.986655902021418e-05,
+      "loss": 0.8328,
+      "step": 264840
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 2.9865797370786175e-05,
+      "loss": 0.736,
+      "step": 264850
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 2.9865035721358176e-05,
+      "loss": 0.7717,
+      "step": 264860
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 2.9864274071930177e-05,
+      "loss": 0.7652,
+      "step": 264870
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 2.986351242250217e-05,
+      "loss": 0.8344,
+      "step": 264880
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 2.9862750773074172e-05,
+      "loss": 0.8754,
+      "step": 264890
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 2.9861989123646166e-05,
+      "loss": 0.7643,
+      "step": 264900
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 2.9861227474218167e-05,
+      "loss": 0.8844,
+      "step": 264910
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 2.9860465824790168e-05,
+      "loss": 0.8891,
+      "step": 264920
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 2.9859704175362162e-05,
+      "loss": 0.7713,
+      "step": 264930
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 2.9858942525934163e-05,
+      "loss": 0.913,
+      "step": 264940
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 2.9858180876506164e-05,
+      "loss": 0.7834,
+      "step": 264950
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 2.985741922707816e-05,
+      "loss": 0.8315,
+      "step": 264960
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 2.985665757765016e-05,
+      "loss": 0.8232,
+      "step": 264970
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 2.985589592822216e-05,
+      "loss": 0.7629,
+      "step": 264980
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 2.9855134278794157e-05,
+      "loss": 0.8694,
+      "step": 264990
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 2.9854372629366158e-05,
+      "loss": 0.8642,
+      "step": 265000
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 2.985361097993816e-05,
+      "loss": 0.8243,
+      "step": 265010
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 2.9852849330510153e-05,
+      "loss": 0.8369,
+      "step": 265020
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 2.9852087681082153e-05,
+      "loss": 0.8955,
+      "step": 265030
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 2.9851326031654154e-05,
+      "loss": 0.8454,
+      "step": 265040
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 2.985056438222615e-05,
+      "loss": 0.8788,
+      "step": 265050
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 2.984980273279815e-05,
+      "loss": 0.7214,
+      "step": 265060
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 2.984904108337015e-05,
+      "loss": 0.8434,
+      "step": 265070
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 2.9848279433942144e-05,
+      "loss": 0.8607,
+      "step": 265080
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 2.9847517784514145e-05,
+      "loss": 0.8135,
+      "step": 265090
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 2.9846756135086146e-05,
+      "loss": 0.8638,
+      "step": 265100
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 2.984599448565814e-05,
+      "loss": 0.8112,
+      "step": 265110
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 2.984523283623014e-05,
+      "loss": 0.8255,
+      "step": 265120
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 2.984447118680214e-05,
+      "loss": 0.7543,
+      "step": 265130
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 2.9843709537374136e-05,
+      "loss": 0.8379,
+      "step": 265140
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 2.9842947887946137e-05,
+      "loss": 0.8048,
+      "step": 265150
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 2.9842186238518137e-05,
+      "loss": 0.7804,
+      "step": 265160
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 2.9841424589090135e-05,
+      "loss": 0.8215,
+      "step": 265170
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 2.9840662939662132e-05,
+      "loss": 0.744,
+      "step": 265180
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 2.9839901290234133e-05,
+      "loss": 0.7566,
+      "step": 265190
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 2.983913964080613e-05,
+      "loss": 0.7972,
+      "step": 265200
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 2.983837799137813e-05,
+      "loss": 0.8553,
+      "step": 265210
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 2.9837616341950132e-05,
+      "loss": 0.7861,
+      "step": 265220
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 2.9836854692522126e-05,
+      "loss": 0.7563,
+      "step": 265230
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 2.9836093043094127e-05,
+      "loss": 0.7802,
+      "step": 265240
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 2.9835331393666128e-05,
+      "loss": 0.8403,
+      "step": 265250
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 2.9834569744238122e-05,
+      "loss": 0.8084,
+      "step": 265260
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 2.9833808094810123e-05,
+      "loss": 0.7731,
+      "step": 265270
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 2.9833046445382124e-05,
+      "loss": 0.83,
+      "step": 265280
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 2.9832284795954118e-05,
+      "loss": 0.8249,
+      "step": 265290
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 2.983152314652612e-05,
+      "loss": 0.867,
+      "step": 265300
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 2.983076149709812e-05,
+      "loss": 0.7904,
+      "step": 265310
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 2.9829999847670114e-05,
+      "loss": 0.7862,
+      "step": 265320
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 2.9829238198242115e-05,
+      "loss": 0.7873,
+      "step": 265330
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 2.9828476548814116e-05,
+      "loss": 0.8605,
+      "step": 265340
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 2.982771489938611e-05,
+      "loss": 0.7958,
+      "step": 265350
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 2.982695324995811e-05,
+      "loss": 0.8717,
+      "step": 265360
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 2.9826191600530108e-05,
+      "loss": 0.736,
+      "step": 265370
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 2.982542995110211e-05,
+      "loss": 0.7955,
+      "step": 265380
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 2.982466830167411e-05,
+      "loss": 0.8564,
+      "step": 265390
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 2.9823906652246104e-05,
+      "loss": 0.829,
+      "step": 265400
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 2.9823145002818105e-05,
+      "loss": 0.725,
+      "step": 265410
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 2.9822383353390105e-05,
+      "loss": 0.8131,
+      "step": 265420
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 2.98216217039621e-05,
+      "loss": 0.7381,
+      "step": 265430
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 2.98208600545341e-05,
+      "loss": 0.7515,
+      "step": 265440
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 2.98200984051061e-05,
+      "loss": 0.7438,
+      "step": 265450
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 2.9819336755678095e-05,
+      "loss": 0.8902,
+      "step": 265460
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 2.9818575106250096e-05,
+      "loss": 0.8367,
+      "step": 265470
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 2.9817813456822097e-05,
+      "loss": 0.7914,
+      "step": 265480
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 2.981705180739409e-05,
+      "loss": 0.7858,
+      "step": 265490
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 2.9816290157966092e-05,
+      "loss": 0.85,
+      "step": 265500
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 2.9815528508538093e-05,
+      "loss": 0.8072,
+      "step": 265510
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 2.9814766859110087e-05,
+      "loss": 0.8306,
+      "step": 265520
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 2.9814005209682088e-05,
+      "loss": 0.8216,
+      "step": 265530
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 2.981324356025409e-05,
+      "loss": 0.7899,
+      "step": 265540
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 2.9812481910826086e-05,
+      "loss": 0.8668,
+      "step": 265550
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 2.9811720261398084e-05,
+      "loss": 0.8493,
+      "step": 265560
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 2.9810958611970084e-05,
+      "loss": 0.8287,
+      "step": 265570
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 2.9810196962542082e-05,
+      "loss": 0.8286,
+      "step": 265580
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 2.9809435313114083e-05,
+      "loss": 0.8303,
+      "step": 265590
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 2.9808673663686084e-05,
+      "loss": 0.924,
+      "step": 265600
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 2.9807912014258078e-05,
+      "loss": 0.8102,
+      "step": 265610
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 2.980715036483008e-05,
+      "loss": 0.779,
+      "step": 265620
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 2.980638871540208e-05,
+      "loss": 0.775,
+      "step": 265630
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 2.9805627065974073e-05,
+      "loss": 0.7876,
+      "step": 265640
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 2.9804865416546074e-05,
+      "loss": 0.9377,
+      "step": 265650
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 2.9804103767118075e-05,
+      "loss": 0.8627,
+      "step": 265660
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 2.980334211769007e-05,
+      "loss": 0.8727,
+      "step": 265670
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 2.980258046826207e-05,
+      "loss": 0.8492,
+      "step": 265680
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 2.980181881883407e-05,
+      "loss": 0.8167,
+      "step": 265690
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 2.9801057169406065e-05,
+      "loss": 0.8085,
+      "step": 265700
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 2.9800295519978066e-05,
+      "loss": 0.791,
+      "step": 265710
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 2.9799533870550067e-05,
+      "loss": 0.8605,
+      "step": 265720
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 2.979877222112206e-05,
+      "loss": 0.7738,
+      "step": 265730
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 2.979801057169406e-05,
+      "loss": 0.685,
+      "step": 265740
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 2.9797248922266063e-05,
+      "loss": 0.7482,
+      "step": 265750
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 2.979648727283806e-05,
+      "loss": 0.8356,
+      "step": 265760
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 2.9795725623410057e-05,
+      "loss": 0.7937,
+      "step": 265770
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 2.9794963973982058e-05,
+      "loss": 0.7429,
+      "step": 265780
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 2.9794202324554056e-05,
+      "loss": 0.92,
+      "step": 265790
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 2.9793440675126057e-05,
+      "loss": 0.797,
+      "step": 265800
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 2.979267902569805e-05,
+      "loss": 0.8048,
+      "step": 265810
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 2.979191737627005e-05,
+      "loss": 0.7758,
+      "step": 265820
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 2.9791155726842052e-05,
+      "loss": 0.784,
+      "step": 265830
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 2.9790394077414046e-05,
+      "loss": 0.8401,
+      "step": 265840
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 2.9789632427986047e-05,
+      "loss": 0.7886,
+      "step": 265850
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 2.9788870778558048e-05,
+      "loss": 0.7859,
+      "step": 265860
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 2.9788109129130042e-05,
+      "loss": 0.8886,
+      "step": 265870
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 2.9787347479702043e-05,
+      "loss": 0.8145,
+      "step": 265880
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 2.9786585830274044e-05,
+      "loss": 0.7976,
+      "step": 265890
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 2.9785824180846038e-05,
+      "loss": 0.8326,
+      "step": 265900
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 2.978506253141804e-05,
+      "loss": 0.8268,
+      "step": 265910
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 2.978430088199004e-05,
+      "loss": 0.7617,
+      "step": 265920
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 2.9783539232562034e-05,
+      "loss": 0.7706,
+      "step": 265930
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 2.9782777583134035e-05,
+      "loss": 0.7578,
+      "step": 265940
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 2.9782015933706036e-05,
+      "loss": 0.7846,
+      "step": 265950
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 2.9781254284278033e-05,
+      "loss": 0.8103,
+      "step": 265960
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 2.9780492634850034e-05,
+      "loss": 0.9478,
+      "step": 265970
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 2.977973098542203e-05,
+      "loss": 0.8293,
+      "step": 265980
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 2.977896933599403e-05,
+      "loss": 0.8153,
+      "step": 265990
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 2.977820768656603e-05,
+      "loss": 0.8456,
+      "step": 266000
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 2.977744603713803e-05,
+      "loss": 0.7476,
+      "step": 266010
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 2.9776684387710025e-05,
+      "loss": 0.7771,
+      "step": 266020
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 2.9775922738282025e-05,
+      "loss": 0.7521,
+      "step": 266030
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 2.9775161088854026e-05,
+      "loss": 0.8164,
+      "step": 266040
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 2.977439943942602e-05,
+      "loss": 0.8245,
+      "step": 266050
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 2.977363778999802e-05,
+      "loss": 0.7294,
+      "step": 266060
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 2.9772876140570022e-05,
+      "loss": 0.803,
+      "step": 266070
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 2.9772114491142016e-05,
+      "loss": 0.8847,
+      "step": 266080
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 2.9771352841714017e-05,
+      "loss": 0.9118,
+      "step": 266090
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 2.9770591192286018e-05,
+      "loss": 0.7054,
+      "step": 266100
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 2.9769829542858012e-05,
+      "loss": 0.7798,
+      "step": 266110
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 2.9769067893430013e-05,
+      "loss": 0.781,
+      "step": 266120
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 2.9768306244002014e-05,
+      "loss": 0.8891,
+      "step": 266130
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 2.976754459457401e-05,
+      "loss": 0.8228,
+      "step": 266140
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 2.976678294514601e-05,
+      "loss": 0.8369,
+      "step": 266150
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 2.976602129571801e-05,
+      "loss": 0.7126,
+      "step": 266160
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 2.9765259646290007e-05,
+      "loss": 0.7826,
+      "step": 266170
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 2.9764497996862008e-05,
+      "loss": 0.7607,
+      "step": 266180
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 2.976373634743401e-05,
+      "loss": 0.7668,
+      "step": 266190
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 2.9762974698006003e-05,
+      "loss": 0.753,
+      "step": 266200
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 2.9762213048578004e-05,
+      "loss": 0.7601,
+      "step": 266210
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 2.9761451399150004e-05,
+      "loss": 0.7706,
+      "step": 266220
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 2.9760689749722e-05,
+      "loss": 0.7325,
+      "step": 266230
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 2.9759928100294e-05,
+      "loss": 0.7712,
+      "step": 266240
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 2.9759166450866e-05,
+      "loss": 0.801,
+      "step": 266250
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 2.9758404801437994e-05,
+      "loss": 0.7517,
+      "step": 266260
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 2.9757643152009995e-05,
+      "loss": 0.7567,
+      "step": 266270
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 2.975688150258199e-05,
+      "loss": 0.8032,
+      "step": 266280
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 2.975611985315399e-05,
+      "loss": 0.7606,
+      "step": 266290
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 2.975535820372599e-05,
+      "loss": 0.9059,
+      "step": 266300
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 2.9754596554297985e-05,
+      "loss": 0.7518,
+      "step": 266310
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 2.9753834904869986e-05,
+      "loss": 0.8195,
+      "step": 266320
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 2.9753073255441987e-05,
+      "loss": 0.8501,
+      "step": 266330
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 2.9752311606013984e-05,
+      "loss": 0.8216,
+      "step": 266340
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 2.9751549956585985e-05,
+      "loss": 0.7795,
+      "step": 266350
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 2.9750788307157983e-05,
+      "loss": 0.8015,
+      "step": 266360
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 2.975002665772998e-05,
+      "loss": 0.8204,
+      "step": 266370
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 2.974926500830198e-05,
+      "loss": 0.7876,
+      "step": 266380
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 2.9748503358873982e-05,
+      "loss": 0.8285,
+      "step": 266390
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 2.9747741709445976e-05,
+      "loss": 0.7312,
+      "step": 266400
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 2.9746980060017977e-05,
+      "loss": 0.9349,
+      "step": 266410
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 2.9746218410589977e-05,
+      "loss": 0.7868,
+      "step": 266420
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 2.974545676116197e-05,
+      "loss": 0.779,
+      "step": 266430
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 2.9744695111733972e-05,
+      "loss": 0.7685,
+      "step": 266440
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 2.9743933462305973e-05,
+      "loss": 0.8013,
+      "step": 266450
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 2.9743171812877967e-05,
+      "loss": 0.7858,
+      "step": 266460
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 2.9742410163449968e-05,
+      "loss": 0.8645,
+      "step": 266470
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 2.974164851402197e-05,
+      "loss": 0.723,
+      "step": 266480
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 2.9740886864593963e-05,
+      "loss": 0.7192,
+      "step": 266490
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 2.9740125215165964e-05,
+      "loss": 0.7654,
+      "step": 266500
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 2.9739363565737965e-05,
+      "loss": 0.7881,
+      "step": 266510
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 2.973860191630996e-05,
+      "loss": 0.9528,
+      "step": 266520
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 2.973784026688196e-05,
+      "loss": 0.7472,
+      "step": 266530
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 2.973707861745396e-05,
+      "loss": 0.8017,
+      "step": 266540
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 2.9736316968025958e-05,
+      "loss": 0.9051,
+      "step": 266550
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 2.973555531859796e-05,
+      "loss": 0.8731,
+      "step": 266560
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 2.9734793669169956e-05,
+      "loss": 0.7471,
+      "step": 266570
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 2.9734032019741954e-05,
+      "loss": 0.8093,
+      "step": 266580
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 2.9733270370313955e-05,
+      "loss": 0.7342,
+      "step": 266590
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 2.9732508720885956e-05,
+      "loss": 0.8184,
+      "step": 266600
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 2.973174707145795e-05,
+      "loss": 0.7736,
+      "step": 266610
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 2.973098542202995e-05,
+      "loss": 0.7638,
+      "step": 266620
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 2.973022377260195e-05,
+      "loss": 0.8362,
+      "step": 266630
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 2.9729462123173945e-05,
+      "loss": 0.8436,
+      "step": 266640
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 2.9728700473745946e-05,
+      "loss": 0.9106,
+      "step": 266650
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 2.9727938824317947e-05,
+      "loss": 0.7595,
+      "step": 266660
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 2.972717717488994e-05,
+      "loss": 0.7389,
+      "step": 266670
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 2.9726415525461942e-05,
+      "loss": 0.7541,
+      "step": 266680
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 2.9725653876033943e-05,
+      "loss": 0.8132,
+      "step": 266690
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 2.9724892226605937e-05,
+      "loss": 0.7932,
+      "step": 266700
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 2.9724130577177938e-05,
+      "loss": 0.791,
+      "step": 266710
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 2.972336892774994e-05,
+      "loss": 0.7571,
+      "step": 266720
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 2.9722607278321933e-05,
+      "loss": 0.7136,
+      "step": 266730
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 2.9721845628893934e-05,
+      "loss": 0.7533,
+      "step": 266740
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 2.972108397946593e-05,
+      "loss": 0.8413,
+      "step": 266750
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 2.9720322330037932e-05,
+      "loss": 0.8138,
+      "step": 266760
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 2.9719560680609933e-05,
+      "loss": 0.7871,
+      "step": 266770
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 2.9718799031181927e-05,
+      "loss": 0.7681,
+      "step": 266780
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 2.9718037381753928e-05,
+      "loss": 0.7791,
+      "step": 266790
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 2.971727573232593e-05,
+      "loss": 0.8352,
+      "step": 266800
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 2.9716514082897923e-05,
+      "loss": 0.7929,
+      "step": 266810
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 2.9715752433469924e-05,
+      "loss": 0.8507,
+      "step": 266820
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 2.9714990784041924e-05,
+      "loss": 0.9791,
+      "step": 266830
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 2.971422913461392e-05,
+      "loss": 0.8485,
+      "step": 266840
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 2.971346748518592e-05,
+      "loss": 0.7843,
+      "step": 266850
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 2.971270583575792e-05,
+      "loss": 0.7755,
+      "step": 266860
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 2.9711944186329914e-05,
+      "loss": 0.806,
+      "step": 266870
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 2.9711182536901915e-05,
+      "loss": 0.7796,
+      "step": 266880
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 2.9710420887473916e-05,
+      "loss": 0.8766,
+      "step": 266890
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 2.970965923804591e-05,
+      "loss": 0.8059,
+      "step": 266900
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 2.970889758861791e-05,
+      "loss": 0.7649,
+      "step": 266910
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 2.9708135939189912e-05,
+      "loss": 0.8077,
+      "step": 266920
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 2.970737428976191e-05,
+      "loss": 0.7861,
+      "step": 266930
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 2.970661264033391e-05,
+      "loss": 0.916,
+      "step": 266940
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 2.9705850990905908e-05,
+      "loss": 0.8153,
+      "step": 266950
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 2.9705089341477905e-05,
+      "loss": 0.811,
+      "step": 266960
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 2.9704327692049906e-05,
+      "loss": 0.758,
+      "step": 266970
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 2.9703566042621907e-05,
+      "loss": 0.8912,
+      "step": 266980
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 2.97028043931939e-05,
+      "loss": 0.8394,
+      "step": 266990
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 2.9702042743765902e-05,
+      "loss": 0.737,
+      "step": 267000
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 2.9701281094337903e-05,
+      "loss": 0.9103,
+      "step": 267010
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 2.9700519444909897e-05,
+      "loss": 0.7975,
+      "step": 267020
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 2.9699757795481897e-05,
+      "loss": 0.8127,
+      "step": 267030
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 2.96989961460539e-05,
+      "loss": 0.8887,
+      "step": 267040
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 2.9698234496625892e-05,
+      "loss": 0.704,
+      "step": 267050
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 2.9697472847197893e-05,
+      "loss": 0.7042,
+      "step": 267060
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 2.9696711197769894e-05,
+      "loss": 0.8153,
+      "step": 267070
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 2.9695949548341888e-05,
+      "loss": 0.8412,
+      "step": 267080
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 2.969518789891389e-05,
+      "loss": 0.6812,
+      "step": 267090
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 2.969442624948589e-05,
+      "loss": 0.829,
+      "step": 267100
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 2.9693664600057884e-05,
+      "loss": 0.7935,
+      "step": 267110
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 2.9692902950629885e-05,
+      "loss": 0.7724,
+      "step": 267120
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 2.9692141301201886e-05,
+      "loss": 0.7382,
+      "step": 267130
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 2.9691379651773883e-05,
+      "loss": 0.7196,
+      "step": 267140
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 2.9690618002345884e-05,
+      "loss": 0.737,
+      "step": 267150
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 2.968985635291788e-05,
+      "loss": 0.732,
+      "step": 267160
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 2.968909470348988e-05,
+      "loss": 0.7858,
+      "step": 267170
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 2.968833305406188e-05,
+      "loss": 0.8616,
+      "step": 267180
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 2.968757140463388e-05,
+      "loss": 0.822,
+      "step": 267190
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 2.9686809755205875e-05,
+      "loss": 0.8045,
+      "step": 267200
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 2.9686048105777876e-05,
+      "loss": 0.8516,
+      "step": 267210
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 2.968528645634987e-05,
+      "loss": 0.7242,
+      "step": 267220
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 2.968452480692187e-05,
+      "loss": 0.778,
+      "step": 267230
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 2.968376315749387e-05,
+      "loss": 0.7757,
+      "step": 267240
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 2.9683001508065865e-05,
+      "loss": 0.6993,
+      "step": 267250
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 2.9682239858637866e-05,
+      "loss": 0.7646,
+      "step": 267260
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 2.9681478209209867e-05,
+      "loss": 0.8913,
+      "step": 267270
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 2.968071655978186e-05,
+      "loss": 0.8082,
+      "step": 267280
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 2.9679954910353862e-05,
+      "loss": 0.7845,
+      "step": 267290
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 2.9679193260925863e-05,
+      "loss": 0.7614,
+      "step": 267300
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 2.967843161149786e-05,
+      "loss": 0.7431,
+      "step": 267310
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 2.9677669962069858e-05,
+      "loss": 0.8238,
+      "step": 267320
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 2.967690831264186e-05,
+      "loss": 0.8445,
+      "step": 267330
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 2.9676146663213856e-05,
+      "loss": 0.747,
+      "step": 267340
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 2.9675385013785857e-05,
+      "loss": 0.8128,
+      "step": 267350
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 2.9674623364357858e-05,
+      "loss": 0.7347,
+      "step": 267360
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 2.9673861714929852e-05,
+      "loss": 0.7513,
+      "step": 267370
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 2.9673100065501853e-05,
+      "loss": 0.8174,
+      "step": 267380
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 2.9672338416073854e-05,
+      "loss": 0.8383,
+      "step": 267390
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 2.9671576766645848e-05,
+      "loss": 0.7507,
+      "step": 267400
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 2.967081511721785e-05,
+      "loss": 0.7164,
+      "step": 267410
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 2.967005346778985e-05,
+      "loss": 0.7055,
+      "step": 267420
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 2.9669291818361844e-05,
+      "loss": 0.8316,
+      "step": 267430
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 2.9668530168933844e-05,
+      "loss": 0.7628,
+      "step": 267440
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 2.9667768519505845e-05,
+      "loss": 0.7524,
+      "step": 267450
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 2.966700687007784e-05,
+      "loss": 0.7978,
+      "step": 267460
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 2.966624522064984e-05,
+      "loss": 0.8396,
+      "step": 267470
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 2.966548357122184e-05,
+      "loss": 0.761,
+      "step": 267480
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 2.9664721921793835e-05,
+      "loss": 0.7305,
+      "step": 267490
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 2.9663960272365836e-05,
+      "loss": 0.809,
+      "step": 267500
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 2.9663198622937837e-05,
+      "loss": 0.8778,
+      "step": 267510
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 2.9662436973509834e-05,
+      "loss": 0.8139,
+      "step": 267520
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 2.9661675324081832e-05,
+      "loss": 0.8231,
+      "step": 267530
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 2.9660913674653833e-05,
+      "loss": 0.7875,
+      "step": 267540
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 2.966015202522583e-05,
+      "loss": 0.854,
+      "step": 267550
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 2.965939037579783e-05,
+      "loss": 0.7753,
+      "step": 267560
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 2.9658628726369832e-05,
+      "loss": 0.716,
+      "step": 267570
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 2.9657867076941826e-05,
+      "loss": 0.8296,
+      "step": 267580
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 2.9657105427513827e-05,
+      "loss": 0.7766,
+      "step": 267590
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 2.9656343778085828e-05,
+      "loss": 0.8375,
+      "step": 267600
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 2.9655582128657822e-05,
+      "loss": 0.7949,
+      "step": 267610
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 2.9654820479229823e-05,
+      "loss": 0.9553,
+      "step": 267620
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 2.9654058829801823e-05,
+      "loss": 0.7699,
+      "step": 267630
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 2.9653297180373818e-05,
+      "loss": 0.7425,
+      "step": 267640
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 2.965253553094582e-05,
+      "loss": 0.7771,
+      "step": 267650
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 2.9651773881517812e-05,
+      "loss": 0.8518,
+      "step": 267660
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 2.9651012232089813e-05,
+      "loss": 0.8648,
+      "step": 267670
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 2.9650250582661814e-05,
+      "loss": 0.818,
+      "step": 267680
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 2.964948893323381e-05,
+      "loss": 0.7861,
+      "step": 267690
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 2.964872728380581e-05,
+      "loss": 0.7491,
+      "step": 267700
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 2.964796563437781e-05,
+      "loss": 0.9031,
+      "step": 267710
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 2.9647203984949807e-05,
+      "loss": 0.8018,
+      "step": 267720
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 2.9646442335521808e-05,
+      "loss": 0.7132,
+      "step": 267730
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 2.964568068609381e-05,
+      "loss": 0.8819,
+      "step": 267740
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 2.9644919036665803e-05,
+      "loss": 0.765,
+      "step": 267750
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 2.9644157387237804e-05,
+      "loss": 0.7652,
+      "step": 267760
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 2.9643395737809805e-05,
+      "loss": 0.8253,
+      "step": 267770
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 2.96426340883818e-05,
+      "loss": 0.8204,
+      "step": 267780
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 2.96418724389538e-05,
+      "loss": 0.8028,
+      "step": 267790
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 2.96411107895258e-05,
+      "loss": 0.7976,
+      "step": 267800
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 2.9640349140097795e-05,
+      "loss": 0.7244,
+      "step": 267810
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 2.9639587490669796e-05,
+      "loss": 0.829,
+      "step": 267820
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 2.9638825841241796e-05,
+      "loss": 0.8456,
+      "step": 267830
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 2.963806419181379e-05,
+      "loss": 0.8432,
+      "step": 267840
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 2.963730254238579e-05,
+      "loss": 0.8034,
+      "step": 267850
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 2.9636540892957792e-05,
+      "loss": 0.8355,
+      "step": 267860
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 2.9635779243529786e-05,
+      "loss": 0.8153,
+      "step": 267870
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 2.9635017594101787e-05,
+      "loss": 0.6954,
+      "step": 267880
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 2.9634255944673788e-05,
+      "loss": 0.7407,
+      "step": 267890
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 2.9633494295245786e-05,
+      "loss": 0.842,
+      "step": 267900
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 2.9632732645817783e-05,
+      "loss": 0.793,
+      "step": 267910
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 2.9631970996389784e-05,
+      "loss": 0.7844,
+      "step": 267920
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 2.963120934696178e-05,
+      "loss": 0.7211,
+      "step": 267930
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 2.9630447697533782e-05,
+      "loss": 0.7758,
+      "step": 267940
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 2.9629686048105783e-05,
+      "loss": 0.7939,
+      "step": 267950
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 2.9628924398677777e-05,
+      "loss": 0.8336,
+      "step": 267960
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 2.9628162749249778e-05,
+      "loss": 0.8191,
+      "step": 267970
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 2.962740109982178e-05,
+      "loss": 0.7981,
+      "step": 267980
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 2.9626639450393773e-05,
+      "loss": 0.9018,
+      "step": 267990
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 2.9625877800965774e-05,
+      "loss": 0.7862,
+      "step": 268000
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 2.9625116151537775e-05,
+      "loss": 0.8129,
+      "step": 268010
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 2.962435450210977e-05,
+      "loss": 0.8418,
+      "step": 268020
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 2.962359285268177e-05,
+      "loss": 0.8691,
+      "step": 268030
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 2.962283120325377e-05,
+      "loss": 0.8247,
+      "step": 268040
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 2.9622069553825764e-05,
+      "loss": 0.7212,
+      "step": 268050
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 2.9621307904397765e-05,
+      "loss": 0.8033,
+      "step": 268060
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 2.9620546254969766e-05,
+      "loss": 0.8668,
+      "step": 268070
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 2.961978460554176e-05,
+      "loss": 0.8339,
+      "step": 268080
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 2.961902295611376e-05,
+      "loss": 0.837,
+      "step": 268090
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 2.9618261306685762e-05,
+      "loss": 0.78,
+      "step": 268100
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 2.961749965725776e-05,
+      "loss": 0.9086,
+      "step": 268110
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 2.9616738007829757e-05,
+      "loss": 0.7669,
+      "step": 268120
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 2.9615976358401754e-05,
+      "loss": 0.824,
+      "step": 268130
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 2.9615214708973755e-05,
+      "loss": 0.8728,
+      "step": 268140
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 2.9614453059545756e-05,
+      "loss": 0.7796,
+      "step": 268150
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 2.961369141011775e-05,
+      "loss": 0.8424,
+      "step": 268160
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 2.961292976068975e-05,
+      "loss": 0.8162,
+      "step": 268170
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 2.9612168111261752e-05,
+      "loss": 0.7153,
+      "step": 268180
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 2.9611406461833746e-05,
+      "loss": 0.8348,
+      "step": 268190
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 2.9610644812405747e-05,
+      "loss": 0.7746,
+      "step": 268200
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 2.9609883162977748e-05,
+      "loss": 0.8037,
+      "step": 268210
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 2.9609121513549742e-05,
+      "loss": 0.753,
+      "step": 268220
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 2.9608359864121743e-05,
+      "loss": 0.8891,
+      "step": 268230
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 2.9607598214693743e-05,
+      "loss": 0.831,
+      "step": 268240
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 2.9606836565265738e-05,
+      "loss": 0.7176,
+      "step": 268250
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 2.960607491583774e-05,
+      "loss": 0.761,
+      "step": 268260
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 2.960531326640974e-05,
+      "loss": 0.7074,
+      "step": 268270
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 2.9604551616981733e-05,
+      "loss": 0.8165,
+      "step": 268280
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 2.9603789967553734e-05,
+      "loss": 0.7678,
+      "step": 268290
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 2.9603028318125735e-05,
+      "loss": 0.8264,
+      "step": 268300
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 2.9602266668697732e-05,
+      "loss": 0.8104,
+      "step": 268310
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 2.9601505019269733e-05,
+      "loss": 0.7693,
+      "step": 268320
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 2.960074336984173e-05,
+      "loss": 0.8417,
+      "step": 268330
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 2.9599981720413728e-05,
+      "loss": 0.864,
+      "step": 268340
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 2.959922007098573e-05,
+      "loss": 0.7601,
+      "step": 268350
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 2.959845842155773e-05,
+      "loss": 0.7203,
+      "step": 268360
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 2.9597696772129724e-05,
+      "loss": 0.7217,
+      "step": 268370
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 2.9596935122701725e-05,
+      "loss": 0.8681,
+      "step": 268380
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 2.9596173473273726e-05,
+      "loss": 0.8001,
+      "step": 268390
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 2.959541182384572e-05,
+      "loss": 0.8658,
+      "step": 268400
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 2.959465017441772e-05,
+      "loss": 0.7993,
+      "step": 268410
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 2.959388852498972e-05,
+      "loss": 0.7992,
+      "step": 268420
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 2.9593126875561716e-05,
+      "loss": 0.8702,
+      "step": 268430
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 2.9592365226133716e-05,
+      "loss": 0.7891,
+      "step": 268440
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 2.9591603576705717e-05,
+      "loss": 0.7918,
+      "step": 268450
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 2.959084192727771e-05,
+      "loss": 0.6921,
+      "step": 268460
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 2.9590080277849712e-05,
+      "loss": 0.8094,
+      "step": 268470
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 2.9589318628421713e-05,
+      "loss": 0.696,
+      "step": 268480
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 2.958855697899371e-05,
+      "loss": 0.8037,
+      "step": 268490
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 2.9587795329565708e-05,
+      "loss": 0.7944,
+      "step": 268500
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 2.958703368013771e-05,
+      "loss": 0.8094,
+      "step": 268510
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 2.9586272030709706e-05,
+      "loss": 0.7675,
+      "step": 268520
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 2.9585510381281707e-05,
+      "loss": 0.7432,
+      "step": 268530
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 2.9584748731853708e-05,
+      "loss": 0.8298,
+      "step": 268540
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 2.9583987082425702e-05,
+      "loss": 0.7772,
+      "step": 268550
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 2.9583225432997703e-05,
+      "loss": 0.7925,
+      "step": 268560
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 2.9582463783569704e-05,
+      "loss": 0.786,
+      "step": 268570
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 2.9581702134141698e-05,
+      "loss": 0.8712,
+      "step": 268580
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 2.95809404847137e-05,
+      "loss": 0.7512,
+      "step": 268590
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 2.9580178835285693e-05,
+      "loss": 0.883,
+      "step": 268600
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 2.9579417185857694e-05,
+      "loss": 0.761,
+      "step": 268610
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 2.9578655536429695e-05,
+      "loss": 0.8135,
+      "step": 268620
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 2.957789388700169e-05,
+      "loss": 0.7374,
+      "step": 268630
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 2.957713223757369e-05,
+      "loss": 0.8498,
+      "step": 268640
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 2.957637058814569e-05,
+      "loss": 0.8152,
+      "step": 268650
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 2.9575608938717684e-05,
+      "loss": 0.7305,
+      "step": 268660
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 2.9574847289289685e-05,
+      "loss": 0.7911,
+      "step": 268670
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 2.9574085639861686e-05,
+      "loss": 0.7904,
+      "step": 268680
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 2.9573323990433684e-05,
+      "loss": 0.8001,
+      "step": 268690
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 2.9572562341005685e-05,
+      "loss": 0.7858,
+      "step": 268700
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 2.9571800691577682e-05,
+      "loss": 0.8145,
+      "step": 268710
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 2.957103904214968e-05,
+      "loss": 0.8704,
+      "step": 268720
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 2.957027739272168e-05,
+      "loss": 0.6705,
+      "step": 268730
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 2.956951574329368e-05,
+      "loss": 0.9295,
+      "step": 268740
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 2.9568754093865675e-05,
+      "loss": 0.694,
+      "step": 268750
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 2.9567992444437676e-05,
+      "loss": 0.7919,
+      "step": 268760
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 2.9567230795009677e-05,
+      "loss": 0.7916,
+      "step": 268770
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 2.956646914558167e-05,
+      "loss": 0.7071,
+      "step": 268780
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 2.9565707496153672e-05,
+      "loss": 0.8001,
+      "step": 268790
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 2.9564945846725673e-05,
+      "loss": 0.8296,
+      "step": 268800
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 2.9564184197297667e-05,
+      "loss": 0.8504,
+      "step": 268810
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 2.9563422547869668e-05,
+      "loss": 0.7991,
+      "step": 268820
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 2.956266089844167e-05,
+      "loss": 0.7079,
+      "step": 268830
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 2.9561899249013663e-05,
+      "loss": 0.8002,
+      "step": 268840
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 2.9561137599585663e-05,
+      "loss": 0.8339,
+      "step": 268850
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 2.9560375950157664e-05,
+      "loss": 0.7899,
+      "step": 268860
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 2.955961430072966e-05,
+      "loss": 0.7873,
+      "step": 268870
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 2.955885265130166e-05,
+      "loss": 0.8757,
+      "step": 268880
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 2.955809100187366e-05,
+      "loss": 0.8178,
+      "step": 268890
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 2.9557329352445658e-05,
+      "loss": 0.8598,
+      "step": 268900
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 2.955656770301766e-05,
+      "loss": 0.8807,
+      "step": 268910
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 2.9555806053589656e-05,
+      "loss": 0.8303,
+      "step": 268920
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 2.9555044404161653e-05,
+      "loss": 0.9213,
+      "step": 268930
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 2.9554282754733654e-05,
+      "loss": 0.7569,
+      "step": 268940
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 2.9553521105305655e-05,
+      "loss": 0.8044,
+      "step": 268950
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 2.955275945587765e-05,
+      "loss": 0.7359,
+      "step": 268960
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 2.955199780644965e-05,
+      "loss": 0.7982,
+      "step": 268970
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 2.955123615702165e-05,
+      "loss": 0.7164,
+      "step": 268980
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 2.9550474507593645e-05,
+      "loss": 0.8434,
+      "step": 268990
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 2.9549712858165646e-05,
+      "loss": 0.8936,
+      "step": 269000
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 2.9548951208737647e-05,
+      "loss": 0.7304,
+      "step": 269010
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 2.954818955930964e-05,
+      "loss": 0.8139,
+      "step": 269020
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 2.954742790988164e-05,
+      "loss": 0.9134,
+      "step": 269030
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 2.9546666260453636e-05,
+      "loss": 0.8197,
+      "step": 269040
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 2.9545904611025637e-05,
+      "loss": 0.843,
+      "step": 269050
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 2.9545142961597637e-05,
+      "loss": 0.7454,
+      "step": 269060
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 2.9544381312169635e-05,
+      "loss": 0.6961,
+      "step": 269070
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 2.9543619662741632e-05,
+      "loss": 0.8779,
+      "step": 269080
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 2.9542858013313633e-05,
+      "loss": 0.7582,
+      "step": 269090
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 2.954209636388563e-05,
+      "loss": 0.7307,
+      "step": 269100
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 2.954133471445763e-05,
+      "loss": 0.7811,
+      "step": 269110
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 2.9540573065029632e-05,
+      "loss": 0.7697,
+      "step": 269120
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 2.9539811415601626e-05,
+      "loss": 0.806,
+      "step": 269130
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 2.9539049766173627e-05,
+      "loss": 0.7753,
+      "step": 269140
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 2.9538288116745628e-05,
+      "loss": 0.7976,
+      "step": 269150
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 2.9537526467317622e-05,
+      "loss": 0.8297,
+      "step": 269160
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 2.9536764817889623e-05,
+      "loss": 0.7773,
+      "step": 269170
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 2.9536003168461624e-05,
+      "loss": 0.8515,
+      "step": 269180
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 2.9535241519033618e-05,
+      "loss": 0.7471,
+      "step": 269190
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 2.953447986960562e-05,
+      "loss": 0.7588,
+      "step": 269200
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 2.953371822017762e-05,
+      "loss": 0.9094,
+      "step": 269210
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 2.9532956570749614e-05,
+      "loss": 0.7766,
+      "step": 269220
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 2.9532194921321615e-05,
+      "loss": 0.8351,
+      "step": 269230
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 2.9531433271893615e-05,
+      "loss": 0.7643,
+      "step": 269240
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 2.953067162246561e-05,
+      "loss": 0.7885,
+      "step": 269250
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 2.952990997303761e-05,
+      "loss": 0.8053,
+      "step": 269260
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 2.952914832360961e-05,
+      "loss": 0.7695,
+      "step": 269270
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 2.952838667418161e-05,
+      "loss": 0.7689,
+      "step": 269280
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 2.952762502475361e-05,
+      "loss": 0.7782,
+      "step": 269290
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 2.9526863375325607e-05,
+      "loss": 0.8535,
+      "step": 269300
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 2.9526101725897605e-05,
+      "loss": 0.7346,
+      "step": 269310
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 2.9525340076469605e-05,
+      "loss": 0.7658,
+      "step": 269320
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 2.9524578427041606e-05,
+      "loss": 0.8183,
+      "step": 269330
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 2.95238167776136e-05,
+      "loss": 0.8607,
+      "step": 269340
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 2.95230551281856e-05,
+      "loss": 0.6985,
+      "step": 269350
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 2.9522293478757602e-05,
+      "loss": 0.8055,
+      "step": 269360
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 2.9521531829329596e-05,
+      "loss": 0.7327,
+      "step": 269370
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 2.9520770179901597e-05,
+      "loss": 0.7529,
+      "step": 269380
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 2.9520008530473598e-05,
+      "loss": 0.7505,
+      "step": 269390
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 2.9519246881045592e-05,
+      "loss": 0.7352,
+      "step": 269400
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 2.9518485231617593e-05,
+      "loss": 0.8043,
+      "step": 269410
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 2.9517723582189594e-05,
+      "loss": 0.7474,
+      "step": 269420
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 2.9516961932761588e-05,
+      "loss": 0.8437,
+      "step": 269430
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 2.951620028333359e-05,
+      "loss": 0.8655,
+      "step": 269440
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 2.951543863390559e-05,
+      "loss": 0.8377,
+      "step": 269450
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 2.9514676984477583e-05,
+      "loss": 0.7657,
+      "step": 269460
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 2.9513915335049584e-05,
+      "loss": 0.8194,
+      "step": 269470
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 2.9513153685621585e-05,
+      "loss": 0.8516,
+      "step": 269480
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 2.9512392036193583e-05,
+      "loss": 0.7542,
+      "step": 269490
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 2.9511630386765583e-05,
+      "loss": 0.7897,
+      "step": 269500
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 2.9510868737337578e-05,
+      "loss": 0.8063,
+      "step": 269510
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 2.951010708790958e-05,
+      "loss": 0.8507,
+      "step": 269520
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 2.950934543848158e-05,
+      "loss": 0.8278,
+      "step": 269530
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 2.9508583789053573e-05,
+      "loss": 0.7813,
+      "step": 269540
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 2.9507822139625574e-05,
+      "loss": 0.736,
+      "step": 269550
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 2.9507060490197575e-05,
+      "loss": 0.8377,
+      "step": 269560
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 2.950629884076957e-05,
+      "loss": 0.7873,
+      "step": 269570
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 2.950553719134157e-05,
+      "loss": 0.7766,
+      "step": 269580
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 2.950477554191357e-05,
+      "loss": 0.8345,
+      "step": 269590
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 2.9504013892485565e-05,
+      "loss": 0.8367,
+      "step": 269600
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 2.9503252243057566e-05,
+      "loss": 0.7711,
+      "step": 269610
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 2.9502490593629567e-05,
+      "loss": 0.7575,
+      "step": 269620
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 2.950172894420156e-05,
+      "loss": 0.8254,
+      "step": 269630
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 2.950096729477356e-05,
+      "loss": 0.8631,
+      "step": 269640
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 2.9500205645345562e-05,
+      "loss": 0.8183,
+      "step": 269650
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 2.949944399591756e-05,
+      "loss": 0.8099,
+      "step": 269660
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 2.9498682346489557e-05,
+      "loss": 0.8585,
+      "step": 269670
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 2.9497920697061558e-05,
+      "loss": 0.8358,
+      "step": 269680
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 2.9497159047633556e-05,
+      "loss": 0.7646,
+      "step": 269690
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 2.9496397398205557e-05,
+      "loss": 0.7678,
+      "step": 269700
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 2.9495635748777557e-05,
+      "loss": 0.7278,
+      "step": 269710
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 2.949487409934955e-05,
+      "loss": 0.7671,
+      "step": 269720
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 2.9494112449921552e-05,
+      "loss": 0.8117,
+      "step": 269730
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 2.9493350800493553e-05,
+      "loss": 0.8574,
+      "step": 269740
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 2.9492589151065547e-05,
+      "loss": 0.8012,
+      "step": 269750
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 2.9491827501637548e-05,
+      "loss": 0.7599,
+      "step": 269760
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 2.949106585220955e-05,
+      "loss": 0.7419,
+      "step": 269770
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 2.9490304202781543e-05,
+      "loss": 0.9152,
+      "step": 269780
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 2.9489542553353544e-05,
+      "loss": 0.8065,
+      "step": 269790
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 2.9488780903925545e-05,
+      "loss": 0.8031,
+      "step": 269800
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 2.948801925449754e-05,
+      "loss": 0.8547,
+      "step": 269810
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 2.948725760506954e-05,
+      "loss": 0.7476,
+      "step": 269820
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 2.948649595564154e-05,
+      "loss": 0.8072,
+      "step": 269830
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 2.9485734306213535e-05,
+      "loss": 0.8137,
+      "step": 269840
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 2.9484972656785535e-05,
+      "loss": 0.7965,
+      "step": 269850
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 2.9484211007357536e-05,
+      "loss": 0.7252,
+      "step": 269860
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 2.9483449357929534e-05,
+      "loss": 0.816,
+      "step": 269870
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 2.948268770850153e-05,
+      "loss": 0.7818,
+      "step": 269880
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 2.9481926059073532e-05,
+      "loss": 0.8167,
+      "step": 269890
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 2.948116440964553e-05,
+      "loss": 0.7651,
+      "step": 269900
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 2.948040276021753e-05,
+      "loss": 0.8191,
+      "step": 269910
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 2.947964111078953e-05,
+      "loss": 0.8665,
+      "step": 269920
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 2.9478879461361525e-05,
+      "loss": 0.8716,
+      "step": 269930
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 2.9478117811933526e-05,
+      "loss": 0.8494,
+      "step": 269940
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 2.9477356162505527e-05,
+      "loss": 0.8847,
+      "step": 269950
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 2.947659451307752e-05,
+      "loss": 0.8338,
+      "step": 269960
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 2.9475832863649522e-05,
+      "loss": 0.9411,
+      "step": 269970
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 2.9475071214221516e-05,
+      "loss": 0.8052,
+      "step": 269980
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 2.9474309564793517e-05,
+      "loss": 0.8076,
+      "step": 269990
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 2.9473547915365518e-05,
+      "loss": 0.8656,
+      "step": 270000
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 2.9472786265937512e-05,
+      "loss": 0.9119,
+      "step": 270010
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 2.9472024616509513e-05,
+      "loss": 0.7525,
+      "step": 270020
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 2.9471262967081514e-05,
+      "loss": 0.7657,
+      "step": 270030
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 2.947050131765351e-05,
+      "loss": 0.7405,
+      "step": 270040
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 2.946973966822551e-05,
+      "loss": 0.7461,
+      "step": 270050
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 2.946897801879751e-05,
+      "loss": 0.8097,
+      "step": 270060
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 2.9468216369369507e-05,
+      "loss": 0.8127,
+      "step": 270070
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 2.9467454719941508e-05,
+      "loss": 0.7932,
+      "step": 270080
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 2.946669307051351e-05,
+      "loss": 0.8463,
+      "step": 270090
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 2.9465931421085503e-05,
+      "loss": 0.8007,
+      "step": 270100
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 2.9465169771657504e-05,
+      "loss": 0.7953,
+      "step": 270110
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 2.9464408122229504e-05,
+      "loss": 0.6865,
+      "step": 270120
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 2.94636464728015e-05,
+      "loss": 0.7538,
+      "step": 270130
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 2.94628848233735e-05,
+      "loss": 0.79,
+      "step": 270140
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 2.94621231739455e-05,
+      "loss": 0.862,
+      "step": 270150
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 2.9461361524517494e-05,
+      "loss": 0.8076,
+      "step": 270160
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 2.9460599875089495e-05,
+      "loss": 0.8532,
+      "step": 270170
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 2.9459838225661496e-05,
+      "loss": 0.7765,
+      "step": 270180
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 2.945907657623349e-05,
+      "loss": 0.8318,
+      "step": 270190
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 2.945831492680549e-05,
+      "loss": 0.7741,
+      "step": 270200
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 2.9457553277377492e-05,
+      "loss": 0.7879,
+      "step": 270210
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 2.9456791627949486e-05,
+      "loss": 0.821,
+      "step": 270220
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 2.9456029978521487e-05,
+      "loss": 0.7743,
+      "step": 270230
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 2.9455268329093488e-05,
+      "loss": 0.7576,
+      "step": 270240
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 2.9454506679665485e-05,
+      "loss": 0.7835,
+      "step": 270250
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 2.9453745030237482e-05,
+      "loss": 0.8309,
+      "step": 270260
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 2.9452983380809483e-05,
+      "loss": 0.8058,
+      "step": 270270
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 2.945222173138148e-05,
+      "loss": 0.757,
+      "step": 270280
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 2.945146008195348e-05,
+      "loss": 0.7954,
+      "step": 270290
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 2.9450698432525482e-05,
+      "loss": 0.8397,
+      "step": 270300
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 2.9449936783097477e-05,
+      "loss": 0.7468,
+      "step": 270310
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 2.9449175133669477e-05,
+      "loss": 0.8017,
+      "step": 270320
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 2.9448413484241478e-05,
+      "loss": 0.7462,
+      "step": 270330
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 2.9447651834813472e-05,
+      "loss": 0.7379,
+      "step": 270340
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 2.9446890185385473e-05,
+      "loss": 0.7955,
+      "step": 270350
+    },
+    {
+      "epoch": 12.35,
+      "learning_rate": 2.9446128535957474e-05,
+      "loss": 0.9219,
+      "step": 270360
+    },
+    {
+      "epoch": 12.35,
+      "learning_rate": 2.9445366886529468e-05,
+      "loss": 0.8527,
+      "step": 270370
+    },
+    {
+      "epoch": 12.35,
+      "learning_rate": 2.944460523710147e-05,
+      "loss": 0.8482,
+      "step": 270380
+    },
+    {
+      "epoch": 12.35,
+      "learning_rate": 2.944384358767347e-05,
+      "loss": 0.7886,
+      "step": 270390
+    },
+    {
+      "epoch": 12.35,
+      "learning_rate": 2.9443081938245464e-05,
+      "loss": 0.892,
+      "step": 270400
+    },
+    {
+      "epoch": 12.35,
+      "learning_rate": 2.9442320288817465e-05,
+      "loss": 0.7886,
+      "step": 270410
+    },
+    {
+      "epoch": 12.35,
+      "learning_rate": 2.944155863938946e-05,
+      "loss": 0.8978,
+      "step": 270420
+    },
+    {
+      "epoch": 12.35,
+      "learning_rate": 2.944079698996146e-05,
+      "loss": 0.7779,
+      "step": 270430
+    },
+    {
+      "epoch": 12.35,
+      "learning_rate": 2.944003534053346e-05,
+      "loss": 0.7579,
+      "step": 270440
+    },
+    {
+      "epoch": 12.35,
+      "learning_rate": 2.9439273691105458e-05,
+      "loss": 0.7721,
+      "step": 270450
+    },
+    {
+      "epoch": 12.35,
+      "learning_rate": 2.943851204167746e-05,
+      "loss": 0.8083,
+      "step": 270460
+    },
+    {
+      "epoch": 12.35,
+      "learning_rate": 2.9437750392249456e-05,
+      "loss": 0.8123,
+      "step": 270470
+    },
+    {
+      "epoch": 12.35,
+      "learning_rate": 2.9436988742821454e-05,
+      "loss": 0.8751,
+      "step": 270480
+    },
+    {
+      "epoch": 12.35,
+      "learning_rate": 2.9436227093393455e-05,
+      "loss": 0.8426,
+      "step": 270490
+    },
+    {
+      "epoch": 12.35,
+      "learning_rate": 2.9435465443965456e-05,
+      "loss": 0.8179,
+      "step": 270500
+    },
+    {
+      "epoch": 12.35,
+      "learning_rate": 2.943470379453745e-05,
+      "loss": 0.8499,
+      "step": 270510
+    },
+    {
+      "epoch": 12.35,
+      "learning_rate": 2.943394214510945e-05,
+      "loss": 0.8558,
+      "step": 270520
+    },
+    {
+      "epoch": 12.35,
+      "learning_rate": 2.943318049568145e-05,
+      "loss": 0.8615,
+      "step": 270530
+    },
+    {
+      "epoch": 12.35,
+      "learning_rate": 2.9432418846253445e-05,
+      "loss": 0.8654,
+      "step": 270540
+    },
+    {
+      "epoch": 12.35,
+      "learning_rate": 2.9431657196825446e-05,
+      "loss": 0.813,
+      "step": 270550
+    },
+    {
+      "epoch": 12.35,
+      "learning_rate": 2.9430895547397447e-05,
+      "loss": 0.8026,
+      "step": 270560
+    },
+    {
+      "epoch": 12.35,
+      "learning_rate": 2.943013389796944e-05,
+      "loss": 0.8327,
+      "step": 270570
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 2.9429372248541442e-05,
+      "loss": 0.7819,
+      "step": 270580
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 2.9428610599113443e-05,
+      "loss": 0.8531,
+      "step": 270590
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 2.9427848949685437e-05,
+      "loss": 0.6766,
+      "step": 270600
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 2.9427087300257438e-05,
+      "loss": 0.7658,
+      "step": 270610
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 2.942632565082944e-05,
+      "loss": 0.8117,
+      "step": 270620
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 2.9425564001401433e-05,
+      "loss": 0.7636,
+      "step": 270630
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 2.9424802351973434e-05,
+      "loss": 0.7806,
+      "step": 270640
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 2.9424040702545434e-05,
+      "loss": 0.7831,
+      "step": 270650
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 2.9423279053117432e-05,
+      "loss": 0.7298,
+      "step": 270660
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 2.9422517403689433e-05,
+      "loss": 0.7218,
+      "step": 270670
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 2.942175575426143e-05,
+      "loss": 0.8158,
+      "step": 270680
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 2.9420994104833428e-05,
+      "loss": 0.7817,
+      "step": 270690
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 2.942023245540543e-05,
+      "loss": 0.7615,
+      "step": 270700
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 2.941947080597743e-05,
+      "loss": 0.8028,
+      "step": 270710
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 2.9418709156549424e-05,
+      "loss": 0.8227,
+      "step": 270720
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 2.9417947507121424e-05,
+      "loss": 0.7367,
+      "step": 270730
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 2.9417185857693425e-05,
+      "loss": 0.7617,
+      "step": 270740
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 2.941642420826542e-05,
+      "loss": 0.8131,
+      "step": 270750
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 2.941566255883742e-05,
+      "loss": 0.7191,
+      "step": 270760
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 2.941490090940942e-05,
+      "loss": 0.8634,
+      "step": 270770
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 2.9414139259981415e-05,
+      "loss": 0.8483,
+      "step": 270780
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 2.9413377610553416e-05,
+      "loss": 0.8225,
+      "step": 270790
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 2.9412615961125417e-05,
+      "loss": 0.7575,
+      "step": 270800
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 2.941185431169741e-05,
+      "loss": 0.8004,
+      "step": 270810
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 2.9411092662269412e-05,
+      "loss": 0.8008,
+      "step": 270820
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 2.9410331012841413e-05,
+      "loss": 0.7321,
+      "step": 270830
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 2.940956936341341e-05,
+      "loss": 0.7753,
+      "step": 270840
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 2.9408807713985408e-05,
+      "loss": 0.8326,
+      "step": 270850
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 2.940804606455741e-05,
+      "loss": 0.7723,
+      "step": 270860
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 2.9407284415129406e-05,
+      "loss": 0.8301,
+      "step": 270870
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 2.9406522765701407e-05,
+      "loss": 0.7797,
+      "step": 270880
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 2.94057611162734e-05,
+      "loss": 0.7731,
+      "step": 270890
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 2.94049994668454e-05,
+      "loss": 0.8361,
+      "step": 270900
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 2.9404237817417402e-05,
+      "loss": 0.8562,
+      "step": 270910
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 2.9403476167989397e-05,
+      "loss": 0.7791,
+      "step": 270920
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 2.9402714518561397e-05,
+      "loss": 0.7644,
+      "step": 270930
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 2.9401952869133398e-05,
+      "loss": 0.7822,
+      "step": 270940
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 2.9401191219705392e-05,
+      "loss": 0.7744,
+      "step": 270950
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 2.9400429570277393e-05,
+      "loss": 0.8502,
+      "step": 270960
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 2.9399667920849394e-05,
+      "loss": 0.7653,
+      "step": 270970
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 2.9398906271421388e-05,
+      "loss": 0.8264,
+      "step": 270980
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 2.939814462199339e-05,
+      "loss": 0.8136,
+      "step": 270990
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 2.939738297256539e-05,
+      "loss": 0.7948,
+      "step": 271000
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 2.9396621323137384e-05,
+      "loss": 0.9017,
+      "step": 271010
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 2.9395859673709385e-05,
+      "loss": 0.798,
+      "step": 271020
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 2.9395098024281386e-05,
+      "loss": 0.7648,
+      "step": 271030
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 2.9394336374853383e-05,
+      "loss": 0.7828,
+      "step": 271040
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 2.9393574725425384e-05,
+      "loss": 0.8847,
+      "step": 271050
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 2.939281307599738e-05,
+      "loss": 0.7274,
+      "step": 271060
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 2.939205142656938e-05,
+      "loss": 0.7741,
+      "step": 271070
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 2.939128977714138e-05,
+      "loss": 0.8884,
+      "step": 271080
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 2.939052812771338e-05,
+      "loss": 0.7788,
+      "step": 271090
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 2.9389766478285375e-05,
+      "loss": 0.778,
+      "step": 271100
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 2.9389004828857376e-05,
+      "loss": 0.7647,
+      "step": 271110
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 2.9388243179429376e-05,
+      "loss": 0.7996,
+      "step": 271120
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 2.938748153000137e-05,
+      "loss": 0.797,
+      "step": 271130
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 2.938671988057337e-05,
+      "loss": 0.7295,
+      "step": 271140
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 2.9385958231145372e-05,
+      "loss": 0.8515,
+      "step": 271150
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 2.9385196581717366e-05,
+      "loss": 0.7325,
+      "step": 271160
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 2.9384434932289367e-05,
+      "loss": 0.7749,
+      "step": 271170
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 2.9383673282861368e-05,
+      "loss": 0.8531,
+      "step": 271180
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 2.9382911633433362e-05,
+      "loss": 0.7829,
+      "step": 271190
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 2.9382149984005363e-05,
+      "loss": 0.7667,
+      "step": 271200
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 2.9381388334577364e-05,
+      "loss": 0.7503,
+      "step": 271210
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 2.9380626685149358e-05,
+      "loss": 0.8817,
+      "step": 271220
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 2.937986503572136e-05,
+      "loss": 0.9079,
+      "step": 271230
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 2.937910338629336e-05,
+      "loss": 0.8155,
+      "step": 271240
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 2.9378341736865357e-05,
+      "loss": 0.7547,
+      "step": 271250
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 2.9377580087437358e-05,
+      "loss": 0.8615,
+      "step": 271260
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 2.9376818438009355e-05,
+      "loss": 0.7696,
+      "step": 271270
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 2.9376056788581353e-05,
+      "loss": 0.8937,
+      "step": 271280
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 2.9375295139153354e-05,
+      "loss": 0.8118,
+      "step": 271290
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 2.9374533489725355e-05,
+      "loss": 0.6443,
+      "step": 271300
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 2.937377184029735e-05,
+      "loss": 0.7633,
+      "step": 271310
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 2.937301019086935e-05,
+      "loss": 0.7827,
+      "step": 271320
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 2.937224854144135e-05,
+      "loss": 0.6848,
+      "step": 271330
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 2.9371486892013344e-05,
+      "loss": 0.7507,
+      "step": 271340
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 2.9370725242585345e-05,
+      "loss": 0.7755,
+      "step": 271350
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 2.936996359315734e-05,
+      "loss": 0.7624,
+      "step": 271360
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 2.936920194372934e-05,
+      "loss": 0.7364,
+      "step": 271370
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 2.936844029430134e-05,
+      "loss": 0.8059,
+      "step": 271380
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 2.9367678644873335e-05,
+      "loss": 0.7351,
+      "step": 271390
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 2.9366916995445336e-05,
+      "loss": 0.8163,
+      "step": 271400
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 2.9366155346017337e-05,
+      "loss": 0.8038,
+      "step": 271410
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 2.9365393696589334e-05,
+      "loss": 0.8053,
+      "step": 271420
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 2.9364632047161332e-05,
+      "loss": 0.8137,
+      "step": 271430
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 2.9363870397733333e-05,
+      "loss": 0.6364,
+      "step": 271440
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 2.936310874830533e-05,
+      "loss": 0.7733,
+      "step": 271450
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 2.936234709887733e-05,
+      "loss": 0.7869,
+      "step": 271460
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 2.9361585449449332e-05,
+      "loss": 0.8421,
+      "step": 271470
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 2.9360823800021326e-05,
+      "loss": 0.7092,
+      "step": 271480
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 2.9360062150593327e-05,
+      "loss": 0.8529,
+      "step": 271490
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 2.9359300501165328e-05,
+      "loss": 0.7808,
+      "step": 271500
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 2.935853885173732e-05,
+      "loss": 0.726,
+      "step": 271510
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 2.9357777202309323e-05,
+      "loss": 0.8132,
+      "step": 271520
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 2.9357015552881323e-05,
+      "loss": 0.8478,
+      "step": 271530
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 2.9356253903453317e-05,
+      "loss": 0.7603,
+      "step": 271540
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 2.9355492254025318e-05,
+      "loss": 0.7983,
+      "step": 271550
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 2.935473060459732e-05,
+      "loss": 0.8044,
+      "step": 271560
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 2.9353968955169313e-05,
+      "loss": 0.878,
+      "step": 271570
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 2.9353207305741314e-05,
+      "loss": 0.7943,
+      "step": 271580
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 2.9352445656313315e-05,
+      "loss": 0.7684,
+      "step": 271590
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 2.935168400688531e-05,
+      "loss": 0.8459,
+      "step": 271600
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 2.935092235745731e-05,
+      "loss": 0.7498,
+      "step": 271610
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 2.935016070802931e-05,
+      "loss": 0.8068,
+      "step": 271620
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 2.9349399058601308e-05,
+      "loss": 0.8547,
+      "step": 271630
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 2.934863740917331e-05,
+      "loss": 0.8342,
+      "step": 271640
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 2.9347875759745307e-05,
+      "loss": 0.7882,
+      "step": 271650
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 2.9347114110317304e-05,
+      "loss": 0.7968,
+      "step": 271660
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 2.9346352460889305e-05,
+      "loss": 0.8478,
+      "step": 271670
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 2.9345590811461306e-05,
+      "loss": 0.732,
+      "step": 271680
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 2.93448291620333e-05,
+      "loss": 0.7884,
+      "step": 271690
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 2.93440675126053e-05,
+      "loss": 0.7441,
+      "step": 271700
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 2.93433058631773e-05,
+      "loss": 0.7513,
+      "step": 271710
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 2.9342544213749296e-05,
+      "loss": 0.8149,
+      "step": 271720
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 2.9341782564321296e-05,
+      "loss": 0.7192,
+      "step": 271730
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 2.9341020914893297e-05,
+      "loss": 0.8353,
+      "step": 271740
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 2.934025926546529e-05,
+      "loss": 0.7659,
+      "step": 271750
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 2.9339497616037292e-05,
+      "loss": 0.8195,
+      "step": 271760
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 2.9338735966609293e-05,
+      "loss": 0.8166,
+      "step": 271770
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 2.9337974317181287e-05,
+      "loss": 0.8807,
+      "step": 271780
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 2.9337212667753288e-05,
+      "loss": 0.8215,
+      "step": 271790
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 2.933645101832529e-05,
+      "loss": 0.8518,
+      "step": 271800
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 2.9335689368897283e-05,
+      "loss": 0.7315,
+      "step": 271810
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 2.9334927719469284e-05,
+      "loss": 0.7918,
+      "step": 271820
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 2.933416607004128e-05,
+      "loss": 0.8138,
+      "step": 271830
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 2.9333404420613282e-05,
+      "loss": 0.7591,
+      "step": 271840
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 2.9332642771185283e-05,
+      "loss": 0.8774,
+      "step": 271850
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 2.9331881121757277e-05,
+      "loss": 0.8503,
+      "step": 271860
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 2.9331119472329278e-05,
+      "loss": 0.7644,
+      "step": 271870
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 2.933035782290128e-05,
+      "loss": 0.8894,
+      "step": 271880
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 2.9329596173473273e-05,
+      "loss": 0.9238,
+      "step": 271890
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 2.9328834524045274e-05,
+      "loss": 0.82,
+      "step": 271900
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 2.9328072874617275e-05,
+      "loss": 0.831,
+      "step": 271910
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 2.932731122518927e-05,
+      "loss": 0.8841,
+      "step": 271920
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 2.932654957576127e-05,
+      "loss": 0.8243,
+      "step": 271930
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 2.932578792633327e-05,
+      "loss": 0.7666,
+      "step": 271940
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 2.9325026276905264e-05,
+      "loss": 0.8423,
+      "step": 271950
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 2.9324264627477265e-05,
+      "loss": 0.8338,
+      "step": 271960
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 2.9323502978049266e-05,
+      "loss": 0.8115,
+      "step": 271970
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 2.932274132862126e-05,
+      "loss": 0.8887,
+      "step": 271980
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 2.932197967919326e-05,
+      "loss": 0.7262,
+      "step": 271990
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 2.9321218029765262e-05,
+      "loss": 0.8218,
+      "step": 272000
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 2.932045638033726e-05,
+      "loss": 0.8581,
+      "step": 272010
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 2.9319694730909257e-05,
+      "loss": 0.7234,
+      "step": 272020
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 2.9318933081481258e-05,
+      "loss": 0.819,
+      "step": 272030
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 2.9318171432053255e-05,
+      "loss": 0.7816,
+      "step": 272040
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 2.9317409782625256e-05,
+      "loss": 0.8404,
+      "step": 272050
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 2.9316648133197257e-05,
+      "loss": 0.756,
+      "step": 272060
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 2.931588648376925e-05,
+      "loss": 0.7659,
+      "step": 272070
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 2.9315124834341252e-05,
+      "loss": 0.9023,
+      "step": 272080
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 2.9314363184913253e-05,
+      "loss": 0.8524,
+      "step": 272090
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 2.9313601535485247e-05,
+      "loss": 0.7502,
+      "step": 272100
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 2.9312839886057248e-05,
+      "loss": 0.7737,
+      "step": 272110
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 2.931207823662925e-05,
+      "loss": 0.9293,
+      "step": 272120
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 2.9311316587201243e-05,
+      "loss": 0.8226,
+      "step": 272130
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 2.9310554937773243e-05,
+      "loss": 0.752,
+      "step": 272140
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 2.9309793288345244e-05,
+      "loss": 0.7993,
+      "step": 272150
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 2.9309031638917238e-05,
+      "loss": 0.8777,
+      "step": 272160
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 2.930826998948924e-05,
+      "loss": 0.7579,
+      "step": 272170
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 2.930750834006124e-05,
+      "loss": 0.7902,
+      "step": 272180
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 2.9306746690633234e-05,
+      "loss": 0.7281,
+      "step": 272190
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 2.9305985041205235e-05,
+      "loss": 0.8848,
+      "step": 272200
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 2.9305223391777236e-05,
+      "loss": 0.8111,
+      "step": 272210
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 2.9304461742349233e-05,
+      "loss": 0.7881,
+      "step": 272220
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 2.930370009292123e-05,
+      "loss": 0.8336,
+      "step": 272230
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 2.930293844349323e-05,
+      "loss": 0.8073,
+      "step": 272240
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 2.930217679406523e-05,
+      "loss": 0.7308,
+      "step": 272250
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 2.930141514463723e-05,
+      "loss": 0.773,
+      "step": 272260
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 2.9300653495209224e-05,
+      "loss": 0.8521,
+      "step": 272270
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 2.9299891845781225e-05,
+      "loss": 0.763,
+      "step": 272280
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 2.9299130196353226e-05,
+      "loss": 0.7984,
+      "step": 272290
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 2.929836854692522e-05,
+      "loss": 0.8236,
+      "step": 272300
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 2.929760689749722e-05,
+      "loss": 0.7931,
+      "step": 272310
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 2.929684524806922e-05,
+      "loss": 0.7988,
+      "step": 272320
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 2.9296083598641216e-05,
+      "loss": 0.7552,
+      "step": 272330
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 2.9295321949213216e-05,
+      "loss": 0.7812,
+      "step": 272340
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 2.9294560299785217e-05,
+      "loss": 0.8105,
+      "step": 272350
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 2.929379865035721e-05,
+      "loss": 0.8431,
+      "step": 272360
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 2.9293037000929212e-05,
+      "loss": 0.8593,
+      "step": 272370
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 2.9292275351501213e-05,
+      "loss": 0.8281,
+      "step": 272380
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 2.929151370207321e-05,
+      "loss": 0.8785,
+      "step": 272390
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 2.9290752052645208e-05,
+      "loss": 0.8439,
+      "step": 272400
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 2.928999040321721e-05,
+      "loss": 0.8082,
+      "step": 272410
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 2.9289228753789206e-05,
+      "loss": 0.8253,
+      "step": 272420
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 2.9288467104361207e-05,
+      "loss": 0.7703,
+      "step": 272430
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 2.9287705454933208e-05,
+      "loss": 0.8405,
+      "step": 272440
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 2.9286943805505202e-05,
+      "loss": 0.8678,
+      "step": 272450
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 2.9286182156077203e-05,
+      "loss": 0.8091,
+      "step": 272460
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 2.9285420506649204e-05,
+      "loss": 0.8916,
+      "step": 272470
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 2.9284658857221198e-05,
+      "loss": 0.8707,
+      "step": 272480
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 2.92838972077932e-05,
+      "loss": 0.8401,
+      "step": 272490
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 2.92831355583652e-05,
+      "loss": 0.7767,
+      "step": 272500
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 2.9282373908937194e-05,
+      "loss": 0.7612,
+      "step": 272510
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 2.9281612259509195e-05,
+      "loss": 0.8761,
+      "step": 272520
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 2.9280850610081195e-05,
+      "loss": 0.8107,
+      "step": 272530
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 2.928008896065319e-05,
+      "loss": 0.8677,
+      "step": 272540
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 2.927932731122519e-05,
+      "loss": 0.7862,
+      "step": 272550
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 2.927856566179719e-05,
+      "loss": 0.9302,
+      "step": 272560
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 2.9277804012369185e-05,
+      "loss": 0.8294,
+      "step": 272570
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 2.9277042362941186e-05,
+      "loss": 0.8378,
+      "step": 272580
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 2.9276280713513187e-05,
+      "loss": 0.7276,
+      "step": 272590
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 2.9275519064085184e-05,
+      "loss": 0.8176,
+      "step": 272600
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 2.9274757414657182e-05,
+      "loss": 0.7303,
+      "step": 272610
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 2.9273995765229183e-05,
+      "loss": 0.8533,
+      "step": 272620
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 2.927323411580118e-05,
+      "loss": 0.7693,
+      "step": 272630
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 2.927247246637318e-05,
+      "loss": 0.8816,
+      "step": 272640
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 2.9271710816945182e-05,
+      "loss": 0.8535,
+      "step": 272650
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 2.9270949167517176e-05,
+      "loss": 0.8668,
+      "step": 272660
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 2.9270187518089177e-05,
+      "loss": 0.7943,
+      "step": 272670
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 2.9269425868661178e-05,
+      "loss": 0.792,
+      "step": 272680
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 2.9268664219233172e-05,
+      "loss": 0.7525,
+      "step": 272690
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 2.9267902569805173e-05,
+      "loss": 0.835,
+      "step": 272700
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 2.9267140920377174e-05,
+      "loss": 0.7722,
+      "step": 272710
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 2.9266379270949168e-05,
+      "loss": 0.805,
+      "step": 272720
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 2.926561762152117e-05,
+      "loss": 0.8318,
+      "step": 272730
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 2.9264855972093163e-05,
+      "loss": 0.759,
+      "step": 272740
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 2.9264094322665163e-05,
+      "loss": 0.8541,
+      "step": 272750
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 2.9263332673237164e-05,
+      "loss": 0.8107,
+      "step": 272760
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 2.926257102380916e-05,
+      "loss": 0.7874,
+      "step": 272770
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 2.926180937438116e-05,
+      "loss": 0.8749,
+      "step": 272780
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 2.926104772495316e-05,
+      "loss": 0.7698,
+      "step": 272790
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 2.9260286075525157e-05,
+      "loss": 0.7661,
+      "step": 272800
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 2.925952442609716e-05,
+      "loss": 0.7736,
+      "step": 272810
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 2.9258762776669156e-05,
+      "loss": 0.9001,
+      "step": 272820
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 2.9258001127241153e-05,
+      "loss": 0.7763,
+      "step": 272830
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 2.9257239477813154e-05,
+      "loss": 0.8808,
+      "step": 272840
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 2.9256477828385155e-05,
+      "loss": 0.8475,
+      "step": 272850
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 2.925571617895715e-05,
+      "loss": 0.8685,
+      "step": 272860
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 2.925495452952915e-05,
+      "loss": 0.7946,
+      "step": 272870
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 2.925419288010115e-05,
+      "loss": 0.7152,
+      "step": 272880
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 2.9253431230673145e-05,
+      "loss": 0.7507,
+      "step": 272890
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 2.9252669581245146e-05,
+      "loss": 0.7372,
+      "step": 272900
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 2.9251907931817147e-05,
+      "loss": 0.9059,
+      "step": 272910
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 2.925114628238914e-05,
+      "loss": 0.8381,
+      "step": 272920
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 2.925038463296114e-05,
+      "loss": 0.7987,
+      "step": 272930
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 2.9249622983533142e-05,
+      "loss": 0.7929,
+      "step": 272940
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 2.9248861334105136e-05,
+      "loss": 0.8215,
+      "step": 272950
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 2.9248099684677137e-05,
+      "loss": 0.8286,
+      "step": 272960
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 2.9247338035249138e-05,
+      "loss": 0.836,
+      "step": 272970
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 2.9246576385821132e-05,
+      "loss": 0.778,
+      "step": 272980
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 2.9245814736393133e-05,
+      "loss": 0.8538,
+      "step": 272990
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 2.9245053086965134e-05,
+      "loss": 0.7612,
+      "step": 273000
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 2.924429143753713e-05,
+      "loss": 0.7739,
+      "step": 273010
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 2.9243529788109132e-05,
+      "loss": 0.8412,
+      "step": 273020
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 2.9242768138681133e-05,
+      "loss": 0.7486,
+      "step": 273030
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 2.9242006489253127e-05,
+      "loss": 0.7916,
+      "step": 273040
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 2.9241244839825128e-05,
+      "loss": 0.7906,
+      "step": 273050
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 2.924048319039713e-05,
+      "loss": 0.8666,
+      "step": 273060
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 2.9239721540969123e-05,
+      "loss": 0.7607,
+      "step": 273070
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 2.9238959891541124e-05,
+      "loss": 0.7824,
+      "step": 273080
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 2.9238198242113125e-05,
+      "loss": 0.912,
+      "step": 273090
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 2.923743659268512e-05,
+      "loss": 0.792,
+      "step": 273100
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 2.923667494325712e-05,
+      "loss": 0.7848,
+      "step": 273110
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 2.923591329382912e-05,
+      "loss": 0.7309,
+      "step": 273120
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 2.9235151644401115e-05,
+      "loss": 0.7606,
+      "step": 273130
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 2.9234389994973115e-05,
+      "loss": 0.7519,
+      "step": 273140
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 2.9233628345545116e-05,
+      "loss": 0.8004,
+      "step": 273150
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 2.923286669611711e-05,
+      "loss": 0.8003,
+      "step": 273160
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 2.923210504668911e-05,
+      "loss": 0.8023,
+      "step": 273170
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 2.9231343397261112e-05,
+      "loss": 0.7844,
+      "step": 273180
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 2.923058174783311e-05,
+      "loss": 0.7848,
+      "step": 273190
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 2.9229820098405107e-05,
+      "loss": 0.8437,
+      "step": 273200
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 2.9229058448977104e-05,
+      "loss": 0.771,
+      "step": 273210
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 2.9228296799549105e-05,
+      "loss": 0.8143,
+      "step": 273220
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 2.9227535150121106e-05,
+      "loss": 0.7691,
+      "step": 273230
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 2.92267735006931e-05,
+      "loss": 0.8343,
+      "step": 273240
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 2.92260118512651e-05,
+      "loss": 0.8483,
+      "step": 273250
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 2.9225250201837102e-05,
+      "loss": 0.8325,
+      "step": 273260
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 2.9224488552409096e-05,
+      "loss": 0.8201,
+      "step": 273270
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 2.9223726902981097e-05,
+      "loss": 0.8354,
+      "step": 273280
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 2.9222965253553098e-05,
+      "loss": 0.8,
+      "step": 273290
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 2.9222203604125092e-05,
+      "loss": 0.9761,
+      "step": 273300
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 2.9221441954697093e-05,
+      "loss": 0.7433,
+      "step": 273310
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 2.9220680305269094e-05,
+      "loss": 0.815,
+      "step": 273320
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 2.9219918655841088e-05,
+      "loss": 0.7505,
+      "step": 273330
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 2.921915700641309e-05,
+      "loss": 0.7852,
+      "step": 273340
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 2.921839535698509e-05,
+      "loss": 0.8051,
+      "step": 273350
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 2.9217633707557083e-05,
+      "loss": 0.8475,
+      "step": 273360
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 2.9216872058129084e-05,
+      "loss": 0.9174,
+      "step": 273370
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 2.9216110408701085e-05,
+      "loss": 0.7969,
+      "step": 273380
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 2.9215348759273083e-05,
+      "loss": 0.8017,
+      "step": 273390
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 2.9214587109845083e-05,
+      "loss": 0.7313,
+      "step": 273400
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 2.921382546041708e-05,
+      "loss": 0.8335,
+      "step": 273410
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 2.921306381098908e-05,
+      "loss": 0.7777,
+      "step": 273420
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 2.921230216156108e-05,
+      "loss": 0.8445,
+      "step": 273430
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 2.921154051213308e-05,
+      "loss": 0.7887,
+      "step": 273440
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 2.9210778862705074e-05,
+      "loss": 0.8459,
+      "step": 273450
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 2.9210017213277075e-05,
+      "loss": 0.8296,
+      "step": 273460
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 2.9209255563849076e-05,
+      "loss": 0.8134,
+      "step": 273470
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 2.920849391442107e-05,
+      "loss": 0.8426,
+      "step": 273480
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 2.920773226499307e-05,
+      "loss": 0.8253,
+      "step": 273490
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 2.920697061556507e-05,
+      "loss": 0.8042,
+      "step": 273500
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 2.9206208966137066e-05,
+      "loss": 0.7872,
+      "step": 273510
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 2.9205447316709067e-05,
+      "loss": 0.7496,
+      "step": 273520
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 2.9204685667281067e-05,
+      "loss": 0.8564,
+      "step": 273530
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 2.920392401785306e-05,
+      "loss": 0.9198,
+      "step": 273540
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 2.9203162368425062e-05,
+      "loss": 0.732,
+      "step": 273550
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 2.9202400718997063e-05,
+      "loss": 0.8519,
+      "step": 273560
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 2.9201639069569057e-05,
+      "loss": 0.8209,
+      "step": 273570
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 2.9200877420141058e-05,
+      "loss": 0.8897,
+      "step": 273580
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 2.920011577071306e-05,
+      "loss": 0.8244,
+      "step": 273590
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 2.9199354121285056e-05,
+      "loss": 0.8229,
+      "step": 273600
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 2.9198592471857057e-05,
+      "loss": 0.8011,
+      "step": 273610
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 2.9197830822429055e-05,
+      "loss": 0.8133,
+      "step": 273620
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 2.9197069173001052e-05,
+      "loss": 0.8066,
+      "step": 273630
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 2.9196307523573053e-05,
+      "loss": 0.711,
+      "step": 273640
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 2.9195545874145047e-05,
+      "loss": 0.7844,
+      "step": 273650
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 2.9194784224717048e-05,
+      "loss": 0.8431,
+      "step": 273660
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 2.919402257528905e-05,
+      "loss": 0.8917,
+      "step": 273670
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 2.9193260925861043e-05,
+      "loss": 0.8082,
+      "step": 273680
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 2.9192499276433044e-05,
+      "loss": 0.8222,
+      "step": 273690
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 2.9191737627005045e-05,
+      "loss": 0.8752,
+      "step": 273700
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 2.919097597757704e-05,
+      "loss": 0.8486,
+      "step": 273710
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 2.919021432814904e-05,
+      "loss": 0.8024,
+      "step": 273720
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 2.918945267872104e-05,
+      "loss": 0.8103,
+      "step": 273730
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 2.9188691029293035e-05,
+      "loss": 0.7946,
+      "step": 273740
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 2.9187929379865035e-05,
+      "loss": 0.7409,
+      "step": 273750
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 2.9187167730437036e-05,
+      "loss": 0.7679,
+      "step": 273760
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 2.9186406081009034e-05,
+      "loss": 0.845,
+      "step": 273770
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 2.9185644431581035e-05,
+      "loss": 0.6931,
+      "step": 273780
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 2.9184882782153032e-05,
+      "loss": 0.8172,
+      "step": 273790
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 2.918412113272503e-05,
+      "loss": 0.8433,
+      "step": 273800
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 2.918335948329703e-05,
+      "loss": 0.8559,
+      "step": 273810
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 2.918259783386903e-05,
+      "loss": 0.804,
+      "step": 273820
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 2.9181836184441025e-05,
+      "loss": 0.9018,
+      "step": 273830
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 2.9181074535013026e-05,
+      "loss": 0.7424,
+      "step": 273840
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 2.9180312885585027e-05,
+      "loss": 0.7779,
+      "step": 273850
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 2.917955123615702e-05,
+      "loss": 0.8532,
+      "step": 273860
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 2.9178789586729022e-05,
+      "loss": 0.8914,
+      "step": 273870
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 2.9178027937301023e-05,
+      "loss": 0.8748,
+      "step": 273880
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 2.9177266287873017e-05,
+      "loss": 0.7587,
+      "step": 273890
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 2.9176504638445018e-05,
+      "loss": 0.848,
+      "step": 273900
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 2.917574298901702e-05,
+      "loss": 0.7914,
+      "step": 273910
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 2.9174981339589013e-05,
+      "loss": 0.8298,
+      "step": 273920
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 2.9174219690161014e-05,
+      "loss": 0.8033,
+      "step": 273930
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 2.9173458040733014e-05,
+      "loss": 0.8305,
+      "step": 273940
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 2.917269639130501e-05,
+      "loss": 0.8036,
+      "step": 273950
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 2.917193474187701e-05,
+      "loss": 0.752,
+      "step": 273960
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 2.917117309244901e-05,
+      "loss": 0.8289,
+      "step": 273970
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 2.9170411443021008e-05,
+      "loss": 0.7436,
+      "step": 273980
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 2.916964979359301e-05,
+      "loss": 0.9124,
+      "step": 273990
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 2.9168888144165006e-05,
+      "loss": 0.8025,
+      "step": 274000
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 2.9168126494737003e-05,
+      "loss": 0.8039,
+      "step": 274010
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 2.9167364845309004e-05,
+      "loss": 0.7807,
+      "step": 274020
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 2.9166603195881005e-05,
+      "loss": 0.8668,
+      "step": 274030
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 2.9165841546453e-05,
+      "loss": 0.8034,
+      "step": 274040
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 2.9165079897025e-05,
+      "loss": 0.7993,
+      "step": 274050
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 2.9164318247597e-05,
+      "loss": 0.7623,
+      "step": 274060
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 2.9163556598168995e-05,
+      "loss": 0.7978,
+      "step": 274070
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 2.9162794948740996e-05,
+      "loss": 0.8648,
+      "step": 274080
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 2.9162033299312997e-05,
+      "loss": 0.809,
+      "step": 274090
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 2.916127164988499e-05,
+      "loss": 0.7831,
+      "step": 274100
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 2.916051000045699e-05,
+      "loss": 0.7484,
+      "step": 274110
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 2.9159748351028986e-05,
+      "loss": 0.8779,
+      "step": 274120
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 2.9158986701600987e-05,
+      "loss": 0.9084,
+      "step": 274130
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 2.9158225052172987e-05,
+      "loss": 0.8119,
+      "step": 274140
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 2.9157463402744985e-05,
+      "loss": 0.747,
+      "step": 274150
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 2.9156701753316982e-05,
+      "loss": 0.7587,
+      "step": 274160
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 2.9155940103888983e-05,
+      "loss": 0.7711,
+      "step": 274170
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 2.915517845446098e-05,
+      "loss": 0.7569,
+      "step": 274180
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 2.915441680503298e-05,
+      "loss": 0.7843,
+      "step": 274190
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 2.9153655155604982e-05,
+      "loss": 0.8236,
+      "step": 274200
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 2.9152893506176976e-05,
+      "loss": 0.8545,
+      "step": 274210
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 2.9152131856748977e-05,
+      "loss": 0.7964,
+      "step": 274220
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 2.9151370207320978e-05,
+      "loss": 0.783,
+      "step": 274230
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 2.9150608557892972e-05,
+      "loss": 0.8985,
+      "step": 274240
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 2.9149846908464973e-05,
+      "loss": 0.7678,
+      "step": 274250
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 2.9149085259036974e-05,
+      "loss": 0.8899,
+      "step": 274260
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 2.9148323609608968e-05,
+      "loss": 0.7544,
+      "step": 274270
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 2.914756196018097e-05,
+      "loss": 0.7709,
+      "step": 274280
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 2.914680031075297e-05,
+      "loss": 0.8067,
+      "step": 274290
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 2.9146038661324964e-05,
+      "loss": 0.8291,
+      "step": 274300
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 2.9145277011896965e-05,
+      "loss": 0.866,
+      "step": 274310
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 2.9144515362468966e-05,
+      "loss": 0.8961,
+      "step": 274320
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 2.914375371304096e-05,
+      "loss": 0.7443,
+      "step": 274330
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 2.914299206361296e-05,
+      "loss": 0.8251,
+      "step": 274340
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 2.914223041418496e-05,
+      "loss": 0.8085,
+      "step": 274350
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 2.914146876475696e-05,
+      "loss": 0.7905,
+      "step": 274360
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 2.9140707115328956e-05,
+      "loss": 0.7949,
+      "step": 274370
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 2.9139945465900957e-05,
+      "loss": 0.8142,
+      "step": 274380
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 2.9139183816472955e-05,
+      "loss": 0.8444,
+      "step": 274390
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 2.9138422167044955e-05,
+      "loss": 0.8106,
+      "step": 274400
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 2.9137660517616956e-05,
+      "loss": 0.8787,
+      "step": 274410
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 2.913689886818895e-05,
+      "loss": 0.7813,
+      "step": 274420
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 2.913613721876095e-05,
+      "loss": 0.7731,
+      "step": 274430
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 2.9135375569332952e-05,
+      "loss": 0.8364,
+      "step": 274440
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 2.9134613919904946e-05,
+      "loss": 0.7201,
+      "step": 274450
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 2.9133852270476947e-05,
+      "loss": 0.7966,
+      "step": 274460
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 2.9133090621048948e-05,
+      "loss": 0.8303,
+      "step": 274470
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 2.9132328971620942e-05,
+      "loss": 0.7638,
+      "step": 274480
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 2.9131567322192943e-05,
+      "loss": 0.8639,
+      "step": 274490
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 2.9130805672764944e-05,
+      "loss": 0.7584,
+      "step": 274500
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 2.9130044023336938e-05,
+      "loss": 0.7806,
+      "step": 274510
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 2.912928237390894e-05,
+      "loss": 0.8529,
+      "step": 274520
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 2.912852072448094e-05,
+      "loss": 0.8753,
+      "step": 274530
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 2.9127759075052934e-05,
+      "loss": 0.7993,
+      "step": 274540
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 2.9126997425624934e-05,
+      "loss": 0.7435,
+      "step": 274550
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 2.9126235776196935e-05,
+      "loss": 0.7642,
+      "step": 274560
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 2.9125474126768933e-05,
+      "loss": 0.8353,
+      "step": 274570
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 2.9124712477340934e-05,
+      "loss": 0.8335,
+      "step": 274580
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 2.9123950827912928e-05,
+      "loss": 0.8402,
+      "step": 274590
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 2.912318917848493e-05,
+      "loss": 0.8384,
+      "step": 274600
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 2.912242752905693e-05,
+      "loss": 0.7841,
+      "step": 274610
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 2.9121665879628923e-05,
+      "loss": 0.8254,
+      "step": 274620
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 2.9120904230200924e-05,
+      "loss": 0.7809,
+      "step": 274630
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 2.9120142580772925e-05,
+      "loss": 0.8504,
+      "step": 274640
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 2.911938093134492e-05,
+      "loss": 0.8602,
+      "step": 274650
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 2.911861928191692e-05,
+      "loss": 0.7709,
+      "step": 274660
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 2.911785763248892e-05,
+      "loss": 0.7305,
+      "step": 274670
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 2.9117095983060915e-05,
+      "loss": 0.797,
+      "step": 274680
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 2.9116334333632916e-05,
+      "loss": 0.7554,
+      "step": 274690
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 2.9115572684204917e-05,
+      "loss": 0.8497,
+      "step": 274700
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 2.911481103477691e-05,
+      "loss": 0.7475,
+      "step": 274710
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 2.911404938534891e-05,
+      "loss": 0.7802,
+      "step": 274720
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 2.9113287735920913e-05,
+      "loss": 0.8756,
+      "step": 274730
+    },
+    {
+      "epoch": 12.55,
+      "learning_rate": 2.911252608649291e-05,
+      "loss": 0.8133,
+      "step": 274740
+    },
+    {
+      "epoch": 12.55,
+      "learning_rate": 2.9111764437064907e-05,
+      "loss": 0.7973,
+      "step": 274750
+    },
+    {
+      "epoch": 12.55,
+      "learning_rate": 2.9111002787636908e-05,
+      "loss": 0.9149,
+      "step": 274760
+    },
+    {
+      "epoch": 12.55,
+      "learning_rate": 2.9110241138208906e-05,
+      "loss": 0.7731,
+      "step": 274770
+    },
+    {
+      "epoch": 12.55,
+      "learning_rate": 2.9109479488780907e-05,
+      "loss": 0.7097,
+      "step": 274780
+    },
+    {
+      "epoch": 12.55,
+      "learning_rate": 2.9108717839352907e-05,
+      "loss": 0.8356,
+      "step": 274790
+    },
+    {
+      "epoch": 12.55,
+      "learning_rate": 2.91079561899249e-05,
+      "loss": 0.8492,
+      "step": 274800
+    },
+    {
+      "epoch": 12.55,
+      "learning_rate": 2.9107194540496902e-05,
+      "loss": 0.7925,
+      "step": 274810
+    },
+    {
+      "epoch": 12.55,
+      "learning_rate": 2.9106432891068903e-05,
+      "loss": 0.814,
+      "step": 274820
+    },
+    {
+      "epoch": 12.55,
+      "learning_rate": 2.9105671241640897e-05,
+      "loss": 0.7822,
+      "step": 274830
+    },
+    {
+      "epoch": 12.55,
+      "learning_rate": 2.9104909592212898e-05,
+      "loss": 0.7603,
+      "step": 274840
+    },
+    {
+      "epoch": 12.55,
+      "learning_rate": 2.91041479427849e-05,
+      "loss": 0.877,
+      "step": 274850
+    },
+    {
+      "epoch": 12.55,
+      "learning_rate": 2.9103386293356893e-05,
+      "loss": 0.9072,
+      "step": 274860
+    },
+    {
+      "epoch": 12.55,
+      "learning_rate": 2.9102624643928894e-05,
+      "loss": 0.7471,
+      "step": 274870
+    },
+    {
+      "epoch": 12.55,
+      "learning_rate": 2.9101862994500895e-05,
+      "loss": 0.8838,
+      "step": 274880
+    },
+    {
+      "epoch": 12.55,
+      "learning_rate": 2.910110134507289e-05,
+      "loss": 0.7425,
+      "step": 274890
+    },
+    {
+      "epoch": 12.55,
+      "learning_rate": 2.910033969564489e-05,
+      "loss": 0.7938,
+      "step": 274900
+    },
+    {
+      "epoch": 12.55,
+      "learning_rate": 2.909957804621689e-05,
+      "loss": 0.8632,
+      "step": 274910
+    },
+    {
+      "epoch": 12.55,
+      "learning_rate": 2.9098816396788885e-05,
+      "loss": 0.7817,
+      "step": 274920
+    },
+    {
+      "epoch": 12.55,
+      "learning_rate": 2.9098054747360886e-05,
+      "loss": 0.8441,
+      "step": 274930
+    },
+    {
+      "epoch": 12.55,
+      "learning_rate": 2.9097293097932886e-05,
+      "loss": 0.7696,
+      "step": 274940
+    },
+    {
+      "epoch": 12.55,
+      "learning_rate": 2.9096531448504884e-05,
+      "loss": 0.8189,
+      "step": 274950
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 2.909576979907688e-05,
+      "loss": 0.716,
+      "step": 274960
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 2.9095008149648882e-05,
+      "loss": 0.886,
+      "step": 274970
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 2.909424650022088e-05,
+      "loss": 0.7854,
+      "step": 274980
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 2.909348485079288e-05,
+      "loss": 0.8309,
+      "step": 274990
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 2.909272320136488e-05,
+      "loss": 0.7672,
+      "step": 275000
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 2.9091961551936875e-05,
+      "loss": 0.8025,
+      "step": 275010
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 2.9091199902508876e-05,
+      "loss": 0.764,
+      "step": 275020
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 2.909043825308087e-05,
+      "loss": 0.7946,
+      "step": 275030
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 2.908967660365287e-05,
+      "loss": 0.8733,
+      "step": 275040
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 2.9088914954224872e-05,
+      "loss": 0.8512,
+      "step": 275050
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 2.9088153304796866e-05,
+      "loss": 0.8432,
+      "step": 275060
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 2.9087391655368867e-05,
+      "loss": 0.8,
+      "step": 275070
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 2.9086630005940868e-05,
+      "loss": 0.8284,
+      "step": 275080
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 2.9085868356512862e-05,
+      "loss": 0.7963,
+      "step": 275090
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 2.9085106707084863e-05,
+      "loss": 0.8475,
+      "step": 275100
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 2.9084345057656864e-05,
+      "loss": 0.8524,
+      "step": 275110
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 2.9083583408228858e-05,
+      "loss": 0.8419,
+      "step": 275120
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 2.908282175880086e-05,
+      "loss": 0.8562,
+      "step": 275130
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 2.908206010937286e-05,
+      "loss": 0.8769,
+      "step": 275140
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 2.9081298459944857e-05,
+      "loss": 0.8662,
+      "step": 275150
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 2.9080536810516858e-05,
+      "loss": 0.7994,
+      "step": 275160
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 2.9079775161088855e-05,
+      "loss": 0.8875,
+      "step": 275170
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 2.9079013511660853e-05,
+      "loss": 0.8391,
+      "step": 275180
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 2.9078251862232854e-05,
+      "loss": 0.7862,
+      "step": 275190
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 2.9077490212804854e-05,
+      "loss": 0.8013,
+      "step": 275200
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 2.907672856337685e-05,
+      "loss": 0.8601,
+      "step": 275210
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 2.907596691394885e-05,
+      "loss": 0.8096,
+      "step": 275220
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 2.907520526452085e-05,
+      "loss": 0.7735,
+      "step": 275230
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 2.9074443615092844e-05,
+      "loss": 0.746,
+      "step": 275240
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 2.9073681965664845e-05,
+      "loss": 0.759,
+      "step": 275250
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 2.9072920316236846e-05,
+      "loss": 0.8335,
+      "step": 275260
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 2.907215866680884e-05,
+      "loss": 0.7974,
+      "step": 275270
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 2.907139701738084e-05,
+      "loss": 0.8099,
+      "step": 275280
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 2.9070635367952842e-05,
+      "loss": 0.9057,
+      "step": 275290
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 2.9069873718524836e-05,
+      "loss": 0.7733,
+      "step": 275300
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 2.9069112069096837e-05,
+      "loss": 0.9183,
+      "step": 275310
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 2.9068350419668838e-05,
+      "loss": 0.8589,
+      "step": 275320
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 2.9067588770240835e-05,
+      "loss": 0.7936,
+      "step": 275330
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 2.9066827120812833e-05,
+      "loss": 0.769,
+      "step": 275340
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 2.9066065471384833e-05,
+      "loss": 0.817,
+      "step": 275350
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 2.906530382195683e-05,
+      "loss": 0.7404,
+      "step": 275360
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 2.9064542172528832e-05,
+      "loss": 0.8642,
+      "step": 275370
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 2.9063780523100833e-05,
+      "loss": 0.8248,
+      "step": 275380
+    },
+    {
+      "epoch": 12.58,
+      "learning_rate": 2.9063018873672827e-05,
+      "loss": 0.7756,
+      "step": 275390
+    },
+    {
+      "epoch": 12.58,
+      "learning_rate": 2.9062257224244827e-05,
+      "loss": 0.8302,
+      "step": 275400
+    },
+    {
+      "epoch": 12.58,
+      "learning_rate": 2.906149557481683e-05,
+      "loss": 0.8213,
+      "step": 275410
+    },
+    {
+      "epoch": 12.58,
+      "learning_rate": 2.9060733925388822e-05,
+      "loss": 0.8671,
+      "step": 275420
+    },
+    {
+      "epoch": 12.58,
+      "learning_rate": 2.9059972275960823e-05,
+      "loss": 0.82,
+      "step": 275430
+    },
+    {
+      "epoch": 12.58,
+      "learning_rate": 2.9059210626532824e-05,
+      "loss": 0.7924,
+      "step": 275440
+    },
+    {
+      "epoch": 12.58,
+      "learning_rate": 2.9058448977104818e-05,
+      "loss": 0.7899,
+      "step": 275450
+    },
+    {
+      "epoch": 12.58,
+      "learning_rate": 2.905768732767682e-05,
+      "loss": 0.8399,
+      "step": 275460
+    },
+    {
+      "epoch": 12.58,
+      "learning_rate": 2.905692567824882e-05,
+      "loss": 0.7709,
+      "step": 275470
+    },
+    {
+      "epoch": 12.58,
+      "learning_rate": 2.9056164028820814e-05,
+      "loss": 0.7642,
+      "step": 275480
+    },
+    {
+      "epoch": 12.58,
+      "learning_rate": 2.9055402379392815e-05,
+      "loss": 0.7953,
+      "step": 275490
+    },
+    {
+      "epoch": 12.58,
+      "learning_rate": 2.905464072996481e-05,
+      "loss": 0.7207,
+      "step": 275500
+    },
+    {
+      "epoch": 12.58,
+      "learning_rate": 2.905387908053681e-05,
+      "loss": 0.8655,
+      "step": 275510
+    },
+    {
+      "epoch": 12.58,
+      "learning_rate": 2.905311743110881e-05,
+      "loss": 0.8149,
+      "step": 275520
+    },
+    {
+      "epoch": 12.58,
+      "learning_rate": 2.9052355781680808e-05,
+      "loss": 0.7796,
+      "step": 275530
+    },
+    {
+      "epoch": 12.58,
+      "learning_rate": 2.905159413225281e-05,
+      "loss": 0.8118,
+      "step": 275540
+    },
+    {
+      "epoch": 12.58,
+      "learning_rate": 2.9050832482824806e-05,
+      "loss": 0.9597,
+      "step": 275550
+    },
+    {
+      "epoch": 12.58,
+      "learning_rate": 2.9050070833396804e-05,
+      "loss": 0.7932,
+      "step": 275560
+    },
+    {
+      "epoch": 12.58,
+      "learning_rate": 2.9049309183968805e-05,
+      "loss": 0.8161,
+      "step": 275570
+    },
+    {
+      "epoch": 12.58,
+      "learning_rate": 2.9048547534540806e-05,
+      "loss": 0.724,
+      "step": 275580
+    },
+    {
+      "epoch": 12.58,
+      "learning_rate": 2.90477858851128e-05,
+      "loss": 0.8136,
+      "step": 275590
+    },
+    {
+      "epoch": 12.58,
+      "learning_rate": 2.90470242356848e-05,
+      "loss": 0.824,
+      "step": 275600
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 2.90462625862568e-05,
+      "loss": 0.8315,
+      "step": 275610
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 2.9045500936828795e-05,
+      "loss": 0.8225,
+      "step": 275620
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 2.9044739287400796e-05,
+      "loss": 0.7614,
+      "step": 275630
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 2.9043977637972797e-05,
+      "loss": 0.7557,
+      "step": 275640
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 2.904321598854479e-05,
+      "loss": 0.8867,
+      "step": 275650
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 2.9042454339116792e-05,
+      "loss": 0.8174,
+      "step": 275660
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 2.9041692689688793e-05,
+      "loss": 0.823,
+      "step": 275670
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 2.9040931040260787e-05,
+      "loss": 0.7671,
+      "step": 275680
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 2.9040169390832788e-05,
+      "loss": 0.8497,
+      "step": 275690
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 2.903940774140479e-05,
+      "loss": 0.8903,
+      "step": 275700
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 2.9038646091976783e-05,
+      "loss": 0.7904,
+      "step": 275710
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 2.9037884442548784e-05,
+      "loss": 0.7984,
+      "step": 275720
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 2.9037122793120785e-05,
+      "loss": 0.8033,
+      "step": 275730
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 2.9036361143692782e-05,
+      "loss": 0.8529,
+      "step": 275740
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 2.9035599494264783e-05,
+      "loss": 0.8683,
+      "step": 275750
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 2.903483784483678e-05,
+      "loss": 0.8545,
+      "step": 275760
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 2.9034076195408778e-05,
+      "loss": 0.8181,
+      "step": 275770
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 2.903331454598078e-05,
+      "loss": 0.7769,
+      "step": 275780
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 2.903255289655278e-05,
+      "loss": 0.8574,
+      "step": 275790
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 2.9031791247124774e-05,
+      "loss": 0.7629,
+      "step": 275800
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 2.9031029597696774e-05,
+      "loss": 0.9852,
+      "step": 275810
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 2.9030267948268775e-05,
+      "loss": 0.7671,
+      "step": 275820
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 2.902950629884077e-05,
+      "loss": 0.8192,
+      "step": 275830
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 2.902874464941277e-05,
+      "loss": 0.7325,
+      "step": 275840
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 2.902798299998477e-05,
+      "loss": 0.7644,
+      "step": 275850
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 2.9027221350556765e-05,
+      "loss": 0.9829,
+      "step": 275860
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 2.9026459701128766e-05,
+      "loss": 0.8469,
+      "step": 275870
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 2.9025698051700767e-05,
+      "loss": 0.7456,
+      "step": 275880
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 2.902493640227276e-05,
+      "loss": 0.7996,
+      "step": 275890
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 2.9024174752844762e-05,
+      "loss": 0.7525,
+      "step": 275900
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 2.9023413103416763e-05,
+      "loss": 0.757,
+      "step": 275910
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 2.9022651453988757e-05,
+      "loss": 0.7813,
+      "step": 275920
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 2.9021889804560758e-05,
+      "loss": 1.0031,
+      "step": 275930
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 2.902112815513276e-05,
+      "loss": 0.8079,
+      "step": 275940
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 2.9020366505704756e-05,
+      "loss": 0.8147,
+      "step": 275950
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 2.9019604856276757e-05,
+      "loss": 0.6881,
+      "step": 275960
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 2.901884320684875e-05,
+      "loss": 0.7526,
+      "step": 275970
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 2.9018081557420752e-05,
+      "loss": 0.7794,
+      "step": 275980
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 2.9017319907992753e-05,
+      "loss": 0.7384,
+      "step": 275990
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 2.9016558258564747e-05,
+      "loss": 0.7938,
+      "step": 276000
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 2.9015796609136748e-05,
+      "loss": 0.7715,
+      "step": 276010
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 2.901503495970875e-05,
+      "loss": 0.92,
+      "step": 276020
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 2.9014273310280742e-05,
+      "loss": 0.843,
+      "step": 276030
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 2.9013511660852743e-05,
+      "loss": 0.7621,
+      "step": 276040
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 2.9012750011424744e-05,
+      "loss": 0.868,
+      "step": 276050
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 2.9011988361996738e-05,
+      "loss": 0.8163,
+      "step": 276060
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 2.901122671256874e-05,
+      "loss": 0.8313,
+      "step": 276070
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 2.901046506314074e-05,
+      "loss": 0.8262,
+      "step": 276080
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 2.9009703413712734e-05,
+      "loss": 0.8112,
+      "step": 276090
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 2.9008941764284735e-05,
+      "loss": 0.7941,
+      "step": 276100
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 2.9008180114856736e-05,
+      "loss": 0.7108,
+      "step": 276110
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 2.9007418465428733e-05,
+      "loss": 0.743,
+      "step": 276120
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 2.9006656816000734e-05,
+      "loss": 0.7731,
+      "step": 276130
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 2.900589516657273e-05,
+      "loss": 0.8343,
+      "step": 276140
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 2.900513351714473e-05,
+      "loss": 0.7928,
+      "step": 276150
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 2.900437186771673e-05,
+      "loss": 0.8242,
+      "step": 276160
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 2.900361021828873e-05,
+      "loss": 0.7114,
+      "step": 276170
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 2.9002848568860725e-05,
+      "loss": 0.8406,
+      "step": 276180
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 2.9002086919432726e-05,
+      "loss": 0.836,
+      "step": 276190
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 2.9001325270004726e-05,
+      "loss": 0.7525,
+      "step": 276200
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 2.900056362057672e-05,
+      "loss": 0.7466,
+      "step": 276210
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 2.899980197114872e-05,
+      "loss": 0.8563,
+      "step": 276220
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 2.8999040321720722e-05,
+      "loss": 0.8005,
+      "step": 276230
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 2.8998278672292716e-05,
+      "loss": 0.7827,
+      "step": 276240
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 2.8997517022864717e-05,
+      "loss": 0.78,
+      "step": 276250
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 2.8996755373436718e-05,
+      "loss": 0.7917,
+      "step": 276260
+    },
+    {
+      "epoch": 12.62,
+      "learning_rate": 2.8995993724008712e-05,
+      "loss": 0.7677,
+      "step": 276270
+    },
+    {
+      "epoch": 12.62,
+      "learning_rate": 2.8995232074580713e-05,
+      "loss": 0.8924,
+      "step": 276280
+    },
+    {
+      "epoch": 12.62,
+      "learning_rate": 2.8994470425152714e-05,
+      "loss": 0.8943,
+      "step": 276290
+    },
+    {
+      "epoch": 12.62,
+      "learning_rate": 2.8993708775724708e-05,
+      "loss": 0.8017,
+      "step": 276300
+    },
+    {
+      "epoch": 12.62,
+      "learning_rate": 2.899294712629671e-05,
+      "loss": 0.7827,
+      "step": 276310
+    },
+    {
+      "epoch": 12.62,
+      "learning_rate": 2.899218547686871e-05,
+      "loss": 0.7772,
+      "step": 276320
+    },
+    {
+      "epoch": 12.62,
+      "learning_rate": 2.8991423827440707e-05,
+      "loss": 0.7974,
+      "step": 276330
+    },
+    {
+      "epoch": 12.62,
+      "learning_rate": 2.8990662178012708e-05,
+      "loss": 0.806,
+      "step": 276340
+    },
+    {
+      "epoch": 12.62,
+      "learning_rate": 2.8989900528584705e-05,
+      "loss": 0.8032,
+      "step": 276350
+    },
+    {
+      "epoch": 12.62,
+      "learning_rate": 2.8989138879156703e-05,
+      "loss": 0.8019,
+      "step": 276360
+    },
+    {
+      "epoch": 12.62,
+      "learning_rate": 2.8988377229728704e-05,
+      "loss": 0.8783,
+      "step": 276370
+    },
+    {
+      "epoch": 12.62,
+      "learning_rate": 2.8987615580300705e-05,
+      "loss": 0.728,
+      "step": 276380
+    },
+    {
+      "epoch": 12.62,
+      "learning_rate": 2.89868539308727e-05,
+      "loss": 0.7656,
+      "step": 276390
+    },
+    {
+      "epoch": 12.62,
+      "learning_rate": 2.89860922814447e-05,
+      "loss": 0.7699,
+      "step": 276400
+    },
+    {
+      "epoch": 12.62,
+      "learning_rate": 2.89853306320167e-05,
+      "loss": 0.8245,
+      "step": 276410
+    },
+    {
+      "epoch": 12.62,
+      "learning_rate": 2.8984568982588694e-05,
+      "loss": 0.8333,
+      "step": 276420
+    },
+    {
+      "epoch": 12.62,
+      "learning_rate": 2.8983807333160695e-05,
+      "loss": 0.8817,
+      "step": 276430
+    },
+    {
+      "epoch": 12.62,
+      "learning_rate": 2.898304568373269e-05,
+      "loss": 0.8795,
+      "step": 276440
+    },
+    {
+      "epoch": 12.62,
+      "learning_rate": 2.898228403430469e-05,
+      "loss": 0.8784,
+      "step": 276450
+    },
+    {
+      "epoch": 12.62,
+      "learning_rate": 2.898152238487669e-05,
+      "loss": 0.8752,
+      "step": 276460
+    },
+    {
+      "epoch": 12.62,
+      "learning_rate": 2.8980760735448685e-05,
+      "loss": 0.8118,
+      "step": 276470
+    },
+    {
+      "epoch": 12.62,
+      "learning_rate": 2.8979999086020686e-05,
+      "loss": 0.7668,
+      "step": 276480
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 2.8979237436592687e-05,
+      "loss": 0.8881,
+      "step": 276490
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 2.8978475787164684e-05,
+      "loss": 0.7214,
+      "step": 276500
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 2.8977714137736682e-05,
+      "loss": 0.7997,
+      "step": 276510
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 2.8976952488308683e-05,
+      "loss": 0.7973,
+      "step": 276520
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 2.897619083888068e-05,
+      "loss": 0.7708,
+      "step": 276530
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 2.897542918945268e-05,
+      "loss": 0.7515,
+      "step": 276540
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 2.8974667540024682e-05,
+      "loss": 0.7573,
+      "step": 276550
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 2.8973905890596676e-05,
+      "loss": 0.8319,
+      "step": 276560
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 2.8973144241168677e-05,
+      "loss": 0.9041,
+      "step": 276570
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 2.8972382591740678e-05,
+      "loss": 0.7477,
+      "step": 276580
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 2.8971620942312672e-05,
+      "loss": 0.7754,
+      "step": 276590
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 2.8970859292884673e-05,
+      "loss": 0.812,
+      "step": 276600
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 2.8970097643456673e-05,
+      "loss": 0.7902,
+      "step": 276610
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 2.8969335994028668e-05,
+      "loss": 0.6927,
+      "step": 276620
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 2.896857434460067e-05,
+      "loss": 0.787,
+      "step": 276630
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 2.896781269517267e-05,
+      "loss": 0.7893,
+      "step": 276640
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 2.8967051045744663e-05,
+      "loss": 0.7737,
+      "step": 276650
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 2.8966289396316664e-05,
+      "loss": 0.7593,
+      "step": 276660
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 2.8965527746888665e-05,
+      "loss": 0.7855,
+      "step": 276670
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 2.896476609746066e-05,
+      "loss": 0.9356,
+      "step": 276680
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 2.896400444803266e-05,
+      "loss": 0.8254,
+      "step": 276690
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 2.896324279860466e-05,
+      "loss": 0.787,
+      "step": 276700
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 2.8962481149176658e-05,
+      "loss": 0.8549,
+      "step": 276710
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 2.8961719499748656e-05,
+      "loss": 0.784,
+      "step": 276720
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 2.8960957850320657e-05,
+      "loss": 0.7801,
+      "step": 276730
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 2.8960196200892654e-05,
+      "loss": 0.7836,
+      "step": 276740
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 2.8959434551464655e-05,
+      "loss": 0.7348,
+      "step": 276750
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 2.8958672902036656e-05,
+      "loss": 0.831,
+      "step": 276760
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 2.895791125260865e-05,
+      "loss": 0.8356,
+      "step": 276770
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 2.895714960318065e-05,
+      "loss": 0.7887,
+      "step": 276780
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 2.895638795375265e-05,
+      "loss": 0.8077,
+      "step": 276790
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 2.8955626304324646e-05,
+      "loss": 0.7736,
+      "step": 276800
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 2.8954864654896646e-05,
+      "loss": 0.8581,
+      "step": 276810
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 2.8954103005468647e-05,
+      "loss": 0.7983,
+      "step": 276820
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 2.895334135604064e-05,
+      "loss": 0.8916,
+      "step": 276830
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 2.8952579706612642e-05,
+      "loss": 0.7557,
+      "step": 276840
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 2.8951818057184643e-05,
+      "loss": 0.843,
+      "step": 276850
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 2.8951056407756637e-05,
+      "loss": 0.78,
+      "step": 276860
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 2.8950294758328638e-05,
+      "loss": 0.7938,
+      "step": 276870
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 2.8949533108900636e-05,
+      "loss": 0.7438,
+      "step": 276880
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 2.8948771459472633e-05,
+      "loss": 0.8115,
+      "step": 276890
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 2.8948009810044634e-05,
+      "loss": 0.7805,
+      "step": 276900
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 2.894724816061663e-05,
+      "loss": 0.9104,
+      "step": 276910
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 2.8946486511188632e-05,
+      "loss": 0.8256,
+      "step": 276920
+    },
+    {
+      "epoch": 12.65,
+      "learning_rate": 2.8945724861760633e-05,
+      "loss": 0.7539,
+      "step": 276930
+    },
+    {
+      "epoch": 12.65,
+      "learning_rate": 2.8944963212332627e-05,
+      "loss": 0.7791,
+      "step": 276940
+    },
+    {
+      "epoch": 12.65,
+      "learning_rate": 2.8944201562904628e-05,
+      "loss": 0.8408,
+      "step": 276950
+    },
+    {
+      "epoch": 12.65,
+      "learning_rate": 2.894343991347663e-05,
+      "loss": 0.8056,
+      "step": 276960
+    },
+    {
+      "epoch": 12.65,
+      "learning_rate": 2.8942678264048623e-05,
+      "loss": 0.8206,
+      "step": 276970
+    },
+    {
+      "epoch": 12.65,
+      "learning_rate": 2.8941916614620624e-05,
+      "loss": 0.8354,
+      "step": 276980
+    },
+    {
+      "epoch": 12.65,
+      "learning_rate": 2.8941154965192625e-05,
+      "loss": 0.8214,
+      "step": 276990
+    },
+    {
+      "epoch": 12.65,
+      "learning_rate": 2.894039331576462e-05,
+      "loss": 0.7482,
+      "step": 277000
+    },
+    {
+      "epoch": 12.65,
+      "learning_rate": 2.893963166633662e-05,
+      "loss": 0.8402,
+      "step": 277010
+    },
+    {
+      "epoch": 12.65,
+      "learning_rate": 2.893887001690862e-05,
+      "loss": 0.7684,
+      "step": 277020
+    },
+    {
+      "epoch": 12.65,
+      "learning_rate": 2.8938108367480614e-05,
+      "loss": 0.7582,
+      "step": 277030
+    },
+    {
+      "epoch": 12.65,
+      "learning_rate": 2.8937346718052615e-05,
+      "loss": 0.859,
+      "step": 277040
+    },
+    {
+      "epoch": 12.65,
+      "learning_rate": 2.8936585068624616e-05,
+      "loss": 0.8602,
+      "step": 277050
+    },
+    {
+      "epoch": 12.65,
+      "learning_rate": 2.893582341919661e-05,
+      "loss": 0.7521,
+      "step": 277060
+    },
+    {
+      "epoch": 12.65,
+      "learning_rate": 2.893506176976861e-05,
+      "loss": 0.7517,
+      "step": 277070
+    },
+    {
+      "epoch": 12.65,
+      "learning_rate": 2.8934300120340612e-05,
+      "loss": 0.817,
+      "step": 277080
+    },
+    {
+      "epoch": 12.65,
+      "learning_rate": 2.893353847091261e-05,
+      "loss": 0.7655,
+      "step": 277090
+    },
+    {
+      "epoch": 12.65,
+      "learning_rate": 2.8932776821484607e-05,
+      "loss": 0.829,
+      "step": 277100
+    },
+    {
+      "epoch": 12.65,
+      "learning_rate": 2.8932015172056608e-05,
+      "loss": 0.8401,
+      "step": 277110
+    },
+    {
+      "epoch": 12.65,
+      "learning_rate": 2.8931253522628605e-05,
+      "loss": 0.8552,
+      "step": 277120
+    },
+    {
+      "epoch": 12.65,
+      "learning_rate": 2.8930491873200606e-05,
+      "loss": 0.8591,
+      "step": 277130
+    },
+    {
+      "epoch": 12.65,
+      "learning_rate": 2.8929730223772607e-05,
+      "loss": 0.884,
+      "step": 277140
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 2.89289685743446e-05,
+      "loss": 0.7869,
+      "step": 277150
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 2.8928206924916602e-05,
+      "loss": 0.8473,
+      "step": 277160
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 2.8927445275488603e-05,
+      "loss": 0.8501,
+      "step": 277170
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 2.8926683626060597e-05,
+      "loss": 0.7653,
+      "step": 277180
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 2.8925921976632598e-05,
+      "loss": 0.781,
+      "step": 277190
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 2.89251603272046e-05,
+      "loss": 0.7893,
+      "step": 277200
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 2.8924398677776593e-05,
+      "loss": 0.7427,
+      "step": 277210
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 2.8923637028348593e-05,
+      "loss": 0.7848,
+      "step": 277220
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 2.8922875378920594e-05,
+      "loss": 0.9002,
+      "step": 277230
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 2.892211372949259e-05,
+      "loss": 0.8734,
+      "step": 277240
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 2.892135208006459e-05,
+      "loss": 0.8203,
+      "step": 277250
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 2.892059043063659e-05,
+      "loss": 0.8802,
+      "step": 277260
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 2.8919828781208584e-05,
+      "loss": 0.7864,
+      "step": 277270
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 2.8919067131780585e-05,
+      "loss": 0.8517,
+      "step": 277280
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 2.8918305482352586e-05,
+      "loss": 0.9332,
+      "step": 277290
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 2.8917543832924583e-05,
+      "loss": 0.744,
+      "step": 277300
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 2.891678218349658e-05,
+      "loss": 0.7722,
+      "step": 277310
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 2.891602053406858e-05,
+      "loss": 0.8252,
+      "step": 277320
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 2.891525888464058e-05,
+      "loss": 0.8565,
+      "step": 277330
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 2.891449723521258e-05,
+      "loss": 0.7252,
+      "step": 277340
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 2.8913735585784574e-05,
+      "loss": 0.7714,
+      "step": 277350
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 2.8912973936356575e-05,
+      "loss": 0.7468,
+      "step": 277360
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 2.8912212286928576e-05,
+      "loss": 0.8527,
+      "step": 277370
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 2.891145063750057e-05,
+      "loss": 0.8449,
+      "step": 277380
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 2.891068898807257e-05,
+      "loss": 0.7899,
+      "step": 277390
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 2.890992733864457e-05,
+      "loss": 0.7888,
+      "step": 277400
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 2.8909165689216566e-05,
+      "loss": 1.0157,
+      "step": 277410
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 2.8908404039788566e-05,
+      "loss": 0.8025,
+      "step": 277420
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 2.8907642390360567e-05,
+      "loss": 0.7916,
+      "step": 277430
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 2.890688074093256e-05,
+      "loss": 0.8077,
+      "step": 277440
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 2.8906119091504562e-05,
+      "loss": 0.7635,
+      "step": 277450
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 2.8905357442076563e-05,
+      "loss": 0.7519,
+      "step": 277460
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 2.8904595792648557e-05,
+      "loss": 0.8611,
+      "step": 277470
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 2.8903834143220558e-05,
+      "loss": 0.7661,
+      "step": 277480
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 2.890307249379256e-05,
+      "loss": 0.821,
+      "step": 277490
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 2.8902310844364556e-05,
+      "loss": 0.8023,
+      "step": 277500
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 2.8901549194936557e-05,
+      "loss": 0.8175,
+      "step": 277510
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 2.8900787545508555e-05,
+      "loss": 0.7835,
+      "step": 277520
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 2.8900025896080552e-05,
+      "loss": 0.8892,
+      "step": 277530
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 2.8899264246652553e-05,
+      "loss": 0.749,
+      "step": 277540
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 2.8898502597224554e-05,
+      "loss": 0.8596,
+      "step": 277550
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 2.8897740947796548e-05,
+      "loss": 0.7446,
+      "step": 277560
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 2.889697929836855e-05,
+      "loss": 0.8303,
+      "step": 277570
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 2.889621764894055e-05,
+      "loss": 0.8458,
+      "step": 277580
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 2.8895455999512544e-05,
+      "loss": 0.7241,
+      "step": 277590
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 2.8894694350084545e-05,
+      "loss": 0.7824,
+      "step": 277600
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 2.8893932700656545e-05,
+      "loss": 0.7816,
+      "step": 277610
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 2.889317105122854e-05,
+      "loss": 0.8226,
+      "step": 277620
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 2.889240940180054e-05,
+      "loss": 0.7176,
+      "step": 277630
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 2.889164775237254e-05,
+      "loss": 0.7534,
+      "step": 277640
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 2.8890886102944535e-05,
+      "loss": 0.7712,
+      "step": 277650
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 2.8890124453516536e-05,
+      "loss": 0.7724,
+      "step": 277660
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 2.8889362804088537e-05,
+      "loss": 0.7956,
+      "step": 277670
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 2.8888601154660535e-05,
+      "loss": 0.7395,
+      "step": 277680
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 2.8887839505232532e-05,
+      "loss": 0.8569,
+      "step": 277690
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 2.8887077855804533e-05,
+      "loss": 0.9156,
+      "step": 277700
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 2.888631620637653e-05,
+      "loss": 0.7852,
+      "step": 277710
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 2.888555455694853e-05,
+      "loss": 0.7249,
+      "step": 277720
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 2.8884792907520532e-05,
+      "loss": 0.7518,
+      "step": 277730
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 2.8884031258092526e-05,
+      "loss": 0.7301,
+      "step": 277740
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 2.8883269608664527e-05,
+      "loss": 0.7922,
+      "step": 277750
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 2.8882507959236528e-05,
+      "loss": 0.7349,
+      "step": 277760
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 2.8881746309808522e-05,
+      "loss": 0.7909,
+      "step": 277770
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 2.8880984660380523e-05,
+      "loss": 0.8377,
+      "step": 277780
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 2.8880223010952524e-05,
+      "loss": 0.7757,
+      "step": 277790
+    },
+    {
+      "epoch": 12.69,
+      "learning_rate": 2.8879461361524518e-05,
+      "loss": 0.7972,
+      "step": 277800
+    },
+    {
+      "epoch": 12.69,
+      "learning_rate": 2.887869971209652e-05,
+      "loss": 0.7851,
+      "step": 277810
+    },
+    {
+      "epoch": 12.69,
+      "learning_rate": 2.8877938062668513e-05,
+      "loss": 0.8717,
+      "step": 277820
+    },
+    {
+      "epoch": 12.69,
+      "learning_rate": 2.8877176413240513e-05,
+      "loss": 0.7609,
+      "step": 277830
+    },
+    {
+      "epoch": 12.69,
+      "learning_rate": 2.8876414763812514e-05,
+      "loss": 0.8253,
+      "step": 277840
+    },
+    {
+      "epoch": 12.69,
+      "learning_rate": 2.887565311438451e-05,
+      "loss": 0.8972,
+      "step": 277850
+    },
+    {
+      "epoch": 12.69,
+      "learning_rate": 2.887489146495651e-05,
+      "loss": 0.7458,
+      "step": 277860
+    },
+    {
+      "epoch": 12.69,
+      "learning_rate": 2.887412981552851e-05,
+      "loss": 0.8071,
+      "step": 277870
+    },
+    {
+      "epoch": 12.69,
+      "learning_rate": 2.8873368166100508e-05,
+      "loss": 0.8363,
+      "step": 277880
+    },
+    {
+      "epoch": 12.69,
+      "learning_rate": 2.887260651667251e-05,
+      "loss": 0.7999,
+      "step": 277890
+    },
+    {
+      "epoch": 12.69,
+      "learning_rate": 2.8871844867244506e-05,
+      "loss": 0.8157,
+      "step": 277900
+    },
+    {
+      "epoch": 12.69,
+      "learning_rate": 2.8871083217816503e-05,
+      "loss": 0.786,
+      "step": 277910
+    },
+    {
+      "epoch": 12.69,
+      "learning_rate": 2.8870321568388504e-05,
+      "loss": 0.8522,
+      "step": 277920
+    },
+    {
+      "epoch": 12.69,
+      "learning_rate": 2.8869559918960505e-05,
+      "loss": 0.8352,
+      "step": 277930
+    },
+    {
+      "epoch": 12.69,
+      "learning_rate": 2.88687982695325e-05,
+      "loss": 0.7738,
+      "step": 277940
+    },
+    {
+      "epoch": 12.69,
+      "learning_rate": 2.88680366201045e-05,
+      "loss": 0.7572,
+      "step": 277950
+    },
+    {
+      "epoch": 12.69,
+      "learning_rate": 2.88672749706765e-05,
+      "loss": 0.813,
+      "step": 277960
+    },
+    {
+      "epoch": 12.69,
+      "learning_rate": 2.8866513321248495e-05,
+      "loss": 0.6873,
+      "step": 277970
+    },
+    {
+      "epoch": 12.69,
+      "learning_rate": 2.8865751671820496e-05,
+      "loss": 0.7874,
+      "step": 277980
+    },
+    {
+      "epoch": 12.69,
+      "learning_rate": 2.8864990022392497e-05,
+      "loss": 0.7941,
+      "step": 277990
+    },
+    {
+      "epoch": 12.69,
+      "learning_rate": 2.886422837296449e-05,
+      "loss": 0.8377,
+      "step": 278000
+    },
+    {
+      "epoch": 12.69,
+      "learning_rate": 2.886346672353649e-05,
+      "loss": 0.8098,
+      "step": 278010
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 2.8862705074108492e-05,
+      "loss": 0.8178,
+      "step": 278020
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 2.8861943424680487e-05,
+      "loss": 0.7961,
+      "step": 278030
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 2.8861181775252487e-05,
+      "loss": 0.8405,
+      "step": 278040
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 2.8860420125824488e-05,
+      "loss": 0.7925,
+      "step": 278050
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 2.8859658476396482e-05,
+      "loss": 0.8627,
+      "step": 278060
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 2.8858896826968483e-05,
+      "loss": 0.8116,
+      "step": 278070
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 2.8858135177540484e-05,
+      "loss": 0.8245,
+      "step": 278080
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 2.885737352811248e-05,
+      "loss": 0.801,
+      "step": 278090
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 2.8856611878684482e-05,
+      "loss": 0.9133,
+      "step": 278100
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 2.885585022925648e-05,
+      "loss": 0.8195,
+      "step": 278110
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 2.8855088579828477e-05,
+      "loss": 0.8023,
+      "step": 278120
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 2.8854326930400478e-05,
+      "loss": 0.9332,
+      "step": 278130
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 2.885356528097248e-05,
+      "loss": 0.9271,
+      "step": 278140
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 2.8852803631544473e-05,
+      "loss": 0.8136,
+      "step": 278150
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 2.8852041982116474e-05,
+      "loss": 0.7767,
+      "step": 278160
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 2.8851280332688475e-05,
+      "loss": 0.8321,
+      "step": 278170
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 2.885051868326047e-05,
+      "loss": 0.9046,
+      "step": 278180
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 2.884975703383247e-05,
+      "loss": 0.7168,
+      "step": 278190
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 2.884899538440447e-05,
+      "loss": 0.8269,
+      "step": 278200
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 2.8848233734976465e-05,
+      "loss": 0.8465,
+      "step": 278210
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 2.8847472085548465e-05,
+      "loss": 0.8022,
+      "step": 278220
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 2.8846710436120466e-05,
+      "loss": 0.777,
+      "step": 278230
+    },
+    {
+      "epoch": 12.71,
+      "learning_rate": 2.884594878669246e-05,
+      "loss": 0.7974,
+      "step": 278240
+    },
+    {
+      "epoch": 12.71,
+      "learning_rate": 2.884518713726446e-05,
+      "loss": 0.8269,
+      "step": 278250
+    },
+    {
+      "epoch": 12.71,
+      "learning_rate": 2.884442548783646e-05,
+      "loss": 0.8061,
+      "step": 278260
+    },
+    {
+      "epoch": 12.71,
+      "learning_rate": 2.8843663838408456e-05,
+      "loss": 0.8219,
+      "step": 278270
+    },
+    {
+      "epoch": 12.71,
+      "learning_rate": 2.8842902188980457e-05,
+      "loss": 0.8112,
+      "step": 278280
+    },
+    {
+      "epoch": 12.71,
+      "learning_rate": 2.8842140539552455e-05,
+      "loss": 0.8619,
+      "step": 278290
+    },
+    {
+      "epoch": 12.71,
+      "learning_rate": 2.8841378890124455e-05,
+      "loss": 0.7867,
+      "step": 278300
+    },
+    {
+      "epoch": 12.71,
+      "learning_rate": 2.8840617240696456e-05,
+      "loss": 0.8223,
+      "step": 278310
+    },
+    {
+      "epoch": 12.71,
+      "learning_rate": 2.883985559126845e-05,
+      "loss": 0.8532,
+      "step": 278320
+    },
+    {
+      "epoch": 12.71,
+      "learning_rate": 2.883909394184045e-05,
+      "loss": 0.7327,
+      "step": 278330
+    },
+    {
+      "epoch": 12.71,
+      "learning_rate": 2.8838332292412452e-05,
+      "loss": 0.9221,
+      "step": 278340
+    },
+    {
+      "epoch": 12.71,
+      "learning_rate": 2.8837570642984446e-05,
+      "loss": 0.8443,
+      "step": 278350
+    },
+    {
+      "epoch": 12.71,
+      "learning_rate": 2.8836808993556447e-05,
+      "loss": 0.7755,
+      "step": 278360
+    },
+    {
+      "epoch": 12.71,
+      "learning_rate": 2.8836047344128448e-05,
+      "loss": 0.8351,
+      "step": 278370
+    },
+    {
+      "epoch": 12.71,
+      "learning_rate": 2.8835285694700442e-05,
+      "loss": 0.838,
+      "step": 278380
+    },
+    {
+      "epoch": 12.71,
+      "learning_rate": 2.8834524045272443e-05,
+      "loss": 0.8726,
+      "step": 278390
+    },
+    {
+      "epoch": 12.71,
+      "learning_rate": 2.8833762395844444e-05,
+      "loss": 0.9049,
+      "step": 278400
+    },
+    {
+      "epoch": 12.71,
+      "learning_rate": 2.8833000746416438e-05,
+      "loss": 0.8547,
+      "step": 278410
+    },
+    {
+      "epoch": 12.71,
+      "learning_rate": 2.883223909698844e-05,
+      "loss": 0.8113,
+      "step": 278420
+    },
+    {
+      "epoch": 12.71,
+      "learning_rate": 2.883147744756044e-05,
+      "loss": 0.7514,
+      "step": 278430
+    },
+    {
+      "epoch": 12.71,
+      "learning_rate": 2.8830715798132433e-05,
+      "loss": 0.7737,
+      "step": 278440
+    },
+    {
+      "epoch": 12.71,
+      "learning_rate": 2.8829954148704434e-05,
+      "loss": 0.859,
+      "step": 278450
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 2.8829192499276435e-05,
+      "loss": 0.8617,
+      "step": 278460
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 2.8828430849848433e-05,
+      "loss": 0.7939,
+      "step": 278470
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 2.8827669200420433e-05,
+      "loss": 0.7314,
+      "step": 278480
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 2.882690755099243e-05,
+      "loss": 0.7398,
+      "step": 278490
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 2.882614590156443e-05,
+      "loss": 0.812,
+      "step": 278500
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 2.882538425213643e-05,
+      "loss": 0.8264,
+      "step": 278510
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 2.882462260270843e-05,
+      "loss": 0.8134,
+      "step": 278520
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 2.8823860953280424e-05,
+      "loss": 0.8839,
+      "step": 278530
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 2.8823099303852425e-05,
+      "loss": 0.8178,
+      "step": 278540
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 2.8822337654424426e-05,
+      "loss": 0.7067,
+      "step": 278550
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 2.882157600499642e-05,
+      "loss": 0.7441,
+      "step": 278560
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 2.882081435556842e-05,
+      "loss": 0.7909,
+      "step": 278570
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 2.8820052706140422e-05,
+      "loss": 0.7216,
+      "step": 278580
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 2.8819291056712416e-05,
+      "loss": 0.7306,
+      "step": 278590
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 2.8818529407284417e-05,
+      "loss": 0.7313,
+      "step": 278600
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 2.8817767757856418e-05,
+      "loss": 0.7836,
+      "step": 278610
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 2.881700610842841e-05,
+      "loss": 0.7581,
+      "step": 278620
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 2.8816244459000412e-05,
+      "loss": 0.8256,
+      "step": 278630
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 2.8815482809572413e-05,
+      "loss": 0.7272,
+      "step": 278640
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 2.8814721160144407e-05,
+      "loss": 0.8367,
+      "step": 278650
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 2.8813959510716408e-05,
+      "loss": 0.8399,
+      "step": 278660
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 2.881319786128841e-05,
+      "loss": 0.7598,
+      "step": 278670
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 2.8812436211860407e-05,
+      "loss": 1.0165,
+      "step": 278680
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 2.8811674562432407e-05,
+      "loss": 0.9004,
+      "step": 278690
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 2.8810912913004405e-05,
+      "loss": 0.8079,
+      "step": 278700
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 2.8810151263576402e-05,
+      "loss": 0.775,
+      "step": 278710
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 2.8809389614148403e-05,
+      "loss": 0.7852,
+      "step": 278720
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 2.8808627964720397e-05,
+      "loss": 0.7776,
+      "step": 278730
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 2.8807866315292398e-05,
+      "loss": 0.8133,
+      "step": 278740
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 2.88071046658644e-05,
+      "loss": 0.8385,
+      "step": 278750
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 2.8806343016436393e-05,
+      "loss": 0.8459,
+      "step": 278760
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 2.8805581367008394e-05,
+      "loss": 0.755,
+      "step": 278770
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 2.8804819717580395e-05,
+      "loss": 0.7616,
+      "step": 278780
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 2.880405806815239e-05,
+      "loss": 0.8572,
+      "step": 278790
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 2.880329641872439e-05,
+      "loss": 0.7591,
+      "step": 278800
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 2.880253476929639e-05,
+      "loss": 0.8335,
+      "step": 278810
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 2.8801773119868385e-05,
+      "loss": 0.8179,
+      "step": 278820
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 2.8801011470440385e-05,
+      "loss": 0.8365,
+      "step": 278830
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 2.8800249821012386e-05,
+      "loss": 0.9246,
+      "step": 278840
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 2.8799488171584384e-05,
+      "loss": 0.8467,
+      "step": 278850
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 2.879872652215638e-05,
+      "loss": 0.7911,
+      "step": 278860
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 2.8797964872728382e-05,
+      "loss": 0.8025,
+      "step": 278870
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 2.879720322330038e-05,
+      "loss": 0.7996,
+      "step": 278880
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 2.879644157387238e-05,
+      "loss": 0.8502,
+      "step": 278890
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 2.879567992444438e-05,
+      "loss": 0.8862,
+      "step": 278900
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 2.8794918275016375e-05,
+      "loss": 0.8616,
+      "step": 278910
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 2.8794156625588376e-05,
+      "loss": 0.8994,
+      "step": 278920
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 2.8793394976160377e-05,
+      "loss": 0.7289,
+      "step": 278930
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 2.879263332673237e-05,
+      "loss": 0.8801,
+      "step": 278940
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 2.8791871677304372e-05,
+      "loss": 0.7658,
+      "step": 278950
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 2.8791110027876373e-05,
+      "loss": 0.7705,
+      "step": 278960
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 2.8790348378448367e-05,
+      "loss": 0.9631,
+      "step": 278970
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 2.8789586729020368e-05,
+      "loss": 0.8307,
+      "step": 278980
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 2.878882507959237e-05,
+      "loss": 0.8875,
+      "step": 278990
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 2.8788063430164363e-05,
+      "loss": 0.9229,
+      "step": 279000
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 2.8787301780736364e-05,
+      "loss": 0.7356,
+      "step": 279010
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 2.8786540131308364e-05,
+      "loss": 0.9366,
+      "step": 279020
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 2.878577848188036e-05,
+      "loss": 0.7946,
+      "step": 279030
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 2.878501683245236e-05,
+      "loss": 0.7469,
+      "step": 279040
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 2.878425518302436e-05,
+      "loss": 0.8437,
+      "step": 279050
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 2.8783493533596358e-05,
+      "loss": 0.7538,
+      "step": 279060
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 2.8782731884168355e-05,
+      "loss": 0.8923,
+      "step": 279070
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 2.8781970234740356e-05,
+      "loss": 0.7427,
+      "step": 279080
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 2.8781208585312354e-05,
+      "loss": 0.9191,
+      "step": 279090
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 2.8780446935884354e-05,
+      "loss": 0.8235,
+      "step": 279100
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 2.8779685286456355e-05,
+      "loss": 0.8035,
+      "step": 279110
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 2.877892363702835e-05,
+      "loss": 0.7092,
+      "step": 279120
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 2.877816198760035e-05,
+      "loss": 0.8563,
+      "step": 279130
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 2.877740033817235e-05,
+      "loss": 0.7595,
+      "step": 279140
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 2.8776638688744345e-05,
+      "loss": 0.7809,
+      "step": 279150
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 2.8775877039316346e-05,
+      "loss": 0.8969,
+      "step": 279160
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 2.8775115389888347e-05,
+      "loss": 0.7478,
+      "step": 279170
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 2.877435374046034e-05,
+      "loss": 0.8262,
+      "step": 279180
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 2.8773592091032342e-05,
+      "loss": 0.7533,
+      "step": 279190
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 2.8772830441604336e-05,
+      "loss": 0.8014,
+      "step": 279200
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 2.8772068792176337e-05,
+      "loss": 0.7745,
+      "step": 279210
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 2.8771307142748338e-05,
+      "loss": 0.8266,
+      "step": 279220
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 2.8770545493320335e-05,
+      "loss": 0.7821,
+      "step": 279230
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 2.8769783843892332e-05,
+      "loss": 0.7854,
+      "step": 279240
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 2.8769022194464333e-05,
+      "loss": 0.781,
+      "step": 279250
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 2.876826054503633e-05,
+      "loss": 0.7781,
+      "step": 279260
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 2.876749889560833e-05,
+      "loss": 0.8666,
+      "step": 279270
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 2.8766737246180332e-05,
+      "loss": 0.87,
+      "step": 279280
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 2.8765975596752327e-05,
+      "loss": 0.8013,
+      "step": 279290
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 2.8765213947324327e-05,
+      "loss": 0.7333,
+      "step": 279300
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 2.8764452297896328e-05,
+      "loss": 0.7297,
+      "step": 279310
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 2.8763690648468322e-05,
+      "loss": 0.922,
+      "step": 279320
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 2.8762928999040323e-05,
+      "loss": 0.7528,
+      "step": 279330
+    },
+    {
+      "epoch": 12.76,
+      "learning_rate": 2.8762167349612324e-05,
+      "loss": 0.7744,
+      "step": 279340
+    },
+    {
+      "epoch": 12.76,
+      "learning_rate": 2.8761405700184318e-05,
+      "loss": 0.8062,
+      "step": 279350
+    },
+    {
+      "epoch": 12.76,
+      "learning_rate": 2.876064405075632e-05,
+      "loss": 0.8279,
+      "step": 279360
+    },
+    {
+      "epoch": 12.76,
+      "learning_rate": 2.875988240132832e-05,
+      "loss": 0.8086,
+      "step": 279370
+    },
+    {
+      "epoch": 12.76,
+      "learning_rate": 2.8759120751900314e-05,
+      "loss": 0.8124,
+      "step": 279380
+    },
+    {
+      "epoch": 12.76,
+      "learning_rate": 2.8758359102472315e-05,
+      "loss": 0.8293,
+      "step": 279390
+    },
+    {
+      "epoch": 12.76,
+      "learning_rate": 2.8757597453044316e-05,
+      "loss": 0.7683,
+      "step": 279400
+    },
+    {
+      "epoch": 12.76,
+      "learning_rate": 2.875683580361631e-05,
+      "loss": 0.8514,
+      "step": 279410
+    },
+    {
+      "epoch": 12.76,
+      "learning_rate": 2.875607415418831e-05,
+      "loss": 0.7703,
+      "step": 279420
+    },
+    {
+      "epoch": 12.76,
+      "learning_rate": 2.875531250476031e-05,
+      "loss": 0.7615,
+      "step": 279430
+    },
+    {
+      "epoch": 12.76,
+      "learning_rate": 2.875455085533231e-05,
+      "loss": 0.7219,
+      "step": 279440
+    },
+    {
+      "epoch": 12.76,
+      "learning_rate": 2.8753789205904306e-05,
+      "loss": 0.7689,
+      "step": 279450
+    },
+    {
+      "epoch": 12.76,
+      "learning_rate": 2.8753027556476307e-05,
+      "loss": 0.8088,
+      "step": 279460
+    },
+    {
+      "epoch": 12.76,
+      "learning_rate": 2.8752265907048305e-05,
+      "loss": 0.8243,
+      "step": 279470
+    },
+    {
+      "epoch": 12.76,
+      "learning_rate": 2.8751504257620306e-05,
+      "loss": 0.8812,
+      "step": 279480
+    },
+    {
+      "epoch": 12.76,
+      "learning_rate": 2.8750742608192306e-05,
+      "loss": 0.8534,
+      "step": 279490
+    },
+    {
+      "epoch": 12.76,
+      "learning_rate": 2.87499809587643e-05,
+      "loss": 0.7829,
+      "step": 279500
+    },
+    {
+      "epoch": 12.76,
+      "learning_rate": 2.87492193093363e-05,
+      "loss": 0.8697,
+      "step": 279510
+    },
+    {
+      "epoch": 12.76,
+      "learning_rate": 2.8748457659908302e-05,
+      "loss": 0.7302,
+      "step": 279520
+    },
+    {
+      "epoch": 12.76,
+      "learning_rate": 2.8747696010480296e-05,
+      "loss": 0.8497,
+      "step": 279530
+    },
+    {
+      "epoch": 12.76,
+      "learning_rate": 2.8746934361052297e-05,
+      "loss": 0.8011,
+      "step": 279540
+    },
+    {
+      "epoch": 12.76,
+      "learning_rate": 2.8746172711624298e-05,
+      "loss": 0.8367,
+      "step": 279550
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 2.8745411062196292e-05,
+      "loss": 0.8143,
+      "step": 279560
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 2.8744649412768293e-05,
+      "loss": 0.7957,
+      "step": 279570
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 2.8743887763340294e-05,
+      "loss": 0.9015,
+      "step": 279580
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 2.8743126113912288e-05,
+      "loss": 0.7694,
+      "step": 279590
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 2.874236446448429e-05,
+      "loss": 0.7935,
+      "step": 279600
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 2.874160281505629e-05,
+      "loss": 0.8397,
+      "step": 279610
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 2.8740841165628284e-05,
+      "loss": 0.729,
+      "step": 279620
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 2.8740079516200284e-05,
+      "loss": 0.752,
+      "step": 279630
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 2.8739317866772282e-05,
+      "loss": 0.8444,
+      "step": 279640
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 2.8738556217344283e-05,
+      "loss": 0.8397,
+      "step": 279650
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 2.873779456791628e-05,
+      "loss": 0.8166,
+      "step": 279660
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 2.8737032918488278e-05,
+      "loss": 0.9185,
+      "step": 279670
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 2.873627126906028e-05,
+      "loss": 0.8498,
+      "step": 279680
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 2.873550961963228e-05,
+      "loss": 0.8945,
+      "step": 279690
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 2.8734747970204274e-05,
+      "loss": 0.7335,
+      "step": 279700
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 2.8733986320776274e-05,
+      "loss": 0.7355,
+      "step": 279710
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 2.8733224671348275e-05,
+      "loss": 0.7626,
+      "step": 279720
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 2.873246302192027e-05,
+      "loss": 0.8271,
+      "step": 279730
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 2.873170137249227e-05,
+      "loss": 0.7046,
+      "step": 279740
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 2.873093972306427e-05,
+      "loss": 0.7654,
+      "step": 279750
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 2.8730178073636265e-05,
+      "loss": 0.8692,
+      "step": 279760
+    },
+    {
+      "epoch": 12.78,
+      "learning_rate": 2.8729416424208266e-05,
+      "loss": 0.8657,
+      "step": 279770
+    },
+    {
+      "epoch": 12.78,
+      "learning_rate": 2.8728654774780267e-05,
+      "loss": 0.8602,
+      "step": 279780
+    },
+    {
+      "epoch": 12.78,
+      "learning_rate": 2.872789312535226e-05,
+      "loss": 0.8327,
+      "step": 279790
+    },
+    {
+      "epoch": 12.78,
+      "learning_rate": 2.8727131475924262e-05,
+      "loss": 0.7719,
+      "step": 279800
+    },
+    {
+      "epoch": 12.78,
+      "learning_rate": 2.8726369826496263e-05,
+      "loss": 0.8182,
+      "step": 279810
+    },
+    {
+      "epoch": 12.78,
+      "learning_rate": 2.8725608177068257e-05,
+      "loss": 0.7839,
+      "step": 279820
+    },
+    {
+      "epoch": 12.78,
+      "learning_rate": 2.8724846527640258e-05,
+      "loss": 0.8356,
+      "step": 279830
+    },
+    {
+      "epoch": 12.78,
+      "learning_rate": 2.872408487821226e-05,
+      "loss": 0.8753,
+      "step": 279840
+    },
+    {
+      "epoch": 12.78,
+      "learning_rate": 2.8723323228784256e-05,
+      "loss": 0.8027,
+      "step": 279850
+    },
+    {
+      "epoch": 12.78,
+      "learning_rate": 2.8722561579356257e-05,
+      "loss": 0.8211,
+      "step": 279860
+    },
+    {
+      "epoch": 12.78,
+      "learning_rate": 2.8721799929928254e-05,
+      "loss": 0.9049,
+      "step": 279870
+    },
+    {
+      "epoch": 12.78,
+      "learning_rate": 2.872103828050025e-05,
+      "loss": 0.7855,
+      "step": 279880
+    },
+    {
+      "epoch": 12.78,
+      "learning_rate": 2.8720276631072252e-05,
+      "loss": 0.804,
+      "step": 279890
+    },
+    {
+      "epoch": 12.78,
+      "learning_rate": 2.8719514981644253e-05,
+      "loss": 0.7594,
+      "step": 279900
+    },
+    {
+      "epoch": 12.78,
+      "learning_rate": 2.8718753332216247e-05,
+      "loss": 0.8372,
+      "step": 279910
+    },
+    {
+      "epoch": 12.78,
+      "learning_rate": 2.8717991682788248e-05,
+      "loss": 0.8628,
+      "step": 279920
+    },
+    {
+      "epoch": 12.78,
+      "learning_rate": 2.871723003336025e-05,
+      "loss": 0.896,
+      "step": 279930
+    },
+    {
+      "epoch": 12.78,
+      "learning_rate": 2.8716468383932243e-05,
+      "loss": 0.8133,
+      "step": 279940
+    },
+    {
+      "epoch": 12.78,
+      "learning_rate": 2.8715706734504244e-05,
+      "loss": 0.9283,
+      "step": 279950
+    },
+    {
+      "epoch": 12.78,
+      "learning_rate": 2.8714945085076245e-05,
+      "loss": 0.8561,
+      "step": 279960
+    },
+    {
+      "epoch": 12.78,
+      "learning_rate": 2.871418343564824e-05,
+      "loss": 0.8682,
+      "step": 279970
+    },
+    {
+      "epoch": 12.78,
+      "learning_rate": 2.871342178622024e-05,
+      "loss": 0.8726,
+      "step": 279980
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 2.871266013679224e-05,
+      "loss": 0.8571,
+      "step": 279990
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 2.8711898487364235e-05,
+      "loss": 0.824,
+      "step": 280000
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 2.8711136837936236e-05,
+      "loss": 0.8027,
+      "step": 280010
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 2.8710375188508237e-05,
+      "loss": 0.8464,
+      "step": 280020
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 2.8709613539080234e-05,
+      "loss": 0.9654,
+      "step": 280030
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 2.870885188965223e-05,
+      "loss": 0.7417,
+      "step": 280040
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 2.8708090240224232e-05,
+      "loss": 0.8617,
+      "step": 280050
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 2.870732859079623e-05,
+      "loss": 0.7754,
+      "step": 280060
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 2.870656694136823e-05,
+      "loss": 0.8412,
+      "step": 280070
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 2.870580529194023e-05,
+      "loss": 0.6986,
+      "step": 280080
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 2.8705043642512226e-05,
+      "loss": 0.7657,
+      "step": 280090
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 2.8704281993084226e-05,
+      "loss": 0.7209,
+      "step": 280100
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 2.870352034365622e-05,
+      "loss": 0.8195,
+      "step": 280110
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 2.870275869422822e-05,
+      "loss": 0.8153,
+      "step": 280120
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 2.8701997044800222e-05,
+      "loss": 0.796,
+      "step": 280130
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 2.8701235395372216e-05,
+      "loss": 0.7211,
+      "step": 280140
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 2.8700473745944217e-05,
+      "loss": 0.8378,
+      "step": 280150
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 2.8699712096516218e-05,
+      "loss": 0.7301,
+      "step": 280160
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 2.8698950447088212e-05,
+      "loss": 0.7446,
+      "step": 280170
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 2.8698188797660213e-05,
+      "loss": 0.8515,
+      "step": 280180
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 2.8697427148232214e-05,
+      "loss": 0.716,
+      "step": 280190
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 2.8696665498804208e-05,
+      "loss": 0.8482,
+      "step": 280200
+    },
+    {
+      "epoch": 12.8,
+      "learning_rate": 2.869590384937621e-05,
+      "loss": 0.7658,
+      "step": 280210
+    },
+    {
+      "epoch": 12.8,
+      "learning_rate": 2.869514219994821e-05,
+      "loss": 0.8473,
+      "step": 280220
+    },
+    {
+      "epoch": 12.8,
+      "learning_rate": 2.8694380550520207e-05,
+      "loss": 0.7234,
+      "step": 280230
+    },
+    {
+      "epoch": 12.8,
+      "learning_rate": 2.8693618901092208e-05,
+      "loss": 0.8317,
+      "step": 280240
+    },
+    {
+      "epoch": 12.8,
+      "learning_rate": 2.8692857251664205e-05,
+      "loss": 0.8194,
+      "step": 280250
+    },
+    {
+      "epoch": 12.8,
+      "learning_rate": 2.8692095602236203e-05,
+      "loss": 0.817,
+      "step": 280260
+    },
+    {
+      "epoch": 12.8,
+      "learning_rate": 2.8691333952808204e-05,
+      "loss": 0.7591,
+      "step": 280270
+    },
+    {
+      "epoch": 12.8,
+      "learning_rate": 2.8690572303380205e-05,
+      "loss": 0.8339,
+      "step": 280280
+    },
+    {
+      "epoch": 12.8,
+      "learning_rate": 2.86898106539522e-05,
+      "loss": 0.8568,
+      "step": 280290
+    },
+    {
+      "epoch": 12.8,
+      "learning_rate": 2.86890490045242e-05,
+      "loss": 0.7869,
+      "step": 280300
+    },
+    {
+      "epoch": 12.8,
+      "learning_rate": 2.86882873550962e-05,
+      "loss": 0.837,
+      "step": 280310
+    },
+    {
+      "epoch": 12.8,
+      "learning_rate": 2.8687525705668194e-05,
+      "loss": 0.8209,
+      "step": 280320
+    },
+    {
+      "epoch": 12.8,
+      "learning_rate": 2.8686764056240195e-05,
+      "loss": 0.7627,
+      "step": 280330
+    },
+    {
+      "epoch": 12.8,
+      "learning_rate": 2.8686002406812196e-05,
+      "loss": 0.7916,
+      "step": 280340
+    },
+    {
+      "epoch": 12.8,
+      "learning_rate": 2.868524075738419e-05,
+      "loss": 0.7916,
+      "step": 280350
+    },
+    {
+      "epoch": 12.8,
+      "learning_rate": 2.868447910795619e-05,
+      "loss": 0.8019,
+      "step": 280360
+    },
+    {
+      "epoch": 12.8,
+      "learning_rate": 2.8683717458528192e-05,
+      "loss": 0.8339,
+      "step": 280370
+    },
+    {
+      "epoch": 12.8,
+      "learning_rate": 2.8682955809100186e-05,
+      "loss": 0.7877,
+      "step": 280380
+    },
+    {
+      "epoch": 12.8,
+      "learning_rate": 2.8682194159672187e-05,
+      "loss": 0.8004,
+      "step": 280390
+    },
+    {
+      "epoch": 12.8,
+      "learning_rate": 2.8681432510244188e-05,
+      "loss": 0.6897,
+      "step": 280400
+    },
+    {
+      "epoch": 12.8,
+      "learning_rate": 2.8680670860816182e-05,
+      "loss": 0.8348,
+      "step": 280410
+    },
+    {
+      "epoch": 12.8,
+      "learning_rate": 2.8679909211388183e-05,
+      "loss": 0.9051,
+      "step": 280420
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 2.8679147561960183e-05,
+      "loss": 0.7531,
+      "step": 280430
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 2.867838591253218e-05,
+      "loss": 0.8851,
+      "step": 280440
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 2.8677624263104182e-05,
+      "loss": 0.9307,
+      "step": 280450
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 2.867686261367618e-05,
+      "loss": 0.8342,
+      "step": 280460
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 2.8676100964248177e-05,
+      "loss": 0.7315,
+      "step": 280470
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 2.8675339314820178e-05,
+      "loss": 0.8078,
+      "step": 280480
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 2.867457766539218e-05,
+      "loss": 0.8301,
+      "step": 280490
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 2.8673816015964173e-05,
+      "loss": 0.8996,
+      "step": 280500
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 2.8673054366536173e-05,
+      "loss": 0.802,
+      "step": 280510
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 2.8672292717108174e-05,
+      "loss": 0.823,
+      "step": 280520
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 2.8671531067680168e-05,
+      "loss": 0.6923,
+      "step": 280530
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 2.867076941825217e-05,
+      "loss": 0.8127,
+      "step": 280540
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 2.867000776882417e-05,
+      "loss": 0.8447,
+      "step": 280550
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 2.8669246119396164e-05,
+      "loss": 0.9114,
+      "step": 280560
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 2.8668484469968165e-05,
+      "loss": 0.8221,
+      "step": 280570
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 2.866772282054016e-05,
+      "loss": 0.8056,
+      "step": 280580
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 2.866696117111216e-05,
+      "loss": 0.8619,
+      "step": 280590
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 2.866619952168416e-05,
+      "loss": 0.7297,
+      "step": 280600
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 2.8665437872256158e-05,
+      "loss": 0.8574,
+      "step": 280610
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 2.8664676222828156e-05,
+      "loss": 0.787,
+      "step": 280620
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 2.8663914573400157e-05,
+      "loss": 0.7864,
+      "step": 280630
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 2.8663152923972154e-05,
+      "loss": 0.7205,
+      "step": 280640
+    },
+    {
+      "epoch": 12.82,
+      "learning_rate": 2.8662391274544155e-05,
+      "loss": 0.836,
+      "step": 280650
+    },
+    {
+      "epoch": 12.82,
+      "learning_rate": 2.8661629625116156e-05,
+      "loss": 0.8247,
+      "step": 280660
+    },
+    {
+      "epoch": 12.82,
+      "learning_rate": 2.866086797568815e-05,
+      "loss": 0.8393,
+      "step": 280670
+    },
+    {
+      "epoch": 12.82,
+      "learning_rate": 2.866010632626015e-05,
+      "loss": 0.9432,
+      "step": 280680
+    },
+    {
+      "epoch": 12.82,
+      "learning_rate": 2.865934467683215e-05,
+      "loss": 0.7182,
+      "step": 280690
+    },
+    {
+      "epoch": 12.82,
+      "learning_rate": 2.8658583027404146e-05,
+      "loss": 0.7121,
+      "step": 280700
+    },
+    {
+      "epoch": 12.82,
+      "learning_rate": 2.8657821377976146e-05,
+      "loss": 0.7763,
+      "step": 280710
+    },
+    {
+      "epoch": 12.82,
+      "learning_rate": 2.8657059728548147e-05,
+      "loss": 0.8524,
+      "step": 280720
+    },
+    {
+      "epoch": 12.82,
+      "learning_rate": 2.865629807912014e-05,
+      "loss": 0.795,
+      "step": 280730
+    },
+    {
+      "epoch": 12.82,
+      "learning_rate": 2.8655536429692142e-05,
+      "loss": 0.864,
+      "step": 280740
+    },
+    {
+      "epoch": 12.82,
+      "learning_rate": 2.8654774780264143e-05,
+      "loss": 0.8441,
+      "step": 280750
+    },
+    {
+      "epoch": 12.82,
+      "learning_rate": 2.8654013130836137e-05,
+      "loss": 0.7752,
+      "step": 280760
+    },
+    {
+      "epoch": 12.82,
+      "learning_rate": 2.8653251481408138e-05,
+      "loss": 0.8082,
+      "step": 280770
+    },
+    {
+      "epoch": 12.82,
+      "learning_rate": 2.865248983198014e-05,
+      "loss": 0.6965,
+      "step": 280780
+    },
+    {
+      "epoch": 12.82,
+      "learning_rate": 2.8651728182552133e-05,
+      "loss": 0.7665,
+      "step": 280790
+    },
+    {
+      "epoch": 12.82,
+      "learning_rate": 2.8650966533124134e-05,
+      "loss": 0.8193,
+      "step": 280800
+    },
+    {
+      "epoch": 12.82,
+      "learning_rate": 2.8650204883696135e-05,
+      "loss": 0.7518,
+      "step": 280810
+    },
+    {
+      "epoch": 12.82,
+      "learning_rate": 2.8649443234268132e-05,
+      "loss": 0.8126,
+      "step": 280820
+    },
+    {
+      "epoch": 12.82,
+      "learning_rate": 2.8648681584840133e-05,
+      "loss": 0.8456,
+      "step": 280830
+    },
+    {
+      "epoch": 12.82,
+      "learning_rate": 2.864791993541213e-05,
+      "loss": 0.8417,
+      "step": 280840
+    },
+    {
+      "epoch": 12.82,
+      "learning_rate": 2.8647158285984128e-05,
+      "loss": 0.7762,
+      "step": 280850
+    },
+    {
+      "epoch": 12.82,
+      "learning_rate": 2.864639663655613e-05,
+      "loss": 0.7301,
+      "step": 280860
+    },
+    {
+      "epoch": 12.83,
+      "learning_rate": 2.864563498712813e-05,
+      "loss": 0.8138,
+      "step": 280870
+    },
+    {
+      "epoch": 12.83,
+      "learning_rate": 2.8644873337700124e-05,
+      "loss": 0.7705,
+      "step": 280880
+    },
+    {
+      "epoch": 12.83,
+      "learning_rate": 2.8644111688272125e-05,
+      "loss": 0.926,
+      "step": 280890
+    },
+    {
+      "epoch": 12.83,
+      "learning_rate": 2.8643350038844125e-05,
+      "loss": 0.8056,
+      "step": 280900
+    },
+    {
+      "epoch": 12.83,
+      "learning_rate": 2.864258838941612e-05,
+      "loss": 0.7768,
+      "step": 280910
+    },
+    {
+      "epoch": 12.83,
+      "learning_rate": 2.864182673998812e-05,
+      "loss": 0.8968,
+      "step": 280920
+    },
+    {
+      "epoch": 12.83,
+      "learning_rate": 2.864106509056012e-05,
+      "loss": 0.7958,
+      "step": 280930
+    },
+    {
+      "epoch": 12.83,
+      "learning_rate": 2.8640303441132115e-05,
+      "loss": 0.8131,
+      "step": 280940
+    },
+    {
+      "epoch": 12.83,
+      "learning_rate": 2.8639541791704116e-05,
+      "loss": 0.8977,
+      "step": 280950
+    },
+    {
+      "epoch": 12.83,
+      "learning_rate": 2.8638780142276117e-05,
+      "loss": 0.8598,
+      "step": 280960
+    },
+    {
+      "epoch": 12.83,
+      "learning_rate": 2.863801849284811e-05,
+      "loss": 0.8041,
+      "step": 280970
+    },
+    {
+      "epoch": 12.83,
+      "learning_rate": 2.8637256843420112e-05,
+      "loss": 0.918,
+      "step": 280980
+    },
+    {
+      "epoch": 12.83,
+      "learning_rate": 2.8636495193992113e-05,
+      "loss": 0.8301,
+      "step": 280990
+    },
+    {
+      "epoch": 12.83,
+      "learning_rate": 2.8635733544564107e-05,
+      "loss": 0.7535,
+      "step": 281000
+    },
+    {
+      "epoch": 12.83,
+      "learning_rate": 2.8634971895136108e-05,
+      "loss": 0.7599,
+      "step": 281010
+    },
+    {
+      "epoch": 12.83,
+      "learning_rate": 2.863421024570811e-05,
+      "loss": 0.8307,
+      "step": 281020
+    },
+    {
+      "epoch": 12.83,
+      "learning_rate": 2.8633448596280106e-05,
+      "loss": 0.8891,
+      "step": 281030
+    },
+    {
+      "epoch": 12.83,
+      "learning_rate": 2.8632686946852107e-05,
+      "loss": 0.7562,
+      "step": 281040
+    },
+    {
+      "epoch": 12.83,
+      "learning_rate": 2.86319252974241e-05,
+      "loss": 0.7804,
+      "step": 281050
+    },
+    {
+      "epoch": 12.83,
+      "learning_rate": 2.8631163647996102e-05,
+      "loss": 0.8609,
+      "step": 281060
+    },
+    {
+      "epoch": 12.83,
+      "learning_rate": 2.8630401998568103e-05,
+      "loss": 0.7798,
+      "step": 281070
+    },
+    {
+      "epoch": 12.83,
+      "learning_rate": 2.8629640349140097e-05,
+      "loss": 0.857,
+      "step": 281080
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 2.8628878699712098e-05,
+      "loss": 0.7266,
+      "step": 281090
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 2.86281170502841e-05,
+      "loss": 0.8166,
+      "step": 281100
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 2.8627355400856093e-05,
+      "loss": 0.7609,
+      "step": 281110
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 2.8626593751428093e-05,
+      "loss": 0.8745,
+      "step": 281120
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 2.8625832102000094e-05,
+      "loss": 0.901,
+      "step": 281130
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 2.8625070452572088e-05,
+      "loss": 0.8209,
+      "step": 281140
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 2.862430880314409e-05,
+      "loss": 0.8751,
+      "step": 281150
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 2.862354715371609e-05,
+      "loss": 0.7805,
+      "step": 281160
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 2.8622785504288084e-05,
+      "loss": 0.8667,
+      "step": 281170
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 2.8622023854860085e-05,
+      "loss": 0.8519,
+      "step": 281180
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 2.8621262205432086e-05,
+      "loss": 0.9712,
+      "step": 281190
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 2.8620500556004083e-05,
+      "loss": 0.7815,
+      "step": 281200
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 2.861973890657608e-05,
+      "loss": 0.8039,
+      "step": 281210
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 2.861897725714808e-05,
+      "loss": 0.8909,
+      "step": 281220
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 2.861821560772008e-05,
+      "loss": 0.8414,
+      "step": 281230
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 2.861745395829208e-05,
+      "loss": 0.8005,
+      "step": 281240
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 2.861669230886408e-05,
+      "loss": 0.8642,
+      "step": 281250
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 2.8615930659436075e-05,
+      "loss": 0.8015,
+      "step": 281260
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 2.8615169010008076e-05,
+      "loss": 0.7513,
+      "step": 281270
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 2.8614407360580077e-05,
+      "loss": 0.8001,
+      "step": 281280
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 2.861364571115207e-05,
+      "loss": 0.8122,
+      "step": 281290
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 2.861288406172407e-05,
+      "loss": 0.8138,
+      "step": 281300
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 2.8612122412296072e-05,
+      "loss": 0.8206,
+      "step": 281310
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 2.8611360762868066e-05,
+      "loss": 0.7919,
+      "step": 281320
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 2.8610599113440067e-05,
+      "loss": 0.7727,
+      "step": 281330
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 2.8609837464012068e-05,
+      "loss": 0.8717,
+      "step": 281340
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 2.8609075814584062e-05,
+      "loss": 0.8231,
+      "step": 281350
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 2.8608314165156063e-05,
+      "loss": 0.8303,
+      "step": 281360
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 2.8607552515728064e-05,
+      "loss": 0.7788,
+      "step": 281370
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 2.8606790866300058e-05,
+      "loss": 0.8929,
+      "step": 281380
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 2.860602921687206e-05,
+      "loss": 0.7379,
+      "step": 281390
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 2.860526756744406e-05,
+      "loss": 0.7238,
+      "step": 281400
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 2.8604505918016057e-05,
+      "loss": 0.7565,
+      "step": 281410
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 2.8603744268588058e-05,
+      "loss": 0.8827,
+      "step": 281420
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 2.8602982619160055e-05,
+      "loss": 0.8245,
+      "step": 281430
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 2.8602220969732053e-05,
+      "loss": 0.8576,
+      "step": 281440
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 2.8601459320304054e-05,
+      "loss": 0.8116,
+      "step": 281450
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 2.8600697670876055e-05,
+      "loss": 0.7945,
+      "step": 281460
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 2.859993602144805e-05,
+      "loss": 0.7899,
+      "step": 281470
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 2.859917437202005e-05,
+      "loss": 0.7827,
+      "step": 281480
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 2.8598412722592044e-05,
+      "loss": 0.8611,
+      "step": 281490
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 2.8597651073164045e-05,
+      "loss": 0.8435,
+      "step": 281500
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 2.8596889423736045e-05,
+      "loss": 0.7429,
+      "step": 281510
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 2.859612777430804e-05,
+      "loss": 0.9463,
+      "step": 281520
+    },
+    {
+      "epoch": 12.86,
+      "learning_rate": 2.859536612488004e-05,
+      "loss": 0.7976,
+      "step": 281530
+    },
+    {
+      "epoch": 12.86,
+      "learning_rate": 2.859460447545204e-05,
+      "loss": 0.7496,
+      "step": 281540
+    },
+    {
+      "epoch": 12.86,
+      "learning_rate": 2.8593842826024035e-05,
+      "loss": 0.8086,
+      "step": 281550
+    },
+    {
+      "epoch": 12.86,
+      "learning_rate": 2.8593081176596036e-05,
+      "loss": 0.9406,
+      "step": 281560
+    },
+    {
+      "epoch": 12.86,
+      "learning_rate": 2.8592319527168037e-05,
+      "loss": 0.786,
+      "step": 281570
+    },
+    {
+      "epoch": 12.86,
+      "learning_rate": 2.8591557877740034e-05,
+      "loss": 0.7849,
+      "step": 281580
+    },
+    {
+      "epoch": 12.86,
+      "learning_rate": 2.8590796228312032e-05,
+      "loss": 0.8605,
+      "step": 281590
+    },
+    {
+      "epoch": 12.86,
+      "learning_rate": 2.8590034578884033e-05,
+      "loss": 0.7636,
+      "step": 281600
+    },
+    {
+      "epoch": 12.86,
+      "learning_rate": 2.858927292945603e-05,
+      "loss": 0.8517,
+      "step": 281610
+    },
+    {
+      "epoch": 12.86,
+      "learning_rate": 2.858851128002803e-05,
+      "loss": 0.8498,
+      "step": 281620
+    },
+    {
+      "epoch": 12.86,
+      "learning_rate": 2.8587749630600032e-05,
+      "loss": 0.8152,
+      "step": 281630
+    },
+    {
+      "epoch": 12.86,
+      "learning_rate": 2.8586987981172026e-05,
+      "loss": 0.7675,
+      "step": 281640
+    },
+    {
+      "epoch": 12.86,
+      "learning_rate": 2.8586226331744027e-05,
+      "loss": 0.7214,
+      "step": 281650
+    },
+    {
+      "epoch": 12.86,
+      "learning_rate": 2.8585464682316028e-05,
+      "loss": 0.772,
+      "step": 281660
+    },
+    {
+      "epoch": 12.86,
+      "learning_rate": 2.8584703032888022e-05,
+      "loss": 0.8347,
+      "step": 281670
+    },
+    {
+      "epoch": 12.86,
+      "learning_rate": 2.8583941383460023e-05,
+      "loss": 0.8136,
+      "step": 281680
+    },
+    {
+      "epoch": 12.86,
+      "learning_rate": 2.8583179734032024e-05,
+      "loss": 0.8204,
+      "step": 281690
+    },
+    {
+      "epoch": 12.86,
+      "learning_rate": 2.8582418084604018e-05,
+      "loss": 0.935,
+      "step": 281700
+    },
+    {
+      "epoch": 12.86,
+      "learning_rate": 2.858165643517602e-05,
+      "loss": 0.9022,
+      "step": 281710
+    },
+    {
+      "epoch": 12.86,
+      "learning_rate": 2.858089478574802e-05,
+      "loss": 0.733,
+      "step": 281720
+    },
+    {
+      "epoch": 12.86,
+      "learning_rate": 2.8580133136320013e-05,
+      "loss": 0.7837,
+      "step": 281730
+    },
+    {
+      "epoch": 12.86,
+      "learning_rate": 2.8579371486892014e-05,
+      "loss": 0.8541,
+      "step": 281740
+    },
+    {
+      "epoch": 12.87,
+      "learning_rate": 2.8578609837464015e-05,
+      "loss": 0.8772,
+      "step": 281750
+    },
+    {
+      "epoch": 12.87,
+      "learning_rate": 2.857784818803601e-05,
+      "loss": 0.8807,
+      "step": 281760
+    },
+    {
+      "epoch": 12.87,
+      "learning_rate": 2.857708653860801e-05,
+      "loss": 0.8071,
+      "step": 281770
+    },
+    {
+      "epoch": 12.87,
+      "learning_rate": 2.857632488918001e-05,
+      "loss": 0.832,
+      "step": 281780
+    },
+    {
+      "epoch": 12.87,
+      "learning_rate": 2.857556323975201e-05,
+      "loss": 0.8296,
+      "step": 281790
+    },
+    {
+      "epoch": 12.87,
+      "learning_rate": 2.8574801590324006e-05,
+      "loss": 0.8118,
+      "step": 281800
+    },
+    {
+      "epoch": 12.87,
+      "learning_rate": 2.8574039940896007e-05,
+      "loss": 0.8071,
+      "step": 281810
+    },
+    {
+      "epoch": 12.87,
+      "learning_rate": 2.8573278291468004e-05,
+      "loss": 0.8629,
+      "step": 281820
+    },
+    {
+      "epoch": 12.87,
+      "learning_rate": 2.8572516642040005e-05,
+      "loss": 0.8412,
+      "step": 281830
+    },
+    {
+      "epoch": 12.87,
+      "learning_rate": 2.8571754992612006e-05,
+      "loss": 0.8107,
+      "step": 281840
+    },
+    {
+      "epoch": 12.87,
+      "learning_rate": 2.8570993343184e-05,
+      "loss": 0.7853,
+      "step": 281850
+    },
+    {
+      "epoch": 12.87,
+      "learning_rate": 2.8570231693756e-05,
+      "loss": 0.7741,
+      "step": 281860
+    },
+    {
+      "epoch": 12.87,
+      "learning_rate": 2.8569470044328e-05,
+      "loss": 0.8045,
+      "step": 281870
+    },
+    {
+      "epoch": 12.87,
+      "learning_rate": 2.8568708394899996e-05,
+      "loss": 0.7226,
+      "step": 281880
+    },
+    {
+      "epoch": 12.87,
+      "learning_rate": 2.8567946745471997e-05,
+      "loss": 0.8869,
+      "step": 281890
+    },
+    {
+      "epoch": 12.87,
+      "learning_rate": 2.8567185096043997e-05,
+      "loss": 0.837,
+      "step": 281900
+    },
+    {
+      "epoch": 12.87,
+      "learning_rate": 2.856642344661599e-05,
+      "loss": 0.7603,
+      "step": 281910
+    },
+    {
+      "epoch": 12.87,
+      "learning_rate": 2.8565661797187992e-05,
+      "loss": 0.9388,
+      "step": 281920
+    },
+    {
+      "epoch": 12.87,
+      "learning_rate": 2.8564900147759993e-05,
+      "loss": 0.811,
+      "step": 281930
+    },
+    {
+      "epoch": 12.87,
+      "learning_rate": 2.8564138498331987e-05,
+      "loss": 0.7613,
+      "step": 281940
+    },
+    {
+      "epoch": 12.87,
+      "learning_rate": 2.8563376848903988e-05,
+      "loss": 0.829,
+      "step": 281950
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 2.8562615199475982e-05,
+      "loss": 0.81,
+      "step": 281960
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 2.8561853550047983e-05,
+      "loss": 0.8756,
+      "step": 281970
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 2.8561091900619984e-05,
+      "loss": 0.8256,
+      "step": 281980
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 2.856033025119198e-05,
+      "loss": 0.8357,
+      "step": 281990
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 2.8559568601763982e-05,
+      "loss": 0.8137,
+      "step": 282000
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 2.855880695233598e-05,
+      "loss": 0.8489,
+      "step": 282010
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 2.8558045302907977e-05,
+      "loss": 0.8623,
+      "step": 282020
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 2.8557283653479978e-05,
+      "loss": 0.8084,
+      "step": 282030
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 2.855652200405198e-05,
+      "loss": 0.7975,
+      "step": 282040
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 2.8555760354623973e-05,
+      "loss": 0.8953,
+      "step": 282050
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 2.8554998705195974e-05,
+      "loss": 0.8566,
+      "step": 282060
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 2.8554237055767975e-05,
+      "loss": 0.8372,
+      "step": 282070
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 2.855347540633997e-05,
+      "loss": 0.8833,
+      "step": 282080
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 2.855271375691197e-05,
+      "loss": 0.8587,
+      "step": 282090
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 2.855195210748397e-05,
+      "loss": 0.8628,
+      "step": 282100
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 2.8551190458055965e-05,
+      "loss": 0.8483,
+      "step": 282110
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 2.8550428808627965e-05,
+      "loss": 0.902,
+      "step": 282120
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 2.8549667159199966e-05,
+      "loss": 0.795,
+      "step": 282130
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 2.854890550977196e-05,
+      "loss": 0.7898,
+      "step": 282140
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 2.854814386034396e-05,
+      "loss": 0.8595,
+      "step": 282150
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 2.8547382210915962e-05,
+      "loss": 0.7822,
+      "step": 282160
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 2.854662056148796e-05,
+      "loss": 0.9019,
+      "step": 282170
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 2.8545858912059957e-05,
+      "loss": 0.852,
+      "step": 282180
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 2.8545097262631958e-05,
+      "loss": 0.8916,
+      "step": 282190
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 2.8544335613203955e-05,
+      "loss": 0.873,
+      "step": 282200
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 2.8543573963775956e-05,
+      "loss": 0.8407,
+      "step": 282210
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 2.8542812314347957e-05,
+      "loss": 0.7177,
+      "step": 282220
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 2.854205066491995e-05,
+      "loss": 0.8725,
+      "step": 282230
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 2.8541289015491952e-05,
+      "loss": 0.7429,
+      "step": 282240
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 2.8540527366063953e-05,
+      "loss": 0.7791,
+      "step": 282250
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 2.8539765716635947e-05,
+      "loss": 0.8045,
+      "step": 282260
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 2.8539004067207948e-05,
+      "loss": 0.7517,
+      "step": 282270
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 2.853824241777995e-05,
+      "loss": 0.7463,
+      "step": 282280
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 2.8537480768351943e-05,
+      "loss": 0.7958,
+      "step": 282290
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 2.8536719118923944e-05,
+      "loss": 0.9161,
+      "step": 282300
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 2.8535957469495944e-05,
+      "loss": 0.8135,
+      "step": 282310
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 2.853519582006794e-05,
+      "loss": 0.8252,
+      "step": 282320
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 2.853443417063994e-05,
+      "loss": 0.7897,
+      "step": 282330
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 2.853367252121194e-05,
+      "loss": 0.8244,
+      "step": 282340
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 2.8532910871783934e-05,
+      "loss": 0.822,
+      "step": 282350
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 2.8532149222355935e-05,
+      "loss": 0.8003,
+      "step": 282360
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 2.8531387572927936e-05,
+      "loss": 0.7542,
+      "step": 282370
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 2.8530625923499933e-05,
+      "loss": 0.8434,
+      "step": 282380
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 2.852986427407193e-05,
+      "loss": 0.8255,
+      "step": 282390
+    },
+    {
+      "epoch": 12.9,
+      "learning_rate": 2.8529102624643932e-05,
+      "loss": 0.8912,
+      "step": 282400
+    },
+    {
+      "epoch": 12.9,
+      "learning_rate": 2.852834097521593e-05,
+      "loss": 0.8463,
+      "step": 282410
+    },
+    {
+      "epoch": 12.9,
+      "learning_rate": 2.852757932578793e-05,
+      "loss": 0.8085,
+      "step": 282420
+    },
+    {
+      "epoch": 12.9,
+      "learning_rate": 2.8526817676359924e-05,
+      "loss": 0.79,
+      "step": 282430
+    },
+    {
+      "epoch": 12.9,
+      "learning_rate": 2.8526056026931925e-05,
+      "loss": 0.7805,
+      "step": 282440
+    },
+    {
+      "epoch": 12.9,
+      "learning_rate": 2.8525294377503926e-05,
+      "loss": 0.8046,
+      "step": 282450
+    },
+    {
+      "epoch": 12.9,
+      "learning_rate": 2.852453272807592e-05,
+      "loss": 0.7612,
+      "step": 282460
+    },
+    {
+      "epoch": 12.9,
+      "learning_rate": 2.852377107864792e-05,
+      "loss": 0.7639,
+      "step": 282470
+    },
+    {
+      "epoch": 12.9,
+      "learning_rate": 2.852300942921992e-05,
+      "loss": 0.7905,
+      "step": 282480
+    },
+    {
+      "epoch": 12.9,
+      "learning_rate": 2.8522247779791916e-05,
+      "loss": 0.7887,
+      "step": 282490
+    },
+    {
+      "epoch": 12.9,
+      "learning_rate": 2.8521486130363917e-05,
+      "loss": 0.7211,
+      "step": 282500
+    },
+    {
+      "epoch": 12.9,
+      "learning_rate": 2.8520724480935917e-05,
+      "loss": 0.9529,
+      "step": 282510
+    },
+    {
+      "epoch": 12.9,
+      "learning_rate": 2.851996283150791e-05,
+      "loss": 0.8633,
+      "step": 282520
+    },
+    {
+      "epoch": 12.9,
+      "learning_rate": 2.8519201182079912e-05,
+      "loss": 0.8133,
+      "step": 282530
+    },
+    {
+      "epoch": 12.9,
+      "learning_rate": 2.8518439532651913e-05,
+      "loss": 0.8654,
+      "step": 282540
+    },
+    {
+      "epoch": 12.9,
+      "learning_rate": 2.8517677883223907e-05,
+      "loss": 0.801,
+      "step": 282550
+    },
+    {
+      "epoch": 12.9,
+      "learning_rate": 2.8516916233795908e-05,
+      "loss": 0.8385,
+      "step": 282560
+    },
+    {
+      "epoch": 12.9,
+      "learning_rate": 2.851615458436791e-05,
+      "loss": 0.8404,
+      "step": 282570
+    },
+    {
+      "epoch": 12.9,
+      "learning_rate": 2.8515392934939906e-05,
+      "loss": 0.7674,
+      "step": 282580
+    },
+    {
+      "epoch": 12.9,
+      "learning_rate": 2.8514631285511907e-05,
+      "loss": 0.8475,
+      "step": 282590
+    },
+    {
+      "epoch": 12.9,
+      "learning_rate": 2.8513869636083905e-05,
+      "loss": 0.8387,
+      "step": 282600
+    },
+    {
+      "epoch": 12.9,
+      "learning_rate": 2.8513107986655902e-05,
+      "loss": 0.853,
+      "step": 282610
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 2.8512346337227903e-05,
+      "loss": 0.8094,
+      "step": 282620
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 2.8511584687799904e-05,
+      "loss": 0.7767,
+      "step": 282630
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 2.8510823038371898e-05,
+      "loss": 0.803,
+      "step": 282640
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 2.85100613889439e-05,
+      "loss": 0.9872,
+      "step": 282650
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 2.85092997395159e-05,
+      "loss": 0.827,
+      "step": 282660
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 2.8508538090087894e-05,
+      "loss": 0.7498,
+      "step": 282670
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 2.8507776440659895e-05,
+      "loss": 0.8784,
+      "step": 282680
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 2.8507014791231896e-05,
+      "loss": 0.8042,
+      "step": 282690
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 2.850625314180389e-05,
+      "loss": 0.842,
+      "step": 282700
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 2.850549149237589e-05,
+      "loss": 0.7888,
+      "step": 282710
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 2.850472984294789e-05,
+      "loss": 0.7361,
+      "step": 282720
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 2.8503968193519885e-05,
+      "loss": 0.8279,
+      "step": 282730
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 2.8503206544091886e-05,
+      "loss": 0.8142,
+      "step": 282740
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 2.8502444894663887e-05,
+      "loss": 0.8895,
+      "step": 282750
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 2.850168324523588e-05,
+      "loss": 0.7554,
+      "step": 282760
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 2.8500921595807882e-05,
+      "loss": 0.742,
+      "step": 282770
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 2.8500159946379883e-05,
+      "loss": 0.8222,
+      "step": 282780
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 2.849939829695188e-05,
+      "loss": 0.7965,
+      "step": 282790
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 2.849863664752388e-05,
+      "loss": 0.8036,
+      "step": 282800
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 2.849787499809588e-05,
+      "loss": 0.8759,
+      "step": 282810
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 2.8497113348667876e-05,
+      "loss": 0.8351,
+      "step": 282820
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 2.8496351699239877e-05,
+      "loss": 0.9284,
+      "step": 282830
+    },
+    {
+      "epoch": 12.92,
+      "learning_rate": 2.8495590049811878e-05,
+      "loss": 0.7288,
+      "step": 282840
+    },
+    {
+      "epoch": 12.92,
+      "learning_rate": 2.8494828400383872e-05,
+      "loss": 0.8369,
+      "step": 282850
+    },
+    {
+      "epoch": 12.92,
+      "learning_rate": 2.8494066750955873e-05,
+      "loss": 0.8603,
+      "step": 282860
+    },
+    {
+      "epoch": 12.92,
+      "learning_rate": 2.8493305101527867e-05,
+      "loss": 0.7007,
+      "step": 282870
+    },
+    {
+      "epoch": 12.92,
+      "learning_rate": 2.8492543452099868e-05,
+      "loss": 0.7866,
+      "step": 282880
+    },
+    {
+      "epoch": 12.92,
+      "learning_rate": 2.849178180267187e-05,
+      "loss": 0.8542,
+      "step": 282890
+    },
+    {
+      "epoch": 12.92,
+      "learning_rate": 2.8491020153243863e-05,
+      "loss": 0.8365,
+      "step": 282900
+    },
+    {
+      "epoch": 12.92,
+      "learning_rate": 2.8490258503815864e-05,
+      "loss": 0.8455,
+      "step": 282910
+    },
+    {
+      "epoch": 12.92,
+      "learning_rate": 2.8489496854387864e-05,
+      "loss": 0.8881,
+      "step": 282920
+    },
+    {
+      "epoch": 12.92,
+      "learning_rate": 2.848873520495986e-05,
+      "loss": 0.8704,
+      "step": 282930
+    },
+    {
+      "epoch": 12.92,
+      "learning_rate": 2.848797355553186e-05,
+      "loss": 0.7924,
+      "step": 282940
+    },
+    {
+      "epoch": 12.92,
+      "learning_rate": 2.848721190610386e-05,
+      "loss": 0.8109,
+      "step": 282950
+    },
+    {
+      "epoch": 12.92,
+      "learning_rate": 2.8486450256675858e-05,
+      "loss": 0.8524,
+      "step": 282960
+    },
+    {
+      "epoch": 12.92,
+      "learning_rate": 2.848568860724786e-05,
+      "loss": 0.845,
+      "step": 282970
+    },
+    {
+      "epoch": 12.92,
+      "learning_rate": 2.8484926957819856e-05,
+      "loss": 0.8429,
+      "step": 282980
+    },
+    {
+      "epoch": 12.92,
+      "learning_rate": 2.8484165308391853e-05,
+      "loss": 0.8054,
+      "step": 282990
+    },
+    {
+      "epoch": 12.92,
+      "learning_rate": 2.8483403658963854e-05,
+      "loss": 0.7867,
+      "step": 283000
+    },
+    {
+      "epoch": 12.92,
+      "learning_rate": 2.8482642009535855e-05,
+      "loss": 0.7679,
+      "step": 283010
+    },
+    {
+      "epoch": 12.92,
+      "learning_rate": 2.848188036010785e-05,
+      "loss": 0.7893,
+      "step": 283020
+    },
+    {
+      "epoch": 12.92,
+      "learning_rate": 2.848111871067985e-05,
+      "loss": 0.8157,
+      "step": 283030
+    },
+    {
+      "epoch": 12.92,
+      "learning_rate": 2.848035706125185e-05,
+      "loss": 0.7769,
+      "step": 283040
+    },
+    {
+      "epoch": 12.92,
+      "learning_rate": 2.8479595411823845e-05,
+      "loss": 0.7792,
+      "step": 283050
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 2.8478833762395846e-05,
+      "loss": 0.9278,
+      "step": 283060
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 2.8478072112967847e-05,
+      "loss": 0.7783,
+      "step": 283070
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 2.847731046353984e-05,
+      "loss": 0.7953,
+      "step": 283080
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 2.847654881411184e-05,
+      "loss": 0.773,
+      "step": 283090
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 2.8475787164683843e-05,
+      "loss": 0.8569,
+      "step": 283100
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 2.8475025515255837e-05,
+      "loss": 0.7736,
+      "step": 283110
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 2.8474263865827837e-05,
+      "loss": 0.8505,
+      "step": 283120
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 2.8473502216399838e-05,
+      "loss": 0.7277,
+      "step": 283130
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 2.8472740566971832e-05,
+      "loss": 0.8462,
+      "step": 283140
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 2.8471978917543833e-05,
+      "loss": 0.7992,
+      "step": 283150
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 2.8471217268115834e-05,
+      "loss": 0.7812,
+      "step": 283160
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 2.847045561868783e-05,
+      "loss": 0.8337,
+      "step": 283170
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 2.8469693969259832e-05,
+      "loss": 0.8056,
+      "step": 283180
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 2.846893231983183e-05,
+      "loss": 0.8449,
+      "step": 283190
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 2.8468170670403827e-05,
+      "loss": 0.8516,
+      "step": 283200
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 2.8467409020975828e-05,
+      "loss": 0.8349,
+      "step": 283210
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 2.846664737154783e-05,
+      "loss": 0.849,
+      "step": 283220
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 2.8465885722119823e-05,
+      "loss": 0.8782,
+      "step": 283230
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 2.8465124072691824e-05,
+      "loss": 0.8117,
+      "step": 283240
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 2.8464362423263825e-05,
+      "loss": 0.8201,
+      "step": 283250
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 2.846360077383582e-05,
+      "loss": 0.7328,
+      "step": 283260
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 2.846283912440782e-05,
+      "loss": 0.8382,
+      "step": 283270
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 2.846207747497982e-05,
+      "loss": 0.8394,
+      "step": 283280
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 2.8461315825551815e-05,
+      "loss": 0.7147,
+      "step": 283290
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 2.8460554176123816e-05,
+      "loss": 0.673,
+      "step": 283300
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 2.8459792526695816e-05,
+      "loss": 0.7903,
+      "step": 283310
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 2.845903087726781e-05,
+      "loss": 0.743,
+      "step": 283320
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 2.845826922783981e-05,
+      "loss": 0.7493,
+      "step": 283330
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 2.845750757841181e-05,
+      "loss": 0.7979,
+      "step": 283340
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 2.8456745928983806e-05,
+      "loss": 0.8229,
+      "step": 283350
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 2.8455984279555807e-05,
+      "loss": 0.753,
+      "step": 283360
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 2.8455222630127805e-05,
+      "loss": 0.8082,
+      "step": 283370
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 2.8454460980699805e-05,
+      "loss": 0.775,
+      "step": 283380
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 2.8453699331271806e-05,
+      "loss": 0.8302,
+      "step": 283390
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 2.84529376818438e-05,
+      "loss": 0.7372,
+      "step": 283400
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 2.84521760324158e-05,
+      "loss": 0.8459,
+      "step": 283410
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 2.8451414382987802e-05,
+      "loss": 0.8563,
+      "step": 283420
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 2.8450652733559796e-05,
+      "loss": 0.8293,
+      "step": 283430
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 2.8449891084131797e-05,
+      "loss": 0.8082,
+      "step": 283440
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 2.8449129434703798e-05,
+      "loss": 0.7459,
+      "step": 283450
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 2.8448367785275792e-05,
+      "loss": 0.8209,
+      "step": 283460
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 2.8447606135847793e-05,
+      "loss": 0.8166,
+      "step": 283470
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 2.8446844486419794e-05,
+      "loss": 0.8083,
+      "step": 283480
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 2.8446082836991788e-05,
+      "loss": 0.8029,
+      "step": 283490
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 2.844532118756379e-05,
+      "loss": 0.7951,
+      "step": 283500
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 2.844455953813579e-05,
+      "loss": 0.792,
+      "step": 283510
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 2.8443797888707784e-05,
+      "loss": 0.7642,
+      "step": 283520
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 2.8443036239279784e-05,
+      "loss": 0.7569,
+      "step": 283530
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 2.8442274589851785e-05,
+      "loss": 0.8783,
+      "step": 283540
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 2.8441512940423783e-05,
+      "loss": 0.8144,
+      "step": 283550
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 2.844075129099578e-05,
+      "loss": 0.8519,
+      "step": 283560
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 2.843998964156778e-05,
+      "loss": 0.8029,
+      "step": 283570
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 2.843922799213978e-05,
+      "loss": 0.804,
+      "step": 283580
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 2.843846634271178e-05,
+      "loss": 0.8211,
+      "step": 283590
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 2.843770469328378e-05,
+      "loss": 0.799,
+      "step": 283600
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 2.8436943043855774e-05,
+      "loss": 0.8094,
+      "step": 283610
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 2.8436181394427775e-05,
+      "loss": 0.7668,
+      "step": 283620
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 2.8435419744999776e-05,
+      "loss": 0.7647,
+      "step": 283630
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 2.843465809557177e-05,
+      "loss": 0.8374,
+      "step": 283640
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 2.843389644614377e-05,
+      "loss": 0.8895,
+      "step": 283650
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 2.8433134796715772e-05,
+      "loss": 0.8689,
+      "step": 283660
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 2.8432373147287766e-05,
+      "loss": 0.7807,
+      "step": 283670
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 2.8431611497859767e-05,
+      "loss": 0.8301,
+      "step": 283680
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 2.8430849848431768e-05,
+      "loss": 0.8361,
+      "step": 283690
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 2.843008819900376e-05,
+      "loss": 0.8224,
+      "step": 283700
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 2.8429326549575763e-05,
+      "loss": 0.8772,
+      "step": 283710
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 2.8428564900147763e-05,
+      "loss": 0.7476,
+      "step": 283720
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 2.8427803250719757e-05,
+      "loss": 0.8779,
+      "step": 283730
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 2.842704160129176e-05,
+      "loss": 0.7519,
+      "step": 283740
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 2.842627995186376e-05,
+      "loss": 0.7753,
+      "step": 283750
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 2.8425518302435757e-05,
+      "loss": 0.8266,
+      "step": 283760
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 2.8424756653007757e-05,
+      "loss": 0.7267,
+      "step": 283770
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 2.8423995003579755e-05,
+      "loss": 0.843,
+      "step": 283780
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 2.8423233354151752e-05,
+      "loss": 0.7658,
+      "step": 283790
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 2.8422471704723753e-05,
+      "loss": 0.7162,
+      "step": 283800
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 2.8421710055295747e-05,
+      "loss": 0.7982,
+      "step": 283810
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 2.8420948405867748e-05,
+      "loss": 0.7338,
+      "step": 283820
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 2.842018675643975e-05,
+      "loss": 0.852,
+      "step": 283830
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 2.8419425107011743e-05,
+      "loss": 0.8183,
+      "step": 283840
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 2.8418663457583744e-05,
+      "loss": 0.8063,
+      "step": 283850
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 2.8417901808155745e-05,
+      "loss": 0.7535,
+      "step": 283860
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 2.841714015872774e-05,
+      "loss": 0.8284,
+      "step": 283870
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 2.841637850929974e-05,
+      "loss": 0.8314,
+      "step": 283880
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 2.841561685987174e-05,
+      "loss": 0.707,
+      "step": 283890
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 2.8414855210443735e-05,
+      "loss": 0.8388,
+      "step": 283900
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 2.8414093561015736e-05,
+      "loss": 0.7843,
+      "step": 283910
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 2.8413331911587736e-05,
+      "loss": 0.8317,
+      "step": 283920
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 2.8412570262159734e-05,
+      "loss": 0.8641,
+      "step": 283930
+    },
+    {
+      "epoch": 12.97,
+      "learning_rate": 2.841180861273173e-05,
+      "loss": 0.6905,
+      "step": 283940
+    },
+    {
+      "epoch": 12.97,
+      "learning_rate": 2.8411046963303732e-05,
+      "loss": 0.8252,
+      "step": 283950
+    },
+    {
+      "epoch": 12.97,
+      "learning_rate": 2.841028531387573e-05,
+      "loss": 0.759,
+      "step": 283960
+    },
+    {
+      "epoch": 12.97,
+      "learning_rate": 2.840952366444773e-05,
+      "loss": 0.804,
+      "step": 283970
+    },
+    {
+      "epoch": 12.97,
+      "learning_rate": 2.840876201501973e-05,
+      "loss": 0.73,
+      "step": 283980
+    },
+    {
+      "epoch": 12.97,
+      "learning_rate": 2.8408000365591725e-05,
+      "loss": 0.7497,
+      "step": 283990
+    },
+    {
+      "epoch": 12.97,
+      "learning_rate": 2.8407238716163726e-05,
+      "loss": 0.8291,
+      "step": 284000
+    },
+    {
+      "epoch": 12.97,
+      "learning_rate": 2.8406477066735727e-05,
+      "loss": 0.8156,
+      "step": 284010
+    },
+    {
+      "epoch": 12.97,
+      "learning_rate": 2.840571541730772e-05,
+      "loss": 0.8342,
+      "step": 284020
+    },
+    {
+      "epoch": 12.97,
+      "learning_rate": 2.8404953767879722e-05,
+      "loss": 0.8218,
+      "step": 284030
+    },
+    {
+      "epoch": 12.97,
+      "learning_rate": 2.8404192118451723e-05,
+      "loss": 0.8105,
+      "step": 284040
+    },
+    {
+      "epoch": 12.97,
+      "learning_rate": 2.8403430469023717e-05,
+      "loss": 0.7623,
+      "step": 284050
+    },
+    {
+      "epoch": 12.97,
+      "learning_rate": 2.8402668819595718e-05,
+      "loss": 0.8202,
+      "step": 284060
+    },
+    {
+      "epoch": 12.97,
+      "learning_rate": 2.840190717016772e-05,
+      "loss": 0.7831,
+      "step": 284070
+    },
+    {
+      "epoch": 12.97,
+      "learning_rate": 2.8401145520739713e-05,
+      "loss": 0.7657,
+      "step": 284080
+    },
+    {
+      "epoch": 12.97,
+      "learning_rate": 2.8400383871311714e-05,
+      "loss": 0.8141,
+      "step": 284090
+    },
+    {
+      "epoch": 12.97,
+      "learning_rate": 2.8399622221883715e-05,
+      "loss": 0.9308,
+      "step": 284100
+    },
+    {
+      "epoch": 12.97,
+      "learning_rate": 2.839886057245571e-05,
+      "loss": 0.7796,
+      "step": 284110
+    },
+    {
+      "epoch": 12.97,
+      "learning_rate": 2.839809892302771e-05,
+      "loss": 0.8791,
+      "step": 284120
+    },
+    {
+      "epoch": 12.97,
+      "learning_rate": 2.839733727359971e-05,
+      "loss": 0.8153,
+      "step": 284130
+    },
+    {
+      "epoch": 12.97,
+      "learning_rate": 2.8396575624171708e-05,
+      "loss": 0.7257,
+      "step": 284140
+    },
+    {
+      "epoch": 12.98,
+      "learning_rate": 2.8395813974743705e-05,
+      "loss": 0.717,
+      "step": 284150
+    },
+    {
+      "epoch": 12.98,
+      "learning_rate": 2.8395052325315706e-05,
+      "loss": 0.9738,
+      "step": 284160
+    },
+    {
+      "epoch": 12.98,
+      "learning_rate": 2.8394290675887704e-05,
+      "loss": 0.8421,
+      "step": 284170
+    },
+    {
+      "epoch": 12.98,
+      "learning_rate": 2.8393529026459704e-05,
+      "loss": 0.7357,
+      "step": 284180
+    },
+    {
+      "epoch": 12.98,
+      "learning_rate": 2.8392767377031705e-05,
+      "loss": 0.8248,
+      "step": 284190
+    },
+    {
+      "epoch": 12.98,
+      "learning_rate": 2.83920057276037e-05,
+      "loss": 0.8471,
+      "step": 284200
+    },
+    {
+      "epoch": 12.98,
+      "learning_rate": 2.83912440781757e-05,
+      "loss": 0.8139,
+      "step": 284210
+    },
+    {
+      "epoch": 12.98,
+      "learning_rate": 2.83904824287477e-05,
+      "loss": 0.7942,
+      "step": 284220
+    },
+    {
+      "epoch": 12.98,
+      "learning_rate": 2.8389720779319695e-05,
+      "loss": 0.7882,
+      "step": 284230
+    },
+    {
+      "epoch": 12.98,
+      "learning_rate": 2.8388959129891696e-05,
+      "loss": 0.8563,
+      "step": 284240
+    },
+    {
+      "epoch": 12.98,
+      "learning_rate": 2.838819748046369e-05,
+      "loss": 0.8891,
+      "step": 284250
+    },
+    {
+      "epoch": 12.98,
+      "learning_rate": 2.838743583103569e-05,
+      "loss": 0.7295,
+      "step": 284260
+    },
+    {
+      "epoch": 12.98,
+      "learning_rate": 2.8386674181607692e-05,
+      "loss": 0.6589,
+      "step": 284270
+    },
+    {
+      "epoch": 12.98,
+      "learning_rate": 2.8385912532179686e-05,
+      "loss": 0.8446,
+      "step": 284280
+    },
+    {
+      "epoch": 12.98,
+      "learning_rate": 2.8385150882751687e-05,
+      "loss": 0.8848,
+      "step": 284290
+    },
+    {
+      "epoch": 12.98,
+      "learning_rate": 2.8384389233323688e-05,
+      "loss": 0.7386,
+      "step": 284300
+    },
+    {
+      "epoch": 12.98,
+      "learning_rate": 2.838362758389568e-05,
+      "loss": 0.7753,
+      "step": 284310
+    },
+    {
+      "epoch": 12.98,
+      "learning_rate": 2.8382865934467683e-05,
+      "loss": 0.8126,
+      "step": 284320
+    },
+    {
+      "epoch": 12.98,
+      "learning_rate": 2.8382104285039683e-05,
+      "loss": 0.7975,
+      "step": 284330
+    },
+    {
+      "epoch": 12.98,
+      "learning_rate": 2.838134263561168e-05,
+      "loss": 0.7971,
+      "step": 284340
+    },
+    {
+      "epoch": 12.98,
+      "learning_rate": 2.8380580986183682e-05,
+      "loss": 0.8131,
+      "step": 284350
+    },
+    {
+      "epoch": 12.98,
+      "learning_rate": 2.837981933675568e-05,
+      "loss": 0.8415,
+      "step": 284360
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 2.8379057687327677e-05,
+      "loss": 0.7849,
+      "step": 284370
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 2.8378296037899677e-05,
+      "loss": 0.7413,
+      "step": 284380
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 2.837753438847168e-05,
+      "loss": 0.8854,
+      "step": 284390
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 2.8376772739043672e-05,
+      "loss": 0.8885,
+      "step": 284400
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 2.8376011089615673e-05,
+      "loss": 0.7851,
+      "step": 284410
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 2.8375249440187674e-05,
+      "loss": 0.8381,
+      "step": 284420
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 2.8374487790759668e-05,
+      "loss": 0.7859,
+      "step": 284430
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 2.837372614133167e-05,
+      "loss": 0.8472,
+      "step": 284440
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 2.837296449190367e-05,
+      "loss": 0.8065,
+      "step": 284450
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 2.8372202842475664e-05,
+      "loss": 0.8034,
+      "step": 284460
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 2.8371441193047665e-05,
+      "loss": 0.8243,
+      "step": 284470
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 2.8370679543619666e-05,
+      "loss": 0.8704,
+      "step": 284480
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 2.836991789419166e-05,
+      "loss": 0.828,
+      "step": 284490
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 2.836915624476366e-05,
+      "loss": 0.7931,
+      "step": 284500
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 2.836839459533566e-05,
+      "loss": 0.8323,
+      "step": 284510
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 2.836763294590766e-05,
+      "loss": 0.805,
+      "step": 284520
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 2.8366871296479656e-05,
+      "loss": 0.8143,
+      "step": 284530
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 2.8366109647051657e-05,
+      "loss": 0.8038,
+      "step": 284540
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 2.8365347997623655e-05,
+      "loss": 0.8597,
+      "step": 284550
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 2.8364586348195656e-05,
+      "loss": 0.8201,
+      "step": 284560
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 2.8363824698767656e-05,
+      "loss": 0.8484,
+      "step": 284570
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 2.836306304933965e-05,
+      "loss": 0.8574,
+      "step": 284580
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 2.836230139991165e-05,
+      "loss": 0.8511,
+      "step": 284590
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 2.8361539750483652e-05,
+      "loss": 0.7628,
+      "step": 284600
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 2.8360778101055646e-05,
+      "loss": 0.939,
+      "step": 284610
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 2.8360016451627647e-05,
+      "loss": 0.7787,
+      "step": 284620
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 2.8359254802199648e-05,
+      "loss": 0.8091,
+      "step": 284630
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 2.8358493152771642e-05,
+      "loss": 0.821,
+      "step": 284640
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 2.8357731503343643e-05,
+      "loss": 0.8414,
+      "step": 284650
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 2.8356969853915644e-05,
+      "loss": 0.7755,
+      "step": 284660
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 2.8356208204487638e-05,
+      "loss": 0.7901,
+      "step": 284670
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 2.835544655505964e-05,
+      "loss": 0.9506,
+      "step": 284680
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 2.835468490563164e-05,
+      "loss": 0.858,
+      "step": 284690
+    },
+    {
+      "epoch": 13.0,
+      "eval_cer": 0.6207169496480857,
+      "eval_em": 0.007473982970671713,
+      "eval_f1": 0.007473982970671713,
+      "eval_loss": 0.896234393119812,
+      "eval_runtime": 1020.8216,
+      "eval_samples_per_second": 10.354,
+      "eval_steps_per_second": 1.295,
+      "eval_wer": 0.9925260170293283,
+      "step": 284696
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 2.8353923256203634e-05,
+      "loss": 0.7716,
+      "step": 284700
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 2.8353161606775635e-05,
+      "loss": 0.7733,
+      "step": 284710
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 2.8352399957347632e-05,
+      "loss": 0.7517,
+      "step": 284720
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 2.8351638307919633e-05,
+      "loss": 0.8079,
+      "step": 284730
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 2.835087665849163e-05,
+      "loss": 0.8857,
+      "step": 284740
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 2.8350115009063628e-05,
+      "loss": 0.823,
+      "step": 284750
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 2.834935335963563e-05,
+      "loss": 0.8029,
+      "step": 284760
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 2.834859171020763e-05,
+      "loss": 0.7768,
+      "step": 284770
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 2.8347830060779624e-05,
+      "loss": 0.7409,
+      "step": 284780
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 2.8347068411351624e-05,
+      "loss": 0.7434,
+      "step": 284790
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 2.8346306761923625e-05,
+      "loss": 0.711,
+      "step": 284800
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 2.834554511249562e-05,
+      "loss": 0.6809,
+      "step": 284810
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 2.834478346306762e-05,
+      "loss": 0.7867,
+      "step": 284820
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 2.834402181363962e-05,
+      "loss": 0.8303,
+      "step": 284830
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 2.8343260164211615e-05,
+      "loss": 0.7024,
+      "step": 284840
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 2.8342498514783616e-05,
+      "loss": 0.7785,
+      "step": 284850
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 2.8341736865355617e-05,
+      "loss": 0.8462,
+      "step": 284860
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 2.834097521592761e-05,
+      "loss": 0.6929,
+      "step": 284870
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 2.8340213566499612e-05,
+      "loss": 0.7587,
+      "step": 284880
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 2.8339451917071613e-05,
+      "loss": 0.7634,
+      "step": 284890
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 2.8338690267643607e-05,
+      "loss": 0.8108,
+      "step": 284900
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 2.8337928618215608e-05,
+      "loss": 0.7589,
+      "step": 284910
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 2.833716696878761e-05,
+      "loss": 0.8323,
+      "step": 284920
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 2.8336405319359606e-05,
+      "loss": 0.7711,
+      "step": 284930
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 2.8335643669931607e-05,
+      "loss": 0.7805,
+      "step": 284940
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 2.8334882020503604e-05,
+      "loss": 0.9331,
+      "step": 284950
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 2.8334120371075602e-05,
+      "loss": 0.8245,
+      "step": 284960
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 2.8333358721647603e-05,
+      "loss": 0.7275,
+      "step": 284970
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 2.8332597072219603e-05,
+      "loss": 0.7682,
+      "step": 284980
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 2.8331835422791598e-05,
+      "loss": 0.8657,
+      "step": 284990
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 2.83310737733636e-05,
+      "loss": 0.7685,
+      "step": 285000
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 2.83303121239356e-05,
+      "loss": 0.7187,
+      "step": 285010
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 2.8329550474507593e-05,
+      "loss": 0.86,
+      "step": 285020
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 2.8328788825079594e-05,
+      "loss": 0.8938,
+      "step": 285030
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 2.8328027175651595e-05,
+      "loss": 0.7473,
+      "step": 285040
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 2.832726552622359e-05,
+      "loss": 0.6934,
+      "step": 285050
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 2.832650387679559e-05,
+      "loss": 0.924,
+      "step": 285060
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 2.832574222736759e-05,
+      "loss": 0.7822,
+      "step": 285070
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 2.8324980577939585e-05,
+      "loss": 0.7635,
+      "step": 285080
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 2.8324218928511586e-05,
+      "loss": 0.7212,
+      "step": 285090
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 2.8323457279083587e-05,
+      "loss": 0.8195,
+      "step": 285100
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 2.832269562965558e-05,
+      "loss": 0.7955,
+      "step": 285110
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 2.832193398022758e-05,
+      "loss": 0.8947,
+      "step": 285120
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 2.8321172330799582e-05,
+      "loss": 0.7364,
+      "step": 285130
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 2.832041068137158e-05,
+      "loss": 0.8792,
+      "step": 285140
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 2.831964903194358e-05,
+      "loss": 0.884,
+      "step": 285150
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 2.8318887382515578e-05,
+      "loss": 0.7619,
+      "step": 285160
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 2.8318125733087576e-05,
+      "loss": 0.776,
+      "step": 285170
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 2.8317364083659576e-05,
+      "loss": 0.8367,
+      "step": 285180
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 2.831660243423157e-05,
+      "loss": 0.7464,
+      "step": 285190
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 2.831584078480357e-05,
+      "loss": 0.8157,
+      "step": 285200
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 2.8315079135375572e-05,
+      "loss": 0.7874,
+      "step": 285210
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 2.8314317485947566e-05,
+      "loss": 0.738,
+      "step": 285220
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 2.8313555836519567e-05,
+      "loss": 0.7319,
+      "step": 285230
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 2.8312794187091568e-05,
+      "loss": 0.7421,
+      "step": 285240
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 2.8312032537663562e-05,
+      "loss": 0.7457,
+      "step": 285250
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 2.8311270888235563e-05,
+      "loss": 0.7713,
+      "step": 285260
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 2.8310509238807564e-05,
+      "loss": 0.7563,
+      "step": 285270
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 2.8309747589379558e-05,
+      "loss": 0.7559,
+      "step": 285280
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 2.830898593995156e-05,
+      "loss": 0.7833,
+      "step": 285290
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 2.830822429052356e-05,
+      "loss": 0.8174,
+      "step": 285300
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 2.8307462641095557e-05,
+      "loss": 0.7625,
+      "step": 285310
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 2.8306700991667558e-05,
+      "loss": 0.7312,
+      "step": 285320
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 2.8305939342239555e-05,
+      "loss": 0.7663,
+      "step": 285330
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 2.8305177692811553e-05,
+      "loss": 0.7711,
+      "step": 285340
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 2.8304416043383554e-05,
+      "loss": 0.8127,
+      "step": 285350
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 2.8303654393955555e-05,
+      "loss": 0.7181,
+      "step": 285360
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 2.830289274452755e-05,
+      "loss": 0.7538,
+      "step": 285370
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 2.830213109509955e-05,
+      "loss": 0.8339,
+      "step": 285380
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 2.830136944567155e-05,
+      "loss": 0.8006,
+      "step": 285390
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 2.8300607796243544e-05,
+      "loss": 0.742,
+      "step": 285400
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 2.8299846146815545e-05,
+      "loss": 0.8386,
+      "step": 285410
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 2.8299084497387546e-05,
+      "loss": 0.8015,
+      "step": 285420
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 2.829832284795954e-05,
+      "loss": 0.8,
+      "step": 285430
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 2.829756119853154e-05,
+      "loss": 0.7665,
+      "step": 285440
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 2.8296799549103542e-05,
+      "loss": 0.7698,
+      "step": 285450
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 2.8296037899675536e-05,
+      "loss": 0.7399,
+      "step": 285460
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 2.8295276250247537e-05,
+      "loss": 0.7683,
+      "step": 285470
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 2.8294514600819538e-05,
+      "loss": 0.8122,
+      "step": 285480
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 2.8293752951391532e-05,
+      "loss": 0.7456,
+      "step": 285490
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 2.8292991301963533e-05,
+      "loss": 0.7259,
+      "step": 285500
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 2.8292229652535534e-05,
+      "loss": 0.7779,
+      "step": 285510
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 2.829146800310753e-05,
+      "loss": 0.7965,
+      "step": 285520
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 2.8290706353679532e-05,
+      "loss": 0.747,
+      "step": 285530
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 2.828994470425153e-05,
+      "loss": 0.749,
+      "step": 285540
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 2.8289183054823527e-05,
+      "loss": 0.7299,
+      "step": 285550
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 2.8288421405395528e-05,
+      "loss": 0.7404,
+      "step": 285560
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 2.828765975596753e-05,
+      "loss": 0.8314,
+      "step": 285570
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 2.8286898106539523e-05,
+      "loss": 0.8449,
+      "step": 285580
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 2.8286136457111523e-05,
+      "loss": 0.8024,
+      "step": 285590
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 2.8285374807683524e-05,
+      "loss": 0.7888,
+      "step": 285600
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 2.828461315825552e-05,
+      "loss": 0.8091,
+      "step": 285610
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 2.828385150882752e-05,
+      "loss": 0.7189,
+      "step": 285620
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 2.828308985939952e-05,
+      "loss": 0.8034,
+      "step": 285630
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 2.8282328209971514e-05,
+      "loss": 0.7699,
+      "step": 285640
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 2.8281566560543515e-05,
+      "loss": 0.8449,
+      "step": 285650
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 2.828080491111551e-05,
+      "loss": 0.8208,
+      "step": 285660
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 2.828004326168751e-05,
+      "loss": 0.7407,
+      "step": 285670
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 2.827928161225951e-05,
+      "loss": 0.813,
+      "step": 285680
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 2.8278519962831508e-05,
+      "loss": 0.7941,
+      "step": 285690
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 2.8277758313403506e-05,
+      "loss": 0.7322,
+      "step": 285700
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 2.8276996663975507e-05,
+      "loss": 0.7313,
+      "step": 285710
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 2.8276235014547504e-05,
+      "loss": 0.8258,
+      "step": 285720
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 2.8275473365119505e-05,
+      "loss": 0.7755,
+      "step": 285730
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 2.8274711715691506e-05,
+      "loss": 0.7528,
+      "step": 285740
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 2.82739500662635e-05,
+      "loss": 0.7327,
+      "step": 285750
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 2.82731884168355e-05,
+      "loss": 0.8221,
+      "step": 285760
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 2.82724267674075e-05,
+      "loss": 0.7295,
+      "step": 285770
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 2.8271665117979496e-05,
+      "loss": 0.8305,
+      "step": 285780
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 2.8270903468551496e-05,
+      "loss": 0.8107,
+      "step": 285790
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 2.8270141819123497e-05,
+      "loss": 0.8547,
+      "step": 285800
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 2.826938016969549e-05,
+      "loss": 0.8218,
+      "step": 285810
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 2.8268618520267492e-05,
+      "loss": 0.7554,
+      "step": 285820
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 2.8267856870839493e-05,
+      "loss": 0.8051,
+      "step": 285830
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 2.8267095221411487e-05,
+      "loss": 0.8204,
+      "step": 285840
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 2.8266333571983488e-05,
+      "loss": 0.6842,
+      "step": 285850
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 2.826557192255549e-05,
+      "loss": 0.7719,
+      "step": 285860
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 2.8264810273127483e-05,
+      "loss": 0.8513,
+      "step": 285870
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 2.8264048623699484e-05,
+      "loss": 0.8613,
+      "step": 285880
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 2.8263286974271485e-05,
+      "loss": 0.8455,
+      "step": 285890
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 2.8262525324843482e-05,
+      "loss": 0.7768,
+      "step": 285900
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 2.826176367541548e-05,
+      "loss": 0.9119,
+      "step": 285910
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 2.826100202598748e-05,
+      "loss": 0.8412,
+      "step": 285920
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 2.8260240376559478e-05,
+      "loss": 0.7266,
+      "step": 285930
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 2.825947872713148e-05,
+      "loss": 0.7417,
+      "step": 285940
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 2.825871707770348e-05,
+      "loss": 0.7821,
+      "step": 285950
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 2.8257955428275474e-05,
+      "loss": 0.7109,
+      "step": 285960
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 2.8257193778847475e-05,
+      "loss": 0.8014,
+      "step": 285970
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 2.8256432129419475e-05,
+      "loss": 0.7187,
+      "step": 285980
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 2.825567047999147e-05,
+      "loss": 0.7435,
+      "step": 285990
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 2.825490883056347e-05,
+      "loss": 0.7631,
+      "step": 286000
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 2.825414718113547e-05,
+      "loss": 0.8413,
+      "step": 286010
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 2.8253385531707465e-05,
+      "loss": 0.8636,
+      "step": 286020
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 2.8252623882279466e-05,
+      "loss": 0.7521,
+      "step": 286030
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 2.8251862232851467e-05,
+      "loss": 0.7641,
+      "step": 286040
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 2.825110058342346e-05,
+      "loss": 0.7669,
+      "step": 286050
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 2.8250338933995462e-05,
+      "loss": 0.8069,
+      "step": 286060
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 2.8249577284567463e-05,
+      "loss": 0.8003,
+      "step": 286070
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 2.8248815635139457e-05,
+      "loss": 0.7848,
+      "step": 286080
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 2.8248053985711458e-05,
+      "loss": 0.7673,
+      "step": 286090
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 2.8247292336283455e-05,
+      "loss": 0.839,
+      "step": 286100
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 2.8246530686855456e-05,
+      "loss": 0.682,
+      "step": 286110
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 2.8245769037427457e-05,
+      "loss": 0.828,
+      "step": 286120
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 2.824500738799945e-05,
+      "loss": 0.8172,
+      "step": 286130
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 2.8244245738571452e-05,
+      "loss": 0.7725,
+      "step": 286140
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 2.8243484089143453e-05,
+      "loss": 0.7833,
+      "step": 286150
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 2.8242722439715447e-05,
+      "loss": 0.6975,
+      "step": 286160
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 2.8241960790287448e-05,
+      "loss": 0.7826,
+      "step": 286170
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 2.824119914085945e-05,
+      "loss": 0.8041,
+      "step": 286180
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 2.8240437491431443e-05,
+      "loss": 0.8457,
+      "step": 286190
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 2.8239675842003443e-05,
+      "loss": 0.6968,
+      "step": 286200
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 2.8238914192575444e-05,
+      "loss": 0.7539,
+      "step": 286210
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 2.823815254314744e-05,
+      "loss": 0.7776,
+      "step": 286220
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 2.823739089371944e-05,
+      "loss": 0.7218,
+      "step": 286230
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 2.823662924429144e-05,
+      "loss": 0.8021,
+      "step": 286240
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 2.8235867594863434e-05,
+      "loss": 0.8872,
+      "step": 286250
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 2.8235105945435435e-05,
+      "loss": 0.7333,
+      "step": 286260
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 2.8234344296007436e-05,
+      "loss": 0.7195,
+      "step": 286270
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 2.8233582646579433e-05,
+      "loss": 0.6972,
+      "step": 286280
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 2.823282099715143e-05,
+      "loss": 0.7945,
+      "step": 286290
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 2.823205934772343e-05,
+      "loss": 0.8351,
+      "step": 286300
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 2.823129769829543e-05,
+      "loss": 0.8691,
+      "step": 286310
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 2.823053604886743e-05,
+      "loss": 0.7028,
+      "step": 286320
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 2.822977439943943e-05,
+      "loss": 0.7026,
+      "step": 286330
+    },
+    {
+      "epoch": 13.08,
+      "learning_rate": 2.8229012750011425e-05,
+      "loss": 0.7691,
+      "step": 286340
+    },
+    {
+      "epoch": 13.08,
+      "learning_rate": 2.8228251100583426e-05,
+      "loss": 0.8536,
+      "step": 286350
+    },
+    {
+      "epoch": 13.08,
+      "learning_rate": 2.8227489451155427e-05,
+      "loss": 0.7364,
+      "step": 286360
+    },
+    {
+      "epoch": 13.08,
+      "learning_rate": 2.822672780172742e-05,
+      "loss": 0.7262,
+      "step": 286370
+    },
+    {
+      "epoch": 13.08,
+      "learning_rate": 2.822596615229942e-05,
+      "loss": 0.7088,
+      "step": 286380
+    },
+    {
+      "epoch": 13.08,
+      "learning_rate": 2.8225204502871422e-05,
+      "loss": 0.7959,
+      "step": 286390
+    },
+    {
+      "epoch": 13.08,
+      "learning_rate": 2.8224442853443417e-05,
+      "loss": 0.7694,
+      "step": 286400
+    },
+    {
+      "epoch": 13.08,
+      "learning_rate": 2.8223681204015417e-05,
+      "loss": 0.7378,
+      "step": 286410
+    },
+    {
+      "epoch": 13.08,
+      "learning_rate": 2.8222919554587418e-05,
+      "loss": 0.8086,
+      "step": 286420
+    },
+    {
+      "epoch": 13.08,
+      "learning_rate": 2.8222157905159412e-05,
+      "loss": 0.7813,
+      "step": 286430
+    },
+    {
+      "epoch": 13.08,
+      "learning_rate": 2.8221396255731413e-05,
+      "loss": 0.8018,
+      "step": 286440
+    },
+    {
+      "epoch": 13.08,
+      "learning_rate": 2.8220634606303414e-05,
+      "loss": 0.7922,
+      "step": 286450
+    },
+    {
+      "epoch": 13.08,
+      "learning_rate": 2.8219872956875408e-05,
+      "loss": 0.7816,
+      "step": 286460
+    },
+    {
+      "epoch": 13.08,
+      "learning_rate": 2.821911130744741e-05,
+      "loss": 0.7411,
+      "step": 286470
+    },
+    {
+      "epoch": 13.08,
+      "learning_rate": 2.821834965801941e-05,
+      "loss": 0.7689,
+      "step": 286480
+    },
+    {
+      "epoch": 13.08,
+      "learning_rate": 2.8217588008591407e-05,
+      "loss": 0.8045,
+      "step": 286490
+    },
+    {
+      "epoch": 13.08,
+      "learning_rate": 2.8216826359163405e-05,
+      "loss": 0.764,
+      "step": 286500
+    },
+    {
+      "epoch": 13.08,
+      "learning_rate": 2.8216064709735406e-05,
+      "loss": 0.7696,
+      "step": 286510
+    },
+    {
+      "epoch": 13.08,
+      "learning_rate": 2.8215303060307403e-05,
+      "loss": 0.8096,
+      "step": 286520
+    },
+    {
+      "epoch": 13.08,
+      "learning_rate": 2.8214541410879404e-05,
+      "loss": 0.7862,
+      "step": 286530
+    },
+    {
+      "epoch": 13.08,
+      "learning_rate": 2.8213779761451405e-05,
+      "loss": 0.8271,
+      "step": 286540
+    },
+    {
+      "epoch": 13.08,
+      "learning_rate": 2.82130181120234e-05,
+      "loss": 0.7125,
+      "step": 286550
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 2.82122564625954e-05,
+      "loss": 0.8534,
+      "step": 286560
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 2.8211494813167394e-05,
+      "loss": 0.7892,
+      "step": 286570
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 2.8210733163739395e-05,
+      "loss": 0.7331,
+      "step": 286580
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 2.8209971514311395e-05,
+      "loss": 0.8117,
+      "step": 286590
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 2.820920986488339e-05,
+      "loss": 0.8441,
+      "step": 286600
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 2.820844821545539e-05,
+      "loss": 0.7931,
+      "step": 286610
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 2.820768656602739e-05,
+      "loss": 0.812,
+      "step": 286620
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 2.8206924916599385e-05,
+      "loss": 0.695,
+      "step": 286630
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 2.8206163267171386e-05,
+      "loss": 0.7381,
+      "step": 286640
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 2.8205401617743387e-05,
+      "loss": 0.7724,
+      "step": 286650
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 2.820463996831538e-05,
+      "loss": 0.7952,
+      "step": 286660
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 2.8203878318887382e-05,
+      "loss": 0.8159,
+      "step": 286670
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 2.8203116669459383e-05,
+      "loss": 0.8184,
+      "step": 286680
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 2.820235502003138e-05,
+      "loss": 0.7515,
+      "step": 286690
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 2.820159337060338e-05,
+      "loss": 0.837,
+      "step": 286700
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 2.820083172117538e-05,
+      "loss": 0.8439,
+      "step": 286710
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 2.8200070071747376e-05,
+      "loss": 0.7989,
+      "step": 286720
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 2.8199308422319377e-05,
+      "loss": 0.8487,
+      "step": 286730
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 2.8198546772891378e-05,
+      "loss": 0.7633,
+      "step": 286740
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 2.8197785123463372e-05,
+      "loss": 0.74,
+      "step": 286750
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 2.8197023474035373e-05,
+      "loss": 0.7029,
+      "step": 286760
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 2.8196261824607374e-05,
+      "loss": 0.8126,
+      "step": 286770
+    },
+    {
+      "epoch": 13.1,
+      "learning_rate": 2.8195500175179368e-05,
+      "loss": 0.7059,
+      "step": 286780
+    },
+    {
+      "epoch": 13.1,
+      "learning_rate": 2.819473852575137e-05,
+      "loss": 0.7687,
+      "step": 286790
+    },
+    {
+      "epoch": 13.1,
+      "learning_rate": 2.819397687632337e-05,
+      "loss": 0.797,
+      "step": 286800
+    },
+    {
+      "epoch": 13.1,
+      "learning_rate": 2.8193215226895363e-05,
+      "loss": 0.7979,
+      "step": 286810
+    },
+    {
+      "epoch": 13.1,
+      "learning_rate": 2.8192453577467364e-05,
+      "loss": 0.7995,
+      "step": 286820
+    },
+    {
+      "epoch": 13.1,
+      "learning_rate": 2.8191691928039365e-05,
+      "loss": 0.7084,
+      "step": 286830
+    },
+    {
+      "epoch": 13.1,
+      "learning_rate": 2.819093027861136e-05,
+      "loss": 0.7772,
+      "step": 286840
+    },
+    {
+      "epoch": 13.1,
+      "learning_rate": 2.819016862918336e-05,
+      "loss": 0.6821,
+      "step": 286850
+    },
+    {
+      "epoch": 13.1,
+      "learning_rate": 2.818940697975536e-05,
+      "loss": 0.7677,
+      "step": 286860
+    },
+    {
+      "epoch": 13.1,
+      "learning_rate": 2.818864533032736e-05,
+      "loss": 0.7414,
+      "step": 286870
+    },
+    {
+      "epoch": 13.1,
+      "learning_rate": 2.8187883680899356e-05,
+      "loss": 0.8267,
+      "step": 286880
+    },
+    {
+      "epoch": 13.1,
+      "learning_rate": 2.8187122031471357e-05,
+      "loss": 0.7251,
+      "step": 286890
+    },
+    {
+      "epoch": 13.1,
+      "learning_rate": 2.8186360382043354e-05,
+      "loss": 0.6732,
+      "step": 286900
+    },
+    {
+      "epoch": 13.1,
+      "learning_rate": 2.8185598732615355e-05,
+      "loss": 0.8214,
+      "step": 286910
+    },
+    {
+      "epoch": 13.1,
+      "learning_rate": 2.8184837083187356e-05,
+      "loss": 0.8994,
+      "step": 286920
+    },
+    {
+      "epoch": 13.1,
+      "learning_rate": 2.818407543375935e-05,
+      "loss": 0.6588,
+      "step": 286930
+    },
+    {
+      "epoch": 13.1,
+      "learning_rate": 2.818331378433135e-05,
+      "loss": 0.8366,
+      "step": 286940
+    },
+    {
+      "epoch": 13.1,
+      "learning_rate": 2.8182552134903352e-05,
+      "loss": 0.7788,
+      "step": 286950
+    },
+    {
+      "epoch": 13.1,
+      "learning_rate": 2.8181790485475346e-05,
+      "loss": 0.7749,
+      "step": 286960
+    },
+    {
+      "epoch": 13.1,
+      "learning_rate": 2.8181028836047347e-05,
+      "loss": 0.7786,
+      "step": 286970
+    },
+    {
+      "epoch": 13.1,
+      "learning_rate": 2.8180267186619347e-05,
+      "loss": 0.7232,
+      "step": 286980
+    },
+    {
+      "epoch": 13.1,
+      "learning_rate": 2.817950553719134e-05,
+      "loss": 0.7753,
+      "step": 286990
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 2.8178743887763342e-05,
+      "loss": 0.7142,
+      "step": 287000
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 2.8177982238335343e-05,
+      "loss": 0.7118,
+      "step": 287010
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 2.8177220588907337e-05,
+      "loss": 0.752,
+      "step": 287020
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 2.8176458939479338e-05,
+      "loss": 0.7935,
+      "step": 287030
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 2.8175697290051332e-05,
+      "loss": 0.8201,
+      "step": 287040
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 2.8174935640623333e-05,
+      "loss": 0.7296,
+      "step": 287050
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 2.8174173991195334e-05,
+      "loss": 0.7899,
+      "step": 287060
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 2.817341234176733e-05,
+      "loss": 0.8356,
+      "step": 287070
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 2.8172650692339332e-05,
+      "loss": 0.8842,
+      "step": 287080
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 2.817188904291133e-05,
+      "loss": 0.7589,
+      "step": 287090
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 2.8171127393483327e-05,
+      "loss": 0.8493,
+      "step": 287100
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 2.8170365744055328e-05,
+      "loss": 0.7548,
+      "step": 287110
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 2.816960409462733e-05,
+      "loss": 0.7014,
+      "step": 287120
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 2.8168842445199323e-05,
+      "loss": 0.8536,
+      "step": 287130
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 2.8168080795771324e-05,
+      "loss": 0.7662,
+      "step": 287140
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 2.8167319146343325e-05,
+      "loss": 0.7548,
+      "step": 287150
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 2.816655749691532e-05,
+      "loss": 0.782,
+      "step": 287160
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 2.816579584748732e-05,
+      "loss": 0.7404,
+      "step": 287170
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 2.816503419805932e-05,
+      "loss": 0.7926,
+      "step": 287180
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 2.8164272548631315e-05,
+      "loss": 0.8624,
+      "step": 287190
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 2.8163510899203315e-05,
+      "loss": 0.8437,
+      "step": 287200
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 2.8162749249775316e-05,
+      "loss": 0.83,
+      "step": 287210
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 2.816198760034731e-05,
+      "loss": 0.7168,
+      "step": 287220
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 2.816122595091931e-05,
+      "loss": 0.8761,
+      "step": 287230
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 2.8160464301491312e-05,
+      "loss": 0.7321,
+      "step": 287240
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 2.8159702652063306e-05,
+      "loss": 0.7156,
+      "step": 287250
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 2.8158941002635307e-05,
+      "loss": 0.8209,
+      "step": 287260
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 2.8158179353207308e-05,
+      "loss": 0.8057,
+      "step": 287270
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 2.8157417703779305e-05,
+      "loss": 0.7941,
+      "step": 287280
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 2.8156656054351306e-05,
+      "loss": 0.7456,
+      "step": 287290
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 2.8155894404923304e-05,
+      "loss": 0.8125,
+      "step": 287300
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 2.81551327554953e-05,
+      "loss": 0.8505,
+      "step": 287310
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 2.8154371106067302e-05,
+      "loss": 0.8162,
+      "step": 287320
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 2.8153609456639303e-05,
+      "loss": 0.8071,
+      "step": 287330
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 2.8152847807211297e-05,
+      "loss": 0.7162,
+      "step": 287340
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 2.8152086157783298e-05,
+      "loss": 0.8627,
+      "step": 287350
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 2.81513245083553e-05,
+      "loss": 0.7291,
+      "step": 287360
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 2.8150562858927293e-05,
+      "loss": 0.7154,
+      "step": 287370
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 2.8149801209499294e-05,
+      "loss": 0.7256,
+      "step": 287380
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 2.8149039560071294e-05,
+      "loss": 0.7979,
+      "step": 287390
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 2.814827791064329e-05,
+      "loss": 0.8576,
+      "step": 287400
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 2.814751626121529e-05,
+      "loss": 0.7914,
+      "step": 287410
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 2.814675461178729e-05,
+      "loss": 0.7911,
+      "step": 287420
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 2.8145992962359284e-05,
+      "loss": 0.7866,
+      "step": 287430
+    },
+    {
+      "epoch": 13.13,
+      "learning_rate": 2.8145231312931285e-05,
+      "loss": 0.757,
+      "step": 287440
+    },
+    {
+      "epoch": 13.13,
+      "learning_rate": 2.8144469663503286e-05,
+      "loss": 0.7098,
+      "step": 287450
+    },
+    {
+      "epoch": 13.13,
+      "learning_rate": 2.814370801407528e-05,
+      "loss": 0.7604,
+      "step": 287460
+    },
+    {
+      "epoch": 13.13,
+      "learning_rate": 2.814294636464728e-05,
+      "loss": 0.8142,
+      "step": 287470
+    },
+    {
+      "epoch": 13.13,
+      "learning_rate": 2.814218471521928e-05,
+      "loss": 0.7519,
+      "step": 287480
+    },
+    {
+      "epoch": 13.13,
+      "learning_rate": 2.814142306579128e-05,
+      "loss": 0.7671,
+      "step": 287490
+    },
+    {
+      "epoch": 13.13,
+      "learning_rate": 2.814066141636328e-05,
+      "loss": 0.7512,
+      "step": 287500
+    },
+    {
+      "epoch": 13.13,
+      "learning_rate": 2.8139899766935274e-05,
+      "loss": 0.8504,
+      "step": 287510
+    },
+    {
+      "epoch": 13.13,
+      "learning_rate": 2.8139138117507275e-05,
+      "loss": 0.7055,
+      "step": 287520
+    },
+    {
+      "epoch": 13.13,
+      "learning_rate": 2.8138376468079276e-05,
+      "loss": 0.7803,
+      "step": 287530
+    },
+    {
+      "epoch": 13.13,
+      "learning_rate": 2.813761481865127e-05,
+      "loss": 0.7985,
+      "step": 287540
+    },
+    {
+      "epoch": 13.13,
+      "learning_rate": 2.813685316922327e-05,
+      "loss": 0.7758,
+      "step": 287550
+    },
+    {
+      "epoch": 13.13,
+      "learning_rate": 2.8136091519795272e-05,
+      "loss": 0.7718,
+      "step": 287560
+    },
+    {
+      "epoch": 13.13,
+      "learning_rate": 2.8135329870367266e-05,
+      "loss": 0.7881,
+      "step": 287570
+    },
+    {
+      "epoch": 13.13,
+      "learning_rate": 2.8134568220939267e-05,
+      "loss": 0.7216,
+      "step": 287580
+    },
+    {
+      "epoch": 13.13,
+      "learning_rate": 2.8133806571511268e-05,
+      "loss": 0.7447,
+      "step": 287590
+    },
+    {
+      "epoch": 13.13,
+      "learning_rate": 2.813304492208326e-05,
+      "loss": 0.7013,
+      "step": 287600
+    },
+    {
+      "epoch": 13.13,
+      "learning_rate": 2.8132283272655262e-05,
+      "loss": 0.788,
+      "step": 287610
+    },
+    {
+      "epoch": 13.13,
+      "learning_rate": 2.8131521623227263e-05,
+      "loss": 0.724,
+      "step": 287620
+    },
+    {
+      "epoch": 13.13,
+      "learning_rate": 2.8130759973799257e-05,
+      "loss": 0.8613,
+      "step": 287630
+    },
+    {
+      "epoch": 13.13,
+      "learning_rate": 2.8129998324371258e-05,
+      "loss": 0.877,
+      "step": 287640
+    },
+    {
+      "epoch": 13.13,
+      "learning_rate": 2.812923667494326e-05,
+      "loss": 0.7394,
+      "step": 287650
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 2.8128475025515257e-05,
+      "loss": 0.7709,
+      "step": 287660
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 2.8127713376087257e-05,
+      "loss": 0.7407,
+      "step": 287670
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 2.8126951726659255e-05,
+      "loss": 0.7141,
+      "step": 287680
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 2.8126190077231252e-05,
+      "loss": 0.7869,
+      "step": 287690
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 2.8125428427803253e-05,
+      "loss": 0.8298,
+      "step": 287700
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 2.8124666778375254e-05,
+      "loss": 0.7202,
+      "step": 287710
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 2.8123905128947248e-05,
+      "loss": 0.8597,
+      "step": 287720
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 2.812314347951925e-05,
+      "loss": 0.7869,
+      "step": 287730
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 2.812238183009125e-05,
+      "loss": 0.7624,
+      "step": 287740
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 2.8121620180663244e-05,
+      "loss": 0.7376,
+      "step": 287750
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 2.8120858531235245e-05,
+      "loss": 0.8328,
+      "step": 287760
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 2.8120096881807246e-05,
+      "loss": 0.7511,
+      "step": 287770
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 2.811933523237924e-05,
+      "loss": 0.7927,
+      "step": 287780
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 2.811857358295124e-05,
+      "loss": 0.776,
+      "step": 287790
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 2.811781193352324e-05,
+      "loss": 0.8285,
+      "step": 287800
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 2.8117050284095235e-05,
+      "loss": 0.7623,
+      "step": 287810
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 2.8116288634667236e-05,
+      "loss": 0.7826,
+      "step": 287820
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 2.8115526985239237e-05,
+      "loss": 0.7965,
+      "step": 287830
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 2.811476533581123e-05,
+      "loss": 0.8208,
+      "step": 287840
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 2.8114003686383232e-05,
+      "loss": 0.8989,
+      "step": 287850
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 2.8113242036955233e-05,
+      "loss": 0.8099,
+      "step": 287860
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 2.811248038752723e-05,
+      "loss": 0.7757,
+      "step": 287870
+    },
+    {
+      "epoch": 13.15,
+      "learning_rate": 2.811171873809923e-05,
+      "loss": 0.8278,
+      "step": 287880
+    },
+    {
+      "epoch": 13.15,
+      "learning_rate": 2.811095708867123e-05,
+      "loss": 0.7532,
+      "step": 287890
+    },
+    {
+      "epoch": 13.15,
+      "learning_rate": 2.8110195439243226e-05,
+      "loss": 0.7855,
+      "step": 287900
+    },
+    {
+      "epoch": 13.15,
+      "learning_rate": 2.8109433789815227e-05,
+      "loss": 0.8107,
+      "step": 287910
+    },
+    {
+      "epoch": 13.15,
+      "learning_rate": 2.8108672140387228e-05,
+      "loss": 0.8954,
+      "step": 287920
+    },
+    {
+      "epoch": 13.15,
+      "learning_rate": 2.8107910490959222e-05,
+      "loss": 0.8447,
+      "step": 287930
+    },
+    {
+      "epoch": 13.15,
+      "learning_rate": 2.8107148841531223e-05,
+      "loss": 0.7754,
+      "step": 287940
+    },
+    {
+      "epoch": 13.15,
+      "learning_rate": 2.8106387192103217e-05,
+      "loss": 0.8442,
+      "step": 287950
+    },
+    {
+      "epoch": 13.15,
+      "learning_rate": 2.8105625542675218e-05,
+      "loss": 0.8314,
+      "step": 287960
+    },
+    {
+      "epoch": 13.15,
+      "learning_rate": 2.810486389324722e-05,
+      "loss": 0.7305,
+      "step": 287970
+    },
+    {
+      "epoch": 13.15,
+      "learning_rate": 2.8104102243819213e-05,
+      "loss": 0.7581,
+      "step": 287980
+    },
+    {
+      "epoch": 13.15,
+      "learning_rate": 2.8103340594391214e-05,
+      "loss": 0.8795,
+      "step": 287990
+    },
+    {
+      "epoch": 13.15,
+      "learning_rate": 2.8102578944963214e-05,
+      "loss": 0.7598,
+      "step": 288000
+    },
+    {
+      "epoch": 13.15,
+      "learning_rate": 2.810181729553521e-05,
+      "loss": 0.677,
+      "step": 288010
+    },
+    {
+      "epoch": 13.15,
+      "learning_rate": 2.810105564610721e-05,
+      "loss": 0.8582,
+      "step": 288020
+    },
+    {
+      "epoch": 13.15,
+      "learning_rate": 2.810029399667921e-05,
+      "loss": 0.8794,
+      "step": 288030
+    },
+    {
+      "epoch": 13.15,
+      "learning_rate": 2.8099532347251208e-05,
+      "loss": 0.8107,
+      "step": 288040
+    },
+    {
+      "epoch": 13.15,
+      "learning_rate": 2.8098770697823205e-05,
+      "loss": 0.7841,
+      "step": 288050
+    },
+    {
+      "epoch": 13.15,
+      "learning_rate": 2.8098009048395206e-05,
+      "loss": 0.8055,
+      "step": 288060
+    },
+    {
+      "epoch": 13.15,
+      "learning_rate": 2.8097247398967204e-05,
+      "loss": 0.7697,
+      "step": 288070
+    },
+    {
+      "epoch": 13.15,
+      "learning_rate": 2.8096485749539204e-05,
+      "loss": 0.8604,
+      "step": 288080
+    },
+    {
+      "epoch": 13.15,
+      "learning_rate": 2.8095724100111205e-05,
+      "loss": 0.7181,
+      "step": 288090
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 2.80949624506832e-05,
+      "loss": 0.7502,
+      "step": 288100
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 2.80942008012552e-05,
+      "loss": 0.7586,
+      "step": 288110
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 2.80934391518272e-05,
+      "loss": 0.7528,
+      "step": 288120
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 2.8092677502399195e-05,
+      "loss": 0.7273,
+      "step": 288130
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 2.8091915852971196e-05,
+      "loss": 0.7628,
+      "step": 288140
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 2.8091154203543197e-05,
+      "loss": 0.8449,
+      "step": 288150
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 2.809039255411519e-05,
+      "loss": 0.8643,
+      "step": 288160
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 2.8089630904687192e-05,
+      "loss": 0.7942,
+      "step": 288170
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 2.8088869255259193e-05,
+      "loss": 0.7737,
+      "step": 288180
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 2.8088107605831187e-05,
+      "loss": 0.8191,
+      "step": 288190
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 2.8087345956403188e-05,
+      "loss": 0.7433,
+      "step": 288200
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 2.808658430697519e-05,
+      "loss": 0.8465,
+      "step": 288210
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 2.8085822657547182e-05,
+      "loss": 0.7641,
+      "step": 288220
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 2.8085061008119183e-05,
+      "loss": 0.8944,
+      "step": 288230
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 2.8084299358691184e-05,
+      "loss": 0.737,
+      "step": 288240
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 2.808353770926318e-05,
+      "loss": 0.8271,
+      "step": 288250
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 2.808277605983518e-05,
+      "loss": 0.7547,
+      "step": 288260
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 2.808201441040718e-05,
+      "loss": 0.8147,
+      "step": 288270
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 2.8081252760979177e-05,
+      "loss": 0.805,
+      "step": 288280
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 2.8080491111551178e-05,
+      "loss": 0.7922,
+      "step": 288290
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 2.807972946212318e-05,
+      "loss": 0.8654,
+      "step": 288300
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 2.8078967812695173e-05,
+      "loss": 0.7301,
+      "step": 288310
+    },
+    {
+      "epoch": 13.17,
+      "learning_rate": 2.8078206163267174e-05,
+      "loss": 0.7969,
+      "step": 288320
+    },
+    {
+      "epoch": 13.17,
+      "learning_rate": 2.8077444513839175e-05,
+      "loss": 0.7098,
+      "step": 288330
+    },
+    {
+      "epoch": 13.17,
+      "learning_rate": 2.807668286441117e-05,
+      "loss": 0.7129,
+      "step": 288340
+    },
+    {
+      "epoch": 13.17,
+      "learning_rate": 2.807592121498317e-05,
+      "loss": 0.7152,
+      "step": 288350
+    },
+    {
+      "epoch": 13.17,
+      "learning_rate": 2.807515956555517e-05,
+      "loss": 0.7694,
+      "step": 288360
+    },
+    {
+      "epoch": 13.17,
+      "learning_rate": 2.8074397916127165e-05,
+      "loss": 0.8216,
+      "step": 288370
+    },
+    {
+      "epoch": 13.17,
+      "learning_rate": 2.8073636266699166e-05,
+      "loss": 0.7567,
+      "step": 288380
+    },
+    {
+      "epoch": 13.17,
+      "learning_rate": 2.8072874617271166e-05,
+      "loss": 0.7487,
+      "step": 288390
+    },
+    {
+      "epoch": 13.17,
+      "learning_rate": 2.807211296784316e-05,
+      "loss": 0.8065,
+      "step": 288400
+    },
+    {
+      "epoch": 13.17,
+      "learning_rate": 2.807135131841516e-05,
+      "loss": 0.686,
+      "step": 288410
+    },
+    {
+      "epoch": 13.17,
+      "learning_rate": 2.807058966898716e-05,
+      "loss": 0.8089,
+      "step": 288420
+    },
+    {
+      "epoch": 13.17,
+      "learning_rate": 2.8069828019559156e-05,
+      "loss": 0.719,
+      "step": 288430
+    },
+    {
+      "epoch": 13.17,
+      "learning_rate": 2.8069066370131157e-05,
+      "loss": 0.7167,
+      "step": 288440
+    },
+    {
+      "epoch": 13.17,
+      "learning_rate": 2.8068304720703155e-05,
+      "loss": 0.8882,
+      "step": 288450
+    },
+    {
+      "epoch": 13.17,
+      "learning_rate": 2.8067543071275156e-05,
+      "loss": 0.7664,
+      "step": 288460
+    },
+    {
+      "epoch": 13.17,
+      "learning_rate": 2.8066781421847156e-05,
+      "loss": 0.8044,
+      "step": 288470
+    },
+    {
+      "epoch": 13.17,
+      "learning_rate": 2.806601977241915e-05,
+      "loss": 0.7492,
+      "step": 288480
+    },
+    {
+      "epoch": 13.17,
+      "learning_rate": 2.806525812299115e-05,
+      "loss": 0.8516,
+      "step": 288490
+    },
+    {
+      "epoch": 13.17,
+      "learning_rate": 2.8064496473563152e-05,
+      "loss": 0.7272,
+      "step": 288500
+    },
+    {
+      "epoch": 13.17,
+      "learning_rate": 2.8063734824135146e-05,
+      "loss": 0.847,
+      "step": 288510
+    },
+    {
+      "epoch": 13.17,
+      "learning_rate": 2.8062973174707147e-05,
+      "loss": 0.8746,
+      "step": 288520
+    },
+    {
+      "epoch": 13.18,
+      "learning_rate": 2.8062211525279148e-05,
+      "loss": 0.8027,
+      "step": 288530
+    },
+    {
+      "epoch": 13.18,
+      "learning_rate": 2.8061449875851142e-05,
+      "loss": 0.7169,
+      "step": 288540
+    },
+    {
+      "epoch": 13.18,
+      "learning_rate": 2.8060688226423143e-05,
+      "loss": 0.7261,
+      "step": 288550
+    },
+    {
+      "epoch": 13.18,
+      "learning_rate": 2.8059926576995144e-05,
+      "loss": 0.8199,
+      "step": 288560
+    },
+    {
+      "epoch": 13.18,
+      "learning_rate": 2.8059164927567138e-05,
+      "loss": 0.7502,
+      "step": 288570
+    },
+    {
+      "epoch": 13.18,
+      "learning_rate": 2.805840327813914e-05,
+      "loss": 0.7048,
+      "step": 288580
+    },
+    {
+      "epoch": 13.18,
+      "learning_rate": 2.805764162871114e-05,
+      "loss": 0.7548,
+      "step": 288590
+    },
+    {
+      "epoch": 13.18,
+      "learning_rate": 2.8056879979283134e-05,
+      "loss": 0.7877,
+      "step": 288600
+    },
+    {
+      "epoch": 13.18,
+      "learning_rate": 2.8056118329855134e-05,
+      "loss": 0.7665,
+      "step": 288610
+    },
+    {
+      "epoch": 13.18,
+      "learning_rate": 2.8055356680427135e-05,
+      "loss": 0.8716,
+      "step": 288620
+    },
+    {
+      "epoch": 13.18,
+      "learning_rate": 2.8054595030999133e-05,
+      "loss": 0.7664,
+      "step": 288630
+    },
+    {
+      "epoch": 13.18,
+      "learning_rate": 2.805383338157113e-05,
+      "loss": 0.7785,
+      "step": 288640
+    },
+    {
+      "epoch": 13.18,
+      "learning_rate": 2.805307173214313e-05,
+      "loss": 0.8362,
+      "step": 288650
+    },
+    {
+      "epoch": 13.18,
+      "learning_rate": 2.805231008271513e-05,
+      "loss": 0.7476,
+      "step": 288660
+    },
+    {
+      "epoch": 13.18,
+      "learning_rate": 2.805154843328713e-05,
+      "loss": 0.8478,
+      "step": 288670
+    },
+    {
+      "epoch": 13.18,
+      "learning_rate": 2.805078678385913e-05,
+      "loss": 0.8313,
+      "step": 288680
+    },
+    {
+      "epoch": 13.18,
+      "learning_rate": 2.8050025134431124e-05,
+      "loss": 0.7705,
+      "step": 288690
+    },
+    {
+      "epoch": 13.18,
+      "learning_rate": 2.8049263485003125e-05,
+      "loss": 0.7642,
+      "step": 288700
+    },
+    {
+      "epoch": 13.18,
+      "learning_rate": 2.8048501835575126e-05,
+      "loss": 0.7747,
+      "step": 288710
+    },
+    {
+      "epoch": 13.18,
+      "learning_rate": 2.804774018614712e-05,
+      "loss": 0.7744,
+      "step": 288720
+    },
+    {
+      "epoch": 13.18,
+      "learning_rate": 2.804697853671912e-05,
+      "loss": 0.7963,
+      "step": 288730
+    },
+    {
+      "epoch": 13.18,
+      "learning_rate": 2.8046216887291122e-05,
+      "loss": 0.7558,
+      "step": 288740
+    },
+    {
+      "epoch": 13.19,
+      "learning_rate": 2.8045455237863116e-05,
+      "loss": 0.7727,
+      "step": 288750
+    },
+    {
+      "epoch": 13.19,
+      "learning_rate": 2.8044693588435117e-05,
+      "loss": 0.7958,
+      "step": 288760
+    },
+    {
+      "epoch": 13.19,
+      "learning_rate": 2.8043931939007118e-05,
+      "loss": 0.8144,
+      "step": 288770
+    },
+    {
+      "epoch": 13.19,
+      "learning_rate": 2.8043170289579112e-05,
+      "loss": 0.858,
+      "step": 288780
+    },
+    {
+      "epoch": 13.19,
+      "learning_rate": 2.8042408640151113e-05,
+      "loss": 0.7369,
+      "step": 288790
+    },
+    {
+      "epoch": 13.19,
+      "learning_rate": 2.8041646990723113e-05,
+      "loss": 0.8613,
+      "step": 288800
+    },
+    {
+      "epoch": 13.19,
+      "learning_rate": 2.8040885341295108e-05,
+      "loss": 0.7531,
+      "step": 288810
+    },
+    {
+      "epoch": 13.19,
+      "learning_rate": 2.804012369186711e-05,
+      "loss": 0.707,
+      "step": 288820
+    },
+    {
+      "epoch": 13.19,
+      "learning_rate": 2.803936204243911e-05,
+      "loss": 0.7868,
+      "step": 288830
+    },
+    {
+      "epoch": 13.19,
+      "learning_rate": 2.8038600393011107e-05,
+      "loss": 0.8527,
+      "step": 288840
+    },
+    {
+      "epoch": 13.19,
+      "learning_rate": 2.8037838743583104e-05,
+      "loss": 0.7603,
+      "step": 288850
+    },
+    {
+      "epoch": 13.19,
+      "learning_rate": 2.80370770941551e-05,
+      "loss": 0.7619,
+      "step": 288860
+    },
+    {
+      "epoch": 13.19,
+      "learning_rate": 2.8036315444727102e-05,
+      "loss": 0.8693,
+      "step": 288870
+    },
+    {
+      "epoch": 13.19,
+      "learning_rate": 2.8035553795299103e-05,
+      "loss": 0.7827,
+      "step": 288880
+    },
+    {
+      "epoch": 13.19,
+      "learning_rate": 2.8034792145871097e-05,
+      "loss": 0.7634,
+      "step": 288890
+    },
+    {
+      "epoch": 13.19,
+      "learning_rate": 2.8034030496443098e-05,
+      "loss": 0.8582,
+      "step": 288900
+    },
+    {
+      "epoch": 13.19,
+      "learning_rate": 2.80332688470151e-05,
+      "loss": 0.7355,
+      "step": 288910
+    },
+    {
+      "epoch": 13.19,
+      "learning_rate": 2.8032507197587093e-05,
+      "loss": 0.7296,
+      "step": 288920
+    },
+    {
+      "epoch": 13.19,
+      "learning_rate": 2.8031745548159094e-05,
+      "loss": 0.8371,
+      "step": 288930
+    },
+    {
+      "epoch": 13.19,
+      "learning_rate": 2.8030983898731095e-05,
+      "loss": 0.8423,
+      "step": 288940
+    },
+    {
+      "epoch": 13.19,
+      "learning_rate": 2.803022224930309e-05,
+      "loss": 0.7657,
+      "step": 288950
+    },
+    {
+      "epoch": 13.19,
+      "learning_rate": 2.802946059987509e-05,
+      "loss": 0.7688,
+      "step": 288960
+    },
+    {
+      "epoch": 13.2,
+      "learning_rate": 2.802869895044709e-05,
+      "loss": 0.7277,
+      "step": 288970
+    },
+    {
+      "epoch": 13.2,
+      "learning_rate": 2.8027937301019085e-05,
+      "loss": 0.7412,
+      "step": 288980
+    },
+    {
+      "epoch": 13.2,
+      "learning_rate": 2.8027175651591086e-05,
+      "loss": 0.7598,
+      "step": 288990
+    },
+    {
+      "epoch": 13.2,
+      "learning_rate": 2.8026414002163087e-05,
+      "loss": 0.769,
+      "step": 289000
+    },
+    {
+      "epoch": 13.2,
+      "learning_rate": 2.802565235273508e-05,
+      "loss": 0.8273,
+      "step": 289010
+    },
+    {
+      "epoch": 13.2,
+      "learning_rate": 2.802489070330708e-05,
+      "loss": 0.8589,
+      "step": 289020
+    },
+    {
+      "epoch": 13.2,
+      "learning_rate": 2.8024129053879082e-05,
+      "loss": 0.6877,
+      "step": 289030
+    },
+    {
+      "epoch": 13.2,
+      "learning_rate": 2.802336740445108e-05,
+      "loss": 0.7851,
+      "step": 289040
+    },
+    {
+      "epoch": 13.2,
+      "learning_rate": 2.802260575502308e-05,
+      "loss": 0.8729,
+      "step": 289050
+    },
+    {
+      "epoch": 13.2,
+      "learning_rate": 2.802184410559508e-05,
+      "loss": 0.8251,
+      "step": 289060
+    },
+    {
+      "epoch": 13.2,
+      "learning_rate": 2.8021082456167076e-05,
+      "loss": 0.748,
+      "step": 289070
+    },
+    {
+      "epoch": 13.2,
+      "learning_rate": 2.8020320806739076e-05,
+      "loss": 0.7182,
+      "step": 289080
+    },
+    {
+      "epoch": 13.2,
+      "learning_rate": 2.8019559157311077e-05,
+      "loss": 0.8181,
+      "step": 289090
+    },
+    {
+      "epoch": 13.2,
+      "learning_rate": 2.801879750788307e-05,
+      "loss": 0.8981,
+      "step": 289100
+    },
+    {
+      "epoch": 13.2,
+      "learning_rate": 2.8018035858455072e-05,
+      "loss": 0.8336,
+      "step": 289110
+    },
+    {
+      "epoch": 13.2,
+      "learning_rate": 2.8017274209027073e-05,
+      "loss": 0.7721,
+      "step": 289120
+    },
+    {
+      "epoch": 13.2,
+      "learning_rate": 2.8016512559599067e-05,
+      "loss": 0.7471,
+      "step": 289130
+    },
+    {
+      "epoch": 13.2,
+      "learning_rate": 2.8015750910171068e-05,
+      "loss": 0.8209,
+      "step": 289140
+    },
+    {
+      "epoch": 13.2,
+      "learning_rate": 2.801498926074307e-05,
+      "loss": 0.7884,
+      "step": 289150
+    },
+    {
+      "epoch": 13.2,
+      "learning_rate": 2.8014227611315063e-05,
+      "loss": 0.7714,
+      "step": 289160
+    },
+    {
+      "epoch": 13.2,
+      "learning_rate": 2.8013465961887064e-05,
+      "loss": 0.84,
+      "step": 289170
+    },
+    {
+      "epoch": 13.2,
+      "learning_rate": 2.8012704312459065e-05,
+      "loss": 0.7835,
+      "step": 289180
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 2.801194266303106e-05,
+      "loss": 0.8074,
+      "step": 289190
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 2.801118101360306e-05,
+      "loss": 0.8729,
+      "step": 289200
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 2.801041936417506e-05,
+      "loss": 0.7254,
+      "step": 289210
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 2.8009657714747058e-05,
+      "loss": 0.83,
+      "step": 289220
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 2.8008896065319055e-05,
+      "loss": 0.7327,
+      "step": 289230
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 2.8008134415891056e-05,
+      "loss": 0.7822,
+      "step": 289240
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 2.8007372766463054e-05,
+      "loss": 0.7855,
+      "step": 289250
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 2.8006611117035055e-05,
+      "loss": 0.8233,
+      "step": 289260
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 2.8005849467607055e-05,
+      "loss": 0.8739,
+      "step": 289270
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 2.800508781817905e-05,
+      "loss": 0.8162,
+      "step": 289280
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 2.800432616875105e-05,
+      "loss": 0.7894,
+      "step": 289290
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 2.800356451932305e-05,
+      "loss": 0.7752,
+      "step": 289300
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 2.8002802869895045e-05,
+      "loss": 0.7251,
+      "step": 289310
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 2.8002041220467046e-05,
+      "loss": 0.8423,
+      "step": 289320
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 2.800127957103904e-05,
+      "loss": 0.7234,
+      "step": 289330
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 2.800051792161104e-05,
+      "loss": 0.8238,
+      "step": 289340
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 2.7999756272183042e-05,
+      "loss": 0.7961,
+      "step": 289350
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 2.7998994622755036e-05,
+      "loss": 0.7489,
+      "step": 289360
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 2.7998232973327037e-05,
+      "loss": 0.7739,
+      "step": 289370
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 2.7997471323899038e-05,
+      "loss": 0.8414,
+      "step": 289380
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 2.7996709674471032e-05,
+      "loss": 0.8486,
+      "step": 289390
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 2.7995948025043033e-05,
+      "loss": 0.8389,
+      "step": 289400
+    },
+    {
+      "epoch": 13.22,
+      "learning_rate": 2.7995186375615033e-05,
+      "loss": 0.85,
+      "step": 289410
+    },
+    {
+      "epoch": 13.22,
+      "learning_rate": 2.799442472618703e-05,
+      "loss": 0.7631,
+      "step": 289420
+    },
+    {
+      "epoch": 13.22,
+      "learning_rate": 2.7993663076759032e-05,
+      "loss": 0.7353,
+      "step": 289430
+    },
+    {
+      "epoch": 13.22,
+      "learning_rate": 2.799290142733103e-05,
+      "loss": 0.7107,
+      "step": 289440
+    },
+    {
+      "epoch": 13.22,
+      "learning_rate": 2.7992139777903027e-05,
+      "loss": 0.8708,
+      "step": 289450
+    },
+    {
+      "epoch": 13.22,
+      "learning_rate": 2.7991378128475028e-05,
+      "loss": 0.8109,
+      "step": 289460
+    },
+    {
+      "epoch": 13.22,
+      "learning_rate": 2.799061647904703e-05,
+      "loss": 0.8447,
+      "step": 289470
+    },
+    {
+      "epoch": 13.22,
+      "learning_rate": 2.7989854829619023e-05,
+      "loss": 0.7625,
+      "step": 289480
+    },
+    {
+      "epoch": 13.22,
+      "learning_rate": 2.7989093180191023e-05,
+      "loss": 0.747,
+      "step": 289490
+    },
+    {
+      "epoch": 13.22,
+      "learning_rate": 2.7988331530763024e-05,
+      "loss": 0.7714,
+      "step": 289500
+    },
+    {
+      "epoch": 13.22,
+      "learning_rate": 2.7987569881335018e-05,
+      "loss": 0.7336,
+      "step": 289510
+    },
+    {
+      "epoch": 13.22,
+      "learning_rate": 2.798680823190702e-05,
+      "loss": 0.8156,
+      "step": 289520
+    },
+    {
+      "epoch": 13.22,
+      "learning_rate": 2.798604658247902e-05,
+      "loss": 0.7705,
+      "step": 289530
+    },
+    {
+      "epoch": 13.22,
+      "learning_rate": 2.7985284933051014e-05,
+      "loss": 0.8068,
+      "step": 289540
+    },
+    {
+      "epoch": 13.22,
+      "learning_rate": 2.7984523283623015e-05,
+      "loss": 0.7949,
+      "step": 289550
+    },
+    {
+      "epoch": 13.22,
+      "learning_rate": 2.7983761634195016e-05,
+      "loss": 0.8066,
+      "step": 289560
+    },
+    {
+      "epoch": 13.22,
+      "learning_rate": 2.798299998476701e-05,
+      "loss": 0.6972,
+      "step": 289570
+    },
+    {
+      "epoch": 13.22,
+      "learning_rate": 2.798223833533901e-05,
+      "loss": 0.7523,
+      "step": 289580
+    },
+    {
+      "epoch": 13.22,
+      "learning_rate": 2.798147668591101e-05,
+      "loss": 0.863,
+      "step": 289590
+    },
+    {
+      "epoch": 13.22,
+      "learning_rate": 2.7980715036483006e-05,
+      "loss": 0.7938,
+      "step": 289600
+    },
+    {
+      "epoch": 13.22,
+      "learning_rate": 2.7979953387055007e-05,
+      "loss": 0.8435,
+      "step": 289610
+    },
+    {
+      "epoch": 13.22,
+      "learning_rate": 2.7979191737627007e-05,
+      "loss": 0.7652,
+      "step": 289620
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 2.7978430088199005e-05,
+      "loss": 0.7749,
+      "step": 289630
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 2.7977668438771006e-05,
+      "loss": 0.7929,
+      "step": 289640
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 2.7976906789343003e-05,
+      "loss": 0.8212,
+      "step": 289650
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 2.7976145139915e-05,
+      "loss": 0.8168,
+      "step": 289660
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 2.7975383490487e-05,
+      "loss": 0.735,
+      "step": 289670
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 2.7974621841059002e-05,
+      "loss": 0.8813,
+      "step": 289680
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 2.7973860191630996e-05,
+      "loss": 0.795,
+      "step": 289690
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 2.7973098542202997e-05,
+      "loss": 0.7497,
+      "step": 289700
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 2.7972336892774998e-05,
+      "loss": 0.7292,
+      "step": 289710
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 2.7971575243346992e-05,
+      "loss": 0.739,
+      "step": 289720
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 2.7970813593918993e-05,
+      "loss": 0.7207,
+      "step": 289730
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 2.7970051944490994e-05,
+      "loss": 0.7924,
+      "step": 289740
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 2.7969290295062988e-05,
+      "loss": 0.8442,
+      "step": 289750
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 2.796852864563499e-05,
+      "loss": 0.756,
+      "step": 289760
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 2.796776699620699e-05,
+      "loss": 0.7951,
+      "step": 289770
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 2.7967005346778984e-05,
+      "loss": 0.8041,
+      "step": 289780
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 2.7966243697350985e-05,
+      "loss": 0.8234,
+      "step": 289790
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 2.7965482047922982e-05,
+      "loss": 0.7644,
+      "step": 289800
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 2.7964720398494983e-05,
+      "loss": 0.8253,
+      "step": 289810
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 2.796395874906698e-05,
+      "loss": 0.8419,
+      "step": 289820
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 2.7963197099638978e-05,
+      "loss": 0.6658,
+      "step": 289830
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 2.796243545021098e-05,
+      "loss": 0.809,
+      "step": 289840
+    },
+    {
+      "epoch": 13.24,
+      "learning_rate": 2.796167380078298e-05,
+      "loss": 0.8208,
+      "step": 289850
+    },
+    {
+      "epoch": 13.24,
+      "learning_rate": 2.7960912151354974e-05,
+      "loss": 0.8931,
+      "step": 289860
+    },
+    {
+      "epoch": 13.24,
+      "learning_rate": 2.7960150501926975e-05,
+      "loss": 0.8001,
+      "step": 289870
+    },
+    {
+      "epoch": 13.24,
+      "learning_rate": 2.7959388852498975e-05,
+      "loss": 0.758,
+      "step": 289880
+    },
+    {
+      "epoch": 13.24,
+      "learning_rate": 2.795862720307097e-05,
+      "loss": 0.805,
+      "step": 289890
+    },
+    {
+      "epoch": 13.24,
+      "learning_rate": 2.795786555364297e-05,
+      "loss": 0.6818,
+      "step": 289900
+    },
+    {
+      "epoch": 13.24,
+      "learning_rate": 2.795710390421497e-05,
+      "loss": 0.829,
+      "step": 289910
+    },
+    {
+      "epoch": 13.24,
+      "learning_rate": 2.7956342254786965e-05,
+      "loss": 0.8845,
+      "step": 289920
+    },
+    {
+      "epoch": 13.24,
+      "learning_rate": 2.7955580605358966e-05,
+      "loss": 0.8341,
+      "step": 289930
+    },
+    {
+      "epoch": 13.24,
+      "learning_rate": 2.7954818955930967e-05,
+      "loss": 0.7749,
+      "step": 289940
+    },
+    {
+      "epoch": 13.24,
+      "learning_rate": 2.795405730650296e-05,
+      "loss": 0.7646,
+      "step": 289950
+    },
+    {
+      "epoch": 13.24,
+      "learning_rate": 2.7953295657074962e-05,
+      "loss": 0.8752,
+      "step": 289960
+    },
+    {
+      "epoch": 13.24,
+      "learning_rate": 2.7952534007646963e-05,
+      "loss": 0.6803,
+      "step": 289970
+    },
+    {
+      "epoch": 13.24,
+      "learning_rate": 2.7951772358218957e-05,
+      "loss": 0.7752,
+      "step": 289980
+    },
+    {
+      "epoch": 13.24,
+      "learning_rate": 2.7951010708790958e-05,
+      "loss": 0.8414,
+      "step": 289990
+    },
+    {
+      "epoch": 13.24,
+      "learning_rate": 2.795024905936296e-05,
+      "loss": 0.7466,
+      "step": 290000
+    },
+    {
+      "epoch": 13.24,
+      "learning_rate": 2.7949487409934956e-05,
+      "loss": 0.7635,
+      "step": 290010
+    },
+    {
+      "epoch": 13.24,
+      "learning_rate": 2.7948725760506957e-05,
+      "loss": 0.8497,
+      "step": 290020
+    },
+    {
+      "epoch": 13.24,
+      "learning_rate": 2.7947964111078954e-05,
+      "loss": 0.7846,
+      "step": 290030
+    },
+    {
+      "epoch": 13.24,
+      "learning_rate": 2.7947202461650952e-05,
+      "loss": 0.6964,
+      "step": 290040
+    },
+    {
+      "epoch": 13.24,
+      "learning_rate": 2.7946440812222953e-05,
+      "loss": 0.7865,
+      "step": 290050
+    },
+    {
+      "epoch": 13.24,
+      "learning_rate": 2.7945679162794954e-05,
+      "loss": 0.8134,
+      "step": 290060
+    },
+    {
+      "epoch": 13.25,
+      "learning_rate": 2.7944917513366948e-05,
+      "loss": 0.8507,
+      "step": 290070
+    },
+    {
+      "epoch": 13.25,
+      "learning_rate": 2.794415586393895e-05,
+      "loss": 0.8314,
+      "step": 290080
+    },
+    {
+      "epoch": 13.25,
+      "learning_rate": 2.794339421451095e-05,
+      "loss": 0.7038,
+      "step": 290090
+    },
+    {
+      "epoch": 13.25,
+      "learning_rate": 2.7942632565082943e-05,
+      "loss": 0.6958,
+      "step": 290100
+    },
+    {
+      "epoch": 13.25,
+      "learning_rate": 2.7941870915654944e-05,
+      "loss": 0.808,
+      "step": 290110
+    },
+    {
+      "epoch": 13.25,
+      "learning_rate": 2.7941109266226945e-05,
+      "loss": 0.7845,
+      "step": 290120
+    },
+    {
+      "epoch": 13.25,
+      "learning_rate": 2.794034761679894e-05,
+      "loss": 0.8712,
+      "step": 290130
+    },
+    {
+      "epoch": 13.25,
+      "learning_rate": 2.793958596737094e-05,
+      "loss": 0.6742,
+      "step": 290140
+    },
+    {
+      "epoch": 13.25,
+      "learning_rate": 2.793882431794294e-05,
+      "loss": 0.8023,
+      "step": 290150
+    },
+    {
+      "epoch": 13.25,
+      "learning_rate": 2.7938062668514935e-05,
+      "loss": 0.7208,
+      "step": 290160
+    },
+    {
+      "epoch": 13.25,
+      "learning_rate": 2.7937301019086936e-05,
+      "loss": 0.7672,
+      "step": 290170
+    },
+    {
+      "epoch": 13.25,
+      "learning_rate": 2.7936539369658937e-05,
+      "loss": 0.7222,
+      "step": 290180
+    },
+    {
+      "epoch": 13.25,
+      "learning_rate": 2.793577772023093e-05,
+      "loss": 0.7821,
+      "step": 290190
+    },
+    {
+      "epoch": 13.25,
+      "learning_rate": 2.793501607080293e-05,
+      "loss": 0.792,
+      "step": 290200
+    },
+    {
+      "epoch": 13.25,
+      "learning_rate": 2.7934254421374932e-05,
+      "loss": 0.833,
+      "step": 290210
+    },
+    {
+      "epoch": 13.25,
+      "learning_rate": 2.793349277194693e-05,
+      "loss": 0.7124,
+      "step": 290220
+    },
+    {
+      "epoch": 13.25,
+      "learning_rate": 2.793273112251893e-05,
+      "loss": 0.8208,
+      "step": 290230
+    },
+    {
+      "epoch": 13.25,
+      "learning_rate": 2.7931969473090928e-05,
+      "loss": 0.7555,
+      "step": 290240
+    },
+    {
+      "epoch": 13.25,
+      "learning_rate": 2.7931207823662926e-05,
+      "loss": 0.793,
+      "step": 290250
+    },
+    {
+      "epoch": 13.25,
+      "learning_rate": 2.7930446174234927e-05,
+      "loss": 0.8437,
+      "step": 290260
+    },
+    {
+      "epoch": 13.25,
+      "learning_rate": 2.792968452480692e-05,
+      "loss": 0.7687,
+      "step": 290270
+    },
+    {
+      "epoch": 13.25,
+      "learning_rate": 2.792892287537892e-05,
+      "loss": 0.8685,
+      "step": 290280
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 2.7928161225950922e-05,
+      "loss": 0.8716,
+      "step": 290290
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 2.7927399576522916e-05,
+      "loss": 0.7676,
+      "step": 290300
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 2.7926637927094917e-05,
+      "loss": 0.7773,
+      "step": 290310
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 2.7925876277666918e-05,
+      "loss": 0.7082,
+      "step": 290320
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 2.7925114628238912e-05,
+      "loss": 0.7821,
+      "step": 290330
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 2.7924352978810913e-05,
+      "loss": 0.7764,
+      "step": 290340
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 2.7923591329382914e-05,
+      "loss": 0.7941,
+      "step": 290350
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 2.7922829679954908e-05,
+      "loss": 0.772,
+      "step": 290360
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 2.792206803052691e-05,
+      "loss": 0.7603,
+      "step": 290370
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 2.792130638109891e-05,
+      "loss": 0.7869,
+      "step": 290380
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 2.7920544731670907e-05,
+      "loss": 0.7384,
+      "step": 290390
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 2.7919783082242905e-05,
+      "loss": 0.8152,
+      "step": 290400
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 2.7919021432814905e-05,
+      "loss": 0.7367,
+      "step": 290410
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 2.7918259783386903e-05,
+      "loss": 0.8589,
+      "step": 290420
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 2.7917498133958904e-05,
+      "loss": 0.7481,
+      "step": 290430
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 2.7916736484530905e-05,
+      "loss": 0.8406,
+      "step": 290440
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 2.79159748351029e-05,
+      "loss": 0.8039,
+      "step": 290450
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 2.79152131856749e-05,
+      "loss": 0.7035,
+      "step": 290460
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 2.79144515362469e-05,
+      "loss": 0.7165,
+      "step": 290470
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 2.7913689886818895e-05,
+      "loss": 0.8328,
+      "step": 290480
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 2.7912928237390895e-05,
+      "loss": 0.8441,
+      "step": 290490
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 2.7912166587962896e-05,
+      "loss": 0.7122,
+      "step": 290500
+    },
+    {
+      "epoch": 13.27,
+      "learning_rate": 2.791140493853489e-05,
+      "loss": 0.6673,
+      "step": 290510
+    },
+    {
+      "epoch": 13.27,
+      "learning_rate": 2.791064328910689e-05,
+      "loss": 0.8112,
+      "step": 290520
+    },
+    {
+      "epoch": 13.27,
+      "learning_rate": 2.7909881639678892e-05,
+      "loss": 0.7331,
+      "step": 290530
+    },
+    {
+      "epoch": 13.27,
+      "learning_rate": 2.7909119990250886e-05,
+      "loss": 0.8019,
+      "step": 290540
+    },
+    {
+      "epoch": 13.27,
+      "learning_rate": 2.7908358340822887e-05,
+      "loss": 0.732,
+      "step": 290550
+    },
+    {
+      "epoch": 13.27,
+      "learning_rate": 2.7907596691394888e-05,
+      "loss": 0.746,
+      "step": 290560
+    },
+    {
+      "epoch": 13.27,
+      "learning_rate": 2.7906835041966882e-05,
+      "loss": 0.8282,
+      "step": 290570
+    },
+    {
+      "epoch": 13.27,
+      "learning_rate": 2.7906073392538883e-05,
+      "loss": 0.7819,
+      "step": 290580
+    },
+    {
+      "epoch": 13.27,
+      "learning_rate": 2.7905311743110884e-05,
+      "loss": 0.8098,
+      "step": 290590
+    },
+    {
+      "epoch": 13.27,
+      "learning_rate": 2.790455009368288e-05,
+      "loss": 0.7262,
+      "step": 290600
+    },
+    {
+      "epoch": 13.27,
+      "learning_rate": 2.7903788444254882e-05,
+      "loss": 0.9058,
+      "step": 290610
+    },
+    {
+      "epoch": 13.27,
+      "learning_rate": 2.790302679482688e-05,
+      "loss": 0.8597,
+      "step": 290620
+    },
+    {
+      "epoch": 13.27,
+      "learning_rate": 2.7902265145398877e-05,
+      "loss": 0.8505,
+      "step": 290630
+    },
+    {
+      "epoch": 13.27,
+      "learning_rate": 2.7901503495970878e-05,
+      "loss": 0.7392,
+      "step": 290640
+    },
+    {
+      "epoch": 13.27,
+      "learning_rate": 2.790074184654288e-05,
+      "loss": 0.8406,
+      "step": 290650
+    },
+    {
+      "epoch": 13.27,
+      "learning_rate": 2.7899980197114873e-05,
+      "loss": 0.8244,
+      "step": 290660
+    },
+    {
+      "epoch": 13.27,
+      "learning_rate": 2.7899218547686874e-05,
+      "loss": 0.7936,
+      "step": 290670
+    },
+    {
+      "epoch": 13.27,
+      "learning_rate": 2.7898456898258874e-05,
+      "loss": 0.7579,
+      "step": 290680
+    },
+    {
+      "epoch": 13.27,
+      "learning_rate": 2.789769524883087e-05,
+      "loss": 0.7169,
+      "step": 290690
+    },
+    {
+      "epoch": 13.27,
+      "learning_rate": 2.789693359940287e-05,
+      "loss": 0.7555,
+      "step": 290700
+    },
+    {
+      "epoch": 13.27,
+      "learning_rate": 2.7896171949974863e-05,
+      "loss": 0.8784,
+      "step": 290710
+    },
+    {
+      "epoch": 13.28,
+      "learning_rate": 2.7895410300546864e-05,
+      "loss": 0.7948,
+      "step": 290720
+    },
+    {
+      "epoch": 13.28,
+      "learning_rate": 2.7894648651118865e-05,
+      "loss": 0.7722,
+      "step": 290730
+    },
+    {
+      "epoch": 13.28,
+      "learning_rate": 2.789388700169086e-05,
+      "loss": 0.7737,
+      "step": 290740
+    },
+    {
+      "epoch": 13.28,
+      "learning_rate": 2.789312535226286e-05,
+      "loss": 0.7389,
+      "step": 290750
+    },
+    {
+      "epoch": 13.28,
+      "learning_rate": 2.789236370283486e-05,
+      "loss": 0.8604,
+      "step": 290760
+    },
+    {
+      "epoch": 13.28,
+      "learning_rate": 2.789160205340686e-05,
+      "loss": 0.7942,
+      "step": 290770
+    },
+    {
+      "epoch": 13.28,
+      "learning_rate": 2.7890840403978856e-05,
+      "loss": 0.803,
+      "step": 290780
+    },
+    {
+      "epoch": 13.28,
+      "learning_rate": 2.7890078754550857e-05,
+      "loss": 0.8952,
+      "step": 290790
+    },
+    {
+      "epoch": 13.28,
+      "learning_rate": 2.7889317105122854e-05,
+      "loss": 0.824,
+      "step": 290800
+    },
+    {
+      "epoch": 13.28,
+      "learning_rate": 2.7888555455694855e-05,
+      "loss": 0.765,
+      "step": 290810
+    },
+    {
+      "epoch": 13.28,
+      "learning_rate": 2.7887793806266856e-05,
+      "loss": 0.856,
+      "step": 290820
+    },
+    {
+      "epoch": 13.28,
+      "learning_rate": 2.788703215683885e-05,
+      "loss": 0.7517,
+      "step": 290830
+    },
+    {
+      "epoch": 13.28,
+      "learning_rate": 2.788627050741085e-05,
+      "loss": 0.8709,
+      "step": 290840
+    },
+    {
+      "epoch": 13.28,
+      "learning_rate": 2.788550885798285e-05,
+      "loss": 0.8673,
+      "step": 290850
+    },
+    {
+      "epoch": 13.28,
+      "learning_rate": 2.7884747208554846e-05,
+      "loss": 0.8425,
+      "step": 290860
+    },
+    {
+      "epoch": 13.28,
+      "learning_rate": 2.7883985559126847e-05,
+      "loss": 0.7991,
+      "step": 290870
+    },
+    {
+      "epoch": 13.28,
+      "learning_rate": 2.7883223909698847e-05,
+      "loss": 0.8239,
+      "step": 290880
+    },
+    {
+      "epoch": 13.28,
+      "learning_rate": 2.788246226027084e-05,
+      "loss": 0.7902,
+      "step": 290890
+    },
+    {
+      "epoch": 13.28,
+      "learning_rate": 2.7881700610842842e-05,
+      "loss": 0.7123,
+      "step": 290900
+    },
+    {
+      "epoch": 13.28,
+      "learning_rate": 2.7880938961414843e-05,
+      "loss": 0.7853,
+      "step": 290910
+    },
+    {
+      "epoch": 13.28,
+      "learning_rate": 2.7880177311986837e-05,
+      "loss": 0.8066,
+      "step": 290920
+    },
+    {
+      "epoch": 13.28,
+      "learning_rate": 2.7879415662558838e-05,
+      "loss": 0.7687,
+      "step": 290930
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 2.787865401313084e-05,
+      "loss": 0.6765,
+      "step": 290940
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 2.7877892363702833e-05,
+      "loss": 0.7625,
+      "step": 290950
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 2.7877130714274834e-05,
+      "loss": 0.7535,
+      "step": 290960
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 2.7876369064846835e-05,
+      "loss": 0.7428,
+      "step": 290970
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 2.7875607415418832e-05,
+      "loss": 0.737,
+      "step": 290980
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 2.787484576599083e-05,
+      "loss": 0.7448,
+      "step": 290990
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 2.787408411656283e-05,
+      "loss": 0.7796,
+      "step": 291000
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 2.7873322467134828e-05,
+      "loss": 0.7889,
+      "step": 291010
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 2.787256081770683e-05,
+      "loss": 0.87,
+      "step": 291020
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 2.787179916827883e-05,
+      "loss": 0.8094,
+      "step": 291030
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 2.7871037518850824e-05,
+      "loss": 0.7914,
+      "step": 291040
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 2.7870275869422825e-05,
+      "loss": 0.8827,
+      "step": 291050
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 2.7869514219994826e-05,
+      "loss": 0.769,
+      "step": 291060
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 2.786875257056682e-05,
+      "loss": 0.7525,
+      "step": 291070
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 2.786799092113882e-05,
+      "loss": 0.8066,
+      "step": 291080
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 2.786722927171082e-05,
+      "loss": 0.7815,
+      "step": 291090
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 2.7866467622282815e-05,
+      "loss": 0.7616,
+      "step": 291100
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 2.7865705972854816e-05,
+      "loss": 0.8089,
+      "step": 291110
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 2.7864944323426817e-05,
+      "loss": 0.7096,
+      "step": 291120
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 2.786418267399881e-05,
+      "loss": 0.8171,
+      "step": 291130
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 2.7863421024570812e-05,
+      "loss": 0.8491,
+      "step": 291140
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 2.7862659375142813e-05,
+      "loss": 0.8539,
+      "step": 291150
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 2.7861897725714807e-05,
+      "loss": 0.8315,
+      "step": 291160
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 2.7861136076286808e-05,
+      "loss": 0.8059,
+      "step": 291170
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 2.7860374426858805e-05,
+      "loss": 0.8228,
+      "step": 291180
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 2.7859612777430806e-05,
+      "loss": 0.7453,
+      "step": 291190
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 2.7858851128002804e-05,
+      "loss": 0.8076,
+      "step": 291200
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 2.78580894785748e-05,
+      "loss": 0.8407,
+      "step": 291210
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 2.7857327829146802e-05,
+      "loss": 0.732,
+      "step": 291220
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 2.7856566179718803e-05,
+      "loss": 0.7573,
+      "step": 291230
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 2.7855804530290797e-05,
+      "loss": 0.7943,
+      "step": 291240
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 2.7855042880862798e-05,
+      "loss": 0.8149,
+      "step": 291250
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 2.78542812314348e-05,
+      "loss": 0.8658,
+      "step": 291260
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 2.7853519582006793e-05,
+      "loss": 0.749,
+      "step": 291270
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 2.7852757932578794e-05,
+      "loss": 0.8487,
+      "step": 291280
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 2.7851996283150794e-05,
+      "loss": 0.898,
+      "step": 291290
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 2.785123463372279e-05,
+      "loss": 0.742,
+      "step": 291300
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 2.785047298429479e-05,
+      "loss": 0.7226,
+      "step": 291310
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 2.784971133486679e-05,
+      "loss": 0.822,
+      "step": 291320
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 2.7848949685438784e-05,
+      "loss": 0.8846,
+      "step": 291330
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 2.7848188036010785e-05,
+      "loss": 0.7719,
+      "step": 291340
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 2.7847426386582786e-05,
+      "loss": 0.8442,
+      "step": 291350
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 2.7846664737154783e-05,
+      "loss": 0.8001,
+      "step": 291360
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 2.784590308772678e-05,
+      "loss": 0.7807,
+      "step": 291370
+    },
+    {
+      "epoch": 13.31,
+      "learning_rate": 2.7845141438298782e-05,
+      "loss": 0.7939,
+      "step": 291380
+    },
+    {
+      "epoch": 13.31,
+      "learning_rate": 2.784437978887078e-05,
+      "loss": 0.8226,
+      "step": 291390
+    },
+    {
+      "epoch": 13.31,
+      "learning_rate": 2.784361813944278e-05,
+      "loss": 0.8396,
+      "step": 291400
+    },
+    {
+      "epoch": 13.31,
+      "learning_rate": 2.784285649001478e-05,
+      "loss": 0.8797,
+      "step": 291410
+    },
+    {
+      "epoch": 13.31,
+      "learning_rate": 2.7842094840586775e-05,
+      "loss": 0.8141,
+      "step": 291420
+    },
+    {
+      "epoch": 13.31,
+      "learning_rate": 2.7841333191158776e-05,
+      "loss": 0.7227,
+      "step": 291430
+    },
+    {
+      "epoch": 13.31,
+      "learning_rate": 2.7840571541730777e-05,
+      "loss": 0.809,
+      "step": 291440
+    },
+    {
+      "epoch": 13.31,
+      "learning_rate": 2.783980989230277e-05,
+      "loss": 0.7845,
+      "step": 291450
+    },
+    {
+      "epoch": 13.31,
+      "learning_rate": 2.783904824287477e-05,
+      "loss": 0.7101,
+      "step": 291460
+    },
+    {
+      "epoch": 13.31,
+      "learning_rate": 2.7838286593446773e-05,
+      "loss": 0.7954,
+      "step": 291470
+    },
+    {
+      "epoch": 13.31,
+      "learning_rate": 2.7837524944018767e-05,
+      "loss": 0.8851,
+      "step": 291480
+    },
+    {
+      "epoch": 13.31,
+      "learning_rate": 2.7836763294590767e-05,
+      "loss": 0.7869,
+      "step": 291490
+    },
+    {
+      "epoch": 13.31,
+      "learning_rate": 2.7836001645162768e-05,
+      "loss": 0.7916,
+      "step": 291500
+    },
+    {
+      "epoch": 13.31,
+      "learning_rate": 2.7835239995734762e-05,
+      "loss": 0.8026,
+      "step": 291510
+    },
+    {
+      "epoch": 13.31,
+      "learning_rate": 2.7834478346306763e-05,
+      "loss": 0.7901,
+      "step": 291520
+    },
+    {
+      "epoch": 13.31,
+      "learning_rate": 2.7833716696878764e-05,
+      "loss": 0.7162,
+      "step": 291530
+    },
+    {
+      "epoch": 13.31,
+      "learning_rate": 2.7832955047450758e-05,
+      "loss": 0.7512,
+      "step": 291540
+    },
+    {
+      "epoch": 13.31,
+      "learning_rate": 2.783219339802276e-05,
+      "loss": 0.7055,
+      "step": 291550
+    },
+    {
+      "epoch": 13.31,
+      "learning_rate": 2.783143174859476e-05,
+      "loss": 0.6622,
+      "step": 291560
+    },
+    {
+      "epoch": 13.31,
+      "learning_rate": 2.7830670099166757e-05,
+      "loss": 0.7311,
+      "step": 291570
+    },
+    {
+      "epoch": 13.31,
+      "learning_rate": 2.7829908449738755e-05,
+      "loss": 0.7097,
+      "step": 291580
+    },
+    {
+      "epoch": 13.31,
+      "learning_rate": 2.7829146800310756e-05,
+      "loss": 0.7512,
+      "step": 291590
+    },
+    {
+      "epoch": 13.32,
+      "learning_rate": 2.7828385150882753e-05,
+      "loss": 0.7134,
+      "step": 291600
+    },
+    {
+      "epoch": 13.32,
+      "learning_rate": 2.7827623501454754e-05,
+      "loss": 0.8198,
+      "step": 291610
+    },
+    {
+      "epoch": 13.32,
+      "learning_rate": 2.7826861852026755e-05,
+      "loss": 0.7866,
+      "step": 291620
+    },
+    {
+      "epoch": 13.32,
+      "learning_rate": 2.782610020259875e-05,
+      "loss": 0.7222,
+      "step": 291630
+    },
+    {
+      "epoch": 13.32,
+      "learning_rate": 2.782533855317075e-05,
+      "loss": 0.8013,
+      "step": 291640
+    },
+    {
+      "epoch": 13.32,
+      "learning_rate": 2.7824576903742744e-05,
+      "loss": 0.8017,
+      "step": 291650
+    },
+    {
+      "epoch": 13.32,
+      "learning_rate": 2.7823815254314745e-05,
+      "loss": 0.7628,
+      "step": 291660
+    },
+    {
+      "epoch": 13.32,
+      "learning_rate": 2.7823053604886746e-05,
+      "loss": 0.9401,
+      "step": 291670
+    },
+    {
+      "epoch": 13.32,
+      "learning_rate": 2.782229195545874e-05,
+      "loss": 0.7574,
+      "step": 291680
+    },
+    {
+      "epoch": 13.32,
+      "learning_rate": 2.782153030603074e-05,
+      "loss": 0.7512,
+      "step": 291690
+    },
+    {
+      "epoch": 13.32,
+      "learning_rate": 2.782076865660274e-05,
+      "loss": 0.7429,
+      "step": 291700
+    },
+    {
+      "epoch": 13.32,
+      "learning_rate": 2.7820007007174735e-05,
+      "loss": 0.7805,
+      "step": 291710
+    },
+    {
+      "epoch": 13.32,
+      "learning_rate": 2.7819245357746736e-05,
+      "loss": 0.7481,
+      "step": 291720
+    },
+    {
+      "epoch": 13.32,
+      "learning_rate": 2.7818483708318737e-05,
+      "loss": 0.7881,
+      "step": 291730
+    },
+    {
+      "epoch": 13.32,
+      "learning_rate": 2.781772205889073e-05,
+      "loss": 0.7584,
+      "step": 291740
+    },
+    {
+      "epoch": 13.32,
+      "learning_rate": 2.7816960409462732e-05,
+      "loss": 0.8119,
+      "step": 291750
+    },
+    {
+      "epoch": 13.32,
+      "learning_rate": 2.7816198760034733e-05,
+      "loss": 0.7871,
+      "step": 291760
+    },
+    {
+      "epoch": 13.32,
+      "learning_rate": 2.781543711060673e-05,
+      "loss": 0.7894,
+      "step": 291770
+    },
+    {
+      "epoch": 13.32,
+      "learning_rate": 2.781467546117873e-05,
+      "loss": 0.8497,
+      "step": 291780
+    },
+    {
+      "epoch": 13.32,
+      "learning_rate": 2.781391381175073e-05,
+      "loss": 0.828,
+      "step": 291790
+    },
+    {
+      "epoch": 13.32,
+      "learning_rate": 2.7813152162322726e-05,
+      "loss": 0.8052,
+      "step": 291800
+    },
+    {
+      "epoch": 13.32,
+      "learning_rate": 2.7812390512894727e-05,
+      "loss": 0.9316,
+      "step": 291810
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 2.7811628863466728e-05,
+      "loss": 0.7507,
+      "step": 291820
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 2.7810867214038722e-05,
+      "loss": 0.785,
+      "step": 291830
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 2.7810105564610723e-05,
+      "loss": 0.7511,
+      "step": 291840
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 2.7809343915182724e-05,
+      "loss": 0.812,
+      "step": 291850
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 2.7808582265754718e-05,
+      "loss": 0.8657,
+      "step": 291860
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 2.780782061632672e-05,
+      "loss": 0.8041,
+      "step": 291870
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 2.780705896689872e-05,
+      "loss": 0.6954,
+      "step": 291880
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 2.7806297317470714e-05,
+      "loss": 0.7354,
+      "step": 291890
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 2.7805535668042714e-05,
+      "loss": 0.7945,
+      "step": 291900
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 2.7804774018614715e-05,
+      "loss": 0.831,
+      "step": 291910
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 2.780401236918671e-05,
+      "loss": 0.8517,
+      "step": 291920
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 2.780325071975871e-05,
+      "loss": 0.7697,
+      "step": 291930
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 2.780248907033071e-05,
+      "loss": 0.8725,
+      "step": 291940
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 2.7801727420902705e-05,
+      "loss": 0.9009,
+      "step": 291950
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 2.7800965771474706e-05,
+      "loss": 0.8438,
+      "step": 291960
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 2.7800204122046707e-05,
+      "loss": 0.8331,
+      "step": 291970
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 2.7799442472618704e-05,
+      "loss": 0.6946,
+      "step": 291980
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 2.7798680823190705e-05,
+      "loss": 0.7796,
+      "step": 291990
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 2.7797919173762703e-05,
+      "loss": 0.7107,
+      "step": 292000
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 2.77971575243347e-05,
+      "loss": 0.8274,
+      "step": 292010
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 2.77963958749067e-05,
+      "loss": 0.8019,
+      "step": 292020
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 2.7795634225478702e-05,
+      "loss": 0.76,
+      "step": 292030
+    },
+    {
+      "epoch": 13.34,
+      "learning_rate": 2.7794872576050696e-05,
+      "loss": 0.734,
+      "step": 292040
+    },
+    {
+      "epoch": 13.34,
+      "learning_rate": 2.7794110926622697e-05,
+      "loss": 0.7614,
+      "step": 292050
+    },
+    {
+      "epoch": 13.34,
+      "learning_rate": 2.7793349277194698e-05,
+      "loss": 0.7229,
+      "step": 292060
+    },
+    {
+      "epoch": 13.34,
+      "learning_rate": 2.779258762776669e-05,
+      "loss": 0.782,
+      "step": 292070
+    },
+    {
+      "epoch": 13.34,
+      "learning_rate": 2.7791825978338693e-05,
+      "loss": 0.7682,
+      "step": 292080
+    },
+    {
+      "epoch": 13.34,
+      "learning_rate": 2.7791064328910687e-05,
+      "loss": 0.8247,
+      "step": 292090
+    },
+    {
+      "epoch": 13.34,
+      "learning_rate": 2.7790302679482687e-05,
+      "loss": 0.7192,
+      "step": 292100
+    },
+    {
+      "epoch": 13.34,
+      "learning_rate": 2.7789541030054688e-05,
+      "loss": 0.8221,
+      "step": 292110
+    },
+    {
+      "epoch": 13.34,
+      "learning_rate": 2.7788779380626682e-05,
+      "loss": 0.9057,
+      "step": 292120
+    },
+    {
+      "epoch": 13.34,
+      "learning_rate": 2.7788017731198683e-05,
+      "loss": 0.7356,
+      "step": 292130
+    },
+    {
+      "epoch": 13.34,
+      "learning_rate": 2.7787256081770684e-05,
+      "loss": 0.7307,
+      "step": 292140
+    },
+    {
+      "epoch": 13.34,
+      "learning_rate": 2.778649443234268e-05,
+      "loss": 0.821,
+      "step": 292150
+    },
+    {
+      "epoch": 13.34,
+      "learning_rate": 2.7785732782914682e-05,
+      "loss": 0.7521,
+      "step": 292160
+    },
+    {
+      "epoch": 13.34,
+      "learning_rate": 2.778497113348668e-05,
+      "loss": 0.7413,
+      "step": 292170
+    },
+    {
+      "epoch": 13.34,
+      "learning_rate": 2.7784209484058677e-05,
+      "loss": 0.8198,
+      "step": 292180
+    },
+    {
+      "epoch": 13.34,
+      "learning_rate": 2.7783447834630678e-05,
+      "loss": 0.7741,
+      "step": 292190
+    },
+    {
+      "epoch": 13.34,
+      "learning_rate": 2.778268618520268e-05,
+      "loss": 0.8157,
+      "step": 292200
+    },
+    {
+      "epoch": 13.34,
+      "learning_rate": 2.7781924535774673e-05,
+      "loss": 0.8657,
+      "step": 292210
+    },
+    {
+      "epoch": 13.34,
+      "learning_rate": 2.7781162886346674e-05,
+      "loss": 0.8637,
+      "step": 292220
+    },
+    {
+      "epoch": 13.34,
+      "learning_rate": 2.7780401236918675e-05,
+      "loss": 0.7409,
+      "step": 292230
+    },
+    {
+      "epoch": 13.34,
+      "learning_rate": 2.777963958749067e-05,
+      "loss": 0.7615,
+      "step": 292240
+    },
+    {
+      "epoch": 13.34,
+      "learning_rate": 2.777887793806267e-05,
+      "loss": 0.7942,
+      "step": 292250
+    },
+    {
+      "epoch": 13.35,
+      "learning_rate": 2.777811628863467e-05,
+      "loss": 0.7748,
+      "step": 292260
+    },
+    {
+      "epoch": 13.35,
+      "learning_rate": 2.7777354639206665e-05,
+      "loss": 0.7227,
+      "step": 292270
+    },
+    {
+      "epoch": 13.35,
+      "learning_rate": 2.7776592989778666e-05,
+      "loss": 0.7862,
+      "step": 292280
+    },
+    {
+      "epoch": 13.35,
+      "learning_rate": 2.7775831340350666e-05,
+      "loss": 0.7277,
+      "step": 292290
+    },
+    {
+      "epoch": 13.35,
+      "learning_rate": 2.777506969092266e-05,
+      "loss": 0.757,
+      "step": 292300
+    },
+    {
+      "epoch": 13.35,
+      "learning_rate": 2.777430804149466e-05,
+      "loss": 0.7678,
+      "step": 292310
+    },
+    {
+      "epoch": 13.35,
+      "learning_rate": 2.7773546392066662e-05,
+      "loss": 0.7428,
+      "step": 292320
+    },
+    {
+      "epoch": 13.35,
+      "learning_rate": 2.7772784742638656e-05,
+      "loss": 0.9031,
+      "step": 292330
+    },
+    {
+      "epoch": 13.35,
+      "learning_rate": 2.7772023093210657e-05,
+      "loss": 0.7596,
+      "step": 292340
+    },
+    {
+      "epoch": 13.35,
+      "learning_rate": 2.7771261443782658e-05,
+      "loss": 0.7814,
+      "step": 292350
+    },
+    {
+      "epoch": 13.35,
+      "learning_rate": 2.7770499794354655e-05,
+      "loss": 0.7547,
+      "step": 292360
+    },
+    {
+      "epoch": 13.35,
+      "learning_rate": 2.7769738144926656e-05,
+      "loss": 0.7398,
+      "step": 292370
+    },
+    {
+      "epoch": 13.35,
+      "learning_rate": 2.7768976495498654e-05,
+      "loss": 0.8987,
+      "step": 292380
+    },
+    {
+      "epoch": 13.35,
+      "learning_rate": 2.776821484607065e-05,
+      "loss": 0.8064,
+      "step": 292390
+    },
+    {
+      "epoch": 13.35,
+      "learning_rate": 2.7767453196642652e-05,
+      "loss": 0.7838,
+      "step": 292400
+    },
+    {
+      "epoch": 13.35,
+      "learning_rate": 2.7766691547214653e-05,
+      "loss": 0.8979,
+      "step": 292410
+    },
+    {
+      "epoch": 13.35,
+      "learning_rate": 2.7765929897786647e-05,
+      "loss": 0.7691,
+      "step": 292420
+    },
+    {
+      "epoch": 13.35,
+      "learning_rate": 2.7765168248358648e-05,
+      "loss": 0.7809,
+      "step": 292430
+    },
+    {
+      "epoch": 13.35,
+      "learning_rate": 2.776440659893065e-05,
+      "loss": 0.8199,
+      "step": 292440
+    },
+    {
+      "epoch": 13.35,
+      "learning_rate": 2.7763644949502643e-05,
+      "loss": 0.7263,
+      "step": 292450
+    },
+    {
+      "epoch": 13.35,
+      "learning_rate": 2.7762883300074644e-05,
+      "loss": 0.8006,
+      "step": 292460
+    },
+    {
+      "epoch": 13.35,
+      "learning_rate": 2.7762121650646645e-05,
+      "loss": 0.7612,
+      "step": 292470
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 2.776136000121864e-05,
+      "loss": 0.7584,
+      "step": 292480
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 2.776059835179064e-05,
+      "loss": 0.7233,
+      "step": 292490
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 2.775983670236264e-05,
+      "loss": 0.8528,
+      "step": 292500
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 2.7759075052934634e-05,
+      "loss": 0.8468,
+      "step": 292510
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 2.7758313403506635e-05,
+      "loss": 0.7612,
+      "step": 292520
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 2.7757551754078636e-05,
+      "loss": 0.9139,
+      "step": 292530
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 2.775679010465063e-05,
+      "loss": 0.822,
+      "step": 292540
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 2.775602845522263e-05,
+      "loss": 0.7492,
+      "step": 292550
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 2.775526680579463e-05,
+      "loss": 0.8441,
+      "step": 292560
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 2.775450515636663e-05,
+      "loss": 0.8352,
+      "step": 292570
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 2.775374350693863e-05,
+      "loss": 0.8338,
+      "step": 292580
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 2.7752981857510624e-05,
+      "loss": 0.864,
+      "step": 292590
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 2.7752220208082625e-05,
+      "loss": 0.8696,
+      "step": 292600
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 2.7751458558654626e-05,
+      "loss": 0.8045,
+      "step": 292610
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 2.775069690922662e-05,
+      "loss": 0.7013,
+      "step": 292620
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 2.774993525979862e-05,
+      "loss": 0.6897,
+      "step": 292630
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 2.7749173610370622e-05,
+      "loss": 0.6704,
+      "step": 292640
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 2.7748411960942616e-05,
+      "loss": 0.826,
+      "step": 292650
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 2.7747650311514617e-05,
+      "loss": 0.857,
+      "step": 292660
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 2.7746888662086618e-05,
+      "loss": 0.824,
+      "step": 292670
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 2.774612701265861e-05,
+      "loss": 0.7363,
+      "step": 292680
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 2.7745365363230613e-05,
+      "loss": 0.8244,
+      "step": 292690
+    },
+    {
+      "epoch": 13.37,
+      "learning_rate": 2.7744603713802613e-05,
+      "loss": 0.7785,
+      "step": 292700
+    },
+    {
+      "epoch": 13.37,
+      "learning_rate": 2.7743842064374607e-05,
+      "loss": 0.7452,
+      "step": 292710
+    },
+    {
+      "epoch": 13.37,
+      "learning_rate": 2.774308041494661e-05,
+      "loss": 0.9796,
+      "step": 292720
+    },
+    {
+      "epoch": 13.37,
+      "learning_rate": 2.774231876551861e-05,
+      "loss": 0.7806,
+      "step": 292730
+    },
+    {
+      "epoch": 13.37,
+      "learning_rate": 2.7741557116090607e-05,
+      "loss": 0.794,
+      "step": 292740
+    },
+    {
+      "epoch": 13.37,
+      "learning_rate": 2.7740795466662604e-05,
+      "loss": 0.7573,
+      "step": 292750
+    },
+    {
+      "epoch": 13.37,
+      "learning_rate": 2.7740033817234605e-05,
+      "loss": 0.8398,
+      "step": 292760
+    },
+    {
+      "epoch": 13.37,
+      "learning_rate": 2.7739272167806602e-05,
+      "loss": 0.8228,
+      "step": 292770
+    },
+    {
+      "epoch": 13.37,
+      "learning_rate": 2.7738510518378603e-05,
+      "loss": 0.7233,
+      "step": 292780
+    },
+    {
+      "epoch": 13.37,
+      "learning_rate": 2.7737748868950604e-05,
+      "loss": 0.8397,
+      "step": 292790
+    },
+    {
+      "epoch": 13.37,
+      "learning_rate": 2.7736987219522598e-05,
+      "loss": 0.7969,
+      "step": 292800
+    },
+    {
+      "epoch": 13.37,
+      "learning_rate": 2.77362255700946e-05,
+      "loss": 0.8671,
+      "step": 292810
+    },
+    {
+      "epoch": 13.37,
+      "learning_rate": 2.77354639206666e-05,
+      "loss": 0.789,
+      "step": 292820
+    },
+    {
+      "epoch": 13.37,
+      "learning_rate": 2.7734702271238594e-05,
+      "loss": 0.8427,
+      "step": 292830
+    },
+    {
+      "epoch": 13.37,
+      "learning_rate": 2.7733940621810595e-05,
+      "loss": 0.7544,
+      "step": 292840
+    },
+    {
+      "epoch": 13.37,
+      "learning_rate": 2.7733178972382596e-05,
+      "loss": 0.8085,
+      "step": 292850
+    },
+    {
+      "epoch": 13.37,
+      "learning_rate": 2.773241732295459e-05,
+      "loss": 0.7294,
+      "step": 292860
+    },
+    {
+      "epoch": 13.37,
+      "learning_rate": 2.773165567352659e-05,
+      "loss": 0.7943,
+      "step": 292870
+    },
+    {
+      "epoch": 13.37,
+      "learning_rate": 2.773089402409859e-05,
+      "loss": 0.735,
+      "step": 292880
+    },
+    {
+      "epoch": 13.37,
+      "learning_rate": 2.7730132374670586e-05,
+      "loss": 0.7766,
+      "step": 292890
+    },
+    {
+      "epoch": 13.37,
+      "learning_rate": 2.7729370725242586e-05,
+      "loss": 0.7813,
+      "step": 292900
+    },
+    {
+      "epoch": 13.38,
+      "learning_rate": 2.7728609075814587e-05,
+      "loss": 0.8172,
+      "step": 292910
+    },
+    {
+      "epoch": 13.38,
+      "learning_rate": 2.772784742638658e-05,
+      "loss": 0.7694,
+      "step": 292920
+    },
+    {
+      "epoch": 13.38,
+      "learning_rate": 2.7727085776958582e-05,
+      "loss": 0.8259,
+      "step": 292930
+    },
+    {
+      "epoch": 13.38,
+      "learning_rate": 2.7726324127530583e-05,
+      "loss": 0.8894,
+      "step": 292940
+    },
+    {
+      "epoch": 13.38,
+      "learning_rate": 2.772556247810258e-05,
+      "loss": 0.7573,
+      "step": 292950
+    },
+    {
+      "epoch": 13.38,
+      "learning_rate": 2.772480082867458e-05,
+      "loss": 0.7699,
+      "step": 292960
+    },
+    {
+      "epoch": 13.38,
+      "learning_rate": 2.772403917924658e-05,
+      "loss": 0.8247,
+      "step": 292970
+    },
+    {
+      "epoch": 13.38,
+      "learning_rate": 2.7723277529818576e-05,
+      "loss": 0.7448,
+      "step": 292980
+    },
+    {
+      "epoch": 13.38,
+      "learning_rate": 2.7722515880390577e-05,
+      "loss": 0.7511,
+      "step": 292990
+    },
+    {
+      "epoch": 13.38,
+      "learning_rate": 2.7721754230962578e-05,
+      "loss": 0.8125,
+      "step": 293000
+    },
+    {
+      "epoch": 13.38,
+      "learning_rate": 2.7720992581534572e-05,
+      "loss": 0.8221,
+      "step": 293010
+    },
+    {
+      "epoch": 13.38,
+      "learning_rate": 2.7720230932106573e-05,
+      "loss": 0.7811,
+      "step": 293020
+    },
+    {
+      "epoch": 13.38,
+      "learning_rate": 2.7719469282678567e-05,
+      "loss": 0.7089,
+      "step": 293030
+    },
+    {
+      "epoch": 13.38,
+      "learning_rate": 2.7718707633250568e-05,
+      "loss": 0.8068,
+      "step": 293040
+    },
+    {
+      "epoch": 13.38,
+      "learning_rate": 2.771794598382257e-05,
+      "loss": 0.7822,
+      "step": 293050
+    },
+    {
+      "epoch": 13.38,
+      "learning_rate": 2.7717184334394563e-05,
+      "loss": 0.8378,
+      "step": 293060
+    },
+    {
+      "epoch": 13.38,
+      "learning_rate": 2.7716422684966564e-05,
+      "loss": 0.8033,
+      "step": 293070
+    },
+    {
+      "epoch": 13.38,
+      "learning_rate": 2.7715661035538565e-05,
+      "loss": 0.8166,
+      "step": 293080
+    },
+    {
+      "epoch": 13.38,
+      "learning_rate": 2.771489938611056e-05,
+      "loss": 0.7997,
+      "step": 293090
+    },
+    {
+      "epoch": 13.38,
+      "learning_rate": 2.771413773668256e-05,
+      "loss": 0.7458,
+      "step": 293100
+    },
+    {
+      "epoch": 13.38,
+      "learning_rate": 2.771337608725456e-05,
+      "loss": 0.7976,
+      "step": 293110
+    },
+    {
+      "epoch": 13.38,
+      "learning_rate": 2.7712614437826558e-05,
+      "loss": 0.8376,
+      "step": 293120
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 2.7711852788398555e-05,
+      "loss": 0.8086,
+      "step": 293130
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 2.7711091138970556e-05,
+      "loss": 0.7878,
+      "step": 293140
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 2.7710329489542554e-05,
+      "loss": 0.7866,
+      "step": 293150
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 2.7709567840114554e-05,
+      "loss": 0.7174,
+      "step": 293160
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 2.7708806190686555e-05,
+      "loss": 0.697,
+      "step": 293170
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 2.770804454125855e-05,
+      "loss": 0.6868,
+      "step": 293180
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 2.770728289183055e-05,
+      "loss": 0.7452,
+      "step": 293190
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 2.770652124240255e-05,
+      "loss": 0.6764,
+      "step": 293200
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 2.7705759592974545e-05,
+      "loss": 0.8145,
+      "step": 293210
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 2.7704997943546546e-05,
+      "loss": 0.8182,
+      "step": 293220
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 2.7704236294118547e-05,
+      "loss": 0.7615,
+      "step": 293230
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 2.770347464469054e-05,
+      "loss": 0.8407,
+      "step": 293240
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 2.7702712995262542e-05,
+      "loss": 0.7596,
+      "step": 293250
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 2.7701951345834543e-05,
+      "loss": 0.7247,
+      "step": 293260
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 2.7701189696406537e-05,
+      "loss": 0.7146,
+      "step": 293270
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 2.7700428046978538e-05,
+      "loss": 0.8229,
+      "step": 293280
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 2.769966639755054e-05,
+      "loss": 0.74,
+      "step": 293290
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 2.7698904748122533e-05,
+      "loss": 0.832,
+      "step": 293300
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 2.7698143098694533e-05,
+      "loss": 0.8409,
+      "step": 293310
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 2.7697381449266534e-05,
+      "loss": 0.8295,
+      "step": 293320
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 2.7696619799838532e-05,
+      "loss": 0.6971,
+      "step": 293330
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 2.769585815041053e-05,
+      "loss": 0.8259,
+      "step": 293340
+    },
+    {
+      "epoch": 13.4,
+      "learning_rate": 2.769509650098253e-05,
+      "loss": 0.8076,
+      "step": 293350
+    },
+    {
+      "epoch": 13.4,
+      "learning_rate": 2.7694334851554527e-05,
+      "loss": 0.8113,
+      "step": 293360
+    },
+    {
+      "epoch": 13.4,
+      "learning_rate": 2.769357320212653e-05,
+      "loss": 0.767,
+      "step": 293370
+    },
+    {
+      "epoch": 13.4,
+      "learning_rate": 2.769281155269853e-05,
+      "loss": 0.8112,
+      "step": 293380
+    },
+    {
+      "epoch": 13.4,
+      "learning_rate": 2.7692049903270523e-05,
+      "loss": 0.7201,
+      "step": 293390
+    },
+    {
+      "epoch": 13.4,
+      "learning_rate": 2.7691288253842524e-05,
+      "loss": 0.8072,
+      "step": 293400
+    },
+    {
+      "epoch": 13.4,
+      "learning_rate": 2.7690526604414525e-05,
+      "loss": 0.8543,
+      "step": 293410
+    },
+    {
+      "epoch": 13.4,
+      "learning_rate": 2.768976495498652e-05,
+      "loss": 0.8113,
+      "step": 293420
+    },
+    {
+      "epoch": 13.4,
+      "learning_rate": 2.768900330555852e-05,
+      "loss": 0.7082,
+      "step": 293430
+    },
+    {
+      "epoch": 13.4,
+      "learning_rate": 2.768824165613052e-05,
+      "loss": 0.7553,
+      "step": 293440
+    },
+    {
+      "epoch": 13.4,
+      "learning_rate": 2.7687480006702515e-05,
+      "loss": 0.9053,
+      "step": 293450
+    },
+    {
+      "epoch": 13.4,
+      "learning_rate": 2.7686718357274516e-05,
+      "loss": 0.8676,
+      "step": 293460
+    },
+    {
+      "epoch": 13.4,
+      "learning_rate": 2.768595670784651e-05,
+      "loss": 0.7477,
+      "step": 293470
+    },
+    {
+      "epoch": 13.4,
+      "learning_rate": 2.768519505841851e-05,
+      "loss": 0.8336,
+      "step": 293480
+    },
+    {
+      "epoch": 13.4,
+      "learning_rate": 2.768443340899051e-05,
+      "loss": 0.746,
+      "step": 293490
+    },
+    {
+      "epoch": 13.4,
+      "learning_rate": 2.7683671759562506e-05,
+      "loss": 0.8137,
+      "step": 293500
+    },
+    {
+      "epoch": 13.4,
+      "learning_rate": 2.7682910110134506e-05,
+      "loss": 0.7055,
+      "step": 293510
+    },
+    {
+      "epoch": 13.4,
+      "learning_rate": 2.7682148460706507e-05,
+      "loss": 0.7805,
+      "step": 293520
+    },
+    {
+      "epoch": 13.4,
+      "learning_rate": 2.7681386811278505e-05,
+      "loss": 0.7889,
+      "step": 293530
+    },
+    {
+      "epoch": 13.4,
+      "learning_rate": 2.7680625161850506e-05,
+      "loss": 0.7321,
+      "step": 293540
+    },
+    {
+      "epoch": 13.4,
+      "learning_rate": 2.7679863512422503e-05,
+      "loss": 0.7934,
+      "step": 293550
+    },
+    {
+      "epoch": 13.4,
+      "learning_rate": 2.76791018629945e-05,
+      "loss": 0.7889,
+      "step": 293560
+    },
+    {
+      "epoch": 13.41,
+      "learning_rate": 2.76783402135665e-05,
+      "loss": 0.7638,
+      "step": 293570
+    },
+    {
+      "epoch": 13.41,
+      "learning_rate": 2.7677578564138502e-05,
+      "loss": 0.7481,
+      "step": 293580
+    },
+    {
+      "epoch": 13.41,
+      "learning_rate": 2.7676816914710496e-05,
+      "loss": 0.7885,
+      "step": 293590
+    },
+    {
+      "epoch": 13.41,
+      "learning_rate": 2.7676055265282497e-05,
+      "loss": 0.7426,
+      "step": 293600
+    },
+    {
+      "epoch": 13.41,
+      "learning_rate": 2.7675293615854498e-05,
+      "loss": 0.704,
+      "step": 293610
+    },
+    {
+      "epoch": 13.41,
+      "learning_rate": 2.7674531966426492e-05,
+      "loss": 0.7787,
+      "step": 293620
+    },
+    {
+      "epoch": 13.41,
+      "learning_rate": 2.7673770316998493e-05,
+      "loss": 0.8495,
+      "step": 293630
+    },
+    {
+      "epoch": 13.41,
+      "learning_rate": 2.7673008667570494e-05,
+      "loss": 0.8388,
+      "step": 293640
+    },
+    {
+      "epoch": 13.41,
+      "learning_rate": 2.7672247018142488e-05,
+      "loss": 0.8254,
+      "step": 293650
+    },
+    {
+      "epoch": 13.41,
+      "learning_rate": 2.767148536871449e-05,
+      "loss": 0.7465,
+      "step": 293660
+    },
+    {
+      "epoch": 13.41,
+      "learning_rate": 2.767072371928649e-05,
+      "loss": 0.8505,
+      "step": 293670
+    },
+    {
+      "epoch": 13.41,
+      "learning_rate": 2.7669962069858484e-05,
+      "loss": 0.7403,
+      "step": 293680
+    },
+    {
+      "epoch": 13.41,
+      "learning_rate": 2.7669200420430485e-05,
+      "loss": 0.8714,
+      "step": 293690
+    },
+    {
+      "epoch": 13.41,
+      "learning_rate": 2.7668438771002485e-05,
+      "loss": 0.8439,
+      "step": 293700
+    },
+    {
+      "epoch": 13.41,
+      "learning_rate": 2.7667677121574483e-05,
+      "loss": 0.8338,
+      "step": 293710
+    },
+    {
+      "epoch": 13.41,
+      "learning_rate": 2.766691547214648e-05,
+      "loss": 0.8394,
+      "step": 293720
+    },
+    {
+      "epoch": 13.41,
+      "learning_rate": 2.766615382271848e-05,
+      "loss": 0.8336,
+      "step": 293730
+    },
+    {
+      "epoch": 13.41,
+      "learning_rate": 2.766539217329048e-05,
+      "loss": 0.7803,
+      "step": 293740
+    },
+    {
+      "epoch": 13.41,
+      "learning_rate": 2.766463052386248e-05,
+      "loss": 0.7632,
+      "step": 293750
+    },
+    {
+      "epoch": 13.41,
+      "learning_rate": 2.766386887443448e-05,
+      "loss": 0.7676,
+      "step": 293760
+    },
+    {
+      "epoch": 13.41,
+      "learning_rate": 2.7663107225006474e-05,
+      "loss": 0.8007,
+      "step": 293770
+    },
+    {
+      "epoch": 13.41,
+      "learning_rate": 2.7662345575578475e-05,
+      "loss": 0.7578,
+      "step": 293780
+    },
+    {
+      "epoch": 13.42,
+      "learning_rate": 2.7661583926150476e-05,
+      "loss": 0.8301,
+      "step": 293790
+    },
+    {
+      "epoch": 13.42,
+      "learning_rate": 2.766082227672247e-05,
+      "loss": 0.6885,
+      "step": 293800
+    },
+    {
+      "epoch": 13.42,
+      "learning_rate": 2.766006062729447e-05,
+      "loss": 0.7637,
+      "step": 293810
+    },
+    {
+      "epoch": 13.42,
+      "learning_rate": 2.7659298977866472e-05,
+      "loss": 0.7429,
+      "step": 293820
+    },
+    {
+      "epoch": 13.42,
+      "learning_rate": 2.7658537328438466e-05,
+      "loss": 0.8136,
+      "step": 293830
+    },
+    {
+      "epoch": 13.42,
+      "learning_rate": 2.7657775679010467e-05,
+      "loss": 0.7421,
+      "step": 293840
+    },
+    {
+      "epoch": 13.42,
+      "learning_rate": 2.7657014029582468e-05,
+      "loss": 0.7486,
+      "step": 293850
+    },
+    {
+      "epoch": 13.42,
+      "learning_rate": 2.7656252380154462e-05,
+      "loss": 0.7662,
+      "step": 293860
+    },
+    {
+      "epoch": 13.42,
+      "learning_rate": 2.7655490730726463e-05,
+      "loss": 0.7165,
+      "step": 293870
+    },
+    {
+      "epoch": 13.42,
+      "learning_rate": 2.7654729081298464e-05,
+      "loss": 0.8769,
+      "step": 293880
+    },
+    {
+      "epoch": 13.42,
+      "learning_rate": 2.7653967431870458e-05,
+      "loss": 0.7968,
+      "step": 293890
+    },
+    {
+      "epoch": 13.42,
+      "learning_rate": 2.765320578244246e-05,
+      "loss": 0.7746,
+      "step": 293900
+    },
+    {
+      "epoch": 13.42,
+      "learning_rate": 2.765244413301446e-05,
+      "loss": 0.7586,
+      "step": 293910
+    },
+    {
+      "epoch": 13.42,
+      "learning_rate": 2.7651682483586457e-05,
+      "loss": 0.8137,
+      "step": 293920
+    },
+    {
+      "epoch": 13.42,
+      "learning_rate": 2.7650920834158454e-05,
+      "loss": 0.8976,
+      "step": 293930
+    },
+    {
+      "epoch": 13.42,
+      "learning_rate": 2.7650159184730452e-05,
+      "loss": 0.8262,
+      "step": 293940
+    },
+    {
+      "epoch": 13.42,
+      "learning_rate": 2.7649397535302453e-05,
+      "loss": 0.8729,
+      "step": 293950
+    },
+    {
+      "epoch": 13.42,
+      "learning_rate": 2.7648635885874453e-05,
+      "loss": 0.778,
+      "step": 293960
+    },
+    {
+      "epoch": 13.42,
+      "learning_rate": 2.7647874236446448e-05,
+      "loss": 0.8725,
+      "step": 293970
+    },
+    {
+      "epoch": 13.42,
+      "learning_rate": 2.764711258701845e-05,
+      "loss": 0.6768,
+      "step": 293980
+    },
+    {
+      "epoch": 13.42,
+      "learning_rate": 2.764635093759045e-05,
+      "loss": 0.7377,
+      "step": 293990
+    },
+    {
+      "epoch": 13.42,
+      "learning_rate": 2.7645589288162443e-05,
+      "loss": 0.8084,
+      "step": 294000
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 2.7644827638734444e-05,
+      "loss": 0.7714,
+      "step": 294010
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 2.7644065989306445e-05,
+      "loss": 0.8086,
+      "step": 294020
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 2.764330433987844e-05,
+      "loss": 0.7323,
+      "step": 294030
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 2.764254269045044e-05,
+      "loss": 0.7736,
+      "step": 294040
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 2.764178104102244e-05,
+      "loss": 0.8135,
+      "step": 294050
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 2.7641019391594435e-05,
+      "loss": 0.642,
+      "step": 294060
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 2.7640257742166436e-05,
+      "loss": 0.7711,
+      "step": 294070
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 2.7639496092738437e-05,
+      "loss": 0.8891,
+      "step": 294080
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 2.763873444331043e-05,
+      "loss": 0.8792,
+      "step": 294090
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 2.763797279388243e-05,
+      "loss": 0.7585,
+      "step": 294100
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 2.7637211144454432e-05,
+      "loss": 0.7489,
+      "step": 294110
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 2.763644949502643e-05,
+      "loss": 0.8139,
+      "step": 294120
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 2.763568784559843e-05,
+      "loss": 0.7696,
+      "step": 294130
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 2.7634926196170428e-05,
+      "loss": 0.8102,
+      "step": 294140
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 2.7634164546742426e-05,
+      "loss": 0.8465,
+      "step": 294150
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 2.7633402897314426e-05,
+      "loss": 0.8674,
+      "step": 294160
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 2.7632641247886427e-05,
+      "loss": 0.7147,
+      "step": 294170
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 2.763187959845842e-05,
+      "loss": 0.8126,
+      "step": 294180
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 2.7631117949030422e-05,
+      "loss": 0.7844,
+      "step": 294190
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 2.7630356299602423e-05,
+      "loss": 0.7163,
+      "step": 294200
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 2.7629594650174417e-05,
+      "loss": 0.7799,
+      "step": 294210
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 2.7628833000746418e-05,
+      "loss": 0.8219,
+      "step": 294220
+    },
+    {
+      "epoch": 13.44,
+      "learning_rate": 2.762807135131842e-05,
+      "loss": 0.7595,
+      "step": 294230
+    },
+    {
+      "epoch": 13.44,
+      "learning_rate": 2.7627309701890413e-05,
+      "loss": 0.7858,
+      "step": 294240
+    },
+    {
+      "epoch": 13.44,
+      "learning_rate": 2.7626548052462414e-05,
+      "loss": 0.8738,
+      "step": 294250
+    },
+    {
+      "epoch": 13.44,
+      "learning_rate": 2.7625786403034415e-05,
+      "loss": 0.8054,
+      "step": 294260
+    },
+    {
+      "epoch": 13.44,
+      "learning_rate": 2.762502475360641e-05,
+      "loss": 0.8027,
+      "step": 294270
+    },
+    {
+      "epoch": 13.44,
+      "learning_rate": 2.762426310417841e-05,
+      "loss": 0.7944,
+      "step": 294280
+    },
+    {
+      "epoch": 13.44,
+      "learning_rate": 2.762350145475041e-05,
+      "loss": 0.7455,
+      "step": 294290
+    },
+    {
+      "epoch": 13.44,
+      "learning_rate": 2.7622739805322405e-05,
+      "loss": 0.8304,
+      "step": 294300
+    },
+    {
+      "epoch": 13.44,
+      "learning_rate": 2.7621978155894405e-05,
+      "loss": 0.7503,
+      "step": 294310
+    },
+    {
+      "epoch": 13.44,
+      "learning_rate": 2.7621216506466406e-05,
+      "loss": 0.7067,
+      "step": 294320
+    },
+    {
+      "epoch": 13.44,
+      "learning_rate": 2.7620454857038404e-05,
+      "loss": 0.7933,
+      "step": 294330
+    },
+    {
+      "epoch": 13.44,
+      "learning_rate": 2.7619693207610405e-05,
+      "loss": 0.6951,
+      "step": 294340
+    },
+    {
+      "epoch": 13.44,
+      "learning_rate": 2.7618931558182402e-05,
+      "loss": 0.7855,
+      "step": 294350
+    },
+    {
+      "epoch": 13.44,
+      "learning_rate": 2.76181699087544e-05,
+      "loss": 0.7902,
+      "step": 294360
+    },
+    {
+      "epoch": 13.44,
+      "learning_rate": 2.76174082593264e-05,
+      "loss": 0.7943,
+      "step": 294370
+    },
+    {
+      "epoch": 13.44,
+      "learning_rate": 2.76166466098984e-05,
+      "loss": 0.7583,
+      "step": 294380
+    },
+    {
+      "epoch": 13.44,
+      "learning_rate": 2.7615884960470395e-05,
+      "loss": 0.6806,
+      "step": 294390
+    },
+    {
+      "epoch": 13.44,
+      "learning_rate": 2.7615123311042396e-05,
+      "loss": 0.8304,
+      "step": 294400
+    },
+    {
+      "epoch": 13.44,
+      "learning_rate": 2.761436166161439e-05,
+      "loss": 0.872,
+      "step": 294410
+    },
+    {
+      "epoch": 13.44,
+      "learning_rate": 2.761360001218639e-05,
+      "loss": 0.74,
+      "step": 294420
+    },
+    {
+      "epoch": 13.44,
+      "learning_rate": 2.7612838362758392e-05,
+      "loss": 0.7849,
+      "step": 294430
+    },
+    {
+      "epoch": 13.44,
+      "learning_rate": 2.7612076713330386e-05,
+      "loss": 0.7678,
+      "step": 294440
+    },
+    {
+      "epoch": 13.45,
+      "learning_rate": 2.7611315063902387e-05,
+      "loss": 0.7161,
+      "step": 294450
+    },
+    {
+      "epoch": 13.45,
+      "learning_rate": 2.7610553414474388e-05,
+      "loss": 0.7803,
+      "step": 294460
+    },
+    {
+      "epoch": 13.45,
+      "learning_rate": 2.7609791765046382e-05,
+      "loss": 0.8407,
+      "step": 294470
+    },
+    {
+      "epoch": 13.45,
+      "learning_rate": 2.7609030115618383e-05,
+      "loss": 0.8504,
+      "step": 294480
+    },
+    {
+      "epoch": 13.45,
+      "learning_rate": 2.7608268466190384e-05,
+      "loss": 0.8307,
+      "step": 294490
+    },
+    {
+      "epoch": 13.45,
+      "learning_rate": 2.760750681676238e-05,
+      "loss": 0.7722,
+      "step": 294500
+    },
+    {
+      "epoch": 13.45,
+      "learning_rate": 2.7606745167334382e-05,
+      "loss": 0.7718,
+      "step": 294510
+    },
+    {
+      "epoch": 13.45,
+      "learning_rate": 2.760598351790638e-05,
+      "loss": 0.7358,
+      "step": 294520
+    },
+    {
+      "epoch": 13.45,
+      "learning_rate": 2.7605221868478377e-05,
+      "loss": 0.8494,
+      "step": 294530
+    },
+    {
+      "epoch": 13.45,
+      "learning_rate": 2.7604460219050378e-05,
+      "loss": 0.8483,
+      "step": 294540
+    },
+    {
+      "epoch": 13.45,
+      "learning_rate": 2.760369856962238e-05,
+      "loss": 0.7891,
+      "step": 294550
+    },
+    {
+      "epoch": 13.45,
+      "learning_rate": 2.7602936920194373e-05,
+      "loss": 0.8701,
+      "step": 294560
+    },
+    {
+      "epoch": 13.45,
+      "learning_rate": 2.7602175270766373e-05,
+      "loss": 0.8404,
+      "step": 294570
+    },
+    {
+      "epoch": 13.45,
+      "learning_rate": 2.7601413621338374e-05,
+      "loss": 0.8418,
+      "step": 294580
+    },
+    {
+      "epoch": 13.45,
+      "learning_rate": 2.760065197191037e-05,
+      "loss": 0.7693,
+      "step": 294590
+    },
+    {
+      "epoch": 13.45,
+      "learning_rate": 2.759989032248237e-05,
+      "loss": 0.7684,
+      "step": 294600
+    },
+    {
+      "epoch": 13.45,
+      "learning_rate": 2.759912867305437e-05,
+      "loss": 0.8398,
+      "step": 294610
+    },
+    {
+      "epoch": 13.45,
+      "learning_rate": 2.7598367023626364e-05,
+      "loss": 0.8296,
+      "step": 294620
+    },
+    {
+      "epoch": 13.45,
+      "learning_rate": 2.7597605374198365e-05,
+      "loss": 0.8334,
+      "step": 294630
+    },
+    {
+      "epoch": 13.45,
+      "learning_rate": 2.7596843724770366e-05,
+      "loss": 0.7351,
+      "step": 294640
+    },
+    {
+      "epoch": 13.45,
+      "learning_rate": 2.759608207534236e-05,
+      "loss": 0.7853,
+      "step": 294650
+    },
+    {
+      "epoch": 13.45,
+      "learning_rate": 2.759532042591436e-05,
+      "loss": 0.7381,
+      "step": 294660
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 2.759455877648636e-05,
+      "loss": 0.7906,
+      "step": 294670
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 2.7593797127058356e-05,
+      "loss": 0.7562,
+      "step": 294680
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 2.7593035477630357e-05,
+      "loss": 0.8143,
+      "step": 294690
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 2.7592273828202357e-05,
+      "loss": 0.8144,
+      "step": 294700
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 2.7591512178774355e-05,
+      "loss": 0.8199,
+      "step": 294710
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 2.7590750529346356e-05,
+      "loss": 0.8449,
+      "step": 294720
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 2.7589988879918353e-05,
+      "loss": 0.8062,
+      "step": 294730
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 2.758922723049035e-05,
+      "loss": 0.8279,
+      "step": 294740
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 2.758846558106235e-05,
+      "loss": 0.8729,
+      "step": 294750
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 2.7587703931634352e-05,
+      "loss": 0.7777,
+      "step": 294760
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 2.7586942282206346e-05,
+      "loss": 0.6965,
+      "step": 294770
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 2.7586180632778347e-05,
+      "loss": 0.8205,
+      "step": 294780
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 2.7585418983350348e-05,
+      "loss": 0.727,
+      "step": 294790
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 2.7584657333922342e-05,
+      "loss": 0.8319,
+      "step": 294800
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 2.7583895684494343e-05,
+      "loss": 0.7494,
+      "step": 294810
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 2.7583134035066344e-05,
+      "loss": 0.7892,
+      "step": 294820
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 2.7582372385638338e-05,
+      "loss": 0.7938,
+      "step": 294830
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 2.758161073621034e-05,
+      "loss": 0.7636,
+      "step": 294840
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 2.758084908678234e-05,
+      "loss": 0.835,
+      "step": 294850
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 2.7580087437354334e-05,
+      "loss": 0.8518,
+      "step": 294860
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 2.7579325787926335e-05,
+      "loss": 0.7365,
+      "step": 294870
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 2.7578564138498332e-05,
+      "loss": 0.7356,
+      "step": 294880
+    },
+    {
+      "epoch": 13.47,
+      "learning_rate": 2.757780248907033e-05,
+      "loss": 0.8438,
+      "step": 294890
+    },
+    {
+      "epoch": 13.47,
+      "learning_rate": 2.757704083964233e-05,
+      "loss": 0.8283,
+      "step": 294900
+    },
+    {
+      "epoch": 13.47,
+      "learning_rate": 2.7576279190214328e-05,
+      "loss": 0.7372,
+      "step": 294910
+    },
+    {
+      "epoch": 13.47,
+      "learning_rate": 2.757551754078633e-05,
+      "loss": 0.74,
+      "step": 294920
+    },
+    {
+      "epoch": 13.47,
+      "learning_rate": 2.757475589135833e-05,
+      "loss": 0.6931,
+      "step": 294930
+    },
+    {
+      "epoch": 13.47,
+      "learning_rate": 2.7573994241930324e-05,
+      "loss": 0.7414,
+      "step": 294940
+    },
+    {
+      "epoch": 13.47,
+      "learning_rate": 2.7573232592502325e-05,
+      "loss": 0.755,
+      "step": 294950
+    },
+    {
+      "epoch": 13.47,
+      "learning_rate": 2.7572470943074325e-05,
+      "loss": 0.7458,
+      "step": 294960
+    },
+    {
+      "epoch": 13.47,
+      "learning_rate": 2.757170929364632e-05,
+      "loss": 0.778,
+      "step": 294970
+    },
+    {
+      "epoch": 13.47,
+      "learning_rate": 2.757094764421832e-05,
+      "loss": 0.8746,
+      "step": 294980
+    },
+    {
+      "epoch": 13.47,
+      "learning_rate": 2.757018599479032e-05,
+      "loss": 0.8411,
+      "step": 294990
+    },
+    {
+      "epoch": 13.47,
+      "learning_rate": 2.7569424345362315e-05,
+      "loss": 0.7487,
+      "step": 295000
+    },
+    {
+      "epoch": 13.47,
+      "learning_rate": 2.7568662695934316e-05,
+      "loss": 0.8305,
+      "step": 295010
+    },
+    {
+      "epoch": 13.47,
+      "learning_rate": 2.7567901046506317e-05,
+      "loss": 0.8699,
+      "step": 295020
+    },
+    {
+      "epoch": 13.47,
+      "learning_rate": 2.756713939707831e-05,
+      "loss": 0.8181,
+      "step": 295030
+    },
+    {
+      "epoch": 13.47,
+      "learning_rate": 2.7566377747650312e-05,
+      "loss": 0.7952,
+      "step": 295040
+    },
+    {
+      "epoch": 13.47,
+      "learning_rate": 2.7565616098222313e-05,
+      "loss": 0.8349,
+      "step": 295050
+    },
+    {
+      "epoch": 13.47,
+      "learning_rate": 2.7564854448794307e-05,
+      "loss": 0.7343,
+      "step": 295060
+    },
+    {
+      "epoch": 13.47,
+      "learning_rate": 2.7564092799366308e-05,
+      "loss": 0.8302,
+      "step": 295070
+    },
+    {
+      "epoch": 13.47,
+      "learning_rate": 2.756333114993831e-05,
+      "loss": 0.7382,
+      "step": 295080
+    },
+    {
+      "epoch": 13.47,
+      "learning_rate": 2.7562569500510306e-05,
+      "loss": 0.7031,
+      "step": 295090
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 2.7561807851082304e-05,
+      "loss": 0.7297,
+      "step": 295100
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 2.7561046201654304e-05,
+      "loss": 0.8019,
+      "step": 295110
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 2.7560284552226302e-05,
+      "loss": 0.8083,
+      "step": 295120
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 2.7559522902798303e-05,
+      "loss": 0.808,
+      "step": 295130
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 2.7558761253370304e-05,
+      "loss": 0.7271,
+      "step": 295140
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 2.7557999603942298e-05,
+      "loss": 0.7744,
+      "step": 295150
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 2.75572379545143e-05,
+      "loss": 0.7009,
+      "step": 295160
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 2.75564763050863e-05,
+      "loss": 0.9008,
+      "step": 295170
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 2.7555714655658293e-05,
+      "loss": 0.8183,
+      "step": 295180
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 2.7554953006230294e-05,
+      "loss": 0.7603,
+      "step": 295190
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 2.7554191356802295e-05,
+      "loss": 0.7886,
+      "step": 295200
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 2.755342970737429e-05,
+      "loss": 0.8921,
+      "step": 295210
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 2.755266805794629e-05,
+      "loss": 0.7945,
+      "step": 295220
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 2.755190640851829e-05,
+      "loss": 0.7204,
+      "step": 295230
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 2.7551144759090285e-05,
+      "loss": 0.7892,
+      "step": 295240
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 2.7550383109662286e-05,
+      "loss": 0.834,
+      "step": 295250
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 2.7549621460234287e-05,
+      "loss": 0.6704,
+      "step": 295260
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 2.754885981080628e-05,
+      "loss": 0.8147,
+      "step": 295270
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 2.754809816137828e-05,
+      "loss": 0.8478,
+      "step": 295280
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 2.7547336511950283e-05,
+      "loss": 0.7123,
+      "step": 295290
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 2.754657486252228e-05,
+      "loss": 0.7908,
+      "step": 295300
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 2.754581321309428e-05,
+      "loss": 0.7424,
+      "step": 295310
+    },
+    {
+      "epoch": 13.49,
+      "learning_rate": 2.7545051563666275e-05,
+      "loss": 0.7456,
+      "step": 295320
+    },
+    {
+      "epoch": 13.49,
+      "learning_rate": 2.7544289914238276e-05,
+      "loss": 0.8114,
+      "step": 295330
+    },
+    {
+      "epoch": 13.49,
+      "learning_rate": 2.7543528264810277e-05,
+      "loss": 0.7255,
+      "step": 295340
+    },
+    {
+      "epoch": 13.49,
+      "learning_rate": 2.754276661538227e-05,
+      "loss": 0.767,
+      "step": 295350
+    },
+    {
+      "epoch": 13.49,
+      "learning_rate": 2.754200496595427e-05,
+      "loss": 0.8245,
+      "step": 295360
+    },
+    {
+      "epoch": 13.49,
+      "learning_rate": 2.7541243316526272e-05,
+      "loss": 0.7752,
+      "step": 295370
+    },
+    {
+      "epoch": 13.49,
+      "learning_rate": 2.7540481667098267e-05,
+      "loss": 0.7967,
+      "step": 295380
+    },
+    {
+      "epoch": 13.49,
+      "learning_rate": 2.7539720017670267e-05,
+      "loss": 0.7443,
+      "step": 295390
+    },
+    {
+      "epoch": 13.49,
+      "learning_rate": 2.7538958368242268e-05,
+      "loss": 0.83,
+      "step": 295400
+    },
+    {
+      "epoch": 13.49,
+      "learning_rate": 2.7538196718814262e-05,
+      "loss": 0.8605,
+      "step": 295410
+    },
+    {
+      "epoch": 13.49,
+      "learning_rate": 2.7537435069386263e-05,
+      "loss": 0.7759,
+      "step": 295420
+    },
+    {
+      "epoch": 13.49,
+      "learning_rate": 2.7536673419958264e-05,
+      "loss": 0.8309,
+      "step": 295430
+    },
+    {
+      "epoch": 13.49,
+      "learning_rate": 2.7535911770530258e-05,
+      "loss": 0.914,
+      "step": 295440
+    },
+    {
+      "epoch": 13.49,
+      "learning_rate": 2.753515012110226e-05,
+      "loss": 0.7779,
+      "step": 295450
+    },
+    {
+      "epoch": 13.49,
+      "learning_rate": 2.753438847167426e-05,
+      "loss": 0.7912,
+      "step": 295460
+    },
+    {
+      "epoch": 13.49,
+      "learning_rate": 2.7533626822246257e-05,
+      "loss": 0.8313,
+      "step": 295470
+    },
+    {
+      "epoch": 13.49,
+      "learning_rate": 2.7532865172818255e-05,
+      "loss": 0.8219,
+      "step": 295480
+    },
+    {
+      "epoch": 13.49,
+      "learning_rate": 2.7532103523390256e-05,
+      "loss": 0.8006,
+      "step": 295490
+    },
+    {
+      "epoch": 13.49,
+      "learning_rate": 2.7531341873962253e-05,
+      "loss": 0.8155,
+      "step": 295500
+    },
+    {
+      "epoch": 13.49,
+      "learning_rate": 2.7530580224534254e-05,
+      "loss": 0.7713,
+      "step": 295510
+    },
+    {
+      "epoch": 13.49,
+      "learning_rate": 2.7529818575106255e-05,
+      "loss": 0.767,
+      "step": 295520
+    },
+    {
+      "epoch": 13.49,
+      "learning_rate": 2.752905692567825e-05,
+      "loss": 0.7816,
+      "step": 295530
+    },
+    {
+      "epoch": 13.5,
+      "learning_rate": 2.752829527625025e-05,
+      "loss": 0.7996,
+      "step": 295540
+    },
+    {
+      "epoch": 13.5,
+      "learning_rate": 2.752753362682225e-05,
+      "loss": 0.8007,
+      "step": 295550
+    },
+    {
+      "epoch": 13.5,
+      "learning_rate": 2.7526771977394245e-05,
+      "loss": 0.8228,
+      "step": 295560
+    },
+    {
+      "epoch": 13.5,
+      "learning_rate": 2.7526010327966245e-05,
+      "loss": 0.8482,
+      "step": 295570
+    },
+    {
+      "epoch": 13.5,
+      "learning_rate": 2.7525248678538246e-05,
+      "loss": 0.6816,
+      "step": 295580
+    },
+    {
+      "epoch": 13.5,
+      "learning_rate": 2.752448702911024e-05,
+      "loss": 0.8403,
+      "step": 295590
+    },
+    {
+      "epoch": 13.5,
+      "learning_rate": 2.752372537968224e-05,
+      "loss": 0.7535,
+      "step": 295600
+    },
+    {
+      "epoch": 13.5,
+      "learning_rate": 2.7522963730254242e-05,
+      "loss": 0.7876,
+      "step": 295610
+    },
+    {
+      "epoch": 13.5,
+      "learning_rate": 2.7522202080826236e-05,
+      "loss": 0.7797,
+      "step": 295620
+    },
+    {
+      "epoch": 13.5,
+      "learning_rate": 2.7521440431398237e-05,
+      "loss": 0.8208,
+      "step": 295630
+    },
+    {
+      "epoch": 13.5,
+      "learning_rate": 2.7520678781970238e-05,
+      "loss": 0.8263,
+      "step": 295640
+    },
+    {
+      "epoch": 13.5,
+      "learning_rate": 2.7519917132542232e-05,
+      "loss": 0.8038,
+      "step": 295650
+    },
+    {
+      "epoch": 13.5,
+      "learning_rate": 2.7519155483114233e-05,
+      "loss": 0.7992,
+      "step": 295660
+    },
+    {
+      "epoch": 13.5,
+      "learning_rate": 2.7518393833686234e-05,
+      "loss": 0.8195,
+      "step": 295670
+    },
+    {
+      "epoch": 13.5,
+      "learning_rate": 2.751763218425823e-05,
+      "loss": 0.7993,
+      "step": 295680
+    },
+    {
+      "epoch": 13.5,
+      "learning_rate": 2.751687053483023e-05,
+      "loss": 0.7168,
+      "step": 295690
+    },
+    {
+      "epoch": 13.5,
+      "learning_rate": 2.751610888540223e-05,
+      "loss": 0.8636,
+      "step": 295700
+    },
+    {
+      "epoch": 13.5,
+      "learning_rate": 2.7515347235974227e-05,
+      "loss": 0.8099,
+      "step": 295710
+    },
+    {
+      "epoch": 13.5,
+      "learning_rate": 2.7514585586546228e-05,
+      "loss": 0.7215,
+      "step": 295720
+    },
+    {
+      "epoch": 13.5,
+      "learning_rate": 2.751382393711823e-05,
+      "loss": 0.8287,
+      "step": 295730
+    },
+    {
+      "epoch": 13.5,
+      "learning_rate": 2.7513062287690223e-05,
+      "loss": 0.864,
+      "step": 295740
+    },
+    {
+      "epoch": 13.5,
+      "learning_rate": 2.7512300638262224e-05,
+      "loss": 0.7673,
+      "step": 295750
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 2.7511538988834224e-05,
+      "loss": 0.8139,
+      "step": 295760
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 2.751077733940622e-05,
+      "loss": 0.7191,
+      "step": 295770
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 2.751001568997822e-05,
+      "loss": 0.7506,
+      "step": 295780
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 2.7509254040550213e-05,
+      "loss": 0.8378,
+      "step": 295790
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 2.7508492391122214e-05,
+      "loss": 0.79,
+      "step": 295800
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 2.7507730741694215e-05,
+      "loss": 0.6949,
+      "step": 295810
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 2.750696909226621e-05,
+      "loss": 0.7401,
+      "step": 295820
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 2.750620744283821e-05,
+      "loss": 0.7037,
+      "step": 295830
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 2.750544579341021e-05,
+      "loss": 0.741,
+      "step": 295840
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 2.7504684143982205e-05,
+      "loss": 0.822,
+      "step": 295850
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 2.7503922494554206e-05,
+      "loss": 0.8695,
+      "step": 295860
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 2.7503160845126207e-05,
+      "loss": 0.7812,
+      "step": 295870
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 2.7502399195698204e-05,
+      "loss": 0.8148,
+      "step": 295880
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 2.7501637546270205e-05,
+      "loss": 0.7811,
+      "step": 295890
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 2.7500875896842203e-05,
+      "loss": 0.7851,
+      "step": 295900
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 2.75001142474142e-05,
+      "loss": 0.8805,
+      "step": 295910
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 2.74993525979862e-05,
+      "loss": 0.7457,
+      "step": 295920
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 2.7498590948558202e-05,
+      "loss": 0.7285,
+      "step": 295930
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 2.7497829299130196e-05,
+      "loss": 0.8093,
+      "step": 295940
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 2.7497067649702197e-05,
+      "loss": 0.8361,
+      "step": 295950
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 2.7496306000274198e-05,
+      "loss": 0.8759,
+      "step": 295960
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 2.749554435084619e-05,
+      "loss": 0.8289,
+      "step": 295970
+    },
+    {
+      "epoch": 13.52,
+      "learning_rate": 2.7494782701418192e-05,
+      "loss": 0.7889,
+      "step": 295980
+    },
+    {
+      "epoch": 13.52,
+      "learning_rate": 2.7494021051990193e-05,
+      "loss": 0.7274,
+      "step": 295990
+    },
+    {
+      "epoch": 13.52,
+      "learning_rate": 2.7493259402562187e-05,
+      "loss": 0.7309,
+      "step": 296000
+    },
+    {
+      "epoch": 13.52,
+      "learning_rate": 2.7492497753134188e-05,
+      "loss": 0.6677,
+      "step": 296010
+    },
+    {
+      "epoch": 13.52,
+      "learning_rate": 2.749173610370619e-05,
+      "loss": 0.8187,
+      "step": 296020
+    },
+    {
+      "epoch": 13.52,
+      "learning_rate": 2.7490974454278183e-05,
+      "loss": 0.7865,
+      "step": 296030
+    },
+    {
+      "epoch": 13.52,
+      "learning_rate": 2.7490212804850184e-05,
+      "loss": 0.7714,
+      "step": 296040
+    },
+    {
+      "epoch": 13.52,
+      "learning_rate": 2.7489451155422185e-05,
+      "loss": 0.7824,
+      "step": 296050
+    },
+    {
+      "epoch": 13.52,
+      "learning_rate": 2.7488689505994182e-05,
+      "loss": 0.779,
+      "step": 296060
+    },
+    {
+      "epoch": 13.52,
+      "learning_rate": 2.748792785656618e-05,
+      "loss": 0.8058,
+      "step": 296070
+    },
+    {
+      "epoch": 13.52,
+      "learning_rate": 2.748716620713818e-05,
+      "loss": 0.6871,
+      "step": 296080
+    },
+    {
+      "epoch": 13.52,
+      "learning_rate": 2.7486404557710178e-05,
+      "loss": 0.7998,
+      "step": 296090
+    },
+    {
+      "epoch": 13.52,
+      "learning_rate": 2.748564290828218e-05,
+      "loss": 0.784,
+      "step": 296100
+    },
+    {
+      "epoch": 13.52,
+      "learning_rate": 2.748488125885418e-05,
+      "loss": 0.8171,
+      "step": 296110
+    },
+    {
+      "epoch": 13.52,
+      "learning_rate": 2.7484119609426174e-05,
+      "loss": 0.7303,
+      "step": 296120
+    },
+    {
+      "epoch": 13.52,
+      "learning_rate": 2.7483357959998175e-05,
+      "loss": 0.821,
+      "step": 296130
+    },
+    {
+      "epoch": 13.52,
+      "learning_rate": 2.7482596310570176e-05,
+      "loss": 0.7738,
+      "step": 296140
+    },
+    {
+      "epoch": 13.52,
+      "learning_rate": 2.748183466114217e-05,
+      "loss": 0.7551,
+      "step": 296150
+    },
+    {
+      "epoch": 13.52,
+      "learning_rate": 2.748107301171417e-05,
+      "loss": 0.7537,
+      "step": 296160
+    },
+    {
+      "epoch": 13.52,
+      "learning_rate": 2.748031136228617e-05,
+      "loss": 0.8745,
+      "step": 296170
+    },
+    {
+      "epoch": 13.52,
+      "learning_rate": 2.7479549712858165e-05,
+      "loss": 0.8574,
+      "step": 296180
+    },
+    {
+      "epoch": 13.52,
+      "learning_rate": 2.7478788063430166e-05,
+      "loss": 0.8639,
+      "step": 296190
+    },
+    {
+      "epoch": 13.53,
+      "learning_rate": 2.7478026414002167e-05,
+      "loss": 0.7425,
+      "step": 296200
+    },
+    {
+      "epoch": 13.53,
+      "learning_rate": 2.747726476457416e-05,
+      "loss": 0.8151,
+      "step": 296210
+    },
+    {
+      "epoch": 13.53,
+      "learning_rate": 2.7476503115146162e-05,
+      "loss": 0.8198,
+      "step": 296220
+    },
+    {
+      "epoch": 13.53,
+      "learning_rate": 2.7475741465718163e-05,
+      "loss": 0.7908,
+      "step": 296230
+    },
+    {
+      "epoch": 13.53,
+      "learning_rate": 2.7474979816290157e-05,
+      "loss": 0.7985,
+      "step": 296240
+    },
+    {
+      "epoch": 13.53,
+      "learning_rate": 2.7474218166862158e-05,
+      "loss": 0.8288,
+      "step": 296250
+    },
+    {
+      "epoch": 13.53,
+      "learning_rate": 2.7473456517434155e-05,
+      "loss": 0.8345,
+      "step": 296260
+    },
+    {
+      "epoch": 13.53,
+      "learning_rate": 2.7472694868006156e-05,
+      "loss": 0.7975,
+      "step": 296270
+    },
+    {
+      "epoch": 13.53,
+      "learning_rate": 2.7471933218578154e-05,
+      "loss": 0.8475,
+      "step": 296280
+    },
+    {
+      "epoch": 13.53,
+      "learning_rate": 2.747117156915015e-05,
+      "loss": 0.7403,
+      "step": 296290
+    },
+    {
+      "epoch": 13.53,
+      "learning_rate": 2.7470409919722152e-05,
+      "loss": 0.7684,
+      "step": 296300
+    },
+    {
+      "epoch": 13.53,
+      "learning_rate": 2.7469648270294153e-05,
+      "loss": 0.8156,
+      "step": 296310
+    },
+    {
+      "epoch": 13.53,
+      "learning_rate": 2.7468886620866147e-05,
+      "loss": 0.7526,
+      "step": 296320
+    },
+    {
+      "epoch": 13.53,
+      "learning_rate": 2.7468124971438148e-05,
+      "loss": 0.7258,
+      "step": 296330
+    },
+    {
+      "epoch": 13.53,
+      "learning_rate": 2.746736332201015e-05,
+      "loss": 0.7837,
+      "step": 296340
+    },
+    {
+      "epoch": 13.53,
+      "learning_rate": 2.7466601672582143e-05,
+      "loss": 0.7894,
+      "step": 296350
+    },
+    {
+      "epoch": 13.53,
+      "learning_rate": 2.7465840023154144e-05,
+      "loss": 0.803,
+      "step": 296360
+    },
+    {
+      "epoch": 13.53,
+      "learning_rate": 2.7465078373726144e-05,
+      "loss": 0.7358,
+      "step": 296370
+    },
+    {
+      "epoch": 13.53,
+      "learning_rate": 2.746431672429814e-05,
+      "loss": 0.7513,
+      "step": 296380
+    },
+    {
+      "epoch": 13.53,
+      "learning_rate": 2.746355507487014e-05,
+      "loss": 0.7731,
+      "step": 296390
+    },
+    {
+      "epoch": 13.53,
+      "learning_rate": 2.746279342544214e-05,
+      "loss": 0.8159,
+      "step": 296400
+    },
+    {
+      "epoch": 13.53,
+      "learning_rate": 2.7462031776014134e-05,
+      "loss": 0.9389,
+      "step": 296410
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 2.7461270126586135e-05,
+      "loss": 0.8716,
+      "step": 296420
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 2.7460508477158136e-05,
+      "loss": 0.8315,
+      "step": 296430
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 2.745974682773013e-05,
+      "loss": 0.7917,
+      "step": 296440
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 2.745898517830213e-05,
+      "loss": 0.7448,
+      "step": 296450
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 2.7458223528874132e-05,
+      "loss": 0.7507,
+      "step": 296460
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 2.745746187944613e-05,
+      "loss": 0.8646,
+      "step": 296470
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 2.745670023001813e-05,
+      "loss": 0.8374,
+      "step": 296480
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 2.7455938580590128e-05,
+      "loss": 0.7892,
+      "step": 296490
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 2.7455176931162125e-05,
+      "loss": 0.7743,
+      "step": 296500
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 2.7454415281734126e-05,
+      "loss": 0.7045,
+      "step": 296510
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 2.7453653632306127e-05,
+      "loss": 0.7652,
+      "step": 296520
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 2.745289198287812e-05,
+      "loss": 0.7824,
+      "step": 296530
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 2.7452130333450122e-05,
+      "loss": 0.7279,
+      "step": 296540
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 2.7451368684022123e-05,
+      "loss": 0.7753,
+      "step": 296550
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 2.7450607034594117e-05,
+      "loss": 0.813,
+      "step": 296560
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 2.7449845385166118e-05,
+      "loss": 0.7044,
+      "step": 296570
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 2.744908373573812e-05,
+      "loss": 0.7247,
+      "step": 296580
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 2.7448322086310112e-05,
+      "loss": 0.7832,
+      "step": 296590
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 2.7447560436882113e-05,
+      "loss": 0.7515,
+      "step": 296600
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 2.7446798787454114e-05,
+      "loss": 0.7546,
+      "step": 296610
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 2.7446037138026108e-05,
+      "loss": 0.8427,
+      "step": 296620
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 2.744527548859811e-05,
+      "loss": 0.898,
+      "step": 296630
+    },
+    {
+      "epoch": 13.55,
+      "learning_rate": 2.744451383917011e-05,
+      "loss": 0.804,
+      "step": 296640
+    },
+    {
+      "epoch": 13.55,
+      "learning_rate": 2.7443752189742104e-05,
+      "loss": 0.8427,
+      "step": 296650
+    },
+    {
+      "epoch": 13.55,
+      "learning_rate": 2.7442990540314105e-05,
+      "loss": 0.7346,
+      "step": 296660
+    },
+    {
+      "epoch": 13.55,
+      "learning_rate": 2.7442228890886106e-05,
+      "loss": 0.7468,
+      "step": 296670
+    },
+    {
+      "epoch": 13.55,
+      "learning_rate": 2.7441467241458103e-05,
+      "loss": 0.8023,
+      "step": 296680
+    },
+    {
+      "epoch": 13.55,
+      "learning_rate": 2.7440705592030104e-05,
+      "loss": 0.8971,
+      "step": 296690
+    },
+    {
+      "epoch": 13.55,
+      "learning_rate": 2.7439943942602098e-05,
+      "loss": 0.8287,
+      "step": 296700
+    },
+    {
+      "epoch": 13.55,
+      "learning_rate": 2.74391822931741e-05,
+      "loss": 0.8132,
+      "step": 296710
+    },
+    {
+      "epoch": 13.55,
+      "learning_rate": 2.74384206437461e-05,
+      "loss": 0.7747,
+      "step": 296720
+    },
+    {
+      "epoch": 13.55,
+      "learning_rate": 2.7437658994318094e-05,
+      "loss": 0.8508,
+      "step": 296730
+    },
+    {
+      "epoch": 13.55,
+      "learning_rate": 2.7436897344890095e-05,
+      "loss": 0.8309,
+      "step": 296740
+    },
+    {
+      "epoch": 13.55,
+      "learning_rate": 2.7436135695462096e-05,
+      "loss": 0.8121,
+      "step": 296750
+    },
+    {
+      "epoch": 13.55,
+      "learning_rate": 2.743537404603409e-05,
+      "loss": 0.8824,
+      "step": 296760
+    },
+    {
+      "epoch": 13.55,
+      "learning_rate": 2.743461239660609e-05,
+      "loss": 0.7172,
+      "step": 296770
+    },
+    {
+      "epoch": 13.55,
+      "learning_rate": 2.743385074717809e-05,
+      "loss": 0.8563,
+      "step": 296780
+    },
+    {
+      "epoch": 13.55,
+      "learning_rate": 2.7433089097750085e-05,
+      "loss": 0.7381,
+      "step": 296790
+    },
+    {
+      "epoch": 13.55,
+      "learning_rate": 2.7432327448322086e-05,
+      "loss": 0.8133,
+      "step": 296800
+    },
+    {
+      "epoch": 13.55,
+      "learning_rate": 2.7431565798894087e-05,
+      "loss": 0.7887,
+      "step": 296810
+    },
+    {
+      "epoch": 13.55,
+      "learning_rate": 2.743080414946608e-05,
+      "loss": 0.7555,
+      "step": 296820
+    },
+    {
+      "epoch": 13.55,
+      "learning_rate": 2.7430042500038082e-05,
+      "loss": 0.8261,
+      "step": 296830
+    },
+    {
+      "epoch": 13.55,
+      "learning_rate": 2.7429280850610083e-05,
+      "loss": 0.6813,
+      "step": 296840
+    },
+    {
+      "epoch": 13.55,
+      "learning_rate": 2.742851920118208e-05,
+      "loss": 0.8093,
+      "step": 296850
+    },
+    {
+      "epoch": 13.56,
+      "learning_rate": 2.742775755175408e-05,
+      "loss": 0.7463,
+      "step": 296860
+    },
+    {
+      "epoch": 13.56,
+      "learning_rate": 2.742699590232608e-05,
+      "loss": 0.818,
+      "step": 296870
+    },
+    {
+      "epoch": 13.56,
+      "learning_rate": 2.7426234252898076e-05,
+      "loss": 0.7868,
+      "step": 296880
+    },
+    {
+      "epoch": 13.56,
+      "learning_rate": 2.7425472603470077e-05,
+      "loss": 0.8351,
+      "step": 296890
+    },
+    {
+      "epoch": 13.56,
+      "learning_rate": 2.7424710954042078e-05,
+      "loss": 0.9349,
+      "step": 296900
+    },
+    {
+      "epoch": 13.56,
+      "learning_rate": 2.7423949304614072e-05,
+      "loss": 0.771,
+      "step": 296910
+    },
+    {
+      "epoch": 13.56,
+      "learning_rate": 2.7423187655186073e-05,
+      "loss": 0.8084,
+      "step": 296920
+    },
+    {
+      "epoch": 13.56,
+      "learning_rate": 2.7422426005758074e-05,
+      "loss": 0.7432,
+      "step": 296930
+    },
+    {
+      "epoch": 13.56,
+      "learning_rate": 2.7421664356330068e-05,
+      "loss": 0.8021,
+      "step": 296940
+    },
+    {
+      "epoch": 13.56,
+      "learning_rate": 2.742090270690207e-05,
+      "loss": 0.7545,
+      "step": 296950
+    },
+    {
+      "epoch": 13.56,
+      "learning_rate": 2.742014105747407e-05,
+      "loss": 0.7756,
+      "step": 296960
+    },
+    {
+      "epoch": 13.56,
+      "learning_rate": 2.7419379408046064e-05,
+      "loss": 0.803,
+      "step": 296970
+    },
+    {
+      "epoch": 13.56,
+      "learning_rate": 2.7418617758618064e-05,
+      "loss": 0.7238,
+      "step": 296980
+    },
+    {
+      "epoch": 13.56,
+      "learning_rate": 2.7417856109190065e-05,
+      "loss": 0.7708,
+      "step": 296990
+    },
+    {
+      "epoch": 13.56,
+      "learning_rate": 2.741709445976206e-05,
+      "loss": 0.7757,
+      "step": 297000
+    },
+    {
+      "epoch": 13.56,
+      "learning_rate": 2.741633281033406e-05,
+      "loss": 0.7238,
+      "step": 297010
+    },
+    {
+      "epoch": 13.56,
+      "learning_rate": 2.741557116090606e-05,
+      "loss": 0.7311,
+      "step": 297020
+    },
+    {
+      "epoch": 13.56,
+      "learning_rate": 2.7414809511478055e-05,
+      "loss": 0.8447,
+      "step": 297030
+    },
+    {
+      "epoch": 13.56,
+      "learning_rate": 2.7414047862050056e-05,
+      "loss": 0.8055,
+      "step": 297040
+    },
+    {
+      "epoch": 13.56,
+      "learning_rate": 2.7413286212622057e-05,
+      "loss": 0.769,
+      "step": 297050
+    },
+    {
+      "epoch": 13.56,
+      "learning_rate": 2.7412524563194054e-05,
+      "loss": 0.7812,
+      "step": 297060
+    },
+    {
+      "epoch": 13.56,
+      "learning_rate": 2.7411762913766055e-05,
+      "loss": 0.7863,
+      "step": 297070
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 2.7411001264338053e-05,
+      "loss": 0.7952,
+      "step": 297080
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 2.741023961491005e-05,
+      "loss": 0.7215,
+      "step": 297090
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 2.740947796548205e-05,
+      "loss": 0.7234,
+      "step": 297100
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 2.7408716316054052e-05,
+      "loss": 0.7822,
+      "step": 297110
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 2.7407954666626046e-05,
+      "loss": 0.7537,
+      "step": 297120
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 2.7407193017198047e-05,
+      "loss": 0.8156,
+      "step": 297130
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 2.7406431367770048e-05,
+      "loss": 0.8271,
+      "step": 297140
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 2.7405669718342042e-05,
+      "loss": 0.8942,
+      "step": 297150
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 2.7404908068914043e-05,
+      "loss": 0.8014,
+      "step": 297160
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 2.7404146419486037e-05,
+      "loss": 0.6865,
+      "step": 297170
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 2.7403384770058038e-05,
+      "loss": 0.8539,
+      "step": 297180
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 2.740262312063004e-05,
+      "loss": 0.8034,
+      "step": 297190
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 2.7401861471202032e-05,
+      "loss": 0.8305,
+      "step": 297200
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 2.7401099821774033e-05,
+      "loss": 0.8087,
+      "step": 297210
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 2.7400338172346034e-05,
+      "loss": 0.8583,
+      "step": 297220
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 2.739957652291803e-05,
+      "loss": 0.8304,
+      "step": 297230
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 2.739881487349003e-05,
+      "loss": 0.7717,
+      "step": 297240
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 2.739805322406203e-05,
+      "loss": 0.7644,
+      "step": 297250
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 2.7397291574634027e-05,
+      "loss": 0.8102,
+      "step": 297260
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 2.7396529925206028e-05,
+      "loss": 0.9067,
+      "step": 297270
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 2.739576827577803e-05,
+      "loss": 0.7435,
+      "step": 297280
+    },
+    {
+      "epoch": 13.58,
+      "learning_rate": 2.7395006626350023e-05,
+      "loss": 0.6989,
+      "step": 297290
+    },
+    {
+      "epoch": 13.58,
+      "learning_rate": 2.7394244976922024e-05,
+      "loss": 0.7989,
+      "step": 297300
+    },
+    {
+      "epoch": 13.58,
+      "learning_rate": 2.7393483327494025e-05,
+      "loss": 0.734,
+      "step": 297310
+    },
+    {
+      "epoch": 13.58,
+      "learning_rate": 2.739272167806602e-05,
+      "loss": 0.742,
+      "step": 297320
+    },
+    {
+      "epoch": 13.58,
+      "learning_rate": 2.739196002863802e-05,
+      "loss": 0.844,
+      "step": 297330
+    },
+    {
+      "epoch": 13.58,
+      "learning_rate": 2.739119837921002e-05,
+      "loss": 0.7239,
+      "step": 297340
+    },
+    {
+      "epoch": 13.58,
+      "learning_rate": 2.7390436729782015e-05,
+      "loss": 0.8501,
+      "step": 297350
+    },
+    {
+      "epoch": 13.58,
+      "learning_rate": 2.7389675080354016e-05,
+      "loss": 0.8519,
+      "step": 297360
+    },
+    {
+      "epoch": 13.58,
+      "learning_rate": 2.7388913430926016e-05,
+      "loss": 0.7638,
+      "step": 297370
+    },
+    {
+      "epoch": 13.58,
+      "learning_rate": 2.738815178149801e-05,
+      "loss": 0.7486,
+      "step": 297380
+    },
+    {
+      "epoch": 13.58,
+      "learning_rate": 2.738739013207001e-05,
+      "loss": 0.8425,
+      "step": 297390
+    },
+    {
+      "epoch": 13.58,
+      "learning_rate": 2.7386628482642012e-05,
+      "loss": 0.757,
+      "step": 297400
+    },
+    {
+      "epoch": 13.58,
+      "learning_rate": 2.7385866833214006e-05,
+      "loss": 0.8622,
+      "step": 297410
+    },
+    {
+      "epoch": 13.58,
+      "learning_rate": 2.7385105183786007e-05,
+      "loss": 0.8416,
+      "step": 297420
+    },
+    {
+      "epoch": 13.58,
+      "learning_rate": 2.7384343534358008e-05,
+      "loss": 0.8386,
+      "step": 297430
+    },
+    {
+      "epoch": 13.58,
+      "learning_rate": 2.7383581884930006e-05,
+      "loss": 0.7882,
+      "step": 297440
+    },
+    {
+      "epoch": 13.58,
+      "learning_rate": 2.7382820235502006e-05,
+      "loss": 0.7626,
+      "step": 297450
+    },
+    {
+      "epoch": 13.58,
+      "learning_rate": 2.7382058586074004e-05,
+      "loss": 0.8009,
+      "step": 297460
+    },
+    {
+      "epoch": 13.58,
+      "learning_rate": 2.7381296936646e-05,
+      "loss": 0.9065,
+      "step": 297470
+    },
+    {
+      "epoch": 13.58,
+      "learning_rate": 2.7380535287218002e-05,
+      "loss": 0.8689,
+      "step": 297480
+    },
+    {
+      "epoch": 13.58,
+      "learning_rate": 2.7379773637790003e-05,
+      "loss": 0.8233,
+      "step": 297490
+    },
+    {
+      "epoch": 13.58,
+      "learning_rate": 2.7379011988361997e-05,
+      "loss": 0.7658,
+      "step": 297500
+    },
+    {
+      "epoch": 13.59,
+      "learning_rate": 2.7378250338933998e-05,
+      "loss": 0.721,
+      "step": 297510
+    },
+    {
+      "epoch": 13.59,
+      "learning_rate": 2.7377488689506e-05,
+      "loss": 0.833,
+      "step": 297520
+    },
+    {
+      "epoch": 13.59,
+      "learning_rate": 2.7376727040077993e-05,
+      "loss": 0.8207,
+      "step": 297530
+    },
+    {
+      "epoch": 13.59,
+      "learning_rate": 2.7375965390649994e-05,
+      "loss": 0.7285,
+      "step": 297540
+    },
+    {
+      "epoch": 13.59,
+      "learning_rate": 2.7375203741221995e-05,
+      "loss": 0.8176,
+      "step": 297550
+    },
+    {
+      "epoch": 13.59,
+      "learning_rate": 2.737444209179399e-05,
+      "loss": 0.7723,
+      "step": 297560
+    },
+    {
+      "epoch": 13.59,
+      "learning_rate": 2.737368044236599e-05,
+      "loss": 0.7326,
+      "step": 297570
+    },
+    {
+      "epoch": 13.59,
+      "learning_rate": 2.737291879293799e-05,
+      "loss": 0.752,
+      "step": 297580
+    },
+    {
+      "epoch": 13.59,
+      "learning_rate": 2.7372157143509984e-05,
+      "loss": 0.8444,
+      "step": 297590
+    },
+    {
+      "epoch": 13.59,
+      "learning_rate": 2.7371395494081985e-05,
+      "loss": 0.7401,
+      "step": 297600
+    },
+    {
+      "epoch": 13.59,
+      "learning_rate": 2.7370633844653986e-05,
+      "loss": 0.7663,
+      "step": 297610
+    },
+    {
+      "epoch": 13.59,
+      "learning_rate": 2.736987219522598e-05,
+      "loss": 0.8729,
+      "step": 297620
+    },
+    {
+      "epoch": 13.59,
+      "learning_rate": 2.736911054579798e-05,
+      "loss": 0.7498,
+      "step": 297630
+    },
+    {
+      "epoch": 13.59,
+      "learning_rate": 2.736834889636998e-05,
+      "loss": 0.8671,
+      "step": 297640
+    },
+    {
+      "epoch": 13.59,
+      "learning_rate": 2.736758724694198e-05,
+      "loss": 0.7666,
+      "step": 297650
+    },
+    {
+      "epoch": 13.59,
+      "learning_rate": 2.736682559751398e-05,
+      "loss": 0.7464,
+      "step": 297660
+    },
+    {
+      "epoch": 13.59,
+      "learning_rate": 2.7366063948085974e-05,
+      "loss": 0.783,
+      "step": 297670
+    },
+    {
+      "epoch": 13.59,
+      "learning_rate": 2.7365302298657975e-05,
+      "loss": 0.7496,
+      "step": 297680
+    },
+    {
+      "epoch": 13.59,
+      "learning_rate": 2.7364540649229976e-05,
+      "loss": 0.7536,
+      "step": 297690
+    },
+    {
+      "epoch": 13.59,
+      "learning_rate": 2.736377899980197e-05,
+      "loss": 0.7208,
+      "step": 297700
+    },
+    {
+      "epoch": 13.59,
+      "learning_rate": 2.736301735037397e-05,
+      "loss": 0.8658,
+      "step": 297710
+    },
+    {
+      "epoch": 13.59,
+      "learning_rate": 2.7362255700945972e-05,
+      "loss": 0.8247,
+      "step": 297720
+    },
+    {
+      "epoch": 13.6,
+      "learning_rate": 2.7361494051517966e-05,
+      "loss": 0.8274,
+      "step": 297730
+    },
+    {
+      "epoch": 13.6,
+      "learning_rate": 2.7360732402089967e-05,
+      "loss": 0.8952,
+      "step": 297740
+    },
+    {
+      "epoch": 13.6,
+      "learning_rate": 2.7359970752661968e-05,
+      "loss": 0.7704,
+      "step": 297750
+    },
+    {
+      "epoch": 13.6,
+      "learning_rate": 2.7359209103233962e-05,
+      "loss": 0.7826,
+      "step": 297760
+    },
+    {
+      "epoch": 13.6,
+      "learning_rate": 2.7358447453805963e-05,
+      "loss": 0.7845,
+      "step": 297770
+    },
+    {
+      "epoch": 13.6,
+      "learning_rate": 2.7357685804377963e-05,
+      "loss": 0.7995,
+      "step": 297780
+    },
+    {
+      "epoch": 13.6,
+      "learning_rate": 2.7356924154949958e-05,
+      "loss": 0.744,
+      "step": 297790
+    },
+    {
+      "epoch": 13.6,
+      "learning_rate": 2.735616250552196e-05,
+      "loss": 0.8395,
+      "step": 297800
+    },
+    {
+      "epoch": 13.6,
+      "learning_rate": 2.735540085609396e-05,
+      "loss": 0.8041,
+      "step": 297810
+    },
+    {
+      "epoch": 13.6,
+      "learning_rate": 2.7354639206665957e-05,
+      "loss": 0.7812,
+      "step": 297820
+    },
+    {
+      "epoch": 13.6,
+      "learning_rate": 2.7353877557237954e-05,
+      "loss": 0.7565,
+      "step": 297830
+    },
+    {
+      "epoch": 13.6,
+      "learning_rate": 2.7353115907809955e-05,
+      "loss": 0.8143,
+      "step": 297840
+    },
+    {
+      "epoch": 13.6,
+      "learning_rate": 2.7352354258381952e-05,
+      "loss": 0.8694,
+      "step": 297850
+    },
+    {
+      "epoch": 13.6,
+      "learning_rate": 2.7351592608953953e-05,
+      "loss": 0.8234,
+      "step": 297860
+    },
+    {
+      "epoch": 13.6,
+      "learning_rate": 2.7350830959525954e-05,
+      "loss": 0.7578,
+      "step": 297870
+    },
+    {
+      "epoch": 13.6,
+      "learning_rate": 2.7350069310097948e-05,
+      "loss": 0.8301,
+      "step": 297880
+    },
+    {
+      "epoch": 13.6,
+      "learning_rate": 2.734930766066995e-05,
+      "loss": 0.7696,
+      "step": 297890
+    },
+    {
+      "epoch": 13.6,
+      "learning_rate": 2.734854601124195e-05,
+      "loss": 0.8118,
+      "step": 297900
+    },
+    {
+      "epoch": 13.6,
+      "learning_rate": 2.7347784361813944e-05,
+      "loss": 0.8183,
+      "step": 297910
+    },
+    {
+      "epoch": 13.6,
+      "learning_rate": 2.7347022712385945e-05,
+      "loss": 0.7436,
+      "step": 297920
+    },
+    {
+      "epoch": 13.6,
+      "learning_rate": 2.7346261062957946e-05,
+      "loss": 0.7831,
+      "step": 297930
+    },
+    {
+      "epoch": 13.6,
+      "learning_rate": 2.734549941352994e-05,
+      "loss": 0.7629,
+      "step": 297940
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 2.734473776410194e-05,
+      "loss": 0.8141,
+      "step": 297950
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 2.734397611467394e-05,
+      "loss": 0.7852,
+      "step": 297960
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 2.7343214465245936e-05,
+      "loss": 0.6684,
+      "step": 297970
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 2.7342452815817937e-05,
+      "loss": 0.7585,
+      "step": 297980
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 2.7341691166389937e-05,
+      "loss": 0.8235,
+      "step": 297990
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 2.734092951696193e-05,
+      "loss": 0.8028,
+      "step": 298000
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 2.7340167867533932e-05,
+      "loss": 0.7877,
+      "step": 298010
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 2.7339406218105933e-05,
+      "loss": 0.7684,
+      "step": 298020
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 2.733864456867793e-05,
+      "loss": 0.7215,
+      "step": 298030
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 2.7337882919249928e-05,
+      "loss": 0.7865,
+      "step": 298040
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 2.733712126982193e-05,
+      "loss": 0.8343,
+      "step": 298050
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 2.7336359620393926e-05,
+      "loss": 0.8788,
+      "step": 298060
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 2.7335597970965927e-05,
+      "loss": 0.7317,
+      "step": 298070
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 2.733483632153792e-05,
+      "loss": 0.7743,
+      "step": 298080
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 2.7334074672109922e-05,
+      "loss": 0.7344,
+      "step": 298090
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 2.7333313022681923e-05,
+      "loss": 0.7989,
+      "step": 298100
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 2.7332551373253917e-05,
+      "loss": 0.6453,
+      "step": 298110
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 2.7331789723825918e-05,
+      "loss": 0.7715,
+      "step": 298120
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 2.733102807439792e-05,
+      "loss": 0.8319,
+      "step": 298130
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 2.7330266424969913e-05,
+      "loss": 0.8241,
+      "step": 298140
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 2.7329504775541914e-05,
+      "loss": 0.864,
+      "step": 298150
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 2.7328743126113915e-05,
+      "loss": 0.6922,
+      "step": 298160
+    },
+    {
+      "epoch": 13.62,
+      "learning_rate": 2.732798147668591e-05,
+      "loss": 0.7799,
+      "step": 298170
+    },
+    {
+      "epoch": 13.62,
+      "learning_rate": 2.732721982725791e-05,
+      "loss": 0.7814,
+      "step": 298180
+    },
+    {
+      "epoch": 13.62,
+      "learning_rate": 2.732645817782991e-05,
+      "loss": 0.7489,
+      "step": 298190
+    },
+    {
+      "epoch": 13.62,
+      "learning_rate": 2.7325696528401908e-05,
+      "loss": 0.7521,
+      "step": 298200
+    },
+    {
+      "epoch": 13.62,
+      "learning_rate": 2.7324934878973905e-05,
+      "loss": 0.8483,
+      "step": 298210
+    },
+    {
+      "epoch": 13.62,
+      "learning_rate": 2.7324173229545906e-05,
+      "loss": 0.7428,
+      "step": 298220
+    },
+    {
+      "epoch": 13.62,
+      "learning_rate": 2.7323411580117904e-05,
+      "loss": 0.7698,
+      "step": 298230
+    },
+    {
+      "epoch": 13.62,
+      "learning_rate": 2.7322649930689905e-05,
+      "loss": 0.8095,
+      "step": 298240
+    },
+    {
+      "epoch": 13.62,
+      "learning_rate": 2.7321888281261905e-05,
+      "loss": 0.8177,
+      "step": 298250
+    },
+    {
+      "epoch": 13.62,
+      "learning_rate": 2.73211266318339e-05,
+      "loss": 0.7379,
+      "step": 298260
+    },
+    {
+      "epoch": 13.62,
+      "learning_rate": 2.73203649824059e-05,
+      "loss": 0.7796,
+      "step": 298270
+    },
+    {
+      "epoch": 13.62,
+      "learning_rate": 2.73196033329779e-05,
+      "loss": 0.8029,
+      "step": 298280
+    },
+    {
+      "epoch": 13.62,
+      "learning_rate": 2.7318841683549895e-05,
+      "loss": 0.8239,
+      "step": 298290
+    },
+    {
+      "epoch": 13.62,
+      "learning_rate": 2.7318080034121896e-05,
+      "loss": 0.8038,
+      "step": 298300
+    },
+    {
+      "epoch": 13.62,
+      "learning_rate": 2.7317318384693897e-05,
+      "loss": 0.8231,
+      "step": 298310
+    },
+    {
+      "epoch": 13.62,
+      "learning_rate": 2.731655673526589e-05,
+      "loss": 0.7939,
+      "step": 298320
+    },
+    {
+      "epoch": 13.62,
+      "learning_rate": 2.7315795085837892e-05,
+      "loss": 0.7909,
+      "step": 298330
+    },
+    {
+      "epoch": 13.62,
+      "learning_rate": 2.7315033436409893e-05,
+      "loss": 0.8326,
+      "step": 298340
+    },
+    {
+      "epoch": 13.62,
+      "learning_rate": 2.7314271786981887e-05,
+      "loss": 0.6877,
+      "step": 298350
+    },
+    {
+      "epoch": 13.62,
+      "learning_rate": 2.7313510137553888e-05,
+      "loss": 0.7425,
+      "step": 298360
+    },
+    {
+      "epoch": 13.62,
+      "learning_rate": 2.731274848812589e-05,
+      "loss": 0.8024,
+      "step": 298370
+    },
+    {
+      "epoch": 13.62,
+      "learning_rate": 2.7311986838697883e-05,
+      "loss": 0.862,
+      "step": 298380
+    },
+    {
+      "epoch": 13.63,
+      "learning_rate": 2.7311225189269883e-05,
+      "loss": 0.831,
+      "step": 298390
+    },
+    {
+      "epoch": 13.63,
+      "learning_rate": 2.7310463539841884e-05,
+      "loss": 0.7504,
+      "step": 298400
+    },
+    {
+      "epoch": 13.63,
+      "learning_rate": 2.7309701890413882e-05,
+      "loss": 0.8168,
+      "step": 298410
+    },
+    {
+      "epoch": 13.63,
+      "learning_rate": 2.730894024098588e-05,
+      "loss": 0.7818,
+      "step": 298420
+    },
+    {
+      "epoch": 13.63,
+      "learning_rate": 2.730817859155788e-05,
+      "loss": 0.8966,
+      "step": 298430
+    },
+    {
+      "epoch": 13.63,
+      "learning_rate": 2.7307416942129878e-05,
+      "loss": 0.8133,
+      "step": 298440
+    },
+    {
+      "epoch": 13.63,
+      "learning_rate": 2.730665529270188e-05,
+      "loss": 0.7614,
+      "step": 298450
+    },
+    {
+      "epoch": 13.63,
+      "learning_rate": 2.730589364327388e-05,
+      "loss": 0.7848,
+      "step": 298460
+    },
+    {
+      "epoch": 13.63,
+      "learning_rate": 2.7305131993845873e-05,
+      "loss": 0.7679,
+      "step": 298470
+    },
+    {
+      "epoch": 13.63,
+      "learning_rate": 2.7304370344417874e-05,
+      "loss": 0.8012,
+      "step": 298480
+    },
+    {
+      "epoch": 13.63,
+      "learning_rate": 2.7303608694989875e-05,
+      "loss": 0.7553,
+      "step": 298490
+    },
+    {
+      "epoch": 13.63,
+      "learning_rate": 2.730284704556187e-05,
+      "loss": 0.7815,
+      "step": 298500
+    },
+    {
+      "epoch": 13.63,
+      "learning_rate": 2.730208539613387e-05,
+      "loss": 0.8877,
+      "step": 298510
+    },
+    {
+      "epoch": 13.63,
+      "learning_rate": 2.730132374670587e-05,
+      "loss": 0.7324,
+      "step": 298520
+    },
+    {
+      "epoch": 13.63,
+      "learning_rate": 2.7300562097277865e-05,
+      "loss": 0.8373,
+      "step": 298530
+    },
+    {
+      "epoch": 13.63,
+      "learning_rate": 2.7299800447849866e-05,
+      "loss": 0.7108,
+      "step": 298540
+    },
+    {
+      "epoch": 13.63,
+      "learning_rate": 2.729903879842186e-05,
+      "loss": 0.7614,
+      "step": 298550
+    },
+    {
+      "epoch": 13.63,
+      "learning_rate": 2.729827714899386e-05,
+      "loss": 0.8666,
+      "step": 298560
+    },
+    {
+      "epoch": 13.63,
+      "learning_rate": 2.729751549956586e-05,
+      "loss": 0.79,
+      "step": 298570
+    },
+    {
+      "epoch": 13.63,
+      "learning_rate": 2.7296753850137856e-05,
+      "loss": 0.792,
+      "step": 298580
+    },
+    {
+      "epoch": 13.63,
+      "learning_rate": 2.7295992200709857e-05,
+      "loss": 0.8973,
+      "step": 298590
+    },
+    {
+      "epoch": 13.63,
+      "learning_rate": 2.7295230551281857e-05,
+      "loss": 0.7437,
+      "step": 298600
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 2.7294468901853855e-05,
+      "loss": 0.7735,
+      "step": 298610
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 2.7293707252425856e-05,
+      "loss": 0.8221,
+      "step": 298620
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 2.7292945602997853e-05,
+      "loss": 0.7972,
+      "step": 298630
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 2.729218395356985e-05,
+      "loss": 0.7596,
+      "step": 298640
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 2.729142230414185e-05,
+      "loss": 0.7865,
+      "step": 298650
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 2.7290660654713852e-05,
+      "loss": 0.8413,
+      "step": 298660
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 2.7289899005285846e-05,
+      "loss": 0.8623,
+      "step": 298670
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 2.7289137355857847e-05,
+      "loss": 0.8473,
+      "step": 298680
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 2.7288375706429848e-05,
+      "loss": 0.844,
+      "step": 298690
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 2.7287614057001842e-05,
+      "loss": 0.8037,
+      "step": 298700
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 2.7286852407573843e-05,
+      "loss": 0.9071,
+      "step": 298710
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 2.7286090758145844e-05,
+      "loss": 0.7338,
+      "step": 298720
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 2.7285329108717838e-05,
+      "loss": 0.7464,
+      "step": 298730
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 2.728456745928984e-05,
+      "loss": 0.7353,
+      "step": 298740
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 2.728380580986184e-05,
+      "loss": 0.8264,
+      "step": 298750
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 2.7283044160433834e-05,
+      "loss": 0.7845,
+      "step": 298760
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 2.7282282511005835e-05,
+      "loss": 0.7296,
+      "step": 298770
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 2.7281520861577835e-05,
+      "loss": 0.8487,
+      "step": 298780
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 2.728075921214983e-05,
+      "loss": 0.8055,
+      "step": 298790
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 2.727999756272183e-05,
+      "loss": 0.886,
+      "step": 298800
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 2.727923591329383e-05,
+      "loss": 0.917,
+      "step": 298810
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 2.727847426386583e-05,
+      "loss": 0.8058,
+      "step": 298820
+    },
+    {
+      "epoch": 13.65,
+      "learning_rate": 2.727771261443783e-05,
+      "loss": 0.7463,
+      "step": 298830
+    },
+    {
+      "epoch": 13.65,
+      "learning_rate": 2.7276950965009827e-05,
+      "loss": 0.7786,
+      "step": 298840
+    },
+    {
+      "epoch": 13.65,
+      "learning_rate": 2.7276189315581825e-05,
+      "loss": 0.8415,
+      "step": 298850
+    },
+    {
+      "epoch": 13.65,
+      "learning_rate": 2.7275427666153825e-05,
+      "loss": 0.8274,
+      "step": 298860
+    },
+    {
+      "epoch": 13.65,
+      "learning_rate": 2.7274666016725826e-05,
+      "loss": 0.8749,
+      "step": 298870
+    },
+    {
+      "epoch": 13.65,
+      "learning_rate": 2.727390436729782e-05,
+      "loss": 0.8346,
+      "step": 298880
+    },
+    {
+      "epoch": 13.65,
+      "learning_rate": 2.727314271786982e-05,
+      "loss": 0.8255,
+      "step": 298890
+    },
+    {
+      "epoch": 13.65,
+      "learning_rate": 2.7272381068441822e-05,
+      "loss": 0.7885,
+      "step": 298900
+    },
+    {
+      "epoch": 13.65,
+      "learning_rate": 2.7271619419013816e-05,
+      "loss": 0.8151,
+      "step": 298910
+    },
+    {
+      "epoch": 13.65,
+      "learning_rate": 2.7270857769585817e-05,
+      "loss": 0.7109,
+      "step": 298920
+    },
+    {
+      "epoch": 13.65,
+      "learning_rate": 2.7270096120157818e-05,
+      "loss": 0.714,
+      "step": 298930
+    },
+    {
+      "epoch": 13.65,
+      "learning_rate": 2.7269334470729812e-05,
+      "loss": 0.695,
+      "step": 298940
+    },
+    {
+      "epoch": 13.65,
+      "learning_rate": 2.7268572821301813e-05,
+      "loss": 0.8181,
+      "step": 298950
+    },
+    {
+      "epoch": 13.65,
+      "learning_rate": 2.7267811171873814e-05,
+      "loss": 0.7601,
+      "step": 298960
+    },
+    {
+      "epoch": 13.65,
+      "learning_rate": 2.7267049522445808e-05,
+      "loss": 0.8071,
+      "step": 298970
+    },
+    {
+      "epoch": 13.65,
+      "learning_rate": 2.726628787301781e-05,
+      "loss": 0.8126,
+      "step": 298980
+    },
+    {
+      "epoch": 13.65,
+      "learning_rate": 2.726552622358981e-05,
+      "loss": 0.7648,
+      "step": 298990
+    },
+    {
+      "epoch": 13.65,
+      "learning_rate": 2.7264764574161807e-05,
+      "loss": 0.8228,
+      "step": 299000
+    },
+    {
+      "epoch": 13.65,
+      "learning_rate": 2.7264002924733804e-05,
+      "loss": 0.7793,
+      "step": 299010
+    },
+    {
+      "epoch": 13.65,
+      "learning_rate": 2.7263241275305802e-05,
+      "loss": 0.7873,
+      "step": 299020
+    },
+    {
+      "epoch": 13.65,
+      "learning_rate": 2.7262479625877803e-05,
+      "loss": 0.847,
+      "step": 299030
+    },
+    {
+      "epoch": 13.65,
+      "learning_rate": 2.7261717976449804e-05,
+      "loss": 0.6924,
+      "step": 299040
+    },
+    {
+      "epoch": 13.66,
+      "learning_rate": 2.7260956327021798e-05,
+      "loss": 0.7728,
+      "step": 299050
+    },
+    {
+      "epoch": 13.66,
+      "learning_rate": 2.72601946775938e-05,
+      "loss": 0.8843,
+      "step": 299060
+    },
+    {
+      "epoch": 13.66,
+      "learning_rate": 2.72594330281658e-05,
+      "loss": 0.7771,
+      "step": 299070
+    },
+    {
+      "epoch": 13.66,
+      "learning_rate": 2.7258671378737793e-05,
+      "loss": 0.7777,
+      "step": 299080
+    },
+    {
+      "epoch": 13.66,
+      "learning_rate": 2.7257909729309794e-05,
+      "loss": 0.7109,
+      "step": 299090
+    },
+    {
+      "epoch": 13.66,
+      "learning_rate": 2.7257148079881795e-05,
+      "loss": 0.7903,
+      "step": 299100
+    },
+    {
+      "epoch": 13.66,
+      "learning_rate": 2.725638643045379e-05,
+      "loss": 0.7873,
+      "step": 299110
+    },
+    {
+      "epoch": 13.66,
+      "learning_rate": 2.725562478102579e-05,
+      "loss": 0.9194,
+      "step": 299120
+    },
+    {
+      "epoch": 13.66,
+      "learning_rate": 2.725486313159779e-05,
+      "loss": 0.8138,
+      "step": 299130
+    },
+    {
+      "epoch": 13.66,
+      "learning_rate": 2.7254101482169785e-05,
+      "loss": 0.8926,
+      "step": 299140
+    },
+    {
+      "epoch": 13.66,
+      "learning_rate": 2.7253339832741786e-05,
+      "loss": 0.7119,
+      "step": 299150
+    },
+    {
+      "epoch": 13.66,
+      "learning_rate": 2.7252578183313787e-05,
+      "loss": 0.8014,
+      "step": 299160
+    },
+    {
+      "epoch": 13.66,
+      "learning_rate": 2.725181653388578e-05,
+      "loss": 0.7644,
+      "step": 299170
+    },
+    {
+      "epoch": 13.66,
+      "learning_rate": 2.725105488445778e-05,
+      "loss": 0.7321,
+      "step": 299180
+    },
+    {
+      "epoch": 13.66,
+      "learning_rate": 2.7250293235029782e-05,
+      "loss": 0.7002,
+      "step": 299190
+    },
+    {
+      "epoch": 13.66,
+      "learning_rate": 2.724953158560178e-05,
+      "loss": 0.7385,
+      "step": 299200
+    },
+    {
+      "epoch": 13.66,
+      "learning_rate": 2.724876993617378e-05,
+      "loss": 0.8541,
+      "step": 299210
+    },
+    {
+      "epoch": 13.66,
+      "learning_rate": 2.7248008286745778e-05,
+      "loss": 0.7894,
+      "step": 299220
+    },
+    {
+      "epoch": 13.66,
+      "learning_rate": 2.7247246637317776e-05,
+      "loss": 0.7807,
+      "step": 299230
+    },
+    {
+      "epoch": 13.66,
+      "learning_rate": 2.7246484987889777e-05,
+      "loss": 0.7686,
+      "step": 299240
+    },
+    {
+      "epoch": 13.66,
+      "learning_rate": 2.7245723338461777e-05,
+      "loss": 0.832,
+      "step": 299250
+    },
+    {
+      "epoch": 13.66,
+      "learning_rate": 2.724496168903377e-05,
+      "loss": 0.828,
+      "step": 299260
+    },
+    {
+      "epoch": 13.67,
+      "learning_rate": 2.7244200039605772e-05,
+      "loss": 0.6763,
+      "step": 299270
+    },
+    {
+      "epoch": 13.67,
+      "learning_rate": 2.7243438390177773e-05,
+      "loss": 0.7977,
+      "step": 299280
+    },
+    {
+      "epoch": 13.67,
+      "learning_rate": 2.7242676740749767e-05,
+      "loss": 0.7151,
+      "step": 299290
+    },
+    {
+      "epoch": 13.67,
+      "learning_rate": 2.7241915091321768e-05,
+      "loss": 0.8042,
+      "step": 299300
+    },
+    {
+      "epoch": 13.67,
+      "learning_rate": 2.724115344189377e-05,
+      "loss": 0.7912,
+      "step": 299310
+    },
+    {
+      "epoch": 13.67,
+      "learning_rate": 2.7240391792465763e-05,
+      "loss": 0.8108,
+      "step": 299320
+    },
+    {
+      "epoch": 13.67,
+      "learning_rate": 2.7239630143037764e-05,
+      "loss": 0.7821,
+      "step": 299330
+    },
+    {
+      "epoch": 13.67,
+      "learning_rate": 2.7238868493609765e-05,
+      "loss": 0.8275,
+      "step": 299340
+    },
+    {
+      "epoch": 13.67,
+      "learning_rate": 2.723810684418176e-05,
+      "loss": 0.7249,
+      "step": 299350
+    },
+    {
+      "epoch": 13.67,
+      "learning_rate": 2.723734519475376e-05,
+      "loss": 0.7237,
+      "step": 299360
+    },
+    {
+      "epoch": 13.67,
+      "learning_rate": 2.723658354532576e-05,
+      "loss": 0.8071,
+      "step": 299370
+    },
+    {
+      "epoch": 13.67,
+      "learning_rate": 2.7235821895897755e-05,
+      "loss": 0.8099,
+      "step": 299380
+    },
+    {
+      "epoch": 13.67,
+      "learning_rate": 2.7235060246469756e-05,
+      "loss": 0.7707,
+      "step": 299390
+    },
+    {
+      "epoch": 13.67,
+      "learning_rate": 2.7234298597041756e-05,
+      "loss": 0.7426,
+      "step": 299400
+    },
+    {
+      "epoch": 13.67,
+      "learning_rate": 2.7233536947613754e-05,
+      "loss": 0.7502,
+      "step": 299410
+    },
+    {
+      "epoch": 13.67,
+      "learning_rate": 2.7232775298185755e-05,
+      "loss": 0.8084,
+      "step": 299420
+    },
+    {
+      "epoch": 13.67,
+      "learning_rate": 2.7232013648757752e-05,
+      "loss": 0.7064,
+      "step": 299430
+    },
+    {
+      "epoch": 13.67,
+      "learning_rate": 2.723125199932975e-05,
+      "loss": 0.8036,
+      "step": 299440
+    },
+    {
+      "epoch": 13.67,
+      "learning_rate": 2.723049034990175e-05,
+      "loss": 0.6757,
+      "step": 299450
+    },
+    {
+      "epoch": 13.67,
+      "learning_rate": 2.722972870047375e-05,
+      "loss": 0.754,
+      "step": 299460
+    },
+    {
+      "epoch": 13.67,
+      "learning_rate": 2.7228967051045745e-05,
+      "loss": 0.7718,
+      "step": 299470
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 2.7228205401617746e-05,
+      "loss": 0.7957,
+      "step": 299480
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 2.722744375218974e-05,
+      "loss": 0.7668,
+      "step": 299490
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 2.722668210276174e-05,
+      "loss": 0.7601,
+      "step": 299500
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 2.7225920453333742e-05,
+      "loss": 0.7514,
+      "step": 299510
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 2.7225158803905736e-05,
+      "loss": 0.7662,
+      "step": 299520
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 2.7224397154477737e-05,
+      "loss": 0.8461,
+      "step": 299530
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 2.7223635505049738e-05,
+      "loss": 0.7871,
+      "step": 299540
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 2.7222873855621732e-05,
+      "loss": 0.8699,
+      "step": 299550
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 2.7222112206193733e-05,
+      "loss": 0.7844,
+      "step": 299560
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 2.7221350556765734e-05,
+      "loss": 0.7868,
+      "step": 299570
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 2.722058890733773e-05,
+      "loss": 0.7295,
+      "step": 299580
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 2.721982725790973e-05,
+      "loss": 0.7572,
+      "step": 299590
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 2.721906560848173e-05,
+      "loss": 0.7452,
+      "step": 299600
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 2.7218303959053727e-05,
+      "loss": 0.804,
+      "step": 299610
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 2.7217542309625728e-05,
+      "loss": 0.8325,
+      "step": 299620
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 2.721678066019773e-05,
+      "loss": 0.8123,
+      "step": 299630
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 2.7216019010769723e-05,
+      "loss": 0.7666,
+      "step": 299640
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 2.7215257361341724e-05,
+      "loss": 0.7919,
+      "step": 299650
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 2.7214495711913724e-05,
+      "loss": 0.8003,
+      "step": 299660
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 2.721373406248572e-05,
+      "loss": 0.7537,
+      "step": 299670
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 2.721297241305772e-05,
+      "loss": 0.817,
+      "step": 299680
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 2.721221076362972e-05,
+      "loss": 0.764,
+      "step": 299690
+    },
+    {
+      "epoch": 13.69,
+      "learning_rate": 2.7211449114201714e-05,
+      "loss": 0.8153,
+      "step": 299700
+    },
+    {
+      "epoch": 13.69,
+      "learning_rate": 2.7210687464773715e-05,
+      "loss": 0.7245,
+      "step": 299710
+    },
+    {
+      "epoch": 13.69,
+      "learning_rate": 2.7209925815345716e-05,
+      "loss": 0.8216,
+      "step": 299720
+    },
+    {
+      "epoch": 13.69,
+      "learning_rate": 2.720916416591771e-05,
+      "loss": 0.7124,
+      "step": 299730
+    },
+    {
+      "epoch": 13.69,
+      "learning_rate": 2.720840251648971e-05,
+      "loss": 0.8441,
+      "step": 299740
+    },
+    {
+      "epoch": 13.69,
+      "learning_rate": 2.7207640867061712e-05,
+      "loss": 0.7843,
+      "step": 299750
+    },
+    {
+      "epoch": 13.69,
+      "learning_rate": 2.7206879217633706e-05,
+      "loss": 0.7854,
+      "step": 299760
+    },
+    {
+      "epoch": 13.69,
+      "learning_rate": 2.7206117568205707e-05,
+      "loss": 0.8072,
+      "step": 299770
+    },
+    {
+      "epoch": 13.69,
+      "learning_rate": 2.7205355918777708e-05,
+      "loss": 0.8055,
+      "step": 299780
+    },
+    {
+      "epoch": 13.69,
+      "learning_rate": 2.7204594269349705e-05,
+      "loss": 0.8445,
+      "step": 299790
+    },
+    {
+      "epoch": 13.69,
+      "learning_rate": 2.7203832619921706e-05,
+      "loss": 0.8184,
+      "step": 299800
+    },
+    {
+      "epoch": 13.69,
+      "learning_rate": 2.7203070970493703e-05,
+      "loss": 0.753,
+      "step": 299810
+    },
+    {
+      "epoch": 13.69,
+      "learning_rate": 2.72023093210657e-05,
+      "loss": 0.7682,
+      "step": 299820
+    },
+    {
+      "epoch": 13.69,
+      "learning_rate": 2.72015476716377e-05,
+      "loss": 0.7934,
+      "step": 299830
+    },
+    {
+      "epoch": 13.69,
+      "learning_rate": 2.7200786022209702e-05,
+      "loss": 0.762,
+      "step": 299840
+    },
+    {
+      "epoch": 13.69,
+      "learning_rate": 2.7200024372781697e-05,
+      "loss": 0.7804,
+      "step": 299850
+    },
+    {
+      "epoch": 13.69,
+      "learning_rate": 2.7199262723353697e-05,
+      "loss": 0.8715,
+      "step": 299860
+    },
+    {
+      "epoch": 13.69,
+      "learning_rate": 2.7198501073925698e-05,
+      "loss": 0.7643,
+      "step": 299870
+    },
+    {
+      "epoch": 13.69,
+      "learning_rate": 2.7197739424497692e-05,
+      "loss": 0.7346,
+      "step": 299880
+    },
+    {
+      "epoch": 13.69,
+      "learning_rate": 2.7196977775069693e-05,
+      "loss": 0.7475,
+      "step": 299890
+    },
+    {
+      "epoch": 13.69,
+      "learning_rate": 2.7196216125641694e-05,
+      "loss": 0.7915,
+      "step": 299900
+    },
+    {
+      "epoch": 13.69,
+      "learning_rate": 2.7195454476213688e-05,
+      "loss": 0.797,
+      "step": 299910
+    },
+    {
+      "epoch": 13.7,
+      "learning_rate": 2.719469282678569e-05,
+      "loss": 0.7451,
+      "step": 299920
+    },
+    {
+      "epoch": 13.7,
+      "learning_rate": 2.7193931177357683e-05,
+      "loss": 0.7341,
+      "step": 299930
+    },
+    {
+      "epoch": 13.7,
+      "learning_rate": 2.7193169527929684e-05,
+      "loss": 0.7697,
+      "step": 299940
+    },
+    {
+      "epoch": 13.7,
+      "learning_rate": 2.7192407878501685e-05,
+      "loss": 0.8206,
+      "step": 299950
+    },
+    {
+      "epoch": 13.7,
+      "learning_rate": 2.7191646229073682e-05,
+      "loss": 0.7966,
+      "step": 299960
+    },
+    {
+      "epoch": 13.7,
+      "learning_rate": 2.719088457964568e-05,
+      "loss": 0.7963,
+      "step": 299970
+    },
+    {
+      "epoch": 13.7,
+      "learning_rate": 2.719012293021768e-05,
+      "loss": 0.8392,
+      "step": 299980
+    },
+    {
+      "epoch": 13.7,
+      "learning_rate": 2.7189361280789678e-05,
+      "loss": 0.8288,
+      "step": 299990
+    },
+    {
+      "epoch": 13.7,
+      "learning_rate": 2.718859963136168e-05,
+      "loss": 0.9214,
+      "step": 300000
+    },
+    {
+      "epoch": 13.7,
+      "learning_rate": 2.718783798193368e-05,
+      "loss": 0.7921,
+      "step": 300010
+    },
+    {
+      "epoch": 13.7,
+      "learning_rate": 2.7187076332505674e-05,
+      "loss": 0.8573,
+      "step": 300020
+    },
+    {
+      "epoch": 13.7,
+      "learning_rate": 2.7186314683077675e-05,
+      "loss": 0.7989,
+      "step": 300030
+    },
+    {
+      "epoch": 13.7,
+      "learning_rate": 2.7185553033649676e-05,
+      "loss": 0.7366,
+      "step": 300040
+    },
+    {
+      "epoch": 13.7,
+      "learning_rate": 2.718479138422167e-05,
+      "loss": 0.8456,
+      "step": 300050
+    },
+    {
+      "epoch": 13.7,
+      "learning_rate": 2.718402973479367e-05,
+      "loss": 0.7487,
+      "step": 300060
+    },
+    {
+      "epoch": 13.7,
+      "learning_rate": 2.718326808536567e-05,
+      "loss": 0.7931,
+      "step": 300070
+    },
+    {
+      "epoch": 13.7,
+      "learning_rate": 2.7182506435937665e-05,
+      "loss": 0.7585,
+      "step": 300080
+    },
+    {
+      "epoch": 13.7,
+      "learning_rate": 2.7181744786509666e-05,
+      "loss": 0.8597,
+      "step": 300090
+    },
+    {
+      "epoch": 13.7,
+      "learning_rate": 2.7180983137081667e-05,
+      "loss": 0.7021,
+      "step": 300100
+    },
+    {
+      "epoch": 13.7,
+      "learning_rate": 2.718022148765366e-05,
+      "loss": 0.8722,
+      "step": 300110
+    },
+    {
+      "epoch": 13.7,
+      "learning_rate": 2.7179459838225662e-05,
+      "loss": 0.7603,
+      "step": 300120
+    },
+    {
+      "epoch": 13.7,
+      "learning_rate": 2.7178698188797663e-05,
+      "loss": 0.8358,
+      "step": 300130
+    },
+    {
+      "epoch": 13.71,
+      "learning_rate": 2.7177936539369657e-05,
+      "loss": 0.7209,
+      "step": 300140
+    },
+    {
+      "epoch": 13.71,
+      "learning_rate": 2.7177174889941658e-05,
+      "loss": 0.8073,
+      "step": 300150
+    },
+    {
+      "epoch": 13.71,
+      "learning_rate": 2.717641324051366e-05,
+      "loss": 0.8702,
+      "step": 300160
+    },
+    {
+      "epoch": 13.71,
+      "learning_rate": 2.7175651591085656e-05,
+      "loss": 0.8217,
+      "step": 300170
+    },
+    {
+      "epoch": 13.71,
+      "learning_rate": 2.7174889941657654e-05,
+      "loss": 0.8794,
+      "step": 300180
+    },
+    {
+      "epoch": 13.71,
+      "learning_rate": 2.7174128292229654e-05,
+      "loss": 0.8181,
+      "step": 300190
+    },
+    {
+      "epoch": 13.71,
+      "learning_rate": 2.7173366642801652e-05,
+      "loss": 0.7787,
+      "step": 300200
+    },
+    {
+      "epoch": 13.71,
+      "learning_rate": 2.7172604993373653e-05,
+      "loss": 0.7795,
+      "step": 300210
+    },
+    {
+      "epoch": 13.71,
+      "learning_rate": 2.7171843343945654e-05,
+      "loss": 0.8103,
+      "step": 300220
+    },
+    {
+      "epoch": 13.71,
+      "learning_rate": 2.7171081694517648e-05,
+      "loss": 0.8273,
+      "step": 300230
+    },
+    {
+      "epoch": 13.71,
+      "learning_rate": 2.717032004508965e-05,
+      "loss": 0.8018,
+      "step": 300240
+    },
+    {
+      "epoch": 13.71,
+      "learning_rate": 2.716955839566165e-05,
+      "loss": 0.7749,
+      "step": 300250
+    },
+    {
+      "epoch": 13.71,
+      "learning_rate": 2.7168796746233644e-05,
+      "loss": 0.7111,
+      "step": 300260
+    },
+    {
+      "epoch": 13.71,
+      "learning_rate": 2.7168035096805644e-05,
+      "loss": 0.8222,
+      "step": 300270
+    },
+    {
+      "epoch": 13.71,
+      "learning_rate": 2.7167273447377645e-05,
+      "loss": 0.756,
+      "step": 300280
+    },
+    {
+      "epoch": 13.71,
+      "learning_rate": 2.716651179794964e-05,
+      "loss": 0.7926,
+      "step": 300290
+    },
+    {
+      "epoch": 13.71,
+      "learning_rate": 2.716575014852164e-05,
+      "loss": 0.7593,
+      "step": 300300
+    },
+    {
+      "epoch": 13.71,
+      "learning_rate": 2.716498849909364e-05,
+      "loss": 0.8388,
+      "step": 300310
+    },
+    {
+      "epoch": 13.71,
+      "learning_rate": 2.7164226849665635e-05,
+      "loss": 0.8132,
+      "step": 300320
+    },
+    {
+      "epoch": 13.71,
+      "learning_rate": 2.7163465200237636e-05,
+      "loss": 0.9018,
+      "step": 300330
+    },
+    {
+      "epoch": 13.71,
+      "learning_rate": 2.7162703550809637e-05,
+      "loss": 0.8526,
+      "step": 300340
+    },
+    {
+      "epoch": 13.71,
+      "learning_rate": 2.716194190138163e-05,
+      "loss": 0.7586,
+      "step": 300350
+    },
+    {
+      "epoch": 13.72,
+      "learning_rate": 2.7161180251953632e-05,
+      "loss": 0.8992,
+      "step": 300360
+    },
+    {
+      "epoch": 13.72,
+      "learning_rate": 2.7160418602525633e-05,
+      "loss": 0.6702,
+      "step": 300370
+    },
+    {
+      "epoch": 13.72,
+      "learning_rate": 2.715965695309763e-05,
+      "loss": 0.7587,
+      "step": 300380
+    },
+    {
+      "epoch": 13.72,
+      "learning_rate": 2.7158895303669628e-05,
+      "loss": 0.8481,
+      "step": 300390
+    },
+    {
+      "epoch": 13.72,
+      "learning_rate": 2.7158133654241625e-05,
+      "loss": 0.8598,
+      "step": 300400
+    },
+    {
+      "epoch": 13.72,
+      "learning_rate": 2.7157372004813626e-05,
+      "loss": 0.7974,
+      "step": 300410
+    },
+    {
+      "epoch": 13.72,
+      "learning_rate": 2.7156610355385627e-05,
+      "loss": 0.7363,
+      "step": 300420
+    },
+    {
+      "epoch": 13.72,
+      "learning_rate": 2.715584870595762e-05,
+      "loss": 0.7542,
+      "step": 300430
+    },
+    {
+      "epoch": 13.72,
+      "learning_rate": 2.715508705652962e-05,
+      "loss": 0.8193,
+      "step": 300440
+    },
+    {
+      "epoch": 13.72,
+      "learning_rate": 2.7154325407101623e-05,
+      "loss": 0.777,
+      "step": 300450
+    },
+    {
+      "epoch": 13.72,
+      "learning_rate": 2.7153563757673617e-05,
+      "loss": 0.7253,
+      "step": 300460
+    },
+    {
+      "epoch": 13.72,
+      "learning_rate": 2.7152802108245617e-05,
+      "loss": 0.8349,
+      "step": 300470
+    },
+    {
+      "epoch": 13.72,
+      "learning_rate": 2.7152040458817618e-05,
+      "loss": 0.8256,
+      "step": 300480
+    },
+    {
+      "epoch": 13.72,
+      "learning_rate": 2.7151278809389612e-05,
+      "loss": 0.8712,
+      "step": 300490
+    },
+    {
+      "epoch": 13.72,
+      "learning_rate": 2.7150517159961613e-05,
+      "loss": 0.8344,
+      "step": 300500
+    },
+    {
+      "epoch": 13.72,
+      "learning_rate": 2.7149755510533614e-05,
+      "loss": 0.8561,
+      "step": 300510
+    },
+    {
+      "epoch": 13.72,
+      "learning_rate": 2.7148993861105608e-05,
+      "loss": 0.8088,
+      "step": 300520
+    },
+    {
+      "epoch": 13.72,
+      "learning_rate": 2.714823221167761e-05,
+      "loss": 0.8456,
+      "step": 300530
+    },
+    {
+      "epoch": 13.72,
+      "learning_rate": 2.714747056224961e-05,
+      "loss": 0.7961,
+      "step": 300540
+    },
+    {
+      "epoch": 13.72,
+      "learning_rate": 2.7146708912821607e-05,
+      "loss": 0.7869,
+      "step": 300550
+    },
+    {
+      "epoch": 13.72,
+      "learning_rate": 2.7145947263393605e-05,
+      "loss": 0.811,
+      "step": 300560
+    },
+    {
+      "epoch": 13.72,
+      "learning_rate": 2.7145185613965606e-05,
+      "loss": 0.7811,
+      "step": 300570
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 2.7144423964537603e-05,
+      "loss": 0.8711,
+      "step": 300580
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 2.7143662315109604e-05,
+      "loss": 0.7509,
+      "step": 300590
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 2.7142900665681605e-05,
+      "loss": 0.7429,
+      "step": 300600
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 2.71421390162536e-05,
+      "loss": 0.7871,
+      "step": 300610
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 2.71413773668256e-05,
+      "loss": 0.7584,
+      "step": 300620
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 2.71406157173976e-05,
+      "loss": 0.6895,
+      "step": 300630
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 2.7139854067969595e-05,
+      "loss": 0.7891,
+      "step": 300640
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 2.7139092418541596e-05,
+      "loss": 0.754,
+      "step": 300650
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 2.7138330769113596e-05,
+      "loss": 0.6952,
+      "step": 300660
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 2.713756911968559e-05,
+      "loss": 0.7929,
+      "step": 300670
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 2.713680747025759e-05,
+      "loss": 0.8055,
+      "step": 300680
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 2.7136045820829592e-05,
+      "loss": 0.8013,
+      "step": 300690
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 2.7135284171401586e-05,
+      "loss": 0.8091,
+      "step": 300700
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 2.7134522521973587e-05,
+      "loss": 0.7791,
+      "step": 300710
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 2.7133760872545588e-05,
+      "loss": 0.7596,
+      "step": 300720
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 2.7132999223117582e-05,
+      "loss": 0.7792,
+      "step": 300730
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 2.7132237573689583e-05,
+      "loss": 0.7658,
+      "step": 300740
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 2.7131475924261584e-05,
+      "loss": 0.7918,
+      "step": 300750
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 2.713071427483358e-05,
+      "loss": 0.7326,
+      "step": 300760
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 2.712995262540558e-05,
+      "loss": 0.7725,
+      "step": 300770
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 2.712919097597758e-05,
+      "loss": 0.7731,
+      "step": 300780
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 2.7128429326549577e-05,
+      "loss": 0.7845,
+      "step": 300790
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 2.7127667677121578e-05,
+      "loss": 0.8202,
+      "step": 300800
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 2.712690602769358e-05,
+      "loss": 0.7838,
+      "step": 300810
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 2.7126144378265573e-05,
+      "loss": 0.7883,
+      "step": 300820
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 2.7125382728837574e-05,
+      "loss": 0.7996,
+      "step": 300830
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 2.7124621079409575e-05,
+      "loss": 0.8489,
+      "step": 300840
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 2.712385942998157e-05,
+      "loss": 0.7556,
+      "step": 300850
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 2.712309778055357e-05,
+      "loss": 0.6988,
+      "step": 300860
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 2.7122336131125564e-05,
+      "loss": 0.8373,
+      "step": 300870
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 2.7121574481697564e-05,
+      "loss": 0.7308,
+      "step": 300880
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 2.7120812832269565e-05,
+      "loss": 0.8076,
+      "step": 300890
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 2.712005118284156e-05,
+      "loss": 0.7844,
+      "step": 300900
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 2.711928953341356e-05,
+      "loss": 0.766,
+      "step": 300910
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 2.711852788398556e-05,
+      "loss": 0.7686,
+      "step": 300920
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 2.7117766234557555e-05,
+      "loss": 0.7831,
+      "step": 300930
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 2.7117004585129556e-05,
+      "loss": 0.9141,
+      "step": 300940
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 2.7116242935701557e-05,
+      "loss": 0.742,
+      "step": 300950
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 2.7115481286273554e-05,
+      "loss": 0.8102,
+      "step": 300960
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 2.7114719636845555e-05,
+      "loss": 0.7261,
+      "step": 300970
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 2.7113957987417553e-05,
+      "loss": 0.7328,
+      "step": 300980
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 2.711319633798955e-05,
+      "loss": 0.7802,
+      "step": 300990
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 2.711243468856155e-05,
+      "loss": 0.8022,
+      "step": 301000
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 2.7111673039133552e-05,
+      "loss": 0.9092,
+      "step": 301010
+    },
+    {
+      "epoch": 13.75,
+      "learning_rate": 2.7110911389705546e-05,
+      "loss": 0.7176,
+      "step": 301020
+    },
+    {
+      "epoch": 13.75,
+      "learning_rate": 2.7110149740277547e-05,
+      "loss": 0.7287,
+      "step": 301030
+    },
+    {
+      "epoch": 13.75,
+      "learning_rate": 2.7109388090849548e-05,
+      "loss": 0.7326,
+      "step": 301040
+    },
+    {
+      "epoch": 13.75,
+      "learning_rate": 2.710862644142154e-05,
+      "loss": 0.8408,
+      "step": 301050
+    },
+    {
+      "epoch": 13.75,
+      "learning_rate": 2.7107864791993543e-05,
+      "loss": 0.7894,
+      "step": 301060
+    },
+    {
+      "epoch": 13.75,
+      "learning_rate": 2.7107103142565543e-05,
+      "loss": 0.9196,
+      "step": 301070
+    },
+    {
+      "epoch": 13.75,
+      "learning_rate": 2.7106341493137537e-05,
+      "loss": 0.795,
+      "step": 301080
+    },
+    {
+      "epoch": 13.75,
+      "learning_rate": 2.7105579843709538e-05,
+      "loss": 0.7468,
+      "step": 301090
+    },
+    {
+      "epoch": 13.75,
+      "learning_rate": 2.710481819428154e-05,
+      "loss": 0.8257,
+      "step": 301100
+    },
+    {
+      "epoch": 13.75,
+      "learning_rate": 2.7104056544853533e-05,
+      "loss": 0.8251,
+      "step": 301110
+    },
+    {
+      "epoch": 13.75,
+      "learning_rate": 2.7103294895425534e-05,
+      "loss": 0.8017,
+      "step": 301120
+    },
+    {
+      "epoch": 13.75,
+      "learning_rate": 2.7102533245997535e-05,
+      "loss": 0.7453,
+      "step": 301130
+    },
+    {
+      "epoch": 13.75,
+      "learning_rate": 2.710177159656953e-05,
+      "loss": 0.812,
+      "step": 301140
+    },
+    {
+      "epoch": 13.75,
+      "learning_rate": 2.710100994714153e-05,
+      "loss": 0.7985,
+      "step": 301150
+    },
+    {
+      "epoch": 13.75,
+      "learning_rate": 2.710024829771353e-05,
+      "loss": 0.7295,
+      "step": 301160
+    },
+    {
+      "epoch": 13.75,
+      "learning_rate": 2.7099486648285528e-05,
+      "loss": 0.7372,
+      "step": 301170
+    },
+    {
+      "epoch": 13.75,
+      "learning_rate": 2.709872499885753e-05,
+      "loss": 0.7641,
+      "step": 301180
+    },
+    {
+      "epoch": 13.75,
+      "learning_rate": 2.7097963349429527e-05,
+      "loss": 0.7312,
+      "step": 301190
+    },
+    {
+      "epoch": 13.75,
+      "learning_rate": 2.7097201700001524e-05,
+      "loss": 0.7622,
+      "step": 301200
+    },
+    {
+      "epoch": 13.75,
+      "learning_rate": 2.7096440050573525e-05,
+      "loss": 0.7583,
+      "step": 301210
+    },
+    {
+      "epoch": 13.75,
+      "learning_rate": 2.7095678401145526e-05,
+      "loss": 0.7728,
+      "step": 301220
+    },
+    {
+      "epoch": 13.75,
+      "learning_rate": 2.709491675171752e-05,
+      "loss": 0.8231,
+      "step": 301230
+    },
+    {
+      "epoch": 13.76,
+      "learning_rate": 2.709415510228952e-05,
+      "loss": 0.7432,
+      "step": 301240
+    },
+    {
+      "epoch": 13.76,
+      "learning_rate": 2.709339345286152e-05,
+      "loss": 0.8261,
+      "step": 301250
+    },
+    {
+      "epoch": 13.76,
+      "learning_rate": 2.7092631803433516e-05,
+      "loss": 0.6928,
+      "step": 301260
+    },
+    {
+      "epoch": 13.76,
+      "learning_rate": 2.7091870154005516e-05,
+      "loss": 0.8042,
+      "step": 301270
+    },
+    {
+      "epoch": 13.76,
+      "learning_rate": 2.7091108504577517e-05,
+      "loss": 0.7128,
+      "step": 301280
+    },
+    {
+      "epoch": 13.76,
+      "learning_rate": 2.709034685514951e-05,
+      "loss": 0.7734,
+      "step": 301290
+    },
+    {
+      "epoch": 13.76,
+      "learning_rate": 2.7089585205721512e-05,
+      "loss": 0.7613,
+      "step": 301300
+    },
+    {
+      "epoch": 13.76,
+      "learning_rate": 2.7088823556293506e-05,
+      "loss": 0.8659,
+      "step": 301310
+    },
+    {
+      "epoch": 13.76,
+      "learning_rate": 2.7088061906865507e-05,
+      "loss": 0.8388,
+      "step": 301320
+    },
+    {
+      "epoch": 13.76,
+      "learning_rate": 2.7087300257437508e-05,
+      "loss": 0.7779,
+      "step": 301330
+    },
+    {
+      "epoch": 13.76,
+      "learning_rate": 2.7086538608009505e-05,
+      "loss": 0.8295,
+      "step": 301340
+    },
+    {
+      "epoch": 13.76,
+      "learning_rate": 2.7085776958581506e-05,
+      "loss": 0.9326,
+      "step": 301350
+    },
+    {
+      "epoch": 13.76,
+      "learning_rate": 2.7085015309153504e-05,
+      "loss": 0.8068,
+      "step": 301360
+    },
+    {
+      "epoch": 13.76,
+      "learning_rate": 2.70842536597255e-05,
+      "loss": 0.7869,
+      "step": 301370
+    },
+    {
+      "epoch": 13.76,
+      "learning_rate": 2.7083492010297502e-05,
+      "loss": 0.7339,
+      "step": 301380
+    },
+    {
+      "epoch": 13.76,
+      "learning_rate": 2.7082730360869503e-05,
+      "loss": 0.7681,
+      "step": 301390
+    },
+    {
+      "epoch": 13.76,
+      "learning_rate": 2.7081968711441497e-05,
+      "loss": 0.7574,
+      "step": 301400
+    },
+    {
+      "epoch": 13.76,
+      "learning_rate": 2.7081207062013498e-05,
+      "loss": 0.7741,
+      "step": 301410
+    },
+    {
+      "epoch": 13.76,
+      "learning_rate": 2.70804454125855e-05,
+      "loss": 0.8817,
+      "step": 301420
+    },
+    {
+      "epoch": 13.76,
+      "learning_rate": 2.7079683763157493e-05,
+      "loss": 0.7599,
+      "step": 301430
+    },
+    {
+      "epoch": 13.76,
+      "learning_rate": 2.7078922113729494e-05,
+      "loss": 0.8287,
+      "step": 301440
+    },
+    {
+      "epoch": 13.76,
+      "learning_rate": 2.7078160464301495e-05,
+      "loss": 0.7162,
+      "step": 301450
+    },
+    {
+      "epoch": 13.77,
+      "learning_rate": 2.707739881487349e-05,
+      "loss": 0.7713,
+      "step": 301460
+    },
+    {
+      "epoch": 13.77,
+      "learning_rate": 2.707663716544549e-05,
+      "loss": 0.7773,
+      "step": 301470
+    },
+    {
+      "epoch": 13.77,
+      "learning_rate": 2.707587551601749e-05,
+      "loss": 0.8481,
+      "step": 301480
+    },
+    {
+      "epoch": 13.77,
+      "learning_rate": 2.7075113866589484e-05,
+      "loss": 0.7964,
+      "step": 301490
+    },
+    {
+      "epoch": 13.77,
+      "learning_rate": 2.7074352217161485e-05,
+      "loss": 0.7915,
+      "step": 301500
+    },
+    {
+      "epoch": 13.77,
+      "learning_rate": 2.7073590567733486e-05,
+      "loss": 0.7307,
+      "step": 301510
+    },
+    {
+      "epoch": 13.77,
+      "learning_rate": 2.707282891830548e-05,
+      "loss": 0.8131,
+      "step": 301520
+    },
+    {
+      "epoch": 13.77,
+      "learning_rate": 2.707206726887748e-05,
+      "loss": 0.6883,
+      "step": 301530
+    },
+    {
+      "epoch": 13.77,
+      "learning_rate": 2.7071305619449482e-05,
+      "loss": 0.7769,
+      "step": 301540
+    },
+    {
+      "epoch": 13.77,
+      "learning_rate": 2.707054397002148e-05,
+      "loss": 0.8334,
+      "step": 301550
+    },
+    {
+      "epoch": 13.77,
+      "learning_rate": 2.706978232059348e-05,
+      "loss": 0.8959,
+      "step": 301560
+    },
+    {
+      "epoch": 13.77,
+      "learning_rate": 2.7069020671165478e-05,
+      "loss": 0.7349,
+      "step": 301570
+    },
+    {
+      "epoch": 13.77,
+      "learning_rate": 2.7068259021737475e-05,
+      "loss": 0.7626,
+      "step": 301580
+    },
+    {
+      "epoch": 13.77,
+      "learning_rate": 2.7067497372309476e-05,
+      "loss": 0.7981,
+      "step": 301590
+    },
+    {
+      "epoch": 13.77,
+      "learning_rate": 2.7066735722881477e-05,
+      "loss": 0.8186,
+      "step": 301600
+    },
+    {
+      "epoch": 13.77,
+      "learning_rate": 2.706597407345347e-05,
+      "loss": 0.821,
+      "step": 301610
+    },
+    {
+      "epoch": 13.77,
+      "learning_rate": 2.7065212424025472e-05,
+      "loss": 0.7388,
+      "step": 301620
+    },
+    {
+      "epoch": 13.77,
+      "learning_rate": 2.7064450774597473e-05,
+      "loss": 0.7949,
+      "step": 301630
+    },
+    {
+      "epoch": 13.77,
+      "learning_rate": 2.7063689125169467e-05,
+      "loss": 0.8348,
+      "step": 301640
+    },
+    {
+      "epoch": 13.77,
+      "learning_rate": 2.7062927475741468e-05,
+      "loss": 0.8171,
+      "step": 301650
+    },
+    {
+      "epoch": 13.77,
+      "learning_rate": 2.706216582631347e-05,
+      "loss": 0.7928,
+      "step": 301660
+    },
+    {
+      "epoch": 13.78,
+      "learning_rate": 2.7061404176885463e-05,
+      "loss": 0.8636,
+      "step": 301670
+    },
+    {
+      "epoch": 13.78,
+      "learning_rate": 2.7060642527457463e-05,
+      "loss": 0.6917,
+      "step": 301680
+    },
+    {
+      "epoch": 13.78,
+      "learning_rate": 2.7059880878029464e-05,
+      "loss": 0.807,
+      "step": 301690
+    },
+    {
+      "epoch": 13.78,
+      "learning_rate": 2.705911922860146e-05,
+      "loss": 0.8172,
+      "step": 301700
+    },
+    {
+      "epoch": 13.78,
+      "learning_rate": 2.705835757917346e-05,
+      "loss": 0.7526,
+      "step": 301710
+    },
+    {
+      "epoch": 13.78,
+      "learning_rate": 2.705759592974546e-05,
+      "loss": 0.7753,
+      "step": 301720
+    },
+    {
+      "epoch": 13.78,
+      "learning_rate": 2.7056834280317454e-05,
+      "loss": 0.7538,
+      "step": 301730
+    },
+    {
+      "epoch": 13.78,
+      "learning_rate": 2.7056072630889455e-05,
+      "loss": 0.7832,
+      "step": 301740
+    },
+    {
+      "epoch": 13.78,
+      "learning_rate": 2.7055310981461456e-05,
+      "loss": 0.9202,
+      "step": 301750
+    },
+    {
+      "epoch": 13.78,
+      "learning_rate": 2.7054549332033453e-05,
+      "loss": 0.8394,
+      "step": 301760
+    },
+    {
+      "epoch": 13.78,
+      "learning_rate": 2.7053787682605454e-05,
+      "loss": 0.73,
+      "step": 301770
+    },
+    {
+      "epoch": 13.78,
+      "learning_rate": 2.7053026033177448e-05,
+      "loss": 0.8019,
+      "step": 301780
+    },
+    {
+      "epoch": 13.78,
+      "learning_rate": 2.705226438374945e-05,
+      "loss": 0.8075,
+      "step": 301790
+    },
+    {
+      "epoch": 13.78,
+      "learning_rate": 2.705150273432145e-05,
+      "loss": 0.8114,
+      "step": 301800
+    },
+    {
+      "epoch": 13.78,
+      "learning_rate": 2.7050741084893444e-05,
+      "loss": 0.8176,
+      "step": 301810
+    },
+    {
+      "epoch": 13.78,
+      "learning_rate": 2.7049979435465445e-05,
+      "loss": 0.7601,
+      "step": 301820
+    },
+    {
+      "epoch": 13.78,
+      "learning_rate": 2.7049217786037446e-05,
+      "loss": 0.9061,
+      "step": 301830
+    },
+    {
+      "epoch": 13.78,
+      "learning_rate": 2.704845613660944e-05,
+      "loss": 0.769,
+      "step": 301840
+    },
+    {
+      "epoch": 13.78,
+      "learning_rate": 2.704769448718144e-05,
+      "loss": 0.7876,
+      "step": 301850
+    },
+    {
+      "epoch": 13.78,
+      "learning_rate": 2.704693283775344e-05,
+      "loss": 0.8529,
+      "step": 301860
+    },
+    {
+      "epoch": 13.78,
+      "learning_rate": 2.7046171188325436e-05,
+      "loss": 0.7706,
+      "step": 301870
+    },
+    {
+      "epoch": 13.78,
+      "learning_rate": 2.7045409538897436e-05,
+      "loss": 0.7772,
+      "step": 301880
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 2.7044647889469437e-05,
+      "loss": 0.9136,
+      "step": 301890
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 2.704388624004143e-05,
+      "loss": 0.7575,
+      "step": 301900
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 2.7043124590613432e-05,
+      "loss": 0.9021,
+      "step": 301910
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 2.7042362941185433e-05,
+      "loss": 0.7958,
+      "step": 301920
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 2.704160129175743e-05,
+      "loss": 0.8533,
+      "step": 301930
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 2.7040839642329428e-05,
+      "loss": 0.8442,
+      "step": 301940
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 2.704007799290143e-05,
+      "loss": 0.803,
+      "step": 301950
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 2.7039316343473426e-05,
+      "loss": 0.8289,
+      "step": 301960
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 2.7038554694045427e-05,
+      "loss": 0.7743,
+      "step": 301970
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 2.7037793044617428e-05,
+      "loss": 0.7888,
+      "step": 301980
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 2.7037031395189422e-05,
+      "loss": 0.7953,
+      "step": 301990
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 2.7036269745761423e-05,
+      "loss": 0.7353,
+      "step": 302000
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 2.7035508096333424e-05,
+      "loss": 0.6828,
+      "step": 302010
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 2.7034746446905418e-05,
+      "loss": 0.816,
+      "step": 302020
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 2.703398479747742e-05,
+      "loss": 0.7759,
+      "step": 302030
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 2.703322314804942e-05,
+      "loss": 0.7264,
+      "step": 302040
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 2.7032461498621414e-05,
+      "loss": 0.8193,
+      "step": 302050
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 2.7031699849193415e-05,
+      "loss": 0.8653,
+      "step": 302060
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 2.7030938199765415e-05,
+      "loss": 0.8191,
+      "step": 302070
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 2.703017655033741e-05,
+      "loss": 0.7164,
+      "step": 302080
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 2.702941490090941e-05,
+      "loss": 0.7528,
+      "step": 302090
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 2.702865325148141e-05,
+      "loss": 0.8506,
+      "step": 302100
+    },
+    {
+      "epoch": 13.8,
+      "learning_rate": 2.7027891602053405e-05,
+      "loss": 0.7488,
+      "step": 302110
+    },
+    {
+      "epoch": 13.8,
+      "learning_rate": 2.7027129952625406e-05,
+      "loss": 0.7425,
+      "step": 302120
+    },
+    {
+      "epoch": 13.8,
+      "learning_rate": 2.7026368303197407e-05,
+      "loss": 0.8523,
+      "step": 302130
+    },
+    {
+      "epoch": 13.8,
+      "learning_rate": 2.7025606653769404e-05,
+      "loss": 0.8155,
+      "step": 302140
+    },
+    {
+      "epoch": 13.8,
+      "learning_rate": 2.7024845004341405e-05,
+      "loss": 0.7538,
+      "step": 302150
+    },
+    {
+      "epoch": 13.8,
+      "learning_rate": 2.7024083354913403e-05,
+      "loss": 0.8202,
+      "step": 302160
+    },
+    {
+      "epoch": 13.8,
+      "learning_rate": 2.70233217054854e-05,
+      "loss": 0.7948,
+      "step": 302170
+    },
+    {
+      "epoch": 13.8,
+      "learning_rate": 2.70225600560574e-05,
+      "loss": 0.7706,
+      "step": 302180
+    },
+    {
+      "epoch": 13.8,
+      "learning_rate": 2.7021798406629402e-05,
+      "loss": 0.7931,
+      "step": 302190
+    },
+    {
+      "epoch": 13.8,
+      "learning_rate": 2.7021036757201396e-05,
+      "loss": 0.8369,
+      "step": 302200
+    },
+    {
+      "epoch": 13.8,
+      "learning_rate": 2.7020275107773397e-05,
+      "loss": 0.776,
+      "step": 302210
+    },
+    {
+      "epoch": 13.8,
+      "learning_rate": 2.7019513458345398e-05,
+      "loss": 0.7997,
+      "step": 302220
+    },
+    {
+      "epoch": 13.8,
+      "learning_rate": 2.7018751808917392e-05,
+      "loss": 0.7681,
+      "step": 302230
+    },
+    {
+      "epoch": 13.8,
+      "learning_rate": 2.7017990159489393e-05,
+      "loss": 0.9016,
+      "step": 302240
+    },
+    {
+      "epoch": 13.8,
+      "learning_rate": 2.7017228510061387e-05,
+      "loss": 0.8685,
+      "step": 302250
+    },
+    {
+      "epoch": 13.8,
+      "learning_rate": 2.7016466860633388e-05,
+      "loss": 0.7915,
+      "step": 302260
+    },
+    {
+      "epoch": 13.8,
+      "learning_rate": 2.701570521120539e-05,
+      "loss": 0.718,
+      "step": 302270
+    },
+    {
+      "epoch": 13.8,
+      "learning_rate": 2.7014943561777383e-05,
+      "loss": 0.7662,
+      "step": 302280
+    },
+    {
+      "epoch": 13.8,
+      "learning_rate": 2.7014181912349383e-05,
+      "loss": 0.7632,
+      "step": 302290
+    },
+    {
+      "epoch": 13.8,
+      "learning_rate": 2.7013420262921384e-05,
+      "loss": 0.7758,
+      "step": 302300
+    },
+    {
+      "epoch": 13.8,
+      "learning_rate": 2.7012658613493382e-05,
+      "loss": 0.819,
+      "step": 302310
+    },
+    {
+      "epoch": 13.8,
+      "learning_rate": 2.701189696406538e-05,
+      "loss": 0.8508,
+      "step": 302320
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 2.701113531463738e-05,
+      "loss": 0.8407,
+      "step": 302330
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 2.7010373665209378e-05,
+      "loss": 0.7969,
+      "step": 302340
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 2.700961201578138e-05,
+      "loss": 0.76,
+      "step": 302350
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 2.700885036635338e-05,
+      "loss": 0.8566,
+      "step": 302360
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 2.7008088716925373e-05,
+      "loss": 0.7906,
+      "step": 302370
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 2.7007327067497374e-05,
+      "loss": 0.7491,
+      "step": 302380
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 2.7006565418069375e-05,
+      "loss": 0.7605,
+      "step": 302390
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 2.700580376864137e-05,
+      "loss": 0.8466,
+      "step": 302400
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 2.700504211921337e-05,
+      "loss": 0.8419,
+      "step": 302410
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 2.700428046978537e-05,
+      "loss": 0.7634,
+      "step": 302420
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 2.7003518820357365e-05,
+      "loss": 0.7232,
+      "step": 302430
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 2.7002757170929366e-05,
+      "loss": 0.8743,
+      "step": 302440
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 2.7001995521501367e-05,
+      "loss": 0.8403,
+      "step": 302450
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 2.700123387207336e-05,
+      "loss": 0.8506,
+      "step": 302460
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 2.700047222264536e-05,
+      "loss": 0.8197,
+      "step": 302470
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 2.6999710573217362e-05,
+      "loss": 0.8621,
+      "step": 302480
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 2.6998948923789356e-05,
+      "loss": 0.787,
+      "step": 302490
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 2.6998187274361357e-05,
+      "loss": 0.7901,
+      "step": 302500
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 2.6997425624933358e-05,
+      "loss": 0.8222,
+      "step": 302510
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 2.6996663975505356e-05,
+      "loss": 0.7944,
+      "step": 302520
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 2.6995902326077353e-05,
+      "loss": 0.8764,
+      "step": 302530
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 2.6995140676649354e-05,
+      "loss": 0.7837,
+      "step": 302540
+    },
+    {
+      "epoch": 13.82,
+      "learning_rate": 2.699437902722135e-05,
+      "loss": 0.7372,
+      "step": 302550
+    },
+    {
+      "epoch": 13.82,
+      "learning_rate": 2.6993617377793352e-05,
+      "loss": 0.7556,
+      "step": 302560
+    },
+    {
+      "epoch": 13.82,
+      "learning_rate": 2.6992855728365353e-05,
+      "loss": 0.9393,
+      "step": 302570
+    },
+    {
+      "epoch": 13.82,
+      "learning_rate": 2.6992094078937347e-05,
+      "loss": 0.6902,
+      "step": 302580
+    },
+    {
+      "epoch": 13.82,
+      "learning_rate": 2.6991332429509348e-05,
+      "loss": 0.7748,
+      "step": 302590
+    },
+    {
+      "epoch": 13.82,
+      "learning_rate": 2.699057078008135e-05,
+      "loss": 0.7218,
+      "step": 302600
+    },
+    {
+      "epoch": 13.82,
+      "learning_rate": 2.6989809130653343e-05,
+      "loss": 0.8606,
+      "step": 302610
+    },
+    {
+      "epoch": 13.82,
+      "learning_rate": 2.6989047481225344e-05,
+      "loss": 0.9073,
+      "step": 302620
+    },
+    {
+      "epoch": 13.82,
+      "learning_rate": 2.6988285831797345e-05,
+      "loss": 0.8825,
+      "step": 302630
+    },
+    {
+      "epoch": 13.82,
+      "learning_rate": 2.698752418236934e-05,
+      "loss": 0.7894,
+      "step": 302640
+    },
+    {
+      "epoch": 13.82,
+      "learning_rate": 2.698676253294134e-05,
+      "loss": 0.8168,
+      "step": 302650
+    },
+    {
+      "epoch": 13.82,
+      "learning_rate": 2.698600088351334e-05,
+      "loss": 0.769,
+      "step": 302660
+    },
+    {
+      "epoch": 13.82,
+      "learning_rate": 2.6985239234085335e-05,
+      "loss": 0.8595,
+      "step": 302670
+    },
+    {
+      "epoch": 13.82,
+      "learning_rate": 2.6984477584657335e-05,
+      "loss": 0.8039,
+      "step": 302680
+    },
+    {
+      "epoch": 13.82,
+      "learning_rate": 2.698371593522933e-05,
+      "loss": 0.8027,
+      "step": 302690
+    },
+    {
+      "epoch": 13.82,
+      "learning_rate": 2.698295428580133e-05,
+      "loss": 0.7976,
+      "step": 302700
+    },
+    {
+      "epoch": 13.82,
+      "learning_rate": 2.698219263637333e-05,
+      "loss": 0.8012,
+      "step": 302710
+    },
+    {
+      "epoch": 13.82,
+      "learning_rate": 2.698143098694533e-05,
+      "loss": 0.8451,
+      "step": 302720
+    },
+    {
+      "epoch": 13.82,
+      "learning_rate": 2.698066933751733e-05,
+      "loss": 0.7356,
+      "step": 302730
+    },
+    {
+      "epoch": 13.82,
+      "learning_rate": 2.6979907688089327e-05,
+      "loss": 0.7988,
+      "step": 302740
+    },
+    {
+      "epoch": 13.82,
+      "learning_rate": 2.6979146038661324e-05,
+      "loss": 0.7948,
+      "step": 302750
+    },
+    {
+      "epoch": 13.82,
+      "learning_rate": 2.6978384389233325e-05,
+      "loss": 0.9,
+      "step": 302760
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 2.6977622739805326e-05,
+      "loss": 0.7827,
+      "step": 302770
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 2.697686109037732e-05,
+      "loss": 0.8606,
+      "step": 302780
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 2.697609944094932e-05,
+      "loss": 0.8267,
+      "step": 302790
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 2.6975337791521322e-05,
+      "loss": 0.7491,
+      "step": 302800
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 2.6974576142093316e-05,
+      "loss": 0.8741,
+      "step": 302810
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 2.6973814492665317e-05,
+      "loss": 0.7627,
+      "step": 302820
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 2.6973052843237318e-05,
+      "loss": 0.7268,
+      "step": 302830
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 2.6972291193809312e-05,
+      "loss": 0.7737,
+      "step": 302840
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 2.6971529544381313e-05,
+      "loss": 0.7541,
+      "step": 302850
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 2.6970767894953314e-05,
+      "loss": 0.7621,
+      "step": 302860
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 2.6970006245525308e-05,
+      "loss": 0.706,
+      "step": 302870
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 2.696924459609731e-05,
+      "loss": 0.8106,
+      "step": 302880
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 2.696848294666931e-05,
+      "loss": 0.7269,
+      "step": 302890
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 2.6967721297241307e-05,
+      "loss": 0.7351,
+      "step": 302900
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 2.6966959647813304e-05,
+      "loss": 0.8003,
+      "step": 302910
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 2.6966197998385305e-05,
+      "loss": 0.795,
+      "step": 302920
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 2.6965436348957303e-05,
+      "loss": 0.7258,
+      "step": 302930
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 2.6964674699529303e-05,
+      "loss": 0.7542,
+      "step": 302940
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 2.6963913050101304e-05,
+      "loss": 0.7494,
+      "step": 302950
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 2.69631514006733e-05,
+      "loss": 0.7701,
+      "step": 302960
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 2.69623897512453e-05,
+      "loss": 0.7908,
+      "step": 302970
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 2.69616281018173e-05,
+      "loss": 0.7407,
+      "step": 302980
+    },
+    {
+      "epoch": 13.84,
+      "learning_rate": 2.6960866452389294e-05,
+      "loss": 0.7706,
+      "step": 302990
+    },
+    {
+      "epoch": 13.84,
+      "learning_rate": 2.6960104802961295e-05,
+      "loss": 0.8772,
+      "step": 303000
+    },
+    {
+      "epoch": 13.84,
+      "learning_rate": 2.6959343153533296e-05,
+      "loss": 0.7279,
+      "step": 303010
+    },
+    {
+      "epoch": 13.84,
+      "learning_rate": 2.695858150410529e-05,
+      "loss": 0.8399,
+      "step": 303020
+    },
+    {
+      "epoch": 13.84,
+      "learning_rate": 2.695781985467729e-05,
+      "loss": 0.8025,
+      "step": 303030
+    },
+    {
+      "epoch": 13.84,
+      "learning_rate": 2.695705820524929e-05,
+      "loss": 0.7771,
+      "step": 303040
+    },
+    {
+      "epoch": 13.84,
+      "learning_rate": 2.6956296555821286e-05,
+      "loss": 0.8103,
+      "step": 303050
+    },
+    {
+      "epoch": 13.84,
+      "learning_rate": 2.6955534906393287e-05,
+      "loss": 0.7907,
+      "step": 303060
+    },
+    {
+      "epoch": 13.84,
+      "learning_rate": 2.6954773256965287e-05,
+      "loss": 0.7882,
+      "step": 303070
+    },
+    {
+      "epoch": 13.84,
+      "learning_rate": 2.695401160753728e-05,
+      "loss": 0.7366,
+      "step": 303080
+    },
+    {
+      "epoch": 13.84,
+      "learning_rate": 2.6953249958109282e-05,
+      "loss": 0.8232,
+      "step": 303090
+    },
+    {
+      "epoch": 13.84,
+      "learning_rate": 2.6952488308681283e-05,
+      "loss": 0.8403,
+      "step": 303100
+    },
+    {
+      "epoch": 13.84,
+      "learning_rate": 2.695172665925328e-05,
+      "loss": 0.8316,
+      "step": 303110
+    },
+    {
+      "epoch": 13.84,
+      "learning_rate": 2.6950965009825278e-05,
+      "loss": 0.8981,
+      "step": 303120
+    },
+    {
+      "epoch": 13.84,
+      "learning_rate": 2.695020336039728e-05,
+      "loss": 0.7005,
+      "step": 303130
+    },
+    {
+      "epoch": 13.84,
+      "learning_rate": 2.6949441710969276e-05,
+      "loss": 0.8128,
+      "step": 303140
+    },
+    {
+      "epoch": 13.84,
+      "learning_rate": 2.6948680061541277e-05,
+      "loss": 0.7217,
+      "step": 303150
+    },
+    {
+      "epoch": 13.84,
+      "learning_rate": 2.694791841211327e-05,
+      "loss": 0.8664,
+      "step": 303160
+    },
+    {
+      "epoch": 13.84,
+      "learning_rate": 2.6947156762685272e-05,
+      "loss": 0.7567,
+      "step": 303170
+    },
+    {
+      "epoch": 13.84,
+      "learning_rate": 2.6946395113257273e-05,
+      "loss": 0.8419,
+      "step": 303180
+    },
+    {
+      "epoch": 13.84,
+      "learning_rate": 2.6945633463829267e-05,
+      "loss": 0.7825,
+      "step": 303190
+    },
+    {
+      "epoch": 13.84,
+      "learning_rate": 2.6944871814401268e-05,
+      "loss": 0.7184,
+      "step": 303200
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 2.694411016497327e-05,
+      "loss": 0.7868,
+      "step": 303210
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 2.6943348515545263e-05,
+      "loss": 0.7709,
+      "step": 303220
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 2.6942586866117264e-05,
+      "loss": 0.8482,
+      "step": 303230
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 2.6941825216689265e-05,
+      "loss": 0.7994,
+      "step": 303240
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 2.694106356726126e-05,
+      "loss": 0.8476,
+      "step": 303250
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 2.694030191783326e-05,
+      "loss": 0.6522,
+      "step": 303260
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 2.693954026840526e-05,
+      "loss": 0.8013,
+      "step": 303270
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 2.6938778618977255e-05,
+      "loss": 0.7759,
+      "step": 303280
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 2.6938016969549255e-05,
+      "loss": 0.9187,
+      "step": 303290
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 2.6937255320121256e-05,
+      "loss": 0.7465,
+      "step": 303300
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 2.6936493670693254e-05,
+      "loss": 0.9204,
+      "step": 303310
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 2.6935732021265255e-05,
+      "loss": 0.7892,
+      "step": 303320
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 2.6934970371837252e-05,
+      "loss": 0.8224,
+      "step": 303330
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 2.693420872240925e-05,
+      "loss": 0.7571,
+      "step": 303340
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 2.693344707298125e-05,
+      "loss": 0.7002,
+      "step": 303350
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 2.693268542355325e-05,
+      "loss": 0.7011,
+      "step": 303360
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 2.6931923774125245e-05,
+      "loss": 0.8133,
+      "step": 303370
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 2.6931162124697246e-05,
+      "loss": 0.8366,
+      "step": 303380
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 2.6930400475269247e-05,
+      "loss": 0.8819,
+      "step": 303390
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 2.692963882584124e-05,
+      "loss": 0.7361,
+      "step": 303400
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 2.6928877176413242e-05,
+      "loss": 0.7385,
+      "step": 303410
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 2.6928115526985243e-05,
+      "loss": 0.745,
+      "step": 303420
+    },
+    {
+      "epoch": 13.86,
+      "learning_rate": 2.6927353877557237e-05,
+      "loss": 0.7852,
+      "step": 303430
+    },
+    {
+      "epoch": 13.86,
+      "learning_rate": 2.6926592228129238e-05,
+      "loss": 0.788,
+      "step": 303440
+    },
+    {
+      "epoch": 13.86,
+      "learning_rate": 2.692583057870124e-05,
+      "loss": 0.8663,
+      "step": 303450
+    },
+    {
+      "epoch": 13.86,
+      "learning_rate": 2.6925068929273233e-05,
+      "loss": 0.827,
+      "step": 303460
+    },
+    {
+      "epoch": 13.86,
+      "learning_rate": 2.6924307279845234e-05,
+      "loss": 0.8262,
+      "step": 303470
+    },
+    {
+      "epoch": 13.86,
+      "learning_rate": 2.6923545630417234e-05,
+      "loss": 0.8205,
+      "step": 303480
+    },
+    {
+      "epoch": 13.86,
+      "learning_rate": 2.692278398098923e-05,
+      "loss": 0.7774,
+      "step": 303490
+    },
+    {
+      "epoch": 13.86,
+      "learning_rate": 2.692202233156123e-05,
+      "loss": 0.82,
+      "step": 303500
+    },
+    {
+      "epoch": 13.86,
+      "learning_rate": 2.692126068213323e-05,
+      "loss": 0.7946,
+      "step": 303510
+    },
+    {
+      "epoch": 13.86,
+      "learning_rate": 2.6920499032705228e-05,
+      "loss": 0.8944,
+      "step": 303520
+    },
+    {
+      "epoch": 13.86,
+      "learning_rate": 2.691973738327723e-05,
+      "loss": 0.7219,
+      "step": 303530
+    },
+    {
+      "epoch": 13.86,
+      "learning_rate": 2.6918975733849226e-05,
+      "loss": 0.6514,
+      "step": 303540
+    },
+    {
+      "epoch": 13.86,
+      "learning_rate": 2.6918214084421223e-05,
+      "loss": 0.8106,
+      "step": 303550
+    },
+    {
+      "epoch": 13.86,
+      "learning_rate": 2.6917452434993224e-05,
+      "loss": 0.8123,
+      "step": 303560
+    },
+    {
+      "epoch": 13.86,
+      "learning_rate": 2.6916690785565225e-05,
+      "loss": 0.8,
+      "step": 303570
+    },
+    {
+      "epoch": 13.86,
+      "learning_rate": 2.691592913613722e-05,
+      "loss": 0.7214,
+      "step": 303580
+    },
+    {
+      "epoch": 13.86,
+      "learning_rate": 2.691516748670922e-05,
+      "loss": 0.7791,
+      "step": 303590
+    },
+    {
+      "epoch": 13.86,
+      "learning_rate": 2.691440583728122e-05,
+      "loss": 0.8166,
+      "step": 303600
+    },
+    {
+      "epoch": 13.86,
+      "learning_rate": 2.6913644187853215e-05,
+      "loss": 0.8366,
+      "step": 303610
+    },
+    {
+      "epoch": 13.86,
+      "learning_rate": 2.6912882538425216e-05,
+      "loss": 0.7706,
+      "step": 303620
+    },
+    {
+      "epoch": 13.86,
+      "learning_rate": 2.691212088899721e-05,
+      "loss": 0.7702,
+      "step": 303630
+    },
+    {
+      "epoch": 13.86,
+      "learning_rate": 2.691135923956921e-05,
+      "loss": 0.8233,
+      "step": 303640
+    },
+    {
+      "epoch": 13.87,
+      "learning_rate": 2.691059759014121e-05,
+      "loss": 0.8531,
+      "step": 303650
+    },
+    {
+      "epoch": 13.87,
+      "learning_rate": 2.6909835940713206e-05,
+      "loss": 0.8397,
+      "step": 303660
+    },
+    {
+      "epoch": 13.87,
+      "learning_rate": 2.6909074291285207e-05,
+      "loss": 0.788,
+      "step": 303670
+    },
+    {
+      "epoch": 13.87,
+      "learning_rate": 2.6908312641857207e-05,
+      "loss": 0.7285,
+      "step": 303680
+    },
+    {
+      "epoch": 13.87,
+      "learning_rate": 2.6907550992429205e-05,
+      "loss": 0.7665,
+      "step": 303690
+    },
+    {
+      "epoch": 13.87,
+      "learning_rate": 2.6906789343001206e-05,
+      "loss": 0.7734,
+      "step": 303700
+    },
+    {
+      "epoch": 13.87,
+      "learning_rate": 2.6906027693573203e-05,
+      "loss": 0.7995,
+      "step": 303710
+    },
+    {
+      "epoch": 13.87,
+      "learning_rate": 2.69052660441452e-05,
+      "loss": 0.8184,
+      "step": 303720
+    },
+    {
+      "epoch": 13.87,
+      "learning_rate": 2.69045043947172e-05,
+      "loss": 0.8426,
+      "step": 303730
+    },
+    {
+      "epoch": 13.87,
+      "learning_rate": 2.6903742745289202e-05,
+      "loss": 0.8206,
+      "step": 303740
+    },
+    {
+      "epoch": 13.87,
+      "learning_rate": 2.6902981095861196e-05,
+      "loss": 0.8234,
+      "step": 303750
+    },
+    {
+      "epoch": 13.87,
+      "learning_rate": 2.6902219446433197e-05,
+      "loss": 0.7905,
+      "step": 303760
+    },
+    {
+      "epoch": 13.87,
+      "learning_rate": 2.6901457797005198e-05,
+      "loss": 0.7797,
+      "step": 303770
+    },
+    {
+      "epoch": 13.87,
+      "learning_rate": 2.6900696147577192e-05,
+      "loss": 0.6984,
+      "step": 303780
+    },
+    {
+      "epoch": 13.87,
+      "learning_rate": 2.6899934498149193e-05,
+      "loss": 0.7546,
+      "step": 303790
+    },
+    {
+      "epoch": 13.87,
+      "learning_rate": 2.6899172848721194e-05,
+      "loss": 0.8924,
+      "step": 303800
+    },
+    {
+      "epoch": 13.87,
+      "learning_rate": 2.6898411199293188e-05,
+      "loss": 0.7575,
+      "step": 303810
+    },
+    {
+      "epoch": 13.87,
+      "learning_rate": 2.689764954986519e-05,
+      "loss": 0.8128,
+      "step": 303820
+    },
+    {
+      "epoch": 13.87,
+      "learning_rate": 2.689688790043719e-05,
+      "loss": 0.6761,
+      "step": 303830
+    },
+    {
+      "epoch": 13.87,
+      "learning_rate": 2.6896126251009184e-05,
+      "loss": 0.7529,
+      "step": 303840
+    },
+    {
+      "epoch": 13.87,
+      "learning_rate": 2.6895364601581185e-05,
+      "loss": 0.8228,
+      "step": 303850
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 2.6894602952153186e-05,
+      "loss": 0.8733,
+      "step": 303860
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 2.689384130272518e-05,
+      "loss": 0.8432,
+      "step": 303870
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 2.689307965329718e-05,
+      "loss": 0.8247,
+      "step": 303880
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 2.689231800386918e-05,
+      "loss": 0.8645,
+      "step": 303890
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 2.689155635444118e-05,
+      "loss": 0.7588,
+      "step": 303900
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 2.689079470501318e-05,
+      "loss": 0.8037,
+      "step": 303910
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 2.6890033055585177e-05,
+      "loss": 0.8393,
+      "step": 303920
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 2.6889271406157175e-05,
+      "loss": 0.7789,
+      "step": 303930
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 2.6888509756729175e-05,
+      "loss": 0.7549,
+      "step": 303940
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 2.6887748107301176e-05,
+      "loss": 0.7682,
+      "step": 303950
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 2.688698645787317e-05,
+      "loss": 0.7885,
+      "step": 303960
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 2.688622480844517e-05,
+      "loss": 0.8011,
+      "step": 303970
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 2.6885463159017172e-05,
+      "loss": 0.839,
+      "step": 303980
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 2.6884701509589166e-05,
+      "loss": 0.8021,
+      "step": 303990
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 2.6883939860161167e-05,
+      "loss": 0.7712,
+      "step": 304000
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 2.6883178210733168e-05,
+      "loss": 0.6909,
+      "step": 304010
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 2.6882416561305162e-05,
+      "loss": 0.752,
+      "step": 304020
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 2.6881654911877163e-05,
+      "loss": 0.7814,
+      "step": 304030
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 2.6880893262449164e-05,
+      "loss": 0.844,
+      "step": 304040
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 2.6880131613021158e-05,
+      "loss": 0.8033,
+      "step": 304050
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 2.687936996359316e-05,
+      "loss": 0.8422,
+      "step": 304060
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 2.687860831416516e-05,
+      "loss": 0.7959,
+      "step": 304070
+    },
+    {
+      "epoch": 13.89,
+      "learning_rate": 2.6877846664737154e-05,
+      "loss": 0.8091,
+      "step": 304080
+    },
+    {
+      "epoch": 13.89,
+      "learning_rate": 2.6877085015309154e-05,
+      "loss": 0.8896,
+      "step": 304090
+    },
+    {
+      "epoch": 13.89,
+      "learning_rate": 2.6876323365881152e-05,
+      "loss": 0.6979,
+      "step": 304100
+    },
+    {
+      "epoch": 13.89,
+      "learning_rate": 2.6875561716453153e-05,
+      "loss": 0.8198,
+      "step": 304110
+    },
+    {
+      "epoch": 13.89,
+      "learning_rate": 2.6874800067025154e-05,
+      "loss": 0.7743,
+      "step": 304120
+    },
+    {
+      "epoch": 13.89,
+      "learning_rate": 2.6874038417597148e-05,
+      "loss": 0.7841,
+      "step": 304130
+    },
+    {
+      "epoch": 13.89,
+      "learning_rate": 2.687327676816915e-05,
+      "loss": 0.7862,
+      "step": 304140
+    },
+    {
+      "epoch": 13.89,
+      "learning_rate": 2.687251511874115e-05,
+      "loss": 0.743,
+      "step": 304150
+    },
+    {
+      "epoch": 13.89,
+      "learning_rate": 2.6871753469313143e-05,
+      "loss": 0.7855,
+      "step": 304160
+    },
+    {
+      "epoch": 13.89,
+      "learning_rate": 2.6870991819885144e-05,
+      "loss": 0.7704,
+      "step": 304170
+    },
+    {
+      "epoch": 13.89,
+      "learning_rate": 2.6870230170457145e-05,
+      "loss": 0.7524,
+      "step": 304180
+    },
+    {
+      "epoch": 13.89,
+      "learning_rate": 2.686946852102914e-05,
+      "loss": 0.7291,
+      "step": 304190
+    },
+    {
+      "epoch": 13.89,
+      "learning_rate": 2.686870687160114e-05,
+      "loss": 0.8316,
+      "step": 304200
+    },
+    {
+      "epoch": 13.89,
+      "learning_rate": 2.686794522217314e-05,
+      "loss": 0.8457,
+      "step": 304210
+    },
+    {
+      "epoch": 13.89,
+      "learning_rate": 2.6867183572745135e-05,
+      "loss": 0.8192,
+      "step": 304220
+    },
+    {
+      "epoch": 13.89,
+      "learning_rate": 2.6866421923317136e-05,
+      "loss": 0.7882,
+      "step": 304230
+    },
+    {
+      "epoch": 13.89,
+      "learning_rate": 2.6865660273889137e-05,
+      "loss": 0.7971,
+      "step": 304240
+    },
+    {
+      "epoch": 13.89,
+      "learning_rate": 2.686489862446113e-05,
+      "loss": 0.83,
+      "step": 304250
+    },
+    {
+      "epoch": 13.89,
+      "learning_rate": 2.686413697503313e-05,
+      "loss": 0.7878,
+      "step": 304260
+    },
+    {
+      "epoch": 13.89,
+      "learning_rate": 2.6863375325605133e-05,
+      "loss": 0.7895,
+      "step": 304270
+    },
+    {
+      "epoch": 13.89,
+      "learning_rate": 2.686261367617713e-05,
+      "loss": 0.7957,
+      "step": 304280
+    },
+    {
+      "epoch": 13.89,
+      "learning_rate": 2.6861852026749127e-05,
+      "loss": 0.8895,
+      "step": 304290
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 2.686109037732113e-05,
+      "loss": 0.7363,
+      "step": 304300
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 2.6860328727893126e-05,
+      "loss": 0.861,
+      "step": 304310
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 2.6859567078465127e-05,
+      "loss": 0.7506,
+      "step": 304320
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 2.6858805429037127e-05,
+      "loss": 0.7977,
+      "step": 304330
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 2.685804377960912e-05,
+      "loss": 0.924,
+      "step": 304340
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 2.6857282130181122e-05,
+      "loss": 0.7334,
+      "step": 304350
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 2.6856520480753123e-05,
+      "loss": 0.8102,
+      "step": 304360
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 2.6855758831325117e-05,
+      "loss": 0.7768,
+      "step": 304370
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 2.6854997181897118e-05,
+      "loss": 0.8392,
+      "step": 304380
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 2.685423553246912e-05,
+      "loss": 0.7382,
+      "step": 304390
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 2.6853473883041113e-05,
+      "loss": 0.7591,
+      "step": 304400
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 2.6852712233613114e-05,
+      "loss": 0.7528,
+      "step": 304410
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 2.6851950584185115e-05,
+      "loss": 0.8854,
+      "step": 304420
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 2.685118893475711e-05,
+      "loss": 0.7513,
+      "step": 304430
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 2.685042728532911e-05,
+      "loss": 0.8898,
+      "step": 304440
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 2.684966563590111e-05,
+      "loss": 0.8326,
+      "step": 304450
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 2.6848903986473105e-05,
+      "loss": 0.8088,
+      "step": 304460
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 2.6848142337045106e-05,
+      "loss": 0.8194,
+      "step": 304470
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 2.6847380687617106e-05,
+      "loss": 0.8111,
+      "step": 304480
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 2.6846619038189104e-05,
+      "loss": 0.7357,
+      "step": 304490
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 2.6845857388761105e-05,
+      "loss": 0.8084,
+      "step": 304500
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 2.6845095739333102e-05,
+      "loss": 0.7456,
+      "step": 304510
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 2.68443340899051e-05,
+      "loss": 0.8244,
+      "step": 304520
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 2.68435724404771e-05,
+      "loss": 0.8285,
+      "step": 304530
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 2.6842810791049095e-05,
+      "loss": 0.7632,
+      "step": 304540
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 2.6842049141621095e-05,
+      "loss": 0.8136,
+      "step": 304550
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 2.6841287492193096e-05,
+      "loss": 0.7137,
+      "step": 304560
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 2.684052584276509e-05,
+      "loss": 0.8798,
+      "step": 304570
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 2.683976419333709e-05,
+      "loss": 0.7908,
+      "step": 304580
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 2.6839002543909092e-05,
+      "loss": 0.7902,
+      "step": 304590
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 2.6838240894481086e-05,
+      "loss": 0.8178,
+      "step": 304600
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 2.6837479245053087e-05,
+      "loss": 0.6454,
+      "step": 304610
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 2.6836717595625088e-05,
+      "loss": 0.8142,
+      "step": 304620
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 2.6835955946197082e-05,
+      "loss": 0.7779,
+      "step": 304630
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 2.6835194296769083e-05,
+      "loss": 0.7886,
+      "step": 304640
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 2.6834432647341084e-05,
+      "loss": 0.8606,
+      "step": 304650
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 2.683367099791308e-05,
+      "loss": 0.8218,
+      "step": 304660
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 2.683290934848508e-05,
+      "loss": 0.7851,
+      "step": 304670
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 2.683214769905708e-05,
+      "loss": 0.8162,
+      "step": 304680
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 2.6831386049629077e-05,
+      "loss": 0.8262,
+      "step": 304690
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 2.6830624400201078e-05,
+      "loss": 0.73,
+      "step": 304700
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 2.682986275077308e-05,
+      "loss": 0.7778,
+      "step": 304710
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 2.6829101101345073e-05,
+      "loss": 0.7946,
+      "step": 304720
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 2.6828339451917074e-05,
+      "loss": 0.7167,
+      "step": 304730
+    },
+    {
+      "epoch": 13.92,
+      "learning_rate": 2.6827577802489074e-05,
+      "loss": 0.8277,
+      "step": 304740
+    },
+    {
+      "epoch": 13.92,
+      "learning_rate": 2.682681615306107e-05,
+      "loss": 0.7623,
+      "step": 304750
+    },
+    {
+      "epoch": 13.92,
+      "learning_rate": 2.682605450363307e-05,
+      "loss": 0.8945,
+      "step": 304760
+    },
+    {
+      "epoch": 13.92,
+      "learning_rate": 2.682529285420507e-05,
+      "loss": 0.8479,
+      "step": 304770
+    },
+    {
+      "epoch": 13.92,
+      "learning_rate": 2.6824531204777064e-05,
+      "loss": 0.8051,
+      "step": 304780
+    },
+    {
+      "epoch": 13.92,
+      "learning_rate": 2.6823769555349065e-05,
+      "loss": 0.8627,
+      "step": 304790
+    },
+    {
+      "epoch": 13.92,
+      "learning_rate": 2.6823007905921066e-05,
+      "loss": 0.8353,
+      "step": 304800
+    },
+    {
+      "epoch": 13.92,
+      "learning_rate": 2.682224625649306e-05,
+      "loss": 0.772,
+      "step": 304810
+    },
+    {
+      "epoch": 13.92,
+      "learning_rate": 2.682148460706506e-05,
+      "loss": 0.7747,
+      "step": 304820
+    },
+    {
+      "epoch": 13.92,
+      "learning_rate": 2.6820722957637062e-05,
+      "loss": 0.8882,
+      "step": 304830
+    },
+    {
+      "epoch": 13.92,
+      "learning_rate": 2.6819961308209056e-05,
+      "loss": 0.8538,
+      "step": 304840
+    },
+    {
+      "epoch": 13.92,
+      "learning_rate": 2.6819199658781057e-05,
+      "loss": 0.8874,
+      "step": 304850
+    },
+    {
+      "epoch": 13.92,
+      "learning_rate": 2.6818438009353058e-05,
+      "loss": 0.8303,
+      "step": 304860
+    },
+    {
+      "epoch": 13.92,
+      "learning_rate": 2.6817676359925055e-05,
+      "loss": 0.7708,
+      "step": 304870
+    },
+    {
+      "epoch": 13.92,
+      "learning_rate": 2.6816914710497053e-05,
+      "loss": 0.73,
+      "step": 304880
+    },
+    {
+      "epoch": 13.92,
+      "learning_rate": 2.6816153061069053e-05,
+      "loss": 0.9441,
+      "step": 304890
+    },
+    {
+      "epoch": 13.92,
+      "learning_rate": 2.681539141164105e-05,
+      "loss": 0.8582,
+      "step": 304900
+    },
+    {
+      "epoch": 13.92,
+      "learning_rate": 2.6814629762213052e-05,
+      "loss": 0.7849,
+      "step": 304910
+    },
+    {
+      "epoch": 13.92,
+      "learning_rate": 2.6813868112785053e-05,
+      "loss": 0.8575,
+      "step": 304920
+    },
+    {
+      "epoch": 13.92,
+      "learning_rate": 2.6813106463357047e-05,
+      "loss": 0.7769,
+      "step": 304930
+    },
+    {
+      "epoch": 13.92,
+      "learning_rate": 2.6812344813929048e-05,
+      "loss": 0.7865,
+      "step": 304940
+    },
+    {
+      "epoch": 13.92,
+      "learning_rate": 2.681158316450105e-05,
+      "loss": 0.7757,
+      "step": 304950
+    },
+    {
+      "epoch": 13.93,
+      "learning_rate": 2.6810821515073042e-05,
+      "loss": 0.821,
+      "step": 304960
+    },
+    {
+      "epoch": 13.93,
+      "learning_rate": 2.6810059865645043e-05,
+      "loss": 0.7069,
+      "step": 304970
+    },
+    {
+      "epoch": 13.93,
+      "learning_rate": 2.6809298216217044e-05,
+      "loss": 0.8336,
+      "step": 304980
+    },
+    {
+      "epoch": 13.93,
+      "learning_rate": 2.6808536566789038e-05,
+      "loss": 0.7645,
+      "step": 304990
+    },
+    {
+      "epoch": 13.93,
+      "learning_rate": 2.680777491736104e-05,
+      "loss": 0.7972,
+      "step": 305000
+    },
+    {
+      "epoch": 13.93,
+      "learning_rate": 2.6807013267933033e-05,
+      "loss": 0.7516,
+      "step": 305010
+    },
+    {
+      "epoch": 13.93,
+      "learning_rate": 2.6806251618505034e-05,
+      "loss": 0.8173,
+      "step": 305020
+    },
+    {
+      "epoch": 13.93,
+      "learning_rate": 2.6805489969077035e-05,
+      "loss": 0.7614,
+      "step": 305030
+    },
+    {
+      "epoch": 13.93,
+      "learning_rate": 2.680472831964903e-05,
+      "loss": 0.8027,
+      "step": 305040
+    },
+    {
+      "epoch": 13.93,
+      "learning_rate": 2.680396667022103e-05,
+      "loss": 0.8491,
+      "step": 305050
+    },
+    {
+      "epoch": 13.93,
+      "learning_rate": 2.680320502079303e-05,
+      "loss": 0.8237,
+      "step": 305060
+    },
+    {
+      "epoch": 13.93,
+      "learning_rate": 2.6802443371365028e-05,
+      "loss": 0.8296,
+      "step": 305070
+    },
+    {
+      "epoch": 13.93,
+      "learning_rate": 2.680168172193703e-05,
+      "loss": 0.8786,
+      "step": 305080
+    },
+    {
+      "epoch": 13.93,
+      "learning_rate": 2.680092007250903e-05,
+      "loss": 0.8472,
+      "step": 305090
+    },
+    {
+      "epoch": 13.93,
+      "learning_rate": 2.6800158423081024e-05,
+      "loss": 0.79,
+      "step": 305100
+    },
+    {
+      "epoch": 13.93,
+      "learning_rate": 2.6799396773653025e-05,
+      "loss": 0.7418,
+      "step": 305110
+    },
+    {
+      "epoch": 13.93,
+      "learning_rate": 2.6798635124225026e-05,
+      "loss": 0.8648,
+      "step": 305120
+    },
+    {
+      "epoch": 13.93,
+      "learning_rate": 2.679787347479702e-05,
+      "loss": 0.8142,
+      "step": 305130
+    },
+    {
+      "epoch": 13.93,
+      "learning_rate": 2.679711182536902e-05,
+      "loss": 0.8185,
+      "step": 305140
+    },
+    {
+      "epoch": 13.93,
+      "learning_rate": 2.679635017594102e-05,
+      "loss": 0.7559,
+      "step": 305150
+    },
+    {
+      "epoch": 13.93,
+      "learning_rate": 2.6795588526513015e-05,
+      "loss": 0.7981,
+      "step": 305160
+    },
+    {
+      "epoch": 13.93,
+      "learning_rate": 2.6794826877085016e-05,
+      "loss": 0.7505,
+      "step": 305170
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 2.6794065227657017e-05,
+      "loss": 0.7701,
+      "step": 305180
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 2.679330357822901e-05,
+      "loss": 0.8393,
+      "step": 305190
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 2.6792541928801012e-05,
+      "loss": 0.7726,
+      "step": 305200
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 2.6791780279373013e-05,
+      "loss": 0.7999,
+      "step": 305210
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 2.6791018629945007e-05,
+      "loss": 0.7852,
+      "step": 305220
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 2.6790256980517008e-05,
+      "loss": 0.8477,
+      "step": 305230
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 2.678949533108901e-05,
+      "loss": 0.8665,
+      "step": 305240
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 2.6788733681661006e-05,
+      "loss": 0.7185,
+      "step": 305250
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 2.6787972032233004e-05,
+      "loss": 0.7468,
+      "step": 305260
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 2.6787210382805005e-05,
+      "loss": 0.787,
+      "step": 305270
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 2.6786448733377002e-05,
+      "loss": 0.917,
+      "step": 305280
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 2.6785687083949003e-05,
+      "loss": 0.7571,
+      "step": 305290
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 2.6784925434521004e-05,
+      "loss": 0.7633,
+      "step": 305300
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 2.6784163785092998e-05,
+      "loss": 0.8921,
+      "step": 305310
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 2.6783402135665e-05,
+      "loss": 0.7773,
+      "step": 305320
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 2.6782640486237e-05,
+      "loss": 0.7618,
+      "step": 305330
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 2.6781878836808994e-05,
+      "loss": 0.7906,
+      "step": 305340
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 2.6781117187380994e-05,
+      "loss": 0.7343,
+      "step": 305350
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 2.6780355537952995e-05,
+      "loss": 0.85,
+      "step": 305360
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 2.677959388852499e-05,
+      "loss": 0.8024,
+      "step": 305370
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 2.677883223909699e-05,
+      "loss": 0.8682,
+      "step": 305380
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 2.677807058966899e-05,
+      "loss": 0.7581,
+      "step": 305390
+    },
+    {
+      "epoch": 13.95,
+      "learning_rate": 2.6777308940240985e-05,
+      "loss": 0.7567,
+      "step": 305400
+    },
+    {
+      "epoch": 13.95,
+      "learning_rate": 2.6776547290812986e-05,
+      "loss": 0.8164,
+      "step": 305410
+    },
+    {
+      "epoch": 13.95,
+      "learning_rate": 2.6775785641384987e-05,
+      "loss": 0.7345,
+      "step": 305420
+    },
+    {
+      "epoch": 13.95,
+      "learning_rate": 2.677502399195698e-05,
+      "loss": 0.7821,
+      "step": 305430
+    },
+    {
+      "epoch": 13.95,
+      "learning_rate": 2.6774262342528982e-05,
+      "loss": 0.8287,
+      "step": 305440
+    },
+    {
+      "epoch": 13.95,
+      "learning_rate": 2.6773500693100983e-05,
+      "loss": 0.7762,
+      "step": 305450
+    },
+    {
+      "epoch": 13.95,
+      "learning_rate": 2.677273904367298e-05,
+      "loss": 0.8005,
+      "step": 305460
+    },
+    {
+      "epoch": 13.95,
+      "learning_rate": 2.6771977394244978e-05,
+      "loss": 0.8475,
+      "step": 305470
+    },
+    {
+      "epoch": 13.95,
+      "learning_rate": 2.6771215744816975e-05,
+      "loss": 0.8594,
+      "step": 305480
+    },
+    {
+      "epoch": 13.95,
+      "learning_rate": 2.6770454095388976e-05,
+      "loss": 0.8981,
+      "step": 305490
+    },
+    {
+      "epoch": 13.95,
+      "learning_rate": 2.6769692445960977e-05,
+      "loss": 0.7093,
+      "step": 305500
+    },
+    {
+      "epoch": 13.95,
+      "learning_rate": 2.676893079653297e-05,
+      "loss": 0.8621,
+      "step": 305510
+    },
+    {
+      "epoch": 13.95,
+      "learning_rate": 2.6768169147104972e-05,
+      "loss": 0.7251,
+      "step": 305520
+    },
+    {
+      "epoch": 13.95,
+      "learning_rate": 2.6767407497676973e-05,
+      "loss": 0.814,
+      "step": 305530
+    },
+    {
+      "epoch": 13.95,
+      "learning_rate": 2.6766645848248967e-05,
+      "loss": 0.8528,
+      "step": 305540
+    },
+    {
+      "epoch": 13.95,
+      "learning_rate": 2.6765884198820968e-05,
+      "loss": 0.8222,
+      "step": 305550
+    },
+    {
+      "epoch": 13.95,
+      "learning_rate": 2.676512254939297e-05,
+      "loss": 0.8465,
+      "step": 305560
+    },
+    {
+      "epoch": 13.95,
+      "learning_rate": 2.6764360899964962e-05,
+      "loss": 0.8091,
+      "step": 305570
+    },
+    {
+      "epoch": 13.95,
+      "learning_rate": 2.6763599250536963e-05,
+      "loss": 0.7479,
+      "step": 305580
+    },
+    {
+      "epoch": 13.95,
+      "learning_rate": 2.6762837601108964e-05,
+      "loss": 0.8002,
+      "step": 305590
+    },
+    {
+      "epoch": 13.95,
+      "learning_rate": 2.6762075951680958e-05,
+      "loss": 0.7435,
+      "step": 305600
+    },
+    {
+      "epoch": 13.95,
+      "learning_rate": 2.676131430225296e-05,
+      "loss": 0.8293,
+      "step": 305610
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 2.676055265282496e-05,
+      "loss": 0.7803,
+      "step": 305620
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 2.6759791003396954e-05,
+      "loss": 0.8279,
+      "step": 305630
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 2.6759029353968955e-05,
+      "loss": 0.7882,
+      "step": 305640
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 2.6758267704540956e-05,
+      "loss": 0.8593,
+      "step": 305650
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 2.6757506055112953e-05,
+      "loss": 0.7883,
+      "step": 305660
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 2.6756744405684954e-05,
+      "loss": 0.7344,
+      "step": 305670
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 2.675598275625695e-05,
+      "loss": 0.8227,
+      "step": 305680
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 2.675522110682895e-05,
+      "loss": 0.8014,
+      "step": 305690
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 2.675445945740095e-05,
+      "loss": 0.7281,
+      "step": 305700
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 2.675369780797295e-05,
+      "loss": 0.8215,
+      "step": 305710
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 2.6752936158544945e-05,
+      "loss": 0.7917,
+      "step": 305720
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 2.6752174509116946e-05,
+      "loss": 0.7125,
+      "step": 305730
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 2.6751412859688946e-05,
+      "loss": 0.7546,
+      "step": 305740
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 2.675065121026094e-05,
+      "loss": 0.7982,
+      "step": 305750
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 2.674988956083294e-05,
+      "loss": 0.7142,
+      "step": 305760
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 2.6749127911404942e-05,
+      "loss": 0.7916,
+      "step": 305770
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 2.6748366261976936e-05,
+      "loss": 0.8258,
+      "step": 305780
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 2.6747604612548937e-05,
+      "loss": 0.7701,
+      "step": 305790
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 2.6746842963120938e-05,
+      "loss": 0.7363,
+      "step": 305800
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 2.6746081313692932e-05,
+      "loss": 0.7797,
+      "step": 305810
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 2.6745319664264933e-05,
+      "loss": 0.7469,
+      "step": 305820
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 2.6744558014836934e-05,
+      "loss": 0.7407,
+      "step": 305830
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 2.674379636540893e-05,
+      "loss": 0.7797,
+      "step": 305840
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 2.674303471598093e-05,
+      "loss": 0.821,
+      "step": 305850
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 2.674227306655293e-05,
+      "loss": 0.7767,
+      "step": 305860
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 2.6741511417124927e-05,
+      "loss": 0.7717,
+      "step": 305870
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 2.6740749767696928e-05,
+      "loss": 0.8851,
+      "step": 305880
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 2.673998811826893e-05,
+      "loss": 0.7978,
+      "step": 305890
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 2.6739226468840923e-05,
+      "loss": 0.8556,
+      "step": 305900
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 2.6738464819412924e-05,
+      "loss": 0.8241,
+      "step": 305910
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 2.6737703169984918e-05,
+      "loss": 0.913,
+      "step": 305920
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 2.673694152055692e-05,
+      "loss": 0.8016,
+      "step": 305930
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 2.673617987112892e-05,
+      "loss": 0.8118,
+      "step": 305940
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 2.6735418221700914e-05,
+      "loss": 0.8642,
+      "step": 305950
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 2.6734656572272914e-05,
+      "loss": 0.7602,
+      "step": 305960
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 2.6733894922844915e-05,
+      "loss": 0.7982,
+      "step": 305970
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 2.673313327341691e-05,
+      "loss": 0.8038,
+      "step": 305980
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 2.673237162398891e-05,
+      "loss": 0.7805,
+      "step": 305990
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 2.673160997456091e-05,
+      "loss": 0.7669,
+      "step": 306000
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 2.6730848325132905e-05,
+      "loss": 0.7516,
+      "step": 306010
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 2.6730086675704906e-05,
+      "loss": 0.8315,
+      "step": 306020
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 2.6729325026276907e-05,
+      "loss": 0.7999,
+      "step": 306030
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 2.6728563376848904e-05,
+      "loss": 0.8395,
+      "step": 306040
+    },
+    {
+      "epoch": 13.98,
+      "learning_rate": 2.6727801727420905e-05,
+      "loss": 0.8226,
+      "step": 306050
+    },
+    {
+      "epoch": 13.98,
+      "learning_rate": 2.6727040077992903e-05,
+      "loss": 0.8341,
+      "step": 306060
+    },
+    {
+      "epoch": 13.98,
+      "learning_rate": 2.67262784285649e-05,
+      "loss": 0.8074,
+      "step": 306070
+    },
+    {
+      "epoch": 13.98,
+      "learning_rate": 2.67255167791369e-05,
+      "loss": 0.8521,
+      "step": 306080
+    },
+    {
+      "epoch": 13.98,
+      "learning_rate": 2.6724755129708902e-05,
+      "loss": 0.814,
+      "step": 306090
+    },
+    {
+      "epoch": 13.98,
+      "learning_rate": 2.6723993480280896e-05,
+      "loss": 0.7852,
+      "step": 306100
+    },
+    {
+      "epoch": 13.98,
+      "learning_rate": 2.6723231830852897e-05,
+      "loss": 0.7513,
+      "step": 306110
+    },
+    {
+      "epoch": 13.98,
+      "learning_rate": 2.6722470181424898e-05,
+      "loss": 0.6741,
+      "step": 306120
+    },
+    {
+      "epoch": 13.98,
+      "learning_rate": 2.6721708531996892e-05,
+      "loss": 0.8059,
+      "step": 306130
+    },
+    {
+      "epoch": 13.98,
+      "learning_rate": 2.6720946882568893e-05,
+      "loss": 0.7864,
+      "step": 306140
+    },
+    {
+      "epoch": 13.98,
+      "learning_rate": 2.6720185233140893e-05,
+      "loss": 0.9029,
+      "step": 306150
+    },
+    {
+      "epoch": 13.98,
+      "learning_rate": 2.6719423583712888e-05,
+      "loss": 0.7219,
+      "step": 306160
+    },
+    {
+      "epoch": 13.98,
+      "learning_rate": 2.671866193428489e-05,
+      "loss": 0.814,
+      "step": 306170
+    },
+    {
+      "epoch": 13.98,
+      "learning_rate": 2.671790028485689e-05,
+      "loss": 0.8888,
+      "step": 306180
+    },
+    {
+      "epoch": 13.98,
+      "learning_rate": 2.6717138635428883e-05,
+      "loss": 0.812,
+      "step": 306190
+    },
+    {
+      "epoch": 13.98,
+      "learning_rate": 2.6716376986000884e-05,
+      "loss": 0.7691,
+      "step": 306200
+    },
+    {
+      "epoch": 13.98,
+      "learning_rate": 2.6715615336572885e-05,
+      "loss": 0.8281,
+      "step": 306210
+    },
+    {
+      "epoch": 13.98,
+      "learning_rate": 2.671485368714488e-05,
+      "loss": 0.6786,
+      "step": 306220
+    },
+    {
+      "epoch": 13.98,
+      "learning_rate": 2.671409203771688e-05,
+      "loss": 0.701,
+      "step": 306230
+    },
+    {
+      "epoch": 13.98,
+      "learning_rate": 2.671333038828888e-05,
+      "loss": 0.833,
+      "step": 306240
+    },
+    {
+      "epoch": 13.98,
+      "learning_rate": 2.6712568738860878e-05,
+      "loss": 0.6975,
+      "step": 306250
+    },
+    {
+      "epoch": 13.98,
+      "learning_rate": 2.671180708943288e-05,
+      "loss": 0.6793,
+      "step": 306260
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 2.6711045440004877e-05,
+      "loss": 0.8955,
+      "step": 306270
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 2.6710283790576874e-05,
+      "loss": 0.8738,
+      "step": 306280
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 2.6709522141148875e-05,
+      "loss": 0.8332,
+      "step": 306290
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 2.6708760491720876e-05,
+      "loss": 0.789,
+      "step": 306300
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 2.670799884229287e-05,
+      "loss": 0.8254,
+      "step": 306310
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 2.670723719286487e-05,
+      "loss": 0.8293,
+      "step": 306320
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 2.670647554343687e-05,
+      "loss": 0.8184,
+      "step": 306330
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 2.6705713894008866e-05,
+      "loss": 0.9129,
+      "step": 306340
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 2.6704952244580867e-05,
+      "loss": 0.8281,
+      "step": 306350
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 2.6704190595152867e-05,
+      "loss": 0.8042,
+      "step": 306360
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 2.670342894572486e-05,
+      "loss": 0.747,
+      "step": 306370
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 2.6702667296296862e-05,
+      "loss": 0.8289,
+      "step": 306380
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 2.6701905646868856e-05,
+      "loss": 0.7946,
+      "step": 306390
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 2.6701143997440857e-05,
+      "loss": 0.934,
+      "step": 306400
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 2.6700382348012858e-05,
+      "loss": 0.8136,
+      "step": 306410
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 2.6699620698584856e-05,
+      "loss": 0.9158,
+      "step": 306420
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 2.6698859049156853e-05,
+      "loss": 0.8088,
+      "step": 306430
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 2.6698097399728854e-05,
+      "loss": 0.6879,
+      "step": 306440
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 2.669733575030085e-05,
+      "loss": 0.7867,
+      "step": 306450
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 2.6696574100872852e-05,
+      "loss": 0.8175,
+      "step": 306460
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 2.6695812451444853e-05,
+      "loss": 0.7565,
+      "step": 306470
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 2.6695050802016847e-05,
+      "loss": 0.8915,
+      "step": 306480
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 2.6694289152588848e-05,
+      "loss": 0.8277,
+      "step": 306490
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 2.669352750316085e-05,
+      "loss": 0.714,
+      "step": 306500
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 2.6692765853732843e-05,
+      "loss": 0.9212,
+      "step": 306510
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 2.6692004204304844e-05,
+      "loss": 0.742,
+      "step": 306520
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 2.6691242554876845e-05,
+      "loss": 0.9145,
+      "step": 306530
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 2.669048090544884e-05,
+      "loss": 0.8439,
+      "step": 306540
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 2.668971925602084e-05,
+      "loss": 0.8097,
+      "step": 306550
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 2.668895760659284e-05,
+      "loss": 0.8316,
+      "step": 306560
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 2.6688195957164834e-05,
+      "loss": 0.8451,
+      "step": 306570
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 2.6687434307736835e-05,
+      "loss": 0.7473,
+      "step": 306580
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 2.6686672658308836e-05,
+      "loss": 0.7891,
+      "step": 306590
+    },
+    {
+      "epoch": 14.0,
+      "eval_cer": 0.6203494065828538,
+      "eval_em": 0.007663197729422895,
+      "eval_f1": 0.007663197729422895,
+      "eval_loss": 0.8980765342712402,
+      "eval_runtime": 1021.0623,
+      "eval_samples_per_second": 10.352,
+      "eval_steps_per_second": 1.295,
+      "eval_wer": 0.9923368022705771,
+      "step": 306596
     }
   ],
   "max_steps": 656970,
   "num_train_epochs": 30,
-  "total_flos": 1.7110728299049062e+18,
+  "total_flos": 2.3955019618668687e+18,
   "trial_name": null,
   "trial_params": null
 }